Jespersen, Otto. Philosophy of Grammar – T01

[The Philosophy of Grammar]

Chapter I
Living Grammar

Speaker and Hearer. Formulas and Free Expressions. Grammatical Types.
Building up of Sentences.

Speaker and Hearer.

The essence of language is human activity — activity on the part of
one individual to make himself understood by another, and activity
on the part of that other to understand what was in the mind of
the first. These two individuals, the producer and the recipient
of language, or as we may more conveniently call them, the speaker
and the hearer, and their relations to one another, should never
be lost sight of if we want to understand the nature of language
and of that part of language which is dealt with in grammar. But in
former times this was often overlooked, and words and forms were
often treated as if they were things or natural objects with an
existence of their own — a conception which may have been to a great
extent fostered through a too exclusive preoccupation with written
or printed words, but which is fundamentally false, as will easily
be seen with a little reflexion.

If the two individuals, the producer and the recipient of language,
are here spoken of as the speaker and the hearer respectively,
this is in consideration of the fact that the spoken and heard word
is the primary form for language, and of far greater importance
than the secondary form used in writing (printing) and reading.
This is evidently true for the countless ages in which mankind had
not yet invented the art of writing or made only a sparing use of
it; but even in our modern newspaper-ridden communities, the
vast majority of us speak infinitely more than we write. At any
rate we shall never be able to understand what language is and
how it develops if we do not continually take into consideration
first and foremost the activity of speaking and hearing, and if we
forget for a moment that writing is only a substitute for speaking.
17A written word is mummified until someone imparts life to it by
transposing it mentally into the corresponding spoken word.

The grammarian must be ever on his guard to avoid the pitfalls
into which the ordinary spelling is apt to lead him. Let me give
a few very elementary instances. The ending for the plural of
substantives and for the third person singular of the present tense
of verbs is in writing the same -s in such words as ends, locks, rises,
but in reality we have three different endings, as seen when we
transcribe them phonetically [endz, lɔks, raiziz]. Similarly the
written ending -ed covers three different spoken endings in sailed,
locked, ended, phonetically [seild, lɔkt, endid]. In the written
language it looks as if the preterits paid and said were formed
in the same w ay, but differently from stayed, but in reality paid and
stayed are formed regularly [peid, steid], whereas said is irregular as
having its vowel shortened [sed]. Where the written language
recognizes only one word there, the spoken language distinguishes
two both as to sound and signification (and grammatical import),
as seen in the sentence “There [ðə] were many people there [ˡðɛˑə].”
Quantity, stress, and intonation, which are very inadequately, if
at all, indicated in the usual spelling, play important parts in the
grammar of the spoken language, and thus we are in many ways
reminded of the important truth that grammar should deal in the
first instance with sounds and only secondarily with letters.

Formulas and Free Expressions.

If after these preliminary remarks we turn our attention to the
psychological side of linguistic activity, it will be well at once to
mention the important distinction between formulas or formular
units and free expressions. Some things in language — in any
language — are of the formula character; that is to say, no one can
change anything in them. A phrase like “How do you do?” is
entirely different from such a phrase as “I gave the boy a lump of
sugar.” In the former everything is fixed: you cannot even change
the stress, saying “How do you do?” or make a pause between
the words, and it is not usual nowadays as in former times to say
“How does your father do?” or “How did you do?” Even
though it may still be possible, after saying “How do you do?” in
the usual way to some of the people present, to alter the stress
and say “And how do you do, little Mary?” the phrase is for all
practical purposes one unchanged and unchangeable formula.
It is the same with “Good morning!”, “Thank you,” “Beg your
pardon,” and other similar expressions. One may indeed analyze
such a formula and show that it consists of several words, but it is
felt and handled as a unit, which may often mean something quite
18different from the meaning of the component words taken separately;
“beg your pardon,” for instance, often means “please
repeat what you said, I did not catch it exactly”; “how do you
do?” is no longer a question requiring an answer, etc.

If, then, free expressions are defined as expressions created on
the spur of the moment after a certain type which has come into
existence in the speaker's subconsciousness as a result of his having
heard many sentences possessing some trait or traits in common,
it follows that the distinction between them and formulas cannot
always be discovered except through a fairly close analysis; to
the hearer the two stand at first on the same footing, and accordingly
formulas can and do play a great part in the formation of types
in the minds of speakers, the more so as many of them are of very
frequent occurrence. Let us take a few more examples.

“Long live the King!” Is this a formula or a free expression?
It is impossible to frame an indefinite number of other sentences on
the same pattern. Combinations such as “Late die the King!”
or “Soon come the train!” are not used nowadays to express a
wish. On the other hand, we may say “Long live the Queen”
or “the President” or “Mr. Johnson.” In other words, the type,
in which an adverb is placed first, then a subjunctive, and lastly a
subject, the whole being the expression of a wish, has totally gone
out of the language as a living force. But those phrases which can
still be used are a survival of that type, and the sentence “Long
live the King” must therefore be analyzed as consisting of a
formula “Long live,” which is living though the type is dead, + a
subject which is variable. We accordingly have a sentence type
whose use is much more restricted in our own days than it was in
older English.

In a paper on ethics by J. Royce I find the principle laid down
“Loyal is that loyally does.” This is at once felt as unnatural,
as the author has taken as a pattern the proverb “Handsome if
that handsome does” without any regard to the fact that
whatever it was at the time when the sentence was first framed,
it is now to all intents and purposes nothing but a formula, as
shown by the use of that without any antecedent and by the word-order.

The distinction between formulas and free expressions pervades
all parts of grammar. Take morphology or accidence: here we
have the same distinction with regard to flexional forms. The
plural eyen was going out of use in the sixteenth century; now
(he form is dead, but once not only that word, but the type according
20to which it was formed, were living elements of the English language.
The only surviving instance of a plural formed through the addition
of -en to the singular is oxen, which is living as a formula, though its
type is extinct. Meanwhile, shoen, fone, eyen, tent have been supplanted
by shoes, foes, eyes, cows; that is, the plural of these words
has been reshaped in accordance with the living type found in
kings, lines, stones, etc. This type is now so universal that all new
words have to conform to it: bicycles, photos, kodaks, aeroplanes,
hooligans, ions, stunts, etc. When eyes was first uttered instead of
eyen, it was an analogical formation on the type of the numerous
words which already had -s in the plural. But now when a child
says eyes for the first time, it is impossible to decide whether he is
reproducing a plural form already heard, or whether he has learned
only the singular eye and then has himself added -s (phonetically
[z]) in accordance with the type he has deduced from numerous
similar words. The result in either case would be the same. If it
were not the fact that the result of the individual's free combination
of existing elements is in the vast majority of instances identical
with the traditional form, the life of any language would be hampered;
a language would be a difficult thing to handle if its speakers
had the burden imposed on them of remembering every little item
separately.

It will be seen that in morphology what was above called a
“type” is the same thing as the principle of what are generally
called regular formations, while irregular forms are “formulas.”

In the theory of word-formation it is customary to distinguish
between productive and unproductive suffixes. An example of a
productive suffix is -ness, because it is possible to form new words
like weariness, closeness, perverseness, etc. On the contrary -lock
in wedlock is unproductive, and so is -th in width, breadth, health, for
Ruskin's attempt to construct a word illth on the analogy of wealth
has met with no success, and no other word with this ending seems
to have come into existence for several hundred years. This is a
further application of what we said above: the type adjective
+ -ness is still living, while wedlock and the words mentioned in -th
are now formulas of a type now extinct. But when the word width
originated, the type was alive. At that far-off time it was possible
to add the ending, which was then something like -iþu, to any
adjective. In course of time, however, the ending dwindled down
to the simple sound þ (th), while the vowel of the first syllable was
modified, with the consequence that the suffix ceased to be productive,
because it was impossible for an ordinary man, who was not
trained in historical grammar, to see that the pairs long: length,
broad: breadth, wide: width, deep: depth, whole: health, dear: dearth,
represented one and the same type of formation. These words
21were, accordingly, handed down traditionally from generation to
generation as units, that is, formulas, and when the want was felt
for a new ‘abstract noun’ (I use here provisionally the ordinary
term for such words), it was no longer the ending -th that was
resorted to, but -ness, because that offered no difficulty, the adjective
entering unchanged into the combination.

With regard to compounds, similar considerations hold good.
Take three old compounds of hūs ‘house,’ hūsbōnde, hūsþing,
hūswīf. These were formed according to the usual type found in
innumerable old compounds; the first framers of them conformed
to the usual rules, and thus they were at first free expressions.
But they were handed down as whole, indivisible words from
generation to generation, and accordingly underwent the usual
sound changes; the long vowel ū was shortened, [s] became voiced
[z] before voiced sounds, [þ] became [t] after [s], [w] and [f] disappeared,
and the vowels of the latter element were obscured, the
result being our present forms husband, husting(s), hussy, phonetically
[hʌzbənd, hʌstiŋz, hʌzi]. The tie, which at first was strong between
these words and hūs, was gradually loosened, the more so because
the long u had here become a diphthong, house. And if there was
a divergence in form, there was as great a divergence in meaning,
the result being that no one except the student of etymology would
ever dream of connecting husband, hustings, or hussy with house.
From the standpoint of the living speech of our own days the three
words are not compound words; they have, in the terminology here
employed, become formulas and are on a par with other disyllabic
words of obscure or forgotten origin, such as sopha or cousin.

With regard to huswif there are, however, different degrees
of isolation from house and wife. Hussy [hʌzi] in the sense
‘bad woman’ has lost all connexion with both; but for the
obsolete sense ‘needle-case’ old dictionaries record various forms
showing conflicting tendencies: huswife [hʌzwaif], hussif [hʌzif],
hussive; and then we have, in the sense of ‘manager of a house,’
housewife, in which the form of both components is intact, but this
appears to be a comparatively recent re-formation, not recognized,
for instance, by Elphinston in 1765. Thus the tendency to make
the old compound into a formula was counteracted more or less
by the actual speech-instinct, which in some applications treated
it as a free expression: in other words, people would go on combining
the two elements without regard to the existence of the
formular compounds, which had become more or less petrified in
sound and in meaning. This phenomenon is far from rare:
grindstone as a formula had become [grinstən] with the usual
shortening of the vowel in both elements, but the result of a free
combination has prevailed in the current pronunciation [graindstoun];
22in waistcoat the new [weistkout] is beginning to be used
instead of the formular [weskot]; fearful is given as sounding
‘ferful’ by eighteenth-century orthoepists, but is now always
[fiəf(u)l]. For other examples see MEG I, 4. 34 ff.

Something similar is seen in words that are not compounds.
In Middle English we find short vowels in many comparatives:
deppre, grettre as against deep, great (greet). Some of these comparatives
became formulas and were handed down as such to new
generations, the only surviving instances being latter and utter,
which have preserved the short vowels because they were isolated
from the positives late and out and acquired a somewhat modified
meaning. But other comparatives were re-formed as free combinations,
thus deeper, greater, and in the same way we have now later
and outer, which are more intimately connected with late and out
than latter and utter are.

Stress presents analogous phenomena. Children, of course,
learn the accentuation as well as the sounds of each word: the
whole of the pronunciation of a word is in so far a formular unit.
But in some words there may be a conflict between two modes of
accentuation, because words may in some instances be formed as
free expressions by the speaker at the moment he wants them.
Adjectives in -able, -ible as a rule have the stress on the fourth
syllable from the ending in consequence of the rhythmic principle
that the vowel which is separated by one (weak) syllable from the
original stress is now stressed, thus ˈdespicable 11 (originally as in
French ˌˌdesˈpicable), ˈcomparable, ˈlamentable, ˈpreferable, etc. In
some of these the rhythmic principle places the stress on the same
syllable as in the corresponding verb: conˈsiderable, ˈviolable.
But in others this is not so, and a free formation, in which the
speaker was thinking of the verb and then would add -able, would
lead to a different accentuation: the adjective corresponding to
acˈcept was ˈacceptable in Shakespeare and some other poets, and
this formula still survives in the reading of the Prayer Book, but
otherwise it now is reshaped as acˈceptable; refutable was [ˈrefjutəbl],
but now it is more usual to say [riˈfjuˑtəbl]; Respectable has given
way to reˈspectable; Shakespeare's and Spencer's ˈdetestable has
been supplanted by deˈtestable, which is Milton's form; in admirable
the new [ədˈmairəbl] has been less successful in supplanting
[ˈædmirəbl], but in a great many adjectives analogy, i.e. free formation,
has prevailed entirely: aˈgreeable, deˈplorable, reˈmarkable,
irreˈsistible. In words with other endings we have the same conflict:
ˈconfessor and confessor, caˈpitalist and ˈcapitalist, deˈmonstrative
23and ˈdemonstrative, etc., sometimes with changes of meaning,
the free formation following not only the accent, but also the
signification of the word from which it is derived, while the formula
has been more or less isolated. (Examples see MEG Ch. V.) The
British advertisement [ədˈvəˑtizmənt] shows the traditional formula,
the American pronunciation [ˌædvəˈtaizmənt] or [ˈædvəˌtaizmənt]
is a free formation on the basis of the verb.

The distinction between a formula and a free combination
also affects word-order. One example may suffice: so long as
some + thing is a free combination of two elements felt as such, another
adjective may be inserted in the usual way: some good thing.
But as soon as something has become a fixed formula, it is inseparable,
and the adjective has to follow: something good. Compare
also the difference between the old “They turned each to other”
and the modern “they turned to each other.”

The coalescence of originally separate elements into a formula
is not always equally complete: in breakfast it is shown not only
by the pronunciation [brekfəst] as against [breik, faˑst], but also
by forms like he breakfasts, breakfasted (formerly breaks fast, broke
fast), but in take place the coalescence is not carried through to the
same extent, and yet this must be recognized as a formula in the
sense ‘come to happen,’ as it is impossible to treat it in the same
way as take with another object, which in some combinations can
be placed first (a book he took) and which can be made the subject in
the passive (the book was taken), neither of which is possible in the
case of take place.

Though it must be admitted that there are doubtful instances
in which it is hard to tell whether we have a formula or not, the
distinction here established between formulas and free combinations
has been shown to pervade the whole domain of linguistic
activity. A formula may be a whole sentence or a group of words,
or it may be one word, or it may be only part of a word, — that is not
important, but it must always be something which to the actual
speech-instinct is a unit which cannot be further analyzed or
decomposed in the way a free combination can. The type or
pattern according to which a formula has been constructed, may
be either an extinct one or a living one; but the type or pattern
according to which a free expression is framed must as a matter of
course be a living one; hence formulas may be regular or irregular,
but free expressions always show a regular formation.

Grammatical Types.

The way in which grammatical types or patterns are created
in the minds of speaking children is really very wonderful, and
24in many cases we see curious effects on the history of languages.
In German the prefix ge-, which at first could be added to any form
of the verb to express completed action, has come to be specially
associated with the past participle. In the verb essen there was,
however, a natural fusion of the vowel of the prefix and the initial
vowel of the verb itself, thus gessen; this was handed down as a
formular unit and was no longer felt to contain the same prefix
as getrunken, gegangen, gesehn and others; in a combination like
ich habe getrunken und gessen it was then felt as if the latter form
was incomplete, and ge- was added: ich habe getrunken und gegessen,
which restored parallelism.

Grammatical habits may thus lead to what from one point of
view may be termed redundancy. We see something similar with
regard to the use of it in many cases. It became an invariable
custom to have a subject before the verb, and therefore a sentence
which did not contain a subject was felt to be incomplete. In
former times no pronoun was felt to be necessary with verbs like
Latin pluit, ningit ‘it rains, it snows,’ etc.; thus Italian still has
piove, nevica, but on the analogy of innumerable such expressions
as I come, he comes, etc., the pronoun it was added in E. it rains,
it snows, and correspondingly in French, German, Danish and
other languages: il pleut, es regnet, det regner. It has been well
remarked that the need for this pronoun was especially felt when it
became the custom to express the difference between affirmation
and question by means of word-order (er kommt, kommt er!), for
now it would be possible in the same way to mark the difference
between es regnet and regnet es?

Verbs like rain, snow had originally no subject, and as it would
be hard even now to define logically what the subject it stands for
and what it means, many scholars 12 look upon it as simply a grammatical
device to make the sentence conform to the type most
generally found. In other cases there is a real subject, yet we are
led for some reason or other to insert the pronoun it. It is possible
to say, for instance, “To find one's way in London is not easy,”
but more often we find it convenient not to introduce the infinitive
at once; in which cases, however, we do not begin with the verb and
say “Is not easy to find one's way in London,” because we are
accustomed to look upon sentences beginning with a verb as interrogative
; so we say “It is not easy,” etc. In the same way it
is possible to say “That Newton was a great genius cannot be
denied,” but if we do not want to place the clause with that first
we have to say “It cannot be denied that Newton was a great
genius.” In these sentences it represents the following infinitive
construction or clause, very much as in “He is a great scoundrel,
25that husband of hers” he represents the words that husband of hers.
Cf. the colloquial: “It is perfectly wonderful the way in which
he remembers things.” It would be awkward to say “She made
that he had committed many offences appear clearly” with the
various grammatical elements arranged as in the usual construction
of make appear (“She made his guilt appear clearly”): this
awkwardness is evaded by using the representative it before the
infinitive: She made it appear clearly that he had committed many
offences. In this way many of the rules concerning the use of it
are seen to be due on the one hand to the speaker's wish to conform
to certain patterns of sentence construction found in innumerable
sentences with other subjects or objects, and on the other hand
to his wish to avoid clumsy combinations which might even sometimes
lead to misunderstandings.

The rules for the use of the auxiliary do in interrogative sentences
are to be explained in a similar way. The universal tendency is
towards having the word-order Subject Verb, but there is a conflicting
tendency to express a question by means of the inverted
order Verb Subject, as in the obsolete “writes he!” (cf. German
“Schreibt er?” and French “Écrit-il?”). Now many interrogative
sentences had the word-order Auxiliary Subject Verb (“Can
he write?” “Will he write?” “Has he written,” etc.), in which
the really significant verb came after the subject just as in ordinary
affirmative sentences: through the creation of the compromise
form “Does he write?” the two conflicting tendencies were reconciled:
from a formal point of view the verb, though an empty one,
preceded the subject to indicate the question, and from another
point of view the subject preceded the real verb. But no auxiliary
is required when the sentence has an interrogative pronoun as
subject (“Who writes?”) because the interrogatory pronoun is
naturally put first, and so the sentence without any does conforms
already to the universal pattern. 13

Building up of Sentences.

Many irregularities in syntax can be explained on the same
principle, e.g. sentences like “Hee that rewards me, heaven reward
him” (Sh.). When a writer uses the pronoun thou, he will have
no difficulty in adding the proper ending -st to the verb if
it follows immediately upon the pronoun; but if it does not
he will be apt to forget it and use the form that is suitable
to the you which may be at the back of his mind. Thus in
Shakespeare (Tp. I. 2. 333) “Thou stroakst me, and made much
of me.” Byron apostrophizes Sulla (Ch. H. IV. 83): “Thou,
who didst subdue Thy country's foes ere thou wouldst pause
to feel The wrath of thy own wrongs, or reap the due Of
hoarded vengeance… thou who with thy frown Annihilated
senates… thou didst lay down,” etc. In Byron such transitions
are not uncommon.

In a similar way the power of if to require a subjunctive is often
exhausted when a second verb comes at some distance from the
conjunction, as in Shakespeare (Hml V. 2. 245) If Hamlet from
himselfe be tane away, And when he's not himselfe, do's wrong
Laertes, Then Hamlet does it not | (Meas. III. 2. 37) if he be a
whoremonger, and comes before him, he were as good go a mile on
his errand | Ruskin: But if the mass of good things be inexhaustible,
and there are horses for everybody, — why is not every beggar
27on horseback? | Mrs. Ward: A woman may chat with whomsoever
she likes, provided it be a time of holiday, and she is not betraying
her art. 14

My chief object in writing this chapter has been to make the
reader realize that language is not exactly what a one-sided occupation
with dictionaries and the usual grammars might lead us to think,
but a set of habits, of habitual actions, and that each word and each
sentence spoken is a complex action on the part of the speaker.
The greater part of these actions are determined by what he has
done previously in similar situations, and that again was determined
chiefly by what he had habitually heard from others. But
in each individual instance, apart from mere formulas, the speaker
has to turn these habits to account to meet a new situation, to
express what has not been expressed previously in every minute
detail; therefore he cannot be a mere slave to habits, but has to
vary them to suit varying needs — and this in course of time may
lead to new turns and new habits; in other words, to new grammatical
forms and usages. Grammar thus becomes a part of
linguistic psychology or psychological linguistics; this, however,
is not the only way in which the study of grammar stands in need
of reshaping and supplementing if it is to avoid the besetting sins
of so many grammarians, pedantry and dogmatism — but that will
form the subject-matter of the following chapters.29

Chapter II
Systematic Grammar

Descriptive and Historical Linguistics. Grammar and Dictionary. Sounds.
Usual Division of Grammar. New System. Morphology.

Descriptive and Historical Linguistics.

There are two ways of treating linguistic phenomena which may be
called the descriptive and the historical. They correspond to what
in physics are called statics and dynamics (or kinetics) and differ
in that the one views phenomena as being in equilibrium, and the
other views them as being in motion. It is the pride of the linguistic
science of the last hundred years or so that it has superseded older
methods by historical grammar, in which phenomena are not only
described, but explained, and it cannot be denied that the new
point of view, by showing the inter-connexion of grammatical
phenomena previously isolated, has obtained many new and important
results. Where formerly we saw only arbitrary rules and
inexplicable exceptions, we now in very many cases see the reasons.
The plural feet from foot was formerly only mentioned as one of
a few exceptions to the rule that plurals in English substantives
were formed in -a: now we know that the long [iˑ] of the plural is
the regular development of Proto-English [œˑ], and that this
[œˑ], wherever it was found, through [eˑ] (still represented in the
E spelling) became [iˑ] in Present English (cp. feed, green, sweet, etc.).
Further, the [œˑ] of fœ·t has been shown to be a mutation of the
original vowel [oˑ], which was preserved in the singular foˑt, where
it has now through a regular raising become [u] in the spoken
language, though the spelling still keeps oo. The mutation in
question was caused by an i in the following syllable; now the
ending in a number of plurals was -iz in Proto-Gothonic (urgermanisch).
Finally this ending, which was dropped after leaving a
trace in the mutated vowel, is seen to be the regular development
of the plural ending found, for instance, in Latin -es. Accordingly
what from the one-sided (static) Modern English point of view is
an isolated fact, is seen to be (dynamically) related to a great
number of other facts in the older stages of the same language
and in other languages of the same family. Irregularities in one
stage are in many instances recognized as survivals of regularities
30in older stages, and a flood of light has been thrown over very much
that had hitherto been veiled in obscurity. This is true not only
of historical linguistics in the stricter sense, but also of comparative
linguistics, which is only another branch of the same science,
supplementing by analogous methods the evidence that is accessible
to us in historical sources, by connecting languages whose common
“ancestor” is lost to tradition.

But, great as have been the triumphs of these new methods,
it should not be forgotten that everything is not said when the
facts of a language are interpreted in the terms of linguistic history.
Even when many irregularities have been traced back to former
regularities, others still remain irregular, however far we dive into
the past; in any case, the earliest accessible stage remains unexplained
and must be taken as it is, for we have now shaken off
the superstition of the first generation of comparative linguists who
imagined that the Aryan (Indo-Germanic) language which is the
basis of our family of languages (grundsprache) was a fair representative
of the primeval language of our earliest ancestors (ursprache).
We can explain many irregularities, but we cannot explain them
away: to the speakers of our modern language they are just as
irregular as if their origin had not been made clear to us. The
distinction between regular and irregular always must be important
to the psychological life of language, for regular forms are those
which speakers use as the basis of new formations, and irregular
forms are those which they will often tend to replace by new forms
created on the principle of analogy.

At any rate, descriptive linguistics can never be rendered
superfluous by historical linguistics, which must always be based
on the description of those stages of the development of a language
which are directly accessible to us. And in the case of a great
many languages only one definite stage is known and can be made
the subject of scientific treatment. On the other hand, in treating
such languages the student will do well never to lose sight of the
lesson taught by those languages which can be investigated historically,
namely that languages are always in a state of flux, that
they are never fixed in every detail, but that in each of them there
are necessarily points that are liable to change even in the course
of a single generation. This is an inevitable consequence of the
very essence of language and of the way in which it is handed down
from one generation to the next.

Grammar and Dictionary.

When we come to consider the best way in which to arrange
linguistic facts, we are at once confronted with the very important
31division between grammar and dictionary (lexicology). Grammar
deals with the general facts of language, and lexicology with special
facts (cf. Sweet, CP 31). 17 That cat denotes that particular animal
is a special fact which concerns that word alone, but the formation
of the plural by adding the sound -s is a general fact because it
concerns a great many other words as well: rats, hats, works, books,
caps, chiefs, etc.

It might be objected that if this be the proper distinction between
grammar and dictionary, the formation of the plural oxen from ox
should form no part of English grammar and should be mentioned
in dictionaries only. This is partly true as shown by the fact that
all dictionaries mention such irregularities under the word concerned,
while they do not trouble to indicate the plural of such
words as cat and the others just mentioned. Similarly with irregular
and regular verbs. Yet such irregularities should not be excluded
from the grammar of a language, as they are necessary to indicate
the limits within which the “general facts” or rules hold good:
if we did not mention oxen, a student might think that oxes was the
real plural of ox. Grammar and dictionary thus in some respects
overlap and deal with the same facts.

We see now that the usual enumeration in grammars of numerals
is really out of place there, but that, on the other hand, such facts
as the formation of ordinals by means of the ending -th and of
20, 30, etc., by means of -ty unquestionably belong to the province
of grammar.

With regard to prepositions, it is quite right that dictionaries
should account for the various uses of at, for, in, etc., just as they
deal fully with the various meanings of the verbs put and set. But
on the other hand prepositions find their proper place in grammars
in so far as there are “general facts” to be mentioned in connexion
with them. I shall mention a few: while prepositions may sometimes
govern dependent interrogatory clauses (“they disagree as to
how he works,” “that depends on what answer she will give”),
they cannot generally govern a clause introduced by that (as they
can in Danish: “der er ingen tvivl om at han er draebt,” literally:
there is no doubt of that he has been killed); the chief exception
is in that (“they differ in that he is generous and she is miserly”).
Therefore sure is treated in two ways in Goldsmith's “Are you sure
of all this, are you sure that nothing ill has befallen my boy?”
Other general facts concern the combination of two prepositions
as in “from behind the bush” (note that to behind is impossible),
the relations between preposition and adverb (as in “climb up a
32tree,” “he is in,” cf. “in his study,” “he steps in,” cf. “he steps
into his study”). Grammar also has to deal with general facts
concerning the ways in which prepositions express rest at a place
and movement to or from a place, as also the relation between the
local and temporal significations of the same preposition; even
more strictly within the province of grammar are those uses of some
prepositions in which they lose their local or temporal signification
and descend into the category of empty or colourless (“pale”)
words or auxiliaries; this is the case with of in “the father of the
boy” (cf. the genitive case in “the boy's father”), “all of them,”
“the City of London,” “that scoundrel of a servant,” etc., and
similarly with to before an infinitive and when it is what many
grammars term a dative equivalent (“I gave a shilling to the boy”
= “I gave the boy a shilling”). But in some cases it may remain
doubtful and to some extent arbitrary what to include in the
grammar and what to reserve for exclusive treatment in the
dictionary.

Now any linguistic phenomenon may be regarded either from
without or from within, either from the outward form or from
the inner meaning. In the first case we take the sound (of a word
or of some other part of a linguistic expression) and then inquire
into the meaning attached to it; in the second case we start from
the signification and ask ourselves what formal expression it has
found in the particular language we are dealing with. If we denote
the outward form by the letter O, and the inner meaning by the
letter I, we may represent the two ways as O → I and I → O
respectively.

In the dictionary we may thus in the first place (O → I) take a
word, say English cat, and then explain what it means, either by a
paraphrase or definition in English, as in a one-language dictionary,
or else by the French translation ‘chat,’ as in a two-language
dictionary. The various meanings of the same word are given,
and in some instances these may in course of time have become so
far differentiated as to constitute practically two or more words,
thus cheer (1) face, (2) food, (3) good humour, (4) applause. In
this part we have to place together words that have the same
sound (homophones or homonyms), e.g. sound (1) what may be
heard, (2) examine, probe, (3) healthy, sane (4) part of the sea. 18

In the second place, by starting from within (I → O) we shall
have a totally different arrangement. We may here try to arrange
all the things and relations signified in a systematic or logical order.
This is easy enough in some few cases, thus in that of the numerals,
33whose place, as we have seen above, is in the dictionary rather
than in the grammar: one, two, three.… But what would be
the best logical arrangement of the words image, picture, photo,
portrait, painting, drawing, sketch? On account of the utter complexity
of the world around us and of the things and thoughts
which language has to express, it is an extremely difficult thing to
make a satisfactory arrangement of the whole vocabulary on a
logical basis; a well-known attempt is made in Roget's Thesaurus
of English Words and Phrases; Bally's arrangement in Traité
de stylistique française Vol. II seems an improvement on Roget's
arrangement, but is far less complete. If in the O → I part all
homophones were placed together, here on the other hand we have
to place synonyms together; thus dog will go with hound, pup,
whelp, cur, mastiff, spaniel, terrier, etc.; way in one signification
with road, path, trail, passage, etc., in another with manner, method,
mode. So again, cheer will be found in one place with repast, food,
provision, meal, etc., in another with approval, sanction, applause,
acclamation, etc. These remarks apply to a one-language dictionary
of the class I → O; in a two-language dictionary we simply start
from some word in the foreign language and give the corresponding
word or words in our own.

As a natural consequence of the difficulty of a systematic arrangement
of all these special facts most dictionaries content themselves
with an arrangement in alphabetical order which is completely
unscientific, but practically convenient. If our alphabet had been
like the Sanskrit alphabet, in which sounds formed by the same
organ are. placed together, the result would, of course, have been
better than with the purely accidental arrangement of the Latin
alphabet, which separates b and p, d and t and throws together
Bounds which have no phonetic similarity at all, consonants and
vowels in complete disorder. It would also be possible to imagine
other arrangements, by which words were placed together if their
sounds were so similar that they might easily be misheard for one
another, thus bag and beg in one place, bag and back in another.
But on the whole no thoroughly satisfactory system is conceivable
in the dictionary part of language.

Anyone accepting, as I have done here, Sweet's dictum that
grammar deals with the general, and the dictionary with the special
facts of language will readily admit that the two fields may sometimes
overlap, and that there are certain things which it will be
necessary or convenient to treat both in the grammar and in the
dictionary. But there exists a whole domain for which it is difficult
to find a place in the twofold system established by that dictum,
namely the theory of the significations of words. No generally
accepted name has been invented for this branch of linguistic
34science: Bréal, one of the pioneers in this field, uses the word
“semantics” (sémantique) from Gr. sēmaino, while others speak
of “semasiology,” and others again (Sayce, J. A. H. Murray) of
“sematology”; Noreen says “semology,” which is rather a
barbarous formation from Gr. sēma, sēmatos, which, by the way,
does not mean ‘signification,’ but ‘sign’; and finally Lady Welby
has an equally objectionable name “signifies.” I shall use Bréal's
word semantics for this study, which has of late years attracted a
good deal of attention. It is a natural consequence of the historical
trend of modern linguistics that much less has been written on
static than on dynamic semantics, i.e. on the way in which the
meanings of words have changed in course of time, but that static
semantics also may present considerable interest, is seen, for
instance, in K. O. Erdmann's book Die bedeutung des wortes.
In spite of the fact that the subject-matter of semantics is the way
in which meanings and changes of meanings may be classified and
brought into a general system, and that this branch of linguistic
science thus deals with “general” and not with “special” facts,
it is not customary to include semantics in grammar (though
this is done in Nyrop's great Grammaire historique), and I may
therefore be excused if I leave semantics out of consideration in
this volume.

Sounds.

If next we proceed to grammar, the first part of nearly all
scientific treatises consists of a theory of sounds without regard to
the meanings that may be attached to them. It is a simple consequence
of the nature of the spoken language that it is possible
to have a theory of human speech-sounds in general, the way in
which they are produced by the organs of speech, and the way
in which they are combined to form syllables and higher units.
By the side of this we have the theory of what is peculiar to the
one particular language with which the grammarian is concerned.
For the general theory of sounds the word phonetics is in common
use, though the same term is often used of the theory of the sounds
of a particular language, as when we speak of “English Phonetics,”
etc. It would, perhaps, be advisable to restrict the word “phonetics”
to universal or general phonetics and to use the word
phonology of the phenomena — peculiar to a particular language
(e.g. “English Phonology”), but this question of terminology is
not very important. Some writers would discriminate between
the two words by using “phonetics” of descriptive (static), and
“phonology” of historical (dynamic) “lautlehre,” but this terminology
is reversed by some (de Saussure, Sechehaye).

It lies outside the scope of this work to say much about phonetics
35or phonology; a few remarks may, however, find their place here.
The arrangement followed in most books on this subject seems to
me very unsystematic; the learner is bewildered at the outset by a
variety of details from many different spheres. In contrast to
this, in my own Fonetik (Danish edition, 1897-99, German edition
Lehrbuch der Phonetik, an English edition in preparation) I
have tried to build up the whole theory more systematically,
thereby also making the subject easier for learners, as I find from
many years' practice in teaching phonetics. My method is to
start first with the smallest units, the elements of sounds, iss
what is produced in one organ of speech, beginning from the lips
and proceeding gradually to the interior speech-organs, and in
each organ taking first the closed position and afterwards the more
open ones; when all the organs have thus been dealt with, I proceed
to the sounds themselves as built up by the simultaneous action
of all the speech-organs, and finally deal with the combination
of sounds.

In treating the phonology of one of our civilized languages it is
necessary to say something about the way in which sounds are
represented in the traditional spelling; especially in historical
phonology sounds and spellings cannot be separately treated,
however important it is never to confound the two things. The
subject may, of course, be viewed from two opposite points of
view: we may start from the spelling and ask what sound is
connected with such and such a spelling, or, inversely, wo may
take the sound and ask how it is represented. The former is the
point of view of the reader, the latter that of the writer.

The definition of Phonetics given above, “the theory of sounds
without regard to meaning” is not strictly correct, for in dealing
with the sounds of any language it is impossible to disregard meaning
altogether. It is important to observe what sounds are used in a
language to distinguish words, i.e. meanings. Two sounds which
are discriminated in one language, because otherwise words denoting
different things would be confounded together, in another language
may not play that rôle, with the result that speakers of that language
are quite indifferent to distinctions which in the first language were
very important. Much of what is usually treated in phonology
might just as well, or even better, find its place in some other part
of the grammar. Grammarians are very seldom quite consistent
in this respect, and I must myself plead guilty to inconsistency,
having in Vol. I of my MEG given some pages to the difference
in stress between substantives and verbs, as in present, object, etc.
But it must be admitted that there are many things in grammar
which may equally well or nearly so be placed at different places
in the system.36

Usual Division of Grammar.

This division with its subdivisions as commonly treated offers
many points for attack. The following survey of the traditional
scheme will show that a consistent system of grammar cannot be
built up on that basis.

In the traditional scheme Morphology is generally divided
into chapters, each dealing with one of the usually recognized
“parts of speech.” Substantives, as the most noble class, are
placed first, then adjectives, etc., prepositions and conjunctions
last. The grammarian has something to say about each of these
classes. In the case of substantives, we get their flexion (inflexion),
i.e. the changes undergone by these words, but nothing is said about
the significance of these changes or the functions of any given form
except what is implied in such names as genitive, plural, etc. The
arrangement is paradigmatic, all the forms of some single word
being placed together; thus there is no attempt to bring together
the same ending if it is found in various paradigms; in OE, for
instance, the dative plural is given separately in each of the several
classes in spite of the fact that it ends in -um in all words.

Next we come to adjectives, where the arrangement is the
same, apart from the fact that (in languages of the same type as
Latin, OE, etc.) many adjectives have separate forms for the three
genders and the paradigms are therefore fuller than those of the
substantives. As the endings, on the other hand, are generally
the same as in the corresponding classes of substantives, much of
what is said in this chapter is necessarily a repetition of what the
reader knows from the first chapter.

If we next proceed to the chapter dealing with numerals, we
shall find a similar treatment of their flexion in so far as numerals
are subject to changes, as is often the case with the early ones.
Irregular flexion is given in full, otherwise we are referred to the
chapter on adjectives. Besides this, however, the grammarian
in this chapter on numerals does what he never dreamed of doing
in the two previous chapters, he gives a complete and orderly
enumeration of all the words belonging to this class. The next
chapter deals with pronouns; these are treated in very much the
37same way as substantives, only with the significant modification
that as in the case of the numerals all pronouns are enumerated,
even if there is nothing peculiar to be told about their forms. Moreover,
these words are classified not according to the method of their
flexion (different “stems,” etc.), as substantives are, but according
to their signification: personal, possessive, demonstrative pronouns,
etc. In many grammars a list of pronominal adverbs
is given in this chapter, though they have nothing to do with
“morphology” proper, as they are not subject to flexional changes.

Verbs, again, are treated in the same manner as substantives,
with no regard either to the signification of the verbs themselves
or to that of the flexional forms, apart from what is implied in the
simple mention of such and such a form as being the first person
singular, or in such names as indicative, subjunctive, etc.

In the adverbs we have only one kind of flexion, comparison.
This, of course, is given, but besides that many grammars here
include a division according to signification, adverbs of time, of
place, of degree, of manner, etc., very much as if in the first chapter
we had had a division of substantives into nouns of time (year,
month, week…), nouns of place (country, town, village…),
etc. Often, too, we have here a division into immediate adverbs
and derived adverbs with rules for the manner in which adverbs
are formed from adjectives, but this evidently belongs to part 2,
Word-formation.

The next class comprises prepositions: as they are unchanged,
and as many grammarians want, nevertheless, to say something
about this class of words, they will in this place give lists of those
prepositions which govern one case and those that govern another,
though it would seem obvious that this should really form part of
one subdivision of the syntax of cases. Finally we have conjunctions
and interjections, and in order to have something to say about these
flexionless words many writers here too will enumerate all of them,
and sometimes arrange them in classes like those of the adverbs.

Next comes the section dealing with word-formation (G. wortbildung,
Fr. derivation). Here it is well worth noticing that in
this section the meaning of each derivative element (prefix, suffix)
is generally given with its form. As for the arrangement, various
systems prevail, some based on the form (first prefixes, then suffixes,
each of these treated separately), some on the signification (formation
of abstract nouns, of agent-nouns, causative verbs, etc.), and
some jumbling together both points of view in the most perplexing
manner. The usual division according to the parts of speech is not
always beneficial: thus in one very good book on English grammar
I find under the substantives the ending -ics (politics, etc.) totally
separated from the adjectives in -ic; while in a third place comes a
38discussion on the substantivizing of adjectives (shown by a plural
in -s) the three things being consequently treated as if they had
nothing to do with one another.

The third part, Syntax, to a very great extent is taken up with
detailing the signification (i.e. function) of those flexional forma
which were dealt with from another point of view in the first part
(cases of nouns, tenses, and moods of verbs, etc.), but not of
those treated in the section Word-formation. In some chapters
on syntax, on the other hand, we find that the formal and functional
sides of each phenomenon are treated in one and the same place
(the construction of sentences, word-order).

It needs no more than this short synopsis of the various chapters
of ordinary grammars to show how inconsistent and confused they
really are; the whole system, if system it can be called, is a survival
from the days when grammatical science was in its infancy, and only
the fact that we have all of us been accustomed to it from our
childhood can account for the vogue it still enjoys. Many grammarians
have modified the system here and there, improving the
arrangement in many details, but as a whole it has not yet been
superseded by a more scientific one. Nor is the task an easy one,
as seen perhaps best by the failure of the two best thought-out
attempts at establishing a consistent system of arrangement of
grammatical facts, those by John Ries (Was ist syntax? Marburg,
1894) and Adolf Noreen (Vårt språk, Stockholm, 1903 ff., not
yet finished). Both books contain many highly ingenious remarks
and much sound criticism of earlier grammarians, but their systems
do not appear to me satisfactory or natural. Instead, however, of
criticizing them, I prefer here to give my own ideas of the subject
and to leave it to others to find out where I agree and where I
disagree with my predecessors. 19

New system.

A consistent system can be arrived at if we take as our main
division what we have already found to constitute the two parts
of the lexicology of a language. In grammar, too, we may start
either from without or from within; 210 in the first part (O → I) we
39take a form as given and then inquire into its meaning or function;
in the second part (I → O) we invert the process and take the meaning
or function and ask how that is expressed in form. The facts
of grammar are the same in the two parts, only the point of view
being different: the treatment is different, and the two parts
supplement each other and together give a complete and perspicuous
survey of the general facts of a language.

Morphology.

In the first part, then, (O → I) we proceed from the form to the
meaning; this part I propose to call Morphology, though the
word thus acquires a somewhat different sense from that usually
given to it. Here things are treated together that are expressed
externally by the same means; in one place we have, for instance,
the ending -s, in another the ending -ed, in a third, mutation, etc.
But it is very important to notice that this does not mean that we
leave the meaning out of account; at each point we have also to
investigate the function or use of such and such an ending or
whatever it may be, which, of course, amounts to the same thing
as answering the question “What does it signify?” In many
instances this can be done simply by giving the name: under -a
in cats we say that it turns the singular cat into a plural; in dealing
with the ending -ed we say that in added, etc., it denotes the second
(passive) participle and the preterit, etc. These may be called
syntactic definitions, and in very simple instances everything
necessary can be said under this head in a few words, while generally
a more detailed analysis must be reserved for the second part of
our grammar. Though Sweet makes practically the same distinction
as I do between the two parts of grammar, I cannot agree
with him when he says (NEG I, 204) that it is “not only possible,
but desirable, to treat form and meaning separately — at least, to
some extent. That part of grammar which concerns itself specially
with forms, and ignores their meaning as much as possible, is called
accidence. That part of grammar which ignores distinction of
form as much as possible, and concentrates itself on their meaning,
is called syntax.” Here I must take exception to the words “ignore
… as much as possible.” It should be the grammarian's task
always to keep the two things in his mind, for sound and signification,
form and function, are inseparable in the life of language, and
it has been the detriment of linguistic science that it has ignored
one side while speaking of the other, and so lost sight of the constant
interplay of sound and sense (see Language, passim).

In an ideal language, combining the greatest expressiveness
with perfect ease and complete freedom from exceptions and
40irregularities as well as from ambiguity, the arrangement of the
grammar would be an easy thing, because the same sound or the
same modification of sounds would always have the same meaning,
and the same signification or function would always be expressed
in the same formal way. This is the case already to a great extent
in the grammar of such artificial languages as Ido, where it is only
necessary once and for all to state the rule that plurality in substantives
is expressed by the ending -i (I → O), or that the ending
-i denotes the plural in substantives (O → I): there is thus perfect
harmony between the morphological and the syntactic way of
expressing the same fact. But our natural languages are otherwise
constructed, they cannot be mapped out by means of straight
lines intersecting one another at right angles like most of the United
States, but are more like Europe with its irregularly curved and
crooked boundaries. Even that comparison does not do justice
to the phenomena of speech, because we have here innumerable
overlappings as if one district belonged at the same time to two or
three different states. We must never lose sight of the fact that one
form may have two or more significations, or no signification at
all, and that one and the same signification or function may be
denoted now by this and now by that formal means, and sometimes
by no form at all. In both parts of the system, therefore, we are
obliged to class together things which are really different, and
to separate things which would seem to belong naturally to
the same class. But it must be our endeavour to frame
our divisions and subdivisions in the most natural manner
possible and to avoid unnecessary repetitions by means of cross-references.

Let me attempt to give a short synopsis of the various subdivisions
of Morphology as I have worked them out in one of the
parts of my Modern English Grammar which have not yet been
printed. Just as in my phonetic books I take first sound elements,
then sounds, and finally sound combinations, I here propose to take
first word elements, then words, and finally word combinations.
It must, however, be conceded that the boundaries between these
divisions are not always clear and indisputable: not in could not
is a separate word, and Americans print can not as two words, but
in England cannot is written in one word; now we cannot, of course,
accept typographical custom as decisive, but the phonetic fusion
with consequent vowel change in can't, don't, won't shows that nt
in these combinations has to be reckoned as a word element and
no longer as a separate word. Inversely the genitive s tends to
become more and more independent of the preceding word, as
shown in the “group genitive” (the King of England's power,
somebody else's hat, Bill Stumps his mark, see ChE Ch. III).41

In the part headed Word Elements we have to speak of each
affix (whether prolix, suffix, or infix) separately, state its form or
forms and define its function or functions. We do not take the
several word classes (parts of speech) and finish one before passing
on to the next, but in speaking of the ending -s, for instance (with
its three phonetically distinct forms [s, z, iz]), we mention first its
function as a sign of the plural in substantives, then as a genitive
sign, then as a mark of the third person singular in the present
tense of verbs, then in the non-adjunct form of possessive pronouns,
e.g. in ours. The ending -n (-en) in a similar way serves to form a
plural in oxen, a non-adjunct possessive in mine, a participle in
beaten, a derived adjective in silken, a derived verb in weaken, etc.
In separate chapters we have to deal with such less conspicuous
word elements as are shown by modifications of the kernel of the
word, thus the voicing of the final consonant to form verbs (halve,
breathe, use from half, breath, use), the mutation (umlaut) to form
the plural (feet from foot) and a verb (feed from food), the apophony
(ablaut) to form the preterit sang and the participle sung from sing,
the change of stress which distinguishes the verb object from the
substantive object; here we may also speak of the change from the
full word that [ðæt] to the empty or pale word spelt in the same way
but pronounced [ðət].

It will probably be objected that by this arrangement we mix
together things from the two distinct provinces of accidence and
word-formation. But on closer inspection it will be seen that it
is hard, not to say impossible, to tell exactly where the boundary
has to be drawn between flexion and word-formation: the formation
of feminine nouns in English (shepherdess) is always taken to
belong to the latter, thus also to some extent in French (maîtresse),
but what are we to say of paysanne from paysan? — is that to be
torn away from bon, bonne, which is counted as flexion and placed
under Accidence? The arrangement here advocated has the
advantage that it brings together what to the naive speech instinct
is identical or similar, and that it opens the eyes of the grammarian
to things which he would otherwise have probably overlooked.
Take, for instance, the various -en-endings, in adjectives, in verbal
derivatives, and in participles: in all these cases -en is found
(whether this means that it is historically preserved or is a later
addition) after the same consonants, while after other consonants
it is not found (i.e. it is in some cases dropped, in others it has
never been added). Note also the parallelism between the adjunct
form in -en and another form without -en: a drunken boy: he is
drunk | ill-gotten wealth: I've got | silken dalliance: clad in silk | in
olden days: the man is old | hidden treasures: it was hid (the original
form, now also hidden) | the maiden queen: an old maid. Now all
42this can be shown to have a curious connexion with the extension
of a great many verbs by means of -en which took place from about
1400 and gave rise not only to the forms happen, listen, frighten,
but also to verbs like broaden, blacken, moisten, which now are
apprehended as formed from adjectives, while originally they were
simply phonetic expansions of existing verbs that had the same
form as the adjectives. (I have not yet published the account
of these phenomena which I promised in MEG I, p. 34.) The new
arrangement brings into focus things which had previously escaped
our attention.

Speaking of word-formation it may not be superfluous here
to enter a protest against the practice prevalent in English grammars
of treating the formatives of Latin words adopted into English as
if they were English formatives. Thus the prefix pre- is given
with such examples as precept, prefer, present, and re- with such
examples as repeat, resist, redeem, redolent, etc., although the part
of the words which remains when we take off the prefix has no
existence as such in English (cept, fer, etc.). This shows that all
these words (although originally formed with the prefixes præ, re)
are in English indivisible “formulas.” Note that in such the first
syllable is pronounced with the short [i] or [e] vowel (cf. prepare,
preparation, repair, reparation), but by the side of such words we
have others with the same written beginning, but pronounced in a
different way, with long [iˑ], and here we have a genuine English
prefix with a signification of its own: presuppose, predetermine,
re-enter, re-open. Only this pre- and this re- deserve a place in
English grammars: the other words belong to the dictionary.
Similar considerations hold good with regard to suffixes: although
there is really an English suffix -ty, we should not include among the
examples of it such a word as beauty [bjuˑti], because there is no
such thing as [bjuˑ] in English (beau [bou] has now nothing to do
with beauty). That beauty is a unit, a formula, is seen by the fact
that the corresponding adjective is beautiful; we may establish the
proportion beautiful: beauty = Ft. beau: beauté (for in the French
word -ti is a living suffix). An English grammar would have to
mention the suffix -ty in safety, certainty, etc., and the change in the
kernel wrought in such instances as reality from real, liability from
liable, etc.

The next part deals with words, mainly the so-called grammatical
words or auxiliaries, whether pronouns, auxiliary verbs, prepositions,
or conjunctions, but only in so far as they are really parts of
grammar, that is “general expressions.” Under will (and the
shorter form 'll in he'll, etc.) we shall thus mention its use to express
(1) volition, (2) futurity, (3) habit. But, as stated above, there
can be here no hard and fast line between grammar and dictionary.43

Finally, in the part devoted to Combination of Words we
shall have to describe each type of word-order and indicate the
rôle it plays in speech. Thus the combination substantive + substantive,
apart from such collocations as Captain Hall, is used in
various kinds of compound substantives, such as mankind, wineglass,
stone wall, cotton dress, bosom friend, womanhater, woman author;
the relations between the two components will have to be specified
both as regards form (stress, also secondarily orthography) and as
regards meaning. Adjective + substantive is chiefly used in such
adjunct groups as red coat, whence compounds of the type blackbird;
but a special kind of compounds is seen in redcoat ‘one who wears
a red coat.’ The combination substantive + verb forms a finite
sentence in father came, where father is the subject. In the inverse
order the substantive may according to circumstances be the
subject (as in the inserted “said Tom” or in the question “Did
Tom?” or after certain adverbs “and so did Tom” or in a conditional
clause without a conjunction “had Tom said that, I should
have believed it”); or the substantive may be the object (as in
“I saw Tom”), etc. All, of course, that I can do here is to sketch
out the bare outlines of the system, leaving the details to be worked
out in future instalments of my Grammar.

Many people probably will wonder at the inclusion of such
things in Morphology, but I venture to think that this is the only
consistent way of dealing with grammatical facts, for word-order
is certainly as much a formal element in building up sentences as
the forms of the words themselves. And with these remarks I
shall leave the first main division of grammar, in which things were
to be looked at from without, from the sound or form. It will be
seen that in our scheme there is no room for the usual paradigms
giving in one place all the forms of the same word, like Latin servus
serve servum servo servi, amo amas amat amamus, etc. Such paradigms
may be useful for learners, 111 and in my system may be given
in an appendix to Morphology, but it should not be overlooked
that from a purely scientific point of view the paradigmatic arrangement
is not one of grammatical form, as it brings together, not the
same forms, but different forms of the same word, which only
belong to one another from a lexical point of view. The arrangement
here advocated is purely grammatical, treating together, in
its first part what may be called grammatical homophones (homomorphs)
and in its second part grammatical synonyms. It will be
remembered that we had the corresponding two classes in the two
divisions of the dictionary.44

Chapter III
Systematic Grammar — continued

Syntax. Universal Grammar? Differences of Languages. What Categories
to Recognize. Syntactic Categories. Syntax and Logic. Notional
Categories.

Syntax.

The second main division of grammar, as we have said, is occupied
with the same phenomena as the first, but from a different point
of view, from the interior or meaning (I → 0). We call this
syntax. The subdivisions will be according to the grammatical
categories, whose rôle and employment in speech is here defined.

One chapter of syntax will deal with Number; it will have
first to recount the several methods of forming the plural (dogs,
oxen, feet, we, those, etc.); this will be done most easily and summarily
by a reference to those paragraphs in our Morphology in
which each ending or other formative is dealt with. Next will
follow an account of everything that is common to all singulars
and to all plurals, no matter how these latter happen to be formed;
thus the plural in “a thousand and one nights” (where Danish
and German have the singular on account of one), the singular in
“more than one man” (= more men than one), cases of attraction,
the ‘generic’ use of singular and plural to denote the whole class
(a cat is a four-footed animal, cats are four-footed animals),
and many other things that could not find their place in the
morphological part.

If we compare these two parts of grammar and remember what
was said above of the two parts of a dictionary, we discover that
the two points of view are really those of the hearer and of the
speaker respectively. In a duologue the hearer encounters certain
sounds and forms, and has to find out their meaning — he moves
from without to within (O → I). The speaker, on the other hand,
starts from certain ideas which he tries to communicate; to him
the meaning is the given thing, and he has to find out how to
express it: he moves from within to without (I → O).

Universal Grammar?

With regard to the categories we have to establish in the
syntactic part of our grammatical system, we must first raise an
extremely important question, namely, are these categories purely
46logical categories, or are they merely linguistic categories? If
the former, then it is evident that they are universal, i.e. belong
to all languages in common; if the latter, then they, or at any rate
some of them, are peculiar to one or more languages as distinct
from the rest. Our question thus is the old one: Can there be
such a thing as a universal (or general) grammar?

The attitude of grammarians with regard to this question has
varied a good deal at different times. Some centuries ago it was
the common belief that grammar was but applied logic, and that
it would therefore be possible to find out the principles underlying
all the various grammars of existing languages; people consequently
tried to eliminate from a language everything that was
not strictly conformable to the rules of logic, and to measure everything
by the canon of their so-called general or philosophical
grammar. Unfortunately they were too often under the delusion
that Latin grammar was the perfect model of logical consistency,
and they therefore laboured to find in every language the distinctions
recognized in Latin. Not unfrequently a priori speculation and
pure logic led them to find in a language what they would never
have dreamt of if it had not been for the Latin grammar in which
they had been steeped from their earliest school-days. This
confusion of logic and Latin grammar with its consequence, a
Procrustean method of dealing with all languages, has been the
most fruitful source of mistakes in the province of grammar. What
Sayce wrote long ago in the article “Grammar” in the ninth edition
of the Encyclopaedia Britannica, “The endeavour to find the distinctions
of Latin grammar in that of English has only resulted in
grotesque errors, and a total misapprehension of the usage of the
English language” — these words are still worth taking to heart,
and should never be forgotten by any grammarian, no matter what
language he is studying.

Such ideas are least to be expected from philologists and linguists;
the latest occurrence I have come across is in Bally (St 156): “la
grammaire qui n'est que la logique appliquée au langage.”

Much more frequently found are such views as the following: “A
universal grammar is no more conceivable than a universal form
of political Constitution or of religion, or than a universal plant
or animal form; the only thing, therefore, that we have to do is
to notice what categories the actually existing languages offer
us, without starting from a ready-made system of categories”
(Steinthal, Charakteristik, 104 f.). Similarly, Benfey says that after
the results achieved by modern linguistics universal and philosophical
grammars have suddenly disappeared so completely that
their methods and views are now only to be traced in such books
as are unaffected by real science (Gesch. d. sprachwiss. 306). And
according to Madvig (1856, p. 20, Kl p. 121), grammatical categories
have nothing to do with the real relations of things in
themselves.

In spite of the aversion thus felt by most modern linguists to
the idea of a grammar arrived at by a process of deductive reasoning
and applicable to all languages, the belief that there are grammatical
notions or categories of a universal character will crop up
here and there in linguistic literature. Thus C. Alphonso Smith,
in his interesting Studies in English Syntax, says (p. 10) that there
is a kind of uniformity of linguistic processes which is not in individual
words, or sounds, or inflexions, but in word relations; that
is, in syntax. “Polynesian words, for example, are not our words,
but the Polynesians have their subjunctive mood, their passive
voice, their array of tenses and cases, because the principles of
syntax are psychical and therefore universal.” And on p. 20:
“One comes almost to believe that the norms of syntax are indestructible,
so persistently do they reappear in unexpected places.”

I am afraid that what is here said about Polynesians is not the
result of a comprehensive study of their languages, but is rather
based on the a priori supposition that no one can dispense with
the syntactic devices mentioned, exactly as the Danish philosopher
Kroman, after establishing a system of nine tenses on a logical basis,
says that “as a matter of course the language of every thinking
nation must have expressions” for all these tenses. A survey
of actually existing languages will show that these have in some
cases much less, in other cases much more, than we should expect,
and that what in one language is expressed in every sentence
with painstaking precision, is in another language left unexpressed
as if it were of no importance whatever. This is especially true
if we come to speak of such things as “the subjunctive mood” —
those languages which have a separate form for it by no means
48apply it to the same purposes, so that even if this mood is known
by the same name in English, German, Danish, French, and Latin,
it is not strictly speaking one and the same thing; it would be
perfectly impossible to give such a definition of the subjunctive
in any of these languages as would assist us in deciding where to
use it and where to use the indicative, still less such a definition
as would at the same time cover its employment in all the languages
mentioned. No wonder, therefore, that there are a great many
languages which have nothing that could be termed a subjunctive
mood, however widely the sense of the word should be stretched.
As a matter of fact, the history of English and Danish shows how
the once flourishing subjunctive has withered more and more,
until it can now be compared only with those rudimentary organs
whose use is problematic or very subordinate indeed.

Differences of Languages.

In comparative lexicology we constantly see how the things
to be represented by words are grouped differently according to
the whims of different languages, what is fused together in one
being separated in another: where English distinguishes between
clock and watch, and French between horloge, pendule, and montre,
German has only one word, uhr (but compensates through being
able by means of compounds to express many more shades:
turmuhr, schlaguhr, wanduhr, stubenuhr, standuhr, stutzuhr, taschenuhr);
where English has prince, German distinguishes between
prinz and fürst; French has café for coffee and café; French
temps corresponds to E. time and weather, and E. time to Fr. temps
and fois — to take only a few obvious examples. It is the same in
grammar, where no two languages have the same groupings and
make the same distinctions. In dealing with the grammar of a
particular language it is therefore important to inquire as carefully
as possible into the distinctions actually made by that language,
without establishing any single category that is not shown by
actual linguistic facts to be recognized by the speech-instinct of
that community or nation. However much the logician may insist
that the superlative is a necessary category which every thinking
nation must be able to express in its language, French has no superlative,
for though le plus pur, le plus fin, le meilleur serve to render
the genuine English superlative the purest, the finest, the best, these
forms are nothing but the comparative made definite by the addition
of the article, and we cannot even say that French has a superlative
consisting of the comparative with the definite article preposed, for
very often we have no definite article, but another determining
word which then has the same effect: mon meilleur ami, etc.49

On the other hand, while French has a real future tense (je
donnerai, etc.), it would be wrong to include a separate future
in the tense system of the English language. Futurity is often
either not expressed at all in the verb (I start to-morrow at six;
cf. also “If he comes”), or it is expressed by means of phrases
which do not signify mere futurity, but something else besides;
in will (he mil start at six) there is an element of volition, in
am to (the congress is to be held next year) an element of destiny,
in may (he may come yet) an element of uncertainty, and in shall
(I shall write to him to-morrow) an element of obligation. It is
true that the original meanings are often nearly obliterated, though
not to the extent to which the original meaning of infinitive + ai
(have to…) is totally forgotten in French futures. The obliteration
is especially strong in shall, as there is no sense of obligation
in “I shall be glad if you can come,” and as shall is hardly ever
used now in the original sense (compare the biblical “thou shalt
not kill” with the modern “you mustn't walk there”), shall
forms the nearest approach in English to a real auxiliary of the
future, and if it were used in all persons, we should have no hesitation
in saying that English had a future tense. But if we were
to recognize “he will come” as a future tense, we might just as
well recognize as future tenses “he may come,” “he is coming,”
“he is going to come,” and other combinations. Thus the objection
is not that will is a separate “word” and that to recognize a
“tense” we must always have a form of a verb in which the kernel
and the nexional ending make up one inseparable unit; nothing
would hinder us from saying that a language had a future tense
if it had an auxiliary (verb or adverb) that really served to indicate
future time, only this would be placed in that part of Morphology
which treats of words, and not, as the French future, in the part
that treats of word elements, — in the Syntax as viewed in this book
that would make no difference.

What Categories to Recognize.

The principle here advocated is that we should recognize in
the syntax of any language only such categories as have found in
that language formal expression, but it will be remembered that
“form” is taken in a very wide sense, including form-words and
word-position. In thus making form the supreme criterion one
should beware, however, of a mistaken notion which might appear
to be the natural outcome of the same principle. We say one
sheep, many sheep: are we then to say that sheep is not a singular
in the first phrase, and not a plural in the second, because it has
the same form, and that this form is rather to be called ‘common
50number’ or ‘no-number’ or something equivalent? It might
be said that cut in “I cut my finger every day” is not in the present
tense, and cut in “I cut my finger yesterday” is not in the past
tense (or preterit), because the form in both sentences is identical.
Further, if we compare “our king's love for his subjects” and
“our kings love their subjects,” we see that the two forms are the
same (apart from the purely conventional distinction made in
writing, but not in speaking, by means of the apostrophe), and a
strict formalist thus would not be entitled to state anything with
regard to the case and number of kings. And what about love?
There is nothing in the form to show us that it is a substantive in
the singular in one phrase and a verb in the plural in the other, and
we should have to invent a separate name for the strange category
thus created. The true moral to be drawn from such examples
is, however, I think, that it is wrong to treat each separate linguistic
item on its own merits; we should rather look at the language as
a whole. Sheep in many sheep is a plural, because in many lambs
and hundreds of other similar cases the English language recognizes
a plural in its substantives; cut in one sentence is in the present
and in the other in the past tense, because a difference at once
arises if we substitute he for I (he cuts, he cut), or another verb for
cut (I tear, I tore); kings in one instance is a genitive singular
and in the other a nominative plural, as seen in “the man's love
for his subjects” and “the men love their subjects,” and finally
love is a substantive and a verb respectively as shown by the form
in such collocations as “our king's admiration for his subjects”
and “our kings admire their subjects.” In other words, while we
should be careful to keep out of the grammar of any language
such distinctions or categories as are found in other languages,
but are not formally expressed in the language in question, we should
be no less averse to deny in a particular case the existence of distinctions
elsewhere made in the same language, because they happen
there to have no outward sign. The question, how many and what
grammatical categories a language distinguishes, must be settled
for the whole of that language, or at any rate for whole classes
of words, by considering what grammatical functions find expression
in form, even if they do not find such expression in all arid every
case where it might be expected: the categories thus established
are then to be applied to the more or less exceptional cases
where there is no external form to guide us. In English, for
instance, we shall have to recognize a plural in substantives,
pronouns, and verbs, but not in adjectives any more than in
adverbs; in Danish, on the other hand, a plural in substantives,
adjectives, and pronouns, but no longer in verbs. There will
be a special reason to remember this principle when we come
51to consider the question how many cases we are to admit in
English.

The principle laid down in the last few paragraphs is not unfrequently
sinned against in grammatical literature. Many writers will
discourse on the facility with which English can turn substantives
into verbs, and vice versa — but English never confounds the two
classes of words, even if it uses the same form now as a substantive,
and now as a verb: a finger and a find are substantives, and finger
and find in you finger this and find that are verbs, in flexion and in
function and everything. An annotator on the passage in Hamlet,
where the ghost is said to go “slow and stately” says with regard
to slow: “Adjectives are often used for adverbs” — no, slow
really is an adverb, just as long in “he stayed long” is an adverb,
even if the form is the same as in “a long stay,” where it is an
adjective. The substantive in five snipe or a few antelope or twenty
sail is often called a singular (sometimes a “collective singular”),
although it is no more a singular than sheep in five sheep: a form
which is always recognized as a plural, probably because grammarians
know that this word has had an unchanged plural from
Old English times. But history really has nothing to do with
our question. Snipe is now one form of the plural of that word
(“the unchanged plural”), and the fact that there exists another
form, snipes, should not make us blind to the real value of the form
snipe.

Syntactic Categories.

We are now in a position to return to the problem of the
possibility of a Universal Grammar. No one ever dreamed of a universal
morphology, for it is clear that all actually found formatives,
as well as their functions and importance, vary from language
to language to such an extent that everything about them must
be reserved for special grammars, with the possible exception of
a few generalities on the rôle of sentence-stress and intonation.
It is only with regard to syntax that people have been inclined to
think that there must be something in common to all human speech,
something immediately based on the nature of human thought,
in other words on logic, and therefore exalted above the accidental
forms of expression found in this or that particular language. We
have already seen that this logical basis is at any rate not
coextensive with the whole province of actual syntax, for many
languages do without a subjunctive mood, or a dative case, some
even without a plural number in their substantives. How far,
then, does this basic logic extend, and what does it mean exactly?

In the system sketched above we found, corresponding to
each separate form, an indication of its syntactic value or function.
52thus for the ending E. -s on the one hand “plural of substantive,”
on the other hand “third person singular present of verb,” etc.
Each of these indications comprised two or more elements, one of
which concerned the “part of speech” or word-class, one denoted
singular or plural number, one the third person, and finally one
the present tense. In English these indications contained comparatively
few elements, but if we take Latin, we shall find that
matters are often more complicated: the ending of bonarum,
for instance, denotes plural, feminine gender, and genitive case,
that of tegerentur plural, third person, imperfect tense, subjunctive
mood, passive voice, and so with other forms. Now it is clear
that though it is impossible, or not always possible, to isolate
these elements from a formal point of view (in animalium, where
is the sign of the plural, and where of the genitive? in fed, where
the indication of the person, of the perfect, of the indicative mood,
of the active voice, etc.?), on the other hand from the syntactic
point of view it is not only possible, but also natural to isolate
them, and to bring together all substantives, all verbs, all singulars,
all genitives, all subjunctives, all first persons, etc. We thus get a
series of isolated syntactic ideas, and we must even go one step further,
for some of these isolated syntactic ideas naturally go together,
forming higher groups or more comprehensive syntactic classes.

In this way substantives, adjectives, verbs, pronouns, etc., together
constitute the division of words into parts of speech or word-classes.

The singular and plural (with the dual) form the category of
number.

The nominative, accusative, dative, genitive, etc., form the
category of cases.

The present, preterit (imperfect, perfect), future, etc., form the
category of tenses.

The indicative, subjunctive, optative, imperative, etc., form the
category of moods.

The active, passive, and middle voice (medium) form the category
of ‘voices’ or ‘turns.’

The first, second, and third persons form, as the name indicates,
the category of persons.

The masculine, feminine, and neuter form the category of
genders.

Syntax and Logic.

We are able to establish all these syntactic ideas and categories
without for one moment stepping outside the province of grammar,
but as soon as we ask the question, what do they stand for, we
at once pass from the sphere of language to the outside world 112
53or to the sphere of thought. Now, some of the categories enumerated
above bear evident relations to something that is found in the
sphere of things: thus the grammatical category of number evidently
corresponds to the distinction found in the outside world
between “one” and “more than one”; to account for the various
grammatical tenses, present, imperfect, etc., one must refer to
the outside notion of “time”; the difference between the three
grammatical persons corresponds to the natural distinction between
the speaker, the person spoken to, and something outside of both.
In some of the other categories the correspondence with something
outside the sphere of speech is not so obvious, and it may be that
those writers who want to establish such correspondence, who
think, for instance, that the grammatical distinction between
substantive and adjective corresponds to an external distinction
between substance and quality, or who try to establish a “logical”
system of cases or moods, are under a fundamental delusion. This
will be examined in some of the following chapters, where we shall
see that such questions involve some very intricate problems.

If we analyze these grammatically, we see that (apart from the
different predicatives) they differ in that one is in the singular,
and the other in the plural number, and that one is in the present
tense, the other in the preterit or past tense. Yet both sentences
predicate something about a whole class, only the class is different
in the two sentences: in the former it is mankind without regard
to sex, in the latter the male part of mankind only, a sex-distinction
being thus implied in what is grammatically a numerical distinction.
And though the tenses are different, no real distinction of time is
meant, for the former truth is not meant to be confined to the
present moment, nor the second to some time in the past. What
is intended in both is a statement that pays no regard to the
54distinction between now and then, something meant to be true for all
time. A logician would have preferred a construction of language
in which both sentences were in the same universal number
(“omnial,” as Bréal calls it) and in the same universal or generic
tense, but the subject of the former in the common gender and that
of the latter in the masculine gender, for then the meaning would
have been unmistakable: “all human beings have been, are,
and always will be mortal,” and “all male human beings have been,
are, and always will be deceitful.” But as a matter of fact, this
is not the way of the English language, and grammar has to state
facts, not desires.

Notional Categories.

We are thus led to recognize that beside, or above, or behind,
the syntactic categories which depend on the structure of each
language as it is actually found, there are some extralingual categories
which are independent of the more or less accidental facts
of existing languages; they are universal in so far as they are
applicable to all languages, though rarely expressed in them in a
clear and unmistakable way. Some of them relate to such facts
of the world without as sex, others to mental states or to logic,
but for want of a better common name for these extralingual
categories I shall use the adjective notional and the substantive
notion. It will be the grammarian's task in each case to investigate
the relation between the notional and the syntactic categories.

Let us take a few French and German examples. Der soldat,
le soldat: male beings, masculine gender; die tochter, la pile:
female beings, feminine gender; der sperling, le cheval: beings
of both sexes, masculine gender; die maus, la souris: beings of
both sexes, feminine gender; das pferd: both sexes, neuter
gender; die schildmache, la sentinelle: male sex, feminine gender;
das weib: female sex, neuter gender; der tisch, le fruit: nonsexual,
masculine gender; die frucht, la table: non-sexual, feminine
gender; das buch: non-sexual, neuter gender. 113 In other departments
it is not possible as here to formulate two sets of terms,
one for the world of reality or universal logic, and one for the world
of grammar, but it should be our endeavour always to keep the
two worlds apart.

Syntactic categories thus, Janus-like, face both ways, towards
form, and towards notion. They stand midway and form the
56connecting link between the world of sounds and the world of
ideas. In speaking (or writing) we start from the right side (C)
of this scheme, and move through syntax (B) to the formal expression
(A): in hearing (or reading) the movement is in the opposite
direction, from A through B to C.

In finding out what categories to recognize in the third division
(C) it is important always to remember that these are to have a
linguistic significance; we want to understand linguistic (grammatical)
phenomena, and consequently it would not do to set
to work as if language did not exist, classifying things or ideas
without regard to their linguistic expression. On the contrary, we
should rather do, mutatis mutandis, what we did above when
establishing our syntactic categories: there we paid the strictest
attention to what had found expression in the forms of the language
examined, and here we must again pay the strictest attention to
the already discovered syntactic categories. It will be the task
of the greater part of this work to attempt a systematic review
of the chief notional categories in so far as they find grammatical
expression, and to investigate the mutual relation of these two
“worlds” in various languages. Often enough we shall find
that grammatical categories are at best symptoms, foreshadowings
of notional categories, and sometimes the ‘notion’ behind a grammatical
phenomenon is as elusive as Kant's ding an sich; and on
the whole we must not expect to arrive at a “universal grammar”
in the sense of the old philosophical grammarians. What we obtain
is the nearest approach to it that modern linguistic science will allow.

Postscript to Chapter III.

The eminent historian of the French language, Ferdinand Brunot, proposes
to revolutionize the teaching of (French) grammar by starting from
within, from the thoughts to be expressed, instead of from the forms. His
great book, La Pensée et la Langue, extremely fertile in new observations
and methodical remarks, was published (Paris, Masson et Cie, 1922) when
more than two-thirds of this volume was written either in its final shape
or in nearly the same shape in which it appears now. It is possible,
though I cannot at present feel it, that my book would have taken a different
shape, had M. Brunot's work appeared before my own convictions had
become settled; as it is now, though I hail him as a powerful ally, I disagree
with him on at least two important points. First, what he advocates as
the proper method (starting from within, from ‘la pensée’) should according
to my view be one of two ways of approaching the facts of language, one
from without to within, and another from within to without. And secondly,
grammar should be kept distinct from dictionary, while M. Brunot in his
lists of synonymous terms too often mixes up the two domains. Nor can I
share his utter contempt for the old theory of “parts of speech,” however
wrong it is in many details.57

Chapter IV
Parts of Speech

Old Systems. Definitions. The Basis of Classification. Language and
Real Life. Proper Names. Actual Meaning of Proper Names.

Old Systems.

It is customary to begin the teaching of grammar by dividing
words into certain classes, generally called “parts of speech”
— substantives, adjectives, verbs, etc. — and by giving definitions
of these classes. The division in the main goes back to the Greek
and Latin grammarians with a few additions and modifications,
but the definitions are very far from having attained the degree
of exactitude found in Euclidean geometry. Most of the definitions
given even in recent books are little better than sham definitions
in which it is extremely easy to pick holes; nor has it been possible
to come to a general arrangement as to what the distinction is to
be based on — whether on form (and form-changes) or on meaning
or on function in the sentence, or on all of these combined.

The most ingenious system in this respect is certainly that of
Varro, who distinguishes four parts of speech, one which has cases
(nouns, nomina), one which has tenses (verbs), one which has both
cases and tenses (participles), and one which has neither (particles).
If this scheme is now generally abandoned, the reason evidently
is that it is so manifestly made to fit Latin (and Greek) only and
that it is not suitable either to modern languages evolved out of
a linguistic structure similar to Latin (English, for instance) or to
languages of a totally different type, such as Eskimo.

This system, again, fits only the ancient languages of our family,
and differs mainly from Varro's scheme in being based on gender
instead of case distinction. Both are equally arbitrary. In
both tense is made the really distinctive feature of verbs, a conception
which has found expression in the German rendering of
verb by zeitwort: but on that showing Chinese has no verbs, while
on the other hand we shall see later that nouns sometimes distinguish
tenses. Other grammarians think that the distinctive feature of
verbs is the personal endings (Steinthal, etc.). But this criterion
would also exclude the Chinese verb from that denomination;
in Danish, again, verbs do not distinguish persons, and it is no help
out of the difficulty to say, as Schleicher does (NV 509) that “verbs
are words which have or have had personal endings,” for it should
not be necessary to know linguistic history to determine what
part of speech a word belongs to.

Definitions.

Let us now cast a glance at some of the definitions found in
J. Hall and E. A. Sonnenschein's Grammar (London, 1902). “Nouns
name. Pronouns identify without naming.” I cannot see that
who in Who killed Cock Robin? identifies; it rather asks some one
else to identify. And none in Then none was for a party — whose
identity is established by that pronoun? “Adjectives are used
with Nouns, to describe, identify or enumerate.” 115 But cannot
adjectives be used without nouns? (the absent are always at fault.
He was angry). On the other hand, is poet in Browning the poet an
adjective? “By means of Verbs something is said about something
or somebody”: You scoundrel — here something is said
about “you” just as much as in You are a scoundrel, and in the
latter sentence it is not the verb are, but the predicative that says
something. “Conjunctions connect groups of words or single
words” — but so does of in a man of honour without being on that
account a conjunction. Not a single one of these definitions is
either exhaustive or cogent. 21659

The Basis of Classification.

Some grammarians, feeling the failure of such definitions as
those just given have been led to despair of solving the difficulty
by the method of examining the meaning of words belonging to
the various classes: and therefore maintain that the only criterion
should be the form of words. This is the line taken, for instance,
by J. Zeitlin (“On the Parts of Speech. The Noun,” in The
English Journal, March 1914), though unfortunately he deals only
with nouns. He takes “form” in rather a wide sense, and says
that “in English the noun does still possess certain formal characteristics
which attach to no other class of words. These are the
prefixing of an article or demonstrative, the use of an inflexional
sign to denote possession and plurality, and union with prepositions
to mark relations originally indicated by inflexional endings.”
He is careful to add that the absence of all the features enumerated
should not exclude a word from being a noun, for this should be
described “as a word which has, or in any given usage may have”
those formal signs.

If form in the strictest sense were taken as the sole test, we
should arrive at the absurd result that must in English, being
indeclinable, belonged to the same class as the, then, for, as, enough,
etc. Our only justification for classing must as a verb is that we
recognize its use in combinations like I must (go), must we (go)? as
parallel to that of I shall (go), shall we (go)? — in other words, that
we take into consideration its meaning and function in the sentence.
And if Zeitlin were to say that the use of must with a
nominative like I is “formal” (in the same way as “union with
prepositions” was one of the “formal” tests by which he recognized
a noun), I should not quarrel with him for taking such things
into account, but perhaps for calling them formal considerations.

In my opinion everything should be kept in view, form, function,
and meaning, but it should be particularly emphasized that form,
which is the most obvious test, may lead to our recognizing some
word-classes in one language which are not distinct classes in
other languages, and that meaning, though very important, is
most difficult to deal with, and especially that it is not possible
to base a classification on short and easily applicable definitions.

We may imagine two extreme types of language structure,
one in which there is always one definite formal criterion in each
word-class, and one in which there are no such outward signs in
60any class. The nearest approach to the former state is found, not
in any of our natural languages, but in an artificial language such
as Esperanto or, still better, Ido, where every common substantive
ends in -o (in the plural in -i), every adjective in -a, every (derived)
adverb in -e, every verb in -r, -s, or -z according to its mood. The
opposite state in which there are no formal signs to show word-classes
is found in Chinese, in which some words can only be used
in certain applications, while others without any outward change
may function now as substantives, now as verbs, now as adverbs,
etc., the value in each case being shown by syntactic rules and the
context.

English here steers a middle course though inclining more
and more to the Chinese system. Take the form round: this
is a substantive in “a round of a ladder,” “he took his daily
round,” an adjective in “a round table,” a verb in “he failed to
round the lamp-post,” an adverb in “come round to-morrow,”
and a preposition in “he walked round the house.” While similarly
may be a substantive (he stayed here for a while), a verb
(to while away time), and a conjunction (while he was away).
Move may be a substantive or a verb, after a preposition, an adverb,
or a conjunction, 117 etc.

On the other hand, we have a great many words which can
belong to one word-class only; admiration, society, life can only
be substantives, polite only an adjective, was, comprehend only
verbs, at only a preposition.

To find out what particular class a given word belongs to,
it is generally of little avail to look at one isolated form. Nor
is there any flexional ending that is the exclusive property of any
single part of speech. The ending -ed (-d) is chiefly found in verbs
(ended, opened, etc.), but it may be also added to substantives to
form adjectives (blue-eyed, moneyed, talented, etc.). Some endings
may be used as tests if we take the meaning of the ending also into
account; thus if an added -s changes the word into a plural, the
word is a substantive, and if it is found in the third person singular,
the word is a verb: this, then, is one of the tests for keeping the
substantive and the verb round apart (many rounds of the ladder;
he rounds the lamp-post). In other cases the use of certain words
in combinations is decisive, thus my and the in “my love for her”
and “the love I bear her,” as against “I love her,” show that
love is a substantive and not a verb as in the last combination (cf.
my admiration, the admiration as against I admire, where admiration
and admire are unambiguous). 21861

It is, however, very important to remark that even if round
and love and a great many other English words belong to more
than one word-class, this is true of the isolated form only: in
each separate case in which the word is used in actual speech it
belongs definitely to one class and to no other. But this is often
overlooked by writers who will say that in the sentence “we tead
at the vicarage” we have a case of a substantive used as a verb.
The truth is that we have a real verb, just as real as dine or eat,
though derived from the substantive tea — and derived without any
distinctive ending in the infinitive (cf. above, p. 52). To form a
verb from another word is not the same thing as using a substantive
as a verb, which is impossible. Dictionaries therefore must recognize
love sb. and love v. as two words, and in the same way tea sb.
and tea v. In such a case as wire they should even recognize
three words, (1) sb. ‘metallic thread,’ (2) ‘to send a message by
wire, to telegraph’ — a verb formed from the first word without
any derivative ending, (3) ‘message, telegram’ — a sb. formed
from the verb without any ending.

In teaching elementary grammar I should not begin with
defining the several parts of speech, least of all by means of the
ordinary definitions, which say so little though seeming to say so
much, but in a more practical way. As a matter of fact the trained
grammarian knows whether a given word is an adjective or a verb
not by referring to such definitions, but in practically the same
way in which we all in seeing an animal know whether it is a cow
or a cat, and children can learn it much as they learn to distinguish
familiar animals, by practice, being shown a sufficient number of
specimens and having their attention drawn successively now to.
this and now to that distinguishing feature. I should take a
piece of connected text, a short story for instance, and first give
it with all the substantives printed in italics. After these have
been pointed out and briefly discussed the pupil will probably
have little difficulty in recognizing a certain number of substantives
of similar meaning and form in another piece in which they are
not marked as such, and may now turn his attention to adjectives,
using the same text as before, this time with the adjectives italicised.
By proceeding in this way through the various classes he will
gradually acquire enough of the “grammatical instinct” to be
62able to understand further lessons in accidence and syntax in his
own and foreign languages.

It is not, however, my purpose here to give advice on elementary
grammatical teaching, but to try to arrive at some scientific understanding
of the logical basis of grammar. This will be best attained,
I think, if we consider what it is that really happens when we talk
of something, and if we examine the relation between the real
world and the way in which we are able to express its phenomena
in language.

Language and Real Life.

Real life everywhere offers us only concretissima: you see
this definite apple, definitely red in one part and yellowish in that
other part, of this definite size and shape and weight and degree
of ripeness, with these definite spots and ruggednesses, in one definite
light and place at this definite moment of this particular day, etc.
As language is totally unable to express all this in corresponding
concreteness, we are obliged for the purpose of communication
to ignore many of these individual and concrete characteristics:
the word “apple” is not only applied to the same apple under
other circumstances, at another time and in another light, but
to a great many other objects as well, which it is convenient to
comprise under the same name because otherwise we should have
an infinite number of individual names and should have to invent
particular names for new objects at every moment of the day.
The world is in constant flux around us and in us, but in order to
grapple with the fleeting reality we create in our thought, or at
any rate in our language, certain more or less fixed points, certain
averages. Reality never presents us with an average object,
but language does, for instead of denoting one actually given thing
a word like apple represents the average of a great many objects
that have something, but of course not everything, in common.
It is, in other words, absolutely necessary for us, if we want to
communicate our impressions and ideas, to have more or less
abstract 119 denominations for class-concepts: apple is abstract
in comparison with any individual apple that comes within our
ken, and so is fruit to an even higher degree, and the same is still
more true of such words as red or yellow and so on: language
everywhere moves in abstract words, only the degree of abstraction
varies infinitely.

Now, if you want to call up a very definite idea in the mind
of your interlocutor you will find that the idea is in itself very
complex, and consists of a great many traits, really more than you
63would be able to enumerate, even if you were to continue to the
end of time. You have to make a selection, and you naturally
select those traits that according to the best of your belief will be
best fitted to call up exactly the same idea in the other man's
mind. More than that, you select also those that will do it in
the easiest way to yourself and to your hearer, and will spare both
of you the trouble of long circuitous expressions. Therefore
instead of a timid gregarious woolly ruminant mammal you say
sheep, instead of male ruler of independent state you say king, etc.
Thus wherever you can, you use single special terms instead of
composite ones. But as special terms are not available for all
composite ideas, you often have to piece together expressions by
means of words each of which renders one of the component traits
of the idea in your mind. Even so, the designation is never exhaustive.
Hence the same man may under various circumstances
be spoken of in totally different ways, and yet the speaker is in
each case understood to refer to the same individual: as “James
Armitage” or simply “Armitage” or “James,” or else as “the
little man in a suit of grey whom we met on the bridge,” or as
“the principal physician at the hospital for women's diseases,” as
“the old Doctor,” as “the Doctor,” as “Her husband,” as “Uncle
James,” as “Uncle,” or simply as “he.” In each case the hearer
supplies from the situation (or context), i.e. from his previous
knowledge, a great many distinctive traits that find no linguistic
expression — most of all in the last-mentioned case, where the
pronoun “he” is the only designation.

Among these designations for the same individual there are
some which are easily seen to have a character of their own, and
we at once single out James and Armitage (and, of course, the combination
James Armitage) as proper names, while we call such
words as man, physician, doctor, husband, uncle, which enter into
some of the other designations, common names, because they are
common to many individuals, or at least to many more, than are
the proper names. Let us now try to consider more closely what
is the essence of proper names.

Proper Names.

A proper name would naturally seem to be a name that can
only be used in speaking of one individual. It is no objection to
this definition that the Pyrenees or the United States are proper names,
for in spite of the plural form by which they are designated this range
of mountains and this political body are looked upon as units, as
individuals: it is not possible to speak of one Pyrenee or of one
United State, but only of one of the Pyrenees, one of the United States.64

A more serious difficulty encounters us when we reflect that
John and Smith by common consent are reckoned among proper
names, and yet it is indubitable that there are many individuals
that are called John, and many that are called Smith, and even a
considerable number that are called John Smith. Rome similarly
is a proper name, yet there are at least five towns of that name in
North America besides the original Rome in Italy. How then are
we to keep up the distinction between proper and common names?

A well-known attempt at a solution is that of John Stuart
Mill (System of Logic, I, Ch. II). According to him proper names are
not connotative; they denote the individuals who are called by
them; but they do not indicate or imply any attributes as belonging
to those individuals, they answer the purpose of showing what
thing it is we are talking about, but not of telling anything about
it. On the other hand, such a name as man, besides denoting
Peter, James, John, and an indefinite number of other individuals,
connotes certain attributes, corporeity, animal life, rationality,
and a certain external form, which for distinction we call the human.
Whenever, therefore, the names given to objects convey any
information, that is, whenever they have any meaning, the meaning
resides not in what whey denote, but in what they connote. The
only names of objects which connote nothing are proper names;
and these have, strictly speaking, no signification.

Similarly a recent Danish writer (H. Bertelsen, Fællesnavne
og egennavne, 1911) says that John is a proper name, because there
is nothing else besides the name that is common to all John's in
contradistinction to Henry's and Richard's, and that while a common
name indicates by singling out something that is peculiar to the
individual persons or things to whom the name is applied, the opposite
is true of a proper name. Accordingly, the distinction has
nothing to do with, or at any rate has no definite relation to, the
number of individuals to whom a name is given. I do not think,
however, that this view gets to the bottom of the problem.

Actual Meaning of Proper Names.

What in my view is of prime importance is the way in which
names are actually employed by speakers and understood by
hearers. Now, every time a proper name is used in actual speech
its value to both speaker and hearer is that of denoting one individual
only, and being restricted to that one definite being. To-day,
in talking to one group of my friends, I may use the name John
about a particular man of that name, but that does not prevent
me from using it to-morrow in different company of a totally
different individual; in both cases, however, the name fulfils its
65purpose of calling up in the mind of the hearer the exact meaning
which I intend. Mill and his followers lay too much stress on
what might be called the dictionary value of the name, and too
little on its contexual value in the particular situation in which
it is spoken or written. It is true that it is quite impossible to
tell the meaning of John when nothing but the name is before us,
but much the same thing may be said of a great many “common
names.” If I am asked to give the meaning of jar or sound or
palm or tract, the only honest answer is, Show me the context,
and I will tell you the meaning. In one connexion pipe is understood
to mean a tobacco-pipe, in another a water-pipe, in a third
a boatswain's whistle, in another one of the tubes of an organ, and
in the same way John, in each separate sentence in which it is
used, has one distinct meaning, which is shown by the context
and situation; and if this meaning is more special in each case
than that of pipe or the other words mentioned, this is only another
side of the important fact that the number of characteristic traits
is greater in the case of a proper name than in the case of a common
name. In Mill's terminology, but in absolute contrast to his view,
I should venture to say that proper names (as actually used)
“connote” the greatest number of attributes.

The first time you hear of a person or read his name in a newspaper,
he is “a mere name” to you, but the more you hear and
see of him the more will the name mean to you. Observe also the
way in which your familiarity with a person in a novel grows the
farther you read. But exactly the same thing happens with a
“common name” that is new to you, say ichneumon: here again,
the meaning or connotation grows along with the growth of your
knowledge. This can only be denied on the assumption that the
connotation of a name is something inherent in the name, something
with an existence independent of any human mind knowing and
using the name: but that is surely absurd and contrary to all right
ideas of the essence of language and human psychology.

If proper names as actually understood did not connote many
attributes, we should be at a loss to understand or explain the
everyday phenomenon of a proper name becoming a common
name. A young Danish girl was asked by a Frenchman what her
father was, and in her ignorance of the French word for ‘sculptor’
got out of the difficulty by saying: “Il est un Thorvaldsen en
miniature.” Oscar Wilde writes: “Every great man nowadays
has his disciples, and it is always Judas who writes the biography”
(Intentions, 81) — a transition to speaking of a Judas. Walter
Pater says that France was about to become an Italy more Italian
than Italy itself (Renaissance, 133). In this way Cæsar became
the general name for Roman emperors, German Kaisers and
66Russian tsars (in Shakespeare's tragedy III. 2. 55, the rabble
shouts: “Liue Brutus, liue, liue… Let him be Cæsar”) — to
mention only a few examples. 120

Logicians, of course, see this, but they dismiss it with some remark
like this (Keynes FL 45): “Proper names, of course, become connotative
when they are used to designate a certain type of person;
for example, a Diogenes, a Thomas, a Don Quixote, a Paul Pry,
a Benedick, a Socrates. But, when so used, such names have really
ceased to be proper names at all; they have come to possess all
the characteristics of general names.” The logician as such with
his predilection for water-tight compartments in the realm of
ideas, is not concerned with what to me as a linguist seems a most
important question, viz. how is it to be explained that a sequence
of sounds with no meaning at all suddenly from non-connotative
becomes connotative, and that this new full meaning is at once
accepted by the whole speaking community?

If we take the view suggested above, this difficulty vanishes at
once. For what has happened is simply this, that out of the
complex of qualities characteristic of the bearer of the name concerned
(and, as I should say, really connoted by the name) one
quality is selected as the best known, and used to characterize some
other being or thing possessed of the same quality. But this is
exactly the same process that we see so very often in common
names, as when a bell-shaped flower is called a bell, however different
it is in other respects from a real bell, or when some politician is
called an old fox, or when we say that pearl, or jewel, of a woman.
The transference in the case of original proper names is due to the
same cause as in the case of common names, viz. their connotativeness,
and the difference between the two classes is thus seen to be
one of degree only.

The difference between Cræsus as applied to the one individual
and as used for a very rich man may be compared to that between
human (connoting everything belonging to man) and humane
(selecting one particular quality).

With our modern European system of composite personal
names we have a transference of names of a somewhat different
kind, when a child through the mere fact of his birth acquires his
father's family name. Here it would be rash to assert that Tymperleys,
for instance, of the same family have nothing in common
but their name; they may sometimes be recognized by their
nose or by their gait, but their common inheritance, physical and
psychical, may be much more extensive, and so the name Tymperley
may get a sense not essentially different from that of such “common
67names” as Yorkshireman, or Frenchman, or negro, or dog. In
some of the latter cases it is difficult to define exactly what the
name “connotes” or by what characteristics we are able to tell
that a person belongs to this or the other class, yet logicians agree
that all these names are connotative. Then why not Tymperley?

It is different, of course, with Christian names, which are given
in a much more arbitrary way. One Maud may have been so
called “after” a rich aunt, and another simply because her parents
thought the name pretty, and the two thus have nothing but the
name in common. The temple of worship and the temple of the
head are in much the same case. (The two Mauds have really
more in common than the two temples, for they are both female
human beings. 121) But that does not affect the main point in my
argument, which is that whenever the name Maud is naturally
used it makes the hearer think of a whole complex of distinctive
qualities or characteristics.

Now it will be said against this view that “the connotation
of a name is not the quality or qualities by which I or anyone
else may happen to recognize the class which it denotes. For
example, I may recognize an Englishman abroad by the cut of
his clothes, or a Frenchman by his pronunciation, or a proctor
by his bands, or a barrister by his wig; but I do not mean any
of these things by these names, nor do they (in Mill's sense) form
any part of the connotation of the names” (Keynes FL 43). This
seems to establish a distinction between essential characteristics
comprised in the “connotation” 222 and unessential or accidental
qualities. But surely no sharp line can be drawn. If I want to
know what is connoted by the names salt and sugar respectively,
is it necessary to apply chemical tests and give the chemical formula
of these two substances, or am I permitted to apply the popular
criterion of tasting them? What qualities are connoted by the
word “dog”? In this and in a great many other cases we apply
class-names without hesitation, though very often we should be
embarrassed if asked what we “mean” by this or that name or
why we apply it in particular instances. Sometimes we recognize
a dog by this, and sometimes by that characteristic, or group of
characteristics, and if we apply the name “dog” to a particular
animal, it means that we feel confident that it possesses the rest
of that complex of traits which together make up dog-nature. 32368

It should also be remembered that what we designate by an
individual name is, if we look very closely into it, merely an abstraction.
Each individual is constantly changing from moment to
moment, and the name serves to comprehend and fix the permanent
elements of the fleeting apparitions, or as it were, reduce them
to a common denominator. Thus we understand sentences like
the following, which are very hard to account for under the assumption
that proper names are strictly non-connotative: he felt
convinced that Jonas was again the Jonas he had known a week
ago, and not the Jonas of the intervening time (Dickens) | there
were days when Sophia was the old Sophia — the forbidding, difficult
Sophia (Bennett) | Anna was astounded by the contrast between
the Titus of Sunday and the Titus of Monday (id.) | The Grasmere
before and after this outrage were two different vales (de Quincey).
In this way, too, we may have a plural of a proper name: Darius
had known England before and after the repeal of the Corn Laws,
and the difference between the two Englands was so strikingly
dramatic… (Bennett).

Linguistically it is utterly impossible to draw a sharp line of
demarcation between proper names and common names. We have
seen transitions from the former to the latter, but the opposite
transition is equally frequent. Only very few proper names have
69always been such (e.g. Rasselas), most of them have originated,
totally or partially, in common names specialized. Is “the Union”
as applied to one particular students' union at Oxford or Cambridge
a proper name? Or the “British Academy” or the “Royal
Insurance Company,” or — from another sphere — “Men and
Women” or “Outspoken Essays” or “Essays and Reviews”
as book-titles? The more arbitrary the name is, the more inclined
we are to recognize it at once as a proper name, but it is no indispensable
condition. The Dover road (meaning ‘the road that
leads to Dover’) is not originally a proper name, while Dover Street
which has no connexion with Dover and might just as well have
been baptized Lincoln Street, is a proper name from the first. But
the Dover Road may in course of time become a proper name, if
the original reason for the name is forgotten and the road has
become an ordinary street; and the transition may to some extent
be marked linguistically by the dropping of the definite article.
One of the London parks is still by many called “the Green Park,”
but others omit the article, and then Green Park is frankly a proper
name; compare also Central Park in New York, New College,
Newcastle. Thus, the absence of the article in English (though not
in Italian or German) becomes one of the exterior marks by which
we may know proper from common names.

In the familiar use of such words as father, mother, cook, nurse
without the article we accordingly have an approximation to
proper names; no doubt they are felt as such by children up to a
certain age, and this is justified if the mother or an aunt in speaking
to the child says father not of her own, but of the child's father.

The specialization which takes place when a common name
becomes a proper name is not different in kind, but only in degree,
from specializations to be observed within the world of common
names. Thus when the Black Forest (or, still more distinctly, the
German name Schwarzwald) has become the name of a particular
mountain range, the relation between this name and the combination
“the black forest” which might be applied as a common name to
some other forest is similar to that between the blackbird and the
black bird. 124

Our inquiry, therefore, has reached this conclusion, that no
sharp line can be drawn between proper and common names, the
70difference being one of degree rather than of kind. A name always
connotes the quality or qualities by which the bearer or bearers
of the name are known, i.e. distinguished from other beings or
things. The more special or specific the thing denoted is, the more
probable is it that the name is chosen arbitrarily, and so much
the more does it approach to, or become, a proper name. If a
speaker wants to call up the idea of some person or thing, he has
at his command in some cases a name specially applied to the individual
concerned, that is, a name which in this particular situation
will be understood as referring to it, or else he has to piece together
by means of other words a composite denomination which is sufficiently
precise for his purpose. The way in which this is done
will be the subject of our consideration in the next chapter.71

Chapter V
Substantives and Adjectives

Survey of Forms. Substance and Quality. Specialization. Interchange of
the Two Classes. Other Combinations.

Survey of Forms.

Among the designations for the same individual which we found
above, p. 64, there were some which contained two elements
that evidently stood in the same relation to each other, viz. little
man, principal physician, old Doctor. Here we call the words
little, principal, and old adjectives, and man, physician, and Doctor
substantives. Adjectives and substantives have much in common,
and there are cases in which it is difficult to tell whether a word
belongs to one or the other class; therefore it is convenient to
have a name that comprises both, and in accordance with the old
Latin terminology which is frequently found also in recent continental
works on grammar, I shall use the word noun (Lat. nomen)
for the larger class of which substantives and adjectives are subdivisions.
English scholars generally use the word noun for what
is here called substantive; but the terminology here adopted
gives us on the one hand the adjective nominal for both classes,
and on the other hand the verb substantivize when we speak, for
instance, of a substantivized adjective.

While in some languages, e.g. Finnish, it seems impossible to
find any criteria in flexion that distinguish substantives from
adjectives, a word like suomalainen being thus simply a noun,
whether we translate it in some connexions as a substantive (Finn,
Finlander) or in others as adjective (Finnish), our own family of
languages distinguishes the two classes of nouns, though with
different degrees of explicitness. In the older languages, Greek,
Latin, etc., the chief formal difference has reference to gender
and is shown by the concord of adjectives with their substantives.
While every substantive is of one definite gender, the adjective
varies, and it is the fact that we say bonus dominus, bona mensa,
bonum templum, that obliges us to recognize substantives and
adjectives as two distinct classes of nouns. Now it is interesting
to note that adjectives are as it were more “orthodox” in their
gender flexions than substantives: we have masculine substantives
72in -a and feminine substantives in -us, but only bonus in the masculine
and bona in the feminine (bonus poeta, bona fagus). On the
whole substantives present many more irregularities in their flexion
(indeclinable or defective words, words in which one stem supplements
another) than adjectives. The same characteristic difference
is still found in German grammar: substantives are more individualistic
and conservative, while adjectives are more subject to
the influence of analogy.

In the Romanic languages, apart from the disappearance of
the neuter gender, the same relations obtain between the two
classes as in Latin, though in spoken French the distinctions
between the masculine and feminine forms have largely been
obliterated — donné and donnée, poli and polie, menu and menue,
grec and grecque being pronounced the same. It is also noteworthy
that there is no invariable rule for the position of adjectives, which
are in some cases placed before, and in others after their substantives.
As a consequence of this, one may here and there be in
doubt which of two collocated words is the substantive and which
the adjective, thus in un savant aveugle, un philosophe grec (see
below); such combinations as un peuple ami, une nation amie
(also une maîtresse femme) may be taken either as a substantive
(peuple, nation, femme) with an adjective, or else as two substantives
joined very much like English boy messenger, woman writer.

In the Gothonic (Germanic) languages similar doubts cannot,
as a rule, exist. At a very early date, adjectives took over some
endings from the pronouns, and then they developed the peculiar
distinction between a strong and a weak declension, the latter
originally an -n-flexion transferred from one class of substantives
and gradually extended to all adjectives and chiefly used after
a defining word, such as the definite article. This state of things
is preserved with some degree of fidelity in German, where we
still have such distinctly adjectival forms as ein alter mann, der
alte mann, alte manner, die alten manner, etc. Icelandic still keeps
the old complicated system of adjective flexion, but the other
Scandinavian languages have greatly simplified it, though retaining
the distinction between strong and weak forms, e.g. Dan. en
gammel mand, den gamle mand ‘an old man, the old man.’

In Old English things were pretty much the same as in German.
But in course of time, phonetic and other developments have
brought about a system that is radically different from the older
one. Some endings, such as those containing r, have completely
disappeared; this has also happened to the endings -e and -en,
which formerly played a very important r6le in both substantives
and adjectives. While -s was formerly used in the genitive of
adjectives in the sg. (m. and n.), it has now been completely
73discarded from the adjectives, which consequently have now only
one form for all cases in both numbers, no matter whether they
are preceded or not by the definite article. On the other hand,
the simplification of substantive flexions, though very radical, has
not been quite so thorough as that of the adjectives. Here the
-s-endings have been especially vigorous, and now form the chief
distinctive feature of substantives, while every trace of the old
Aryan concord has disappeared. Thus we must say that in the
old boy's (gen.) and the old boys' (pi.), we see that old is an adjective,
from its having no ending, and that boys is a substantive, from
the ending -s. When we have the blacks used of the negro race,
the adjective black has become completely substantivized;
similarly the heathens is a substantive, while the heathen continues
to be an adjective, even if it stands alone without any following
substantive, employed in what many grammarians call a “substantival
function.” Accordingly, in Shakespeare, H5 III. 5. 10
“Normans, but bastard Normans, Norman bastards” we have
first the adj. bastard and the subst. Normans, and then the adj.
Norman and the subst. bastards.

Substance and Quality.

This brief survey has shown us that though the formal distinction
between substantive and adjective is not marked with equal
clearness in all the languages considered, there is still a tendency
to make such a distinction. It is also easy to show that where
the two classes are distinguished, the distribution of the words is
always essentially the same: words denoting such ideas as stone,
tree, knife, woman are everywhere substantives, and words for
big, old, bright, grey are everywhere adjectives. This agreement
makes it highly probable that the distinction cannot be purely
accidental t it must have some intrinsic reason, some logical or
psychological (“notional”) foundation, and we shall now proceed
to examine what that foundation is.

An answer very often given is that substantives denote substances
(persons and things), and adjectives qualities found in these
things. This definition is evidently at the root of the name substantive,
but it cannot be said to be completely satisfactory. The
names of many “substances” are so patently derived from some
one quality that the two ideas cannot possibly be separated: the
blacks, eatables, desert, a plain must be called substantives and
are in every respect treated as such in the language. And no
doubt a great many other substantives the origin of which is now
forgotten were at first names of one quality singled out among
others by the speakers. So, linguistically the distinction between
74“substance” and “quality” cannot have any great value. And
from a philosophical point of view it may be said that we know
substances only through their qualities; the essence of any substance
is the sum of all those qualities that we are able to perceive
(or conceive) as in some way connected. While formerly substances
were thought of as realities per se and qualities were
considered as having no existence in themselves, there is perhaps
now a strong tendency in the opposite direction, to look upon the
substance or “substratum” of various qualities as a fiction, rendered
more or less necessary by our habits of thought, and to say
that it is the “qualities” that ultimately constitute the real world,
i.e. everything that can be perceived by us and is of value to us. 125

Whether the reader may be inclined to attach much or little
importance to the arguments just presented, he must acknowledge
that the old definition is powerless to solve the riddle of the
so-called “abstracts” like wisdom, kindness, for though these words
are to all intents and purposes substantives and are treated as
such in all languages, yet they evidently denote the same qualities
as the adjectives wise and kind, and there is nothing substantial
about them. Whatever notional definition one gives of a substantive,
these words make difficulties, and it will be best at the
present moment to leave them out of consideration altogether —
we shall return to them in a following chapter (X).

Specialization.

Apart from “abstracts,” then, I find the solution of our problem
in the view that on the whole substantives are more special
than adjectives, they are applicable to fewer objects than adjectives,
in the parlance of logicians, the extension of a substantive
is less, and its intension is greater than that of an adjective. The
adjective indicates and singles out one quality, one distinguishing
mark, but each substantive suggests, to whoever understands it,
many distinguishing features by which he recognizes the person
or thing in question. What these features are, is not as a rule
indicated in the name itself; even in the case of a descriptive
name one or two salient features only are selected, and the others
are understood: a botanist easily recognizes a bluebell or a blackberry
bush even at a season when the one has no blue flowers and
the others no black berries. 22675

The difference between the two classes is seen very clearly
when the same word may be used in both capacities. We have
a great many substantivized adjectives, but their meaning is
always more special than that of the corresponding adjectives,
compare e.g. a cathedral (une cathédrale, Sp. un catedral), the blacks
(= negroes), natives (both = ‘inhabitants’ and ‘oysters’), sweets,
evergreens, etc. The same is true of those cases where the adjectival
use has disappeared, as in tithe (orig. a numeral, ‘tenth’),
friend (an old participle of a verb ‘to love’), and of such old Latin
or Greek participles as fact, secret, serpent, Orient, horizon.

Inversely, when a substantive is made into an adjective, we
find that its meaning has become less special. Thus the French
rose, mauve, puce, etc., are more general when they stand as colour-indicating
adjectives than as substantives: they can be applied
to more different things, as they now “connote” only one of the
characteristics that go to make up the things they stand for in
their original signification. 127 English examples of the transition
are chief, choice, dainty (orig. ‘a delicacy’), level, kindred (orig.
‘relationship’).

The Latin adjective ridiculus according to Bréal (MSL 6. 171)
is evolved from a neuter substantive ridiculum ‘objet de risée,’
formed in the same way as curriculum, cubiculum, vehiculum.
When applied to persons it took masculine and feminine endings,
ridiculus, ridicula, and it is this formal trait which made it into
an adjective; but at the same time its signification became slightly
more general and eliminated the element of ‘thing.’

A gradual transition from substantive to adjective is seen in
the so-called weak adjectives in Gothonic. As Osthoff has pointed
out, these go back to an old substantive-formation parallel to that
found in Gr. strabōn ‘the squint-eyed man’ corresponding to the
adj. strabos ‘squinting,’ or Lat. Cato Catonis ‘the sly one,’ cp.
adj. catus, Macro cp. adj. macer. In Gothonic this was gradually
extended, but at first these forms, like the Greek and Latin words
mentioned, were nicknames or distinguishing names, thus individual
in their application. As Osthoff says, Latin M. Porcius Cato,
Abudius Rufo, transferred into German, meant something like
M. Porcius der Kluge, Abudius der Rote, just as in OHG we have
76with the same ending Ludowig ther snello, and as we still in German
have the weak form of the adjective in Karl der Grosse, Friederich
der Weise, August der Starke. The definite article was not at first
required, cf. ON Brage Gamle (‘the old one’) and only later Are
enn (hinn) gamle. Thus also in Beowulf beahsele beorhta, originally
to be interpreted as two substantives in apposition, ‘ringhall — the
bright one’; href en blaca ‘raven, the black being.’ A combination
like “þær se goda sæt | Beowulf” is at first like “there the
good one sat, (namely) Beowulf,” parallel to “peer se cyning sæt,
Beowulf,” but later se goda, was connected more directly with
Beowulf or some other substantive; this formation was extended
to neuters (not yet in the oldest English epic) and finally became
the regular way of making an adjective definite before its substantive.
The number of words that require the weak form of
an adjective has been constantly growing, especially in German.
By this gradual development, which has made these forms just
as much real adjectives as the old “strong” forms, the old individualizing
force has been lost, and the words have become more
general in their meaning than they were once, though it may be said
still that (der) gute (mann) is more special than (ein) guter (mann).

Bally (Traité de stylistique française, 305) calls attention to
another effect of substantivizing an adjective: “Vous êtes un
impertinent” est plus familier et plus énergique que “Vous êtes
impertinent.” Here the substantivizing is effected simply by
adding the indefinite article. The same effect is observed in other
languages, compare “He is a bore” with “He is tedious”; “Er
ist ein prahlhans” with “Er ist prahlerisch,” etc. It is the same
with terms of endearment: “You are a dear” is more affectionate
than “You are dear,” which is hardly ever said. The explanation
is obvious: these substantives are more vigorous because they
are more special than the adjectives, though seemingly embodying
the same idea.

It is a simple corollary of our definition that the most special
of substantives, proper names, cannot be turned into adjectives
(or adjuncts, see below) without really losing their character of
proper names and becoming more general. We see this in such
a combination as the Gladstone ministry, which means the ministry
headed by Gladstone, and stands in the same relation to the real
proper name Gladstone as Roman to Rome or English to England.
The general signification is seen even more clearly in such examples
as Brussels sprouts (which may be grown anywhere) or a Japan
table (which means a table lacquered in the way invented in Japan). 12877

Interchange of the Two Classes.

Let us now turn to those cases in which an adjectival and a
substantival element of the same group can more or less naturally
be made to exchange places. Couturat, who is on the whole
inclined to make light of the difference between the two classes
of words, possibly on account of the slight formal difference found
in his own mother-tongue, adduces such examples as: “un sage
sceptique est un sceptique sage, un philosophe grec est un Grec philosophe,”
and says that the difference is only a nuance, according
as one of the qualities is looked upon as more essential or simply
as more important or interesting under the circumstances: for it
is evident that one is a Greek before being a philosopher, “et
néanmoins nous parlons plutôt des philosophes grecs que des Grecs
philosophes” (Revue de Métaphysique et de Morale, 1912, 9).

Now it may be difficult to say which of these two ideas is the
more important or interesting, but if we apply the above-mentioned
criterion we shall easily see why in choosing between the
two ways of designating the Greeks who are philosophers (= the
philosophers who are Greeks), we naturally make philosopher
(the more special idea) the substantive and Greek (the more general
one) the adjective and say the Greek philosophers (les philosophes
grecs) rather than les Grecs philosophes (in English the conversion
is not so complete and the philosophical Greeks does not exactly
cover the French expression). A famous German book is called
“Griechische denker.” “Denkende griechen” would be a much
weaker title, because the adjective denkend is much more vague
in its application than the substantive denker, which at once singles
out those who think more deeply and more professionally than
ordinary “thinking” people.

Another example: Mr. Galsworthy somewhere writes: “Having
been a Conservative Liberal in politics till well past sixty, it was
not until Disraeli's time that he became a Liberal Conservative.”
The words conservative and liberal are made into substantives (and
then take -s in the plural) when they mean members of two political
parties; evidently this is a more special idea than that which is
attached to the same words as general adjectives. 129

If we compare the two expressions a poor Russian and a Russian
pauper, we see first that the substantive Russian is more special
than the corresponding adjective in that it implies the idea
‘man or woman,’ and that on the other hand pauper is more
special than poor, which may be applied to many things besides
human beings: pauper is even more specialized than ‘a poor
78person’ as implying one that is entitled to or receives public
charity. 130

Other Combinations.

The rule of the greater complexity and specialization of substantives
thus holds good wherever we are able directly to compare
two words of closely similar signification; but can it be applied
to other cases — can we say that in any collocation of an adjective
and a substantive the former is always less special than the latter?
In a great many cases we can undoubtedly apply the criterion,
even in its most arithmetical form, by counting how many individuals
each word may be applied to. Napoleon the third: there
are only few Napoleons, but many persons and things that are
third in a series. A new book: there are more new things than
books in existence. An Icelandic peasant: it is true that there
are more peasants in the world than Icelanders, but then the
adjective Icelandic can be applied to a great many things as well
as to persons: Icelandic mountains and waterfalls and sheep and
horses and sweaters, etc., etc. Some of my critics objected to my
example a poor widow, saying that if we substitute rich it was
unfortunately very doubtful whether there were more rich persons
in existence than widows — thus overlooking the fact that rich
may be said of towns, villages, countries, mines, spoils, stores,
rewards, attire, experience, sculpture, repast, cakes, cream, rimes
and so forth. The Atlantic Ocean: the adjective is found, for
instance in Shelley's poems, with the substantives clouds, waves,
and islets. The adjective rare, though meaning ‘not often met
with’ may be used in speaking of innumerable objects, men, stones,
trees, stamps, mental qualities, etc., and thus falls within the
definition. But it must, of course, be conceded that the numerical
test cannot always be applied, as adjectives and substantives
which may be put together are very often by the nature of the
case incommensurable: we speak of a grey stone, but who shall
say whether the word grey or stone is applicable to the greater
number of objects. But applicability to a greater or lesser number
is only one side of what is implied in the words special and general,
and I am inclined to lay more stress on the greater complexity of
qualities denoted by substantives, as against the singling out of
one quality in the case of an adjective. This complexity is so
essential that only in rare cases will it be possible by heaping
79adjective upon adjective to arrive at a complete definition of the
notion evoked by the naming of a substantive: there will always,
as Bertelsen remarks, remain an indefinable x, a kernel which
may be thought of as “bearer” of the qualities which we may
have specified. This again is what underlies the old definition
by means of “substance,” which is thus seen to contain one
element of truth though not the whole truth. If one wants a
metaphorical figure, substantives may be compared to crystallizations
of qualities which in adjectives are found only in the
liquid state.

It must also be mentioned here that our languages contain
a certain number of substantives of a highly general signification,
thing, body, being. But their “general” signification is not of the
same order as that of adjectives: they very often serve as comprehensive
terms for a number of undoubtedly substantival ideas
(all these things, said instead of enumerating books, paper, garments,
etc.) — this use is very frequent in philosophic and abstract
scientific thinking. In everyday speech they may be loosely used
instead of a special substantive which is either not found in the
language or else is momentarily forgotten (cp. such words as
thingummybob, G. dingsda). Otherwise they rarely occur except
in combination with an adjective, and then they are often little
more than a kind of grammatical device for substantivizing the
adjective like the E. one. (Ones, in the new ones, is a substitute
for the substantive mentioned a few moments before; in her young
ones, said of a bird, it supplies the want of a substantive corresponding
to children). This leads to their use in compound
pronouns: something, nothing, quelquechose, ingenting, somebody, etc.
On the other hand, when once a language has a certain way of
forming adjectives, it may extend the type to highly specialized
adjectives, e.g. in a pink-eyed cat, a ten-roomed house, which combinations
have been advanced against my whole theory: there
are more cats than pink-eyed beings, etc. This, however, does
not seem to me to invalidate the general truth of the theory as
here explained: it must be remembered also that the real adjectival
part of such combinations is pink or ten, respectively.

It will be easily understood from what has been said above
that the so-called degrees of comparison (greater, greatest) are as
a rule found only with adjectives: such comparisons necessarily
deal with one quality at a time. The more special an idea is, the
less use will there be for degrees of comparison. And where we
do find in actual usage comparatives or superlatives of substantive
forms they will be seen on closer inspection to single out one
quality and thus to mean the same thing as if they were formed
from real adjectives. Thus Gr. basileuteros, basileutatos ‘more
80(most) of a king, kinglier, kingliest’ (other examples Delbrück,
Synt. 1. 415), Magyar szamár ‘ass,’ szamarabb ‘sillier,’ róka ‘fox,’
rókább ‘slyer.’ Finnish ranta ‘strand,’ rannempi ‘nearer to the
strand,’ syksy ‘autumn,’ syksymänä ‘later in the autumn.’ Cf.
also Paul P § 250.

One final remark: we cannot make the complexity of qualities
or specialization of signification a criterion by which to decide
whether a certain word is a substantive or an adjective: that
must be settled in each case by formal criteria varying from
language to language. What has been attempted in this chapter
is to find whether or no there is anything in the nature of things
or of our thinking that justifies the classification found in so many
languages by which substantives are kept distinct from adjectives.
We cannot, of course, expect to find any sharp or rigid line of
demarcation separating the two classes in the way beloved by
logicians: language-makers, that is ordinary speakers, are not
very accurate thinkers. But neither are they devoid of a certain
natural logic, and however blurred the outlines may sometimes
be, the main general classifications expressed by grammatical
forms will always be found to have some logical foundation. It
is so in the case before us: substantives are broadly distinguished
as having a more special signification, and adjectives as having
a more general signification, because the former connote the possession
of a complexity of qualities, and the latter the possession
of one single quality. 13181

Chapter VI
Parts of Speech — concluded

Pronouns. Verbs. Particles. Summary. Word.

Pronouns.

Pronouns are everywhere recognized as one of the word-classes,
but what constitutes their distinctive peculiarity? The old
definition is embodied in the term itself: pronouns stand instead
of the name of a person or thing. This is expanded by Sweet
(NEG § 196): a pronoun is a substitute for a noun and is used
partly for the sake of brevity, partly to avoid the repetition of a
noun, and partly to avoid the necessity of definite statement.
But this does not suit all cases, and the definition breaks down
in the very first pronoun; it is very unnatural to the unsophisticated
mind to say that “I see you” stands instead of “Otto
Jespersen sees Mary Brown,” on the contrary most people will
say that in Bellum Gallicum the writer uses the word Cæsar instead
of “I.” We may also say “I, Otto Jespersen, hereby declare…,”
which would be preposterous if “I” were simply a substitute for
the name. And grammatically it is very important that “I” is
the first person, and the name is in the third, as shown in many
languages by the form of the verb. Further: no one doubts that
nobody and the interrogative who are pronouns, but it is not easy
to see what nouns they can be said to be substitutes for.

In this way we should get a class of substitute words which
might be subdivided into pro-nouns, pro-adjectives, pro-adverbs,
pro-infinitives, pro-verbs (and pro-sentences as so in the second
instance above), but it could hardly be called a real grammatical
class.

Noreen's treatment of pronouns (VS 5. 63 ff.) is very original
and instructive. He contrasts pronouns with “expressive
sememes” the signification of which is fixed in so far as it is essentially
contained in the linguistic expression itself; pronouns then
are characterized by their signification being variable and essentially
contained in a reference to some circumstance which is found
outside of the linguistic expression itself and is determined by
the whole of the situation. “I” is a pronoun because it signifies
one person when John Brown, and another when Mary Smith
speaks. The consequence is that a great many words and groups
of words are pronouns, according to Noreen, for instance the undersigned;
today; (there were three boys), the biggest one, etc. No
two words could be more pronominal than yes and no (but what
about On the contrary as a reply instead of no?); here is the pronominal
adverb of place of the first person, and there the corresponding
adverb for the second and third persons, and now and
then are the corresponding pronominal adverbs of time (but the
combinations here and there, now and then, meaning ‘in various
not defined places’ and ‘occasionally’ cannot be pronouns according
to Noreen's definition). Further right, left, on Sunday, the
horse (not only the, but both words together), my horse, are all
of them pronouns. Noreen is at some pains (not very successfully)
to prove that such a common “proper name” as John is
not a pronoun though its proper signification wherever it occurs
is determined by the whole situation. And what about father as
used by the child for ‘my father’?

Noreen's class is too comprehensive and too heterogeneous,
and yet it is not easy to see how words like the interrogative who
and what or like some, nothing can fall within the definition. But
the main defect in his treatment of this and of other points to my
mind is due to his building up categories entirely from the “semological”
or what I should call the notional point of view without
regard to the way in which the meaning is expressed in actual
language, that is, without any consideration of formal elements.
If we keep both sides in view we shall find that there is really
some sense in comprising a certain number of shifters (to use the
term I employ in Language, p. 123), reminders (ib. 353), representative
and relational words under one class with the old-established
name of pronouns. It may not be easy to say what is
common to all of them from the notional point of view, but if we
83take each of the traditional sub-classes by itself its notional unity
is manifest: personal pronouns with the corresponding possessives
— demonstrative pronouns — relative pronouns — interrogative pronouns —
indefinite pronouns, though with regard to this last class
the boundaries between a few of them, such as some, and such
adjectives as many, are rather vague; consequently grammarians
disagree as to what words they should include in this sub-class.
This, however, is not essentially different from what we find in any
other grammatical classification: there will always be some borderline
cases. And when we investigate the forms and functions of
these pronouns in various languages we discover that they present
certain features by which they are distinguished from other words.
But these features are not the same in all languages, nor are they
exactly the same with all the pronouns found within the same
one language. Formal and functional anomalies abound in pronouns.
In English we have the distinction between two cases as
in he: him, they: them, and between an adjunct and a non-adjunct
form in my: mine, the sex-distinction in he: she and the similar
distinction who: what, the irregular plural in he, she: they, that:
those, combinations of the type of somebody, something, which are
not found with ordinary adjectives, the use of each without any
accompanying substantive or article, etc. 132 Similar peculiarities
are found in the pronouns of other languages; in French we have,
for instance, the special forms je, me, tu, te, etc., which are only
found in close conjunction with verbal forms.

The term pronoun is sometimes restricted (generally in French
books, but also in the Report of the Joint Committee on Terminology)
to those words which function as what in Ch. VII I shall
call “primary words,” while my is called a “possessive adjective”
and this in this book a “demonstrative adjective.” There is, however,
not the slightest reason for thus tearing asunder my and
mine, or, even worse, his in “his cap was new” and “his was
a new cap” or this in “this book is old” and “this is an old
book” 233 and assigning the same form to two different “parts of
speech,” especially as it then becomes necessary to establish the
same sub-classes of adjectives (possessive, demonstrative) as are
found in pronouns. I should even go so far as to include among
pronouns the so-called pronominal adverbs then, there, thence,
84when, where, whence, etc., which share some of the peculiarities of
pronouns and are evidently formed from them (note also such
formations as whenever, cf. whoever, and somewhere, etc.).

Numerals are often given as a separate part of speech; it
would probably be better to treat them as a separate sub-class
under the pronouns, with which they have some points in common.
One besides being a numeral is, in English as well as in some other
languages, an indefinite pronoun (“one never knows”), cf. also
the combination oneself. Its weak form is the so-called “indefinite
article,” and if its counterpart the “definite article” is justly
reckoned among pronouns, the same should be the case with a, an,
Ft. un, etc. To establish a separate “part of speech” for the
two “articles,” as is done in some grammars, is irrational. E. other
was originally an ordinal meaning ‘second’ as anden still does
in Danish; now it is generally classed among pronouns, and this
is justified by its use in each other, one another. Most numerals
are indeclinable, but in languages where some of them are declined,
these often present anomalies comparable to those found in other
pronouns. If we include numerals among pronouns, we might
include also the indefinite numerals many, few: logically these
stand in the same series as all, some and the negative none, no,
which are always reckoned among indefinite pronouns. But then
we must also include much, little as in much harm, little gold (with
mass-words, cf. Ch. XIV). 134 All these quantifiers, as they might
be called, differ from ordinary qualifying adjectives in being
capable of standing alone (without articles) as “primaries” as
when we say “some (many, all, both, two) were absent,” “all
(much, little) is true”; they are always placed before qualifiers
and cannot be transcribed in the form of a predicative: “a nice
young lady” is the same as “a lady who is nice and young,” but
such a transposition is impossible with “many ladies,” “much
wine,” etc., just as it is impossible with “no ladies,” “what
ladies,” “that wine,” and other pronouns.

A final word may be added about the names of some of the
sub-classes. Relative pronouns: in these days when everything
has been shown to be relative, it would perhaps be possible to
introduce a more pertinent name, e.g. connective or conjunctive
pronouns, as their business is to join sentences in pretty much
the same way as conjunctions do: indeed it may be questioned
whether E. that is not the conjunction rather than a pronoun;
compare the possibility of omitting that: “I know the man (that)
you mentioned” and “I know (that) you mentioned the man,”
and the impossibility of having a preposition before that: “the
man that you spoke about” as against “the man about whom
85you spoke.” — Personal pronouns: if this refers to person in the
sense of ‘human being’ it is improper in cases like G. er, Fr. elle
or E. it applied to a table (der tisch, la table), and even more to
the “impersonal” it, es, il in it rains, es regnet, il pleut. If on
the other hand the name personal is taken to refer to the three
grammatical persons (see Ch. XVI), it may be justly said that only
the two first persons strictly belong here, for all the other pronouns
(this, who, nothing, etc.) are of the third person just as much
as he or she. But it will be difficult to find a better name to substitute
for “personal” pronouns, and the question is not very
important. The delimitation of personal and demonstrative pronouns
sometimes offers difficulties; thus in Dan., where de, dem
formally go with the demonstrative den, det, but functionally are
the plural both of den, det and of han, hun ‘he, she.’

Verbs.

Verbs in most languages, at any rate those of the Aryan,
Semitic, and Ugro-finnic types, have so many distinctive features
that it is quite necessary to recognize them as a separate class of
words, even if here and there one or more of those distinguishing
traits that are generally given as characteristic of verbs may be
found wanting. Such traits are the distinctions of persons (first,
second, third), of tense, of mood, and of voice (cf. above, p. 68).
As for their meaning, verbs are what Sweet calls phenomenon
words and may be broadly divided into those that denote action
(he eats, breathes, kills, speaks, etc.), those that denote some process
(he becomes, grows, loses, dies, etc.), and those that denote some
state or condition (he sleeps, remains, waits, lives, suffers, etc.),
though there are some verbs which it is difficult to include in any
one of these classes (he resists, scorns, pleases). It is nearly always
easy to see whether a given idea is verbal or no, and if we combine
a verb with a pronoun as in the examples given (or with a
noun: the man eats, etc.) we discover that the verb imparts to
the combination a special character of finish and makes it a (more
or less) complete piece of communication — a character which is
wanting if we combine a noun or pronoun with an adjective or
adverb. The verb is a life-giving element, which makes it particularly
valuable in building up sentences: a sentence nearly
always contains a verb, and only exceptionally do we find
combinations without a verb which might be called complete
sentences. Some grammarians even go so far as to require the
presence of a verb in order to call a given piece of communication
a sentence. We shall discuss this question in a later
chapter.86

If now we compare the two combinations the dog barks and
the barking dog, we see that though barks and barking are evidently
closely related and may be called different forms of the same word,
it is only the former combination which is rounded off as a complete
piece of communication, while the barking dog lacks that
peculiar finish and makes us ask: What about that dog? The
sentence-building power is found in all those forms which are often
called “finite” verb forms, but not in such forms as barking or
eaten (participles), nor in infinitives like to bark, to eat. Participles
are really a kind of adjectives formed from verbs, and infinitives
have something in common with substantives, though syntactically
both participles and infinitives retain many of the characteristics
of a verb. From one point of view, therefore, we should
be justified in restricting the name verb to those forms (the finite
forms) that have the eminently verbal power of forming sentences,
and in treating the “verbids” (participles and infinitives) as a
separate class intermediate between nouns and verbs (cf. the old
name participium, i.e. what participates in the character of noun
and verb). Still it must be admitted that it would be somewhat
unnatural to dissociate eat and eaten in such sentences as he is
eating the apple, he will eat the apple, he has eaten the apple from
he eats the apple, he ate the apple; 135 and it is, therefore, preferable
to recognize non-finite forms of verbs by the side of finite forms,
as is done in most grammars.

Particles.

In nearly all grammars adverbs, prepositions, conjunctions,
and interjections are treated as four distinct “parts of speech,”
the difference between them being thus put on a par with that
between substantives, adjectives, pronouns, and verbs. But in
this way the dissimilarities between these words are grossly exaggerated,
and their evident similarities correspondingly obscured,
and I therefore propose to revert to the old terminology by which
these four classes are treated as one called “particles.”

As regards form they are all invariable — apart from the power
that some adverbs possess of forming comparatives and superlatives
in the same way as the adjectives to which they are related.
But in order to estimate the differences in meaning or function
that have led most grammarians to consider them as different
parts of speech, it will be necessary to cast a glance at some words
outside these classes.

Many words are subject to a distinction which is designated
87by different names and therefore not perceived as essentially the
same wherever found, namely that between a word complete in
itself (or used for the moment as such) and one completed by some
addition, generally of a restrictive nature. Thus we have the
complete verb in he sings, he plays, he begins; and the same verb
followed by a complement in he sings a song, he plays the piano,
he begins work. In this case it is usual to call the verb intransitive
in one case and transitive in the other, while the complement is
termed its object. In other verbs where these names are not
generally used, the distinction is really the same: he can is complete;
in he can sing the verb can is completed by the addition
of an infinitive. For this latter distinction we have no settled
term, and the terms used by some, independent and auxiliary
verb, are not quite adequate; for while on the one hand we have
an antiquated use of can with a different kind of complement in
“He could the Bible in the holy tongue,” we have on the other
hand such combinations as “He is able,” “he is able to sing,”
and “he wants to sing.” A further case in point is seen in he
grows, where the verb is complete, and he grows bigger, where it
is complemented by a “predicative”; cp. Troy was and Troy
was a town. Yet in spite of these differences in verbs no one thinks
of assigning them to different parts of speech: sing, play, begin,
can, grow, be are always verbs, whether in a particular combination
they are complete or incomplete.

If now we turn to such words as on or in, we find what is to
my mind an exact parallel to the instances just mentioned in
their employment in combinations like “put your cap on” and
“put your cap on your head,” “he was in” and “he was in the
house”; yet on and in in the former sentences are termed adverbs,
and in the latter prepositions, and these are reckoned as two
different parts of speech. Would it not be more natural to include
them in one class and to say that on and in are sometimes complete
in themselves and sometimes followed by a complement (or object)?
Take other examples: “he climbs up” and “he climbs up a
tree,” “he falls down” and “he falls down the steps” (cf. “he
ascends, or descends” with or without the complement “the
steps” expressed); “he had been there before” and “he had
been there before breakfast.” 136 Is near in “it was near one
o'clock” a preposition or an adverb according to the usual system?
(Cf. the two synonyms almost and about, the former called an adverb,
the latter a preposition.) The close correspondence between the
object of a transitive verb and that of a “preposition” is seen in
those cases in which a preposition is nothing but a verbal form
in a special use, as for example concerning (G. betreffend) and past
88in “he walked past the door at half-past one,” which is simply
the participle passed written in a different way; in “he walked
past” it has no complement.

We may even find the same word used in two ways in the same
sentence, thus “After the Baden business, and he had [= after
he had] dragged off his wife to Champagne, the Duke became
greatly broken” (Thackeray); if this is rare it must be remembered
that it is similarly rare to find one and the same verb in the
same sentence construed first transitively and then intransitively,
or first with a substantive and then with a clause as object.

The examples given above show the same word used now as
a preposition and now as a conjunction, in other cases we have
slight differences as in “because of his absence” and “because he
was absent,” which is historically explained by the origin of because
from by cause (people once said “because that he was absent”).
In other cases, again, a particular word has only one use, either
with an ordinary object or with a clause as its complement:
“during his absence,” “while he was absent.” But this should
not make us hesitate to affirm the essential identity of prepositions
and conjunctions, just as we put all verbs in one class in spite of
the fact that they cannot all take a complementary clause.89

The definition of a conjunction as a sentence-preposition does
not apply to some words which are always reckoned among conjunctions,
such as and in “he and I are great friends,” “she sang
and danced,” and or in “was it blue or green?” etc. The same
words may be used to connect sentences, as in “she sang, and he
danced,” “he is mad, or I am much mistaken.” In both cases
they are coordinating connectives, while prepositions and the
conjunctions hitherto considered are subordinating connectives,
but though this is an important distinction there is no reason on
that account to separate them into two word-classes. And and
with mean nearly the same thing, the chief difference between
them being that the former coordinates and the latter subordinates;
this has some grammatical consequences — notice for example the
form of the verb in “he and his wife are coming” as against “he
with his wife is coming” (“he is coming with his wife”) and the
possessive pronoun in Danish: “han og hans kone kommer,” but
“han kommer med sin kone.” But the slightness of the notional
difference makes people apt to infringe the strict rule, as in Shakespeare's
“Don Alphonso, With other gentlemen of good esteeme
Are journying” (see MEG II, 6.53 ff.). 137 Both, either and neither
are so far peculiar in that they ‘anticipate’ an and, or, nor,
following, but they need not, of course, be considered as a
class apart.

As the last “part of speech” the usual lists give interjections,
under which name are comprised both words which are never
used otherwise (some containing sounds not found in ordinary
words, e.g. an inhaled f produced by sudden pain, or the suction-stop
inadequately written tut, and others formed by means of
ordinary sounds, e.g. hullo, oh), and on the other hand words from
the ordinary language, e.g. Well! Why! Fiddlesticks! Nonsense!
Come! and the Elizabethan Go to! The only thing that these
elements have in common is their ability to stand alone as a complete
“utterance,” otherwise they may be assigned to various
word-classes. They should not therefore be isolated from their
ordinary uses. Those interjections which cannot be used except
as interjections may most conveniently be classed with other
‘particles.’90

Summary.

I have finished my survey of the various word-classes or parts
of speech. It will be seen that while making many criticisms,
especially of the definitions often given, I have still been able to
retain much of the traditional scheme. I cannot go so far as, for
instance, E. Sapir, who says (L 125) that “no logical scheme of
the parts of speech — their number, nature, and necessary confines
— is of the slightest interest to the linguist” because “each language
has its own scheme. Everything depends on the formal
demarcations which it recognizes.”

Here the word-classes are quite fixed and allow of no transposition:
substantives (boy, apple), adjectives (little, green), pronouns
(this, it), verbs (picked, ate), particles (up, and, immediately).

I therefore venture to maintain that the demarcation of these
five classes is consonant with reason, though we are unable to
define them so rigidly as to be left with no doubtful or borderline
cases. Only we must beware of imagining that these classes are
absolutely notional: they are grammatical classes and as such
will vary to some extent — but only to some extent — from language
to language. They may not fit such languages as Eskimo and
Chinese (two extremes) in the same way as they fit Latin or English,
but in these and the other languages which form the chief subject
of this book the old terms substantive, adjective, etc., are indispensable:
they will therefore be retained in the senses and with
the provisos indicated in these chapters.

Word.

What is a word? and what is one word (not two or more)!
These are very difficult problems, which cannot be left untouched
in this volume. 138

Words are linguistic units, but they are not phonetic units:
no merely phonetic analysis of a string of spoken sounds can reveal
to us the number of words it is made up of, or the division between
word and word. This has long been recognized by phoneticians
and is indisputable: a maze sounds exactly like amaze, in sight
like incite, a sister like assist her, Fr. a semblé like assemblé, il
l'emporte like il en porte, etc. Nor can the spelling be decisive,
because spelling is often perfectly arbitrary and dependent on
fashion or, in some countries, on ministerial decrees not always
well advised. Does at any rate change its character, if written,
as it now is occasionally, at anyrate? Or any one, some one if
written anyone, someone? (No one is parallel, but the spelling
noone could never become popular, because it would be read as
noon.) There is hardly sufficient reason for German official spellings
92like miteinander, infolgedessen, zurzeit, etc. In his first books
Barrie wrote the Scottish phrase I suppaud, probably because he
thought it a verb like suppose, but later he was told its origin and
now, if I am not mistaken, writes I'se uphauld (= I shall uphold).
All this shows the difficulty of deciding whether certain combinations
are to be considered two un-amalgamated words or one
amalgamated word.

On the other hand, words are not notional units, for, as Noreen
remarks, the word triangle and the combination three-sided rectilinear
figure have exactly the same meaning, just as “Armitage”
and “the old doctor in the grey suit whom we met on the bridge”
may designate the same man. As, consequently, neither sound
nor meaning in itself shows us what is one word and what is more
than one word, we must look out for grammatical (syntactic)
criteria to decide the question.

In the following cases purely linguistic criteria show that what
was originally two words has become one. G. grossmacht and
Dan. stormagt differ from E. great power as shown by their flexion:
die europäischen grossmächte, de europæiske stormagter, but in
English with a different word-order we say the great European
Powers. 139 The numerals 5 + 10 both in Lat. quindecim and E.
fifteen differ in sound from the uncompounded numerals; Lat.
duodecim also in not having a dative form duobusdecim, etc. Fr.
quinze, douze must, of course, be considered units, even in a higher
degree, because they have lost all similarity with cinq, deux and
dix. Dan. een og tyve ‘one and twenty’ is one word in spite of
the spelling, because the same form is used before a neuter: een
og tyve år (but et år). E. breakfast, vouchsafe were two words until
people began saying he breakfasted, he vouchsafes instead of the
earlier he broke fast, he vouches safe; cp. p. 24. Each other might
claim to be spelt as one word, because it takes a preposition before
the whole combination (with each other) instead of the old construction
each with other. In French je m'en fuis has become je m'enfuis,
and is now rightly so written because the perfect is je me suis enfui;
but the parallel expression je m'en vais is always written separately:
it is true that colloquially je me suis en-allé is often said instead
of the orthodox je m'en suis allé, but the amalgamation cannot
be complete as with enfuis, because the use of different stems
(vais, allé, irai) prevents the fusion into one form. Fr. république,
E. republic, are units, which Lat. res publica cannot be on account
of its flexion: rem publicam. The absence of inner flexion in
G. jedermann, jedermanns, die mitternacht (jeder is originally nom.,
93mitter dat.) shows completed unification, as does also the flexion
in Lat. ipsum instead of eumpse (ipse from is-pse).

In all these cases a complete amalgamation of what was at
first two words must be recognized, because we have unmistakable
linguistic criteria by which to show that native instinct really
treats the combination as a unity; but this is not the case in
E. he loves, which has sometimes been thought to be as much a
unit as Lat. amat (ama-t): in English we can separate the
elements (he never loves) and isolate each of them, while in amat
this is impossible; similarly, Fr. il a aimé is not a unit in the same
way as Lat. amavit, because we can say il n'a pas aimé, a-t-il aimé,
etc. (see my criticism of various scholars, Language, p. 422 ff.).

Sometimes we have the opposite movement, from word-units
to looser combinations. The cohesion between the two elements
of English compound substantives is looser than it was formerly
(and than it is in German and Danish). While G. steinmauer and
Dan. stenmur are in every respect one word, E. stone wall and
similar combinations are now rather to be considered two, stone
being an adjunct and wall a primary. This is shown not only by
the equal (or varying) stress, but also in other ways: by coordination
with adjectives: his personal and party interests | among the
evening and weekly papers | a Yorkshire young lady; by the use
of one: five gold watches, and seven silver ones; by the use of
adverbs: a purely family gathering; by isolation: any position,
whether State or national | things that are dead, second-hand, and
pointless. Some of these first elements have in this way become
so completely adjectival, that they can take the superlative ending
-est (chiefest, choicest), and adverbs can be formed from them
(chiefly, choicely), see MEG II, Ch. XIII (above, 62 note). In
Shakespeare's “so new a fashioned robe” we see how another type
of compound (new-fashioned) is also felt as loosely coherent.

All these considerations, as well as the changes of initial sounds
frequent, for instance, in Keltic languages, and such phenomena
as ON “hann kvaðsk eigi vita” (he said-himself not know, i.e.
he said that he did not know) and many others 140 show how difficult
it is in many cases to say what is one and what is two words.
Isolability in many cases assists us, but it should not be forgotten
that there are words, which we must recognize as such, and which
yet for one reason or another cannot be isolated; thus the Russian
prepositions consisting of a consonant alone, s, v, or French words
like je, tu, le, which never occur alone, although there is, indeed,
no purely phonetic reason against their being isolated. If these
are words, it is because they can be placed in various positions
94with other words, which are undoubtedly complete words; consequently
je, tu, etc., are not themselves parts of words, but whole
words. In the same way an, bei, statt in G. “ich nehme es an,
wir wohnten der versammlung bei, es findet nur selten statt”
are words, and a consistent orthography would have to write
“an zu nehmen, bei zu wohnen, es hat statt gefunden” instead
of the usual forms in one word: the position of the words is the
same as in “gern zu nehmen, dort zu wohnen, er hat etwas gefunden,”
etc. 141

We should never forget that words are nearly always used in
connected speech, where they are more or less closely linked with
other words: these are generally helpful, and often quite indispensable,
to show the particular meaning in which the given word
is to be understood. Isolated words, as we find them in dictionaries
and philological treatises, are abstractions, which in that
form have little to do with real living speech. It is true that in
answers and retorts words occur isolated, even words which cannot
otherwise stand by themselves, e.g. if: “If I were rich enough…”
“Yes, if!” — but then the meaning is understood from what precedes,
exactly as “Yesterday” when said as an answer to the
question “When did she arrive?” means “She arrived yesterday.”
But such isolation must always be considered an exception, not
the rule.

A term is wanted for a combination of words which together
form a sense unit, though they need not always come in immediate
juxtaposition and thus are shown to form not one word but two
or more words. This may be called a phrase, though that term
is used in a different way by other writers. The words puts off
form a phrase, the meaning of which (‘postpones’) cannot be
inferred from that of the words separately; the words may be
separated, e.g. he puts it off. G. wenn auch forms a phrase, e.g. in
wenn er auch reich ist.95

Chapter VII
The Three Ranks

Subordination. Substantives. Adjectives. Pronouns. Verbs. Adverbs.
Word groups. Clauses. Final remarks.

Subordination.

The question of the class into which a word should be put — whether
that of substantives or adjectives, or some other — is one that
concerns the word in itself. Some answer to that question will
therefore be found in dictionaries. 142 We have now to consider
combinations of words, and here we shall find that though a substantive
always remains a substantive and an adjective an adjective,
there is a certain scheme of subordination in connected speech
which is analogous to the distribution of words into ‘parts of
speech,’ without being entirely dependent on it.

In any composite denomination of a thing or person (such as
those to which I referred on p. 64), we always find that there is
one word of supreme importance to which the others are joined
as subordinates. This chief word is defined (qualified, modified)
by another word, which in its turn may be defined (qualified,
modified) by a third word, etc. We are thus led to establish different
“ranks” of words according to their mutual relations as defined
or defining. In the combination extremely hot weather the last
word weather, which is evidently the chief idea, may be called
primary; hot, which defines weather, secondary, and extremely,
which defines hot, tertiary. Though a tertiary word may be further
defined by a (quaternary) word, and this again by a (quinary)
word, and so forth, it is needless to distinguish more than three
ranks, as there are no formal or other traits that distinguish words
of these lower orders from tertiary words. Thus, in the phrase
a certainly not very cleverly worded remark, no one of the words
certainly, not, and very, though defining the following word, is in
any way grammatically different from what it would be as a
tertiary word, as it is in certainly a clever remark, not a clever
remark, a very clever remark.96

If now we compare the combination a furiously barking dog
(a dog barking furiously), in which dog is primary, barking secondary,
and furiously tertiary, with the dog barks furiously, it is evident
that the same subordination obtains in the latter as in the former
combination. Yet there is a fundamental difference between
them, which calls for separate terms for the two kinds of combination:
we shall call the former kind junction, and the latter nexus.
The difference has already been mentioned on p. 87, and there
will be occasion for a fuller discussion of it in Ch. VIII, where we
shall see that there are other types of nexus besides the one seen
in the dog barks. It should be noted that the dog is a primary not
only when it is the subject, as in the dog barks, but also when it is
the object of a verb, as in I see the dog, or of a preposition, as in
he runs after the dog.

As regards terminology, the words primary, secondary, and
tertiary are applicable to nexus as well as to junction, but it will
be useful to have the special names adjunct for a secondary word
in a junction, and adnex for a secondary word in a nexus. For
tertiary we may use the term subjunct, and quaternary words,
in the rare cases in which a special name is needed, may be termed
sub-subjuncts. 143

Just as we may have two (or more) coordinate primaries, e.g.
in the dog and the cat ran away, we may, of course, have two or more
coordinate adjuncts to the same primary: thus, in a nice young
lady the words a, nice, and young equally define lady; compare
also much (II) good (II) white (II) wine (I) with very (III) good (II)
wine (I). Coordinate adjuncts are often joined by means of
connectives, as in a rainy and stormy afternoon | a brilliant, though
lengthy novel. Where there is no connective the last adjunct
often stands in a specially close connexion with the primary as
forming one idea, one compound primary (young-lady), especially
in some fixed combinations (in high good humour, by great good
fortune, MEG II, 15. 15; extreme old age, ib. 12. 47). Sometimes
the first of two adjuncts tends to be subordinate to the second and
thus nearly becomes a subjunct, as in burning hot soup, a shocking
bad nurse. In this way very, which was an adjective (as it still is
in the very day) in Chaucer's a verray parfit gentil knight, has become
first an intermediate between an adjunct and a subjunct, and then
a subjunct which must be classed among adverbs; other examples
MEG II, 15. 2. A somewhat related instance is nice (and) in nice
and warm (15. 29), to which there is a curious parallel in It. bell'e:
Giacosa, Foglie 136 il concerto.… On ci ho bell'e rinunziato |
97ib: 117 Tu l'hai bell'e trovato. Other instances of adjuncts,
where subjuncts might be expected, are Fr. elle est toute surprise |
les fenêtres grandes ouvertes.

Coordinated subjuncts are seen, e.g. in a logically and grammatically
unjustifiable construction | a seldom or never seen form.

In the examples hitherto chosen we have had substantives
as primaries, adjectives as adjuncts, and adverbs as subjuncts;
and there is certainly some degree of correspondence between the
three parts of speech and the three ranks here established. We
might even define substantives as words standing habitually as
primaries, adjectives as words standing habitually as adjuncts, and
adverbs as words standing habitually as subjuncts. But the
correspondence is far from complete, as will be evident from the
following survey: the two things, word-classes and ranks, really
move in two different spheres.

Substantives.

Substantives as Primaries. No further examples are needed.

Substantives as Adjuncts. The old-established way of using
a substantive as an adjunct is by putting it in the genitive case,
e.g. Shelley's poems | the butcher's shop | St. Paul's Cathedral.
But it should be noted that a genitive case may also be a primary
(through what is often called ellipsis), as in “I prefer Keats's
poems to Shelley's | I bought it at the butcher's | St. Paul's is a
fine building.” In English what was the first element of a compound
is now often to be considered an independent word, standing as an
adjunct, thus in stone wall | a silk dress and a cotton one; on the
way in which these words tend to be treated as adjectives, see
p. 94, above. Other examples of substantives as adjuncts
are women writers | a queen bee | boy messengers, and (why not?)
Captain Smith | Doctor Johnson — cf. the non-inflexion in G. Kaiser
Wilhelms Erinnerungen (though with much fluctuation with compound
titles).

In some cases when we want to join two substantival ideas it
is found impossible or impracticable to make one of them into an
adjunct of the other by simple juxtaposition; here languages
often have recourse to the ‘definitive genitive’ or a corresponding
prepositional combination, as in Lat. urbs Romæ (cf. the juxtaposition
in Dan. byen Rom, and on the other hand combinations
like Captain Smith), Fr. la cité de Rome, E. the city of Rome, etc.,
and further the interesting expressions E. a devil of a fellow | that
scoundrel of a servant | his ghost of a voice | G. ein alter schelm von
lohnbedienter (with the exceptional use of the nominative after
von) | Dan. den skurk av en tjener | et vidunder av et barn | det fæ
98til Nielsen | Fr. ce fripon de valet | un amour d'enfant | celui qui
avail un si drôle de nom | It. quel ciarlatano d'un dottore | quel pover
uomo di tuo padre, etc. This is connected with the Scandinavian
use of a possessive pronoun dit fæ ‘you fool’ and to the Spanish
Pobrecitos de nosotros I | Desdichada de mi! Cf. on this and similar
phenomena Grimm, Personenwechsel, Schuchardt Br. 197, Tegner
G. 115 ff., Sandfeld in Dania VII.

Substantives as Subjuncts (subnexes). The use is rare, except
in word groups, where it is extremely frequent (see p. 102). Examples:
emotions, part religious… but part-human (Stevenson) |
the sea went mountains high. In “Come home | I bought it cheap”
home and cheap were originally substantives, but are now generally
called adverbs; cf. also go South.

Adjectives.

Adjectives as Primaries: you had better bow to the impossible
(sg.) | ye have the poor (pl.) always with you (MEG II, Ch. XI) —
but in savages, regulars, Christians, the moderns, etc., we have
real substantives, as shown by the plural ending; so also in
“the child is a dear,” as shown by the article (MEG Ch. IX).
G. beamter is generally reckoned a substantive, but is rather
an adjective primary, as seen from the flexion: der beamte, ein
beamter.

Adjectives as Adjuncts: no examples are here necessary.

Adjectives as Subjuncts. In “a fast moving engine | a long
delayed punishment | a clean shaven face” and similar instances
it is historically more correct to call the italicized words adverbs
(in which the old adverbial ending -e has become mute in the same
way as other weak -e's) rather than adjective subjuncts. On
new-laid eggs, cheerful tempered men, etc., see MEG II, 15. 3, on
burning hot, see p. 97, above.

Pronouns.

Pronouns as Primaries: I am well this is mine | who said
that? | what happened? | nobody knows, etc. (But in a mere,
nobody we have a real substantive, cf. the pl. nobodies.)

Pronouns as Adjuncts: this hat | my hat | what hat? | no
hat, etc.

In some cases there is no formal distinction between pronouns
in these two employments, but in others there is, cf. mine: my |
none: no; thus also in G. mein hut: der meine. Note also “Hier
ist ein umstand (ein ding) rich tig genannt, aber nur éiner (éines).”
99In Fr. we have formal differences in several cases: mon chapeau:
le mien | ce chapeau: celui-ci | quel chapeau: lequel? | chaque:
chacun | quelque: quelqu'un.

Pronouns as Subjuncts. Besides “pronominal adverbs,” which
need no exemplification, we have such instances as “I am that
sleepy (vg.) | the more, the merrier | none too able | I won't stay
any longer | nothing loth | somewhat paler than usual.” 144

Verbs.

Finite forms of verbs can only stand as secondary words
(adnexes), never either as primaries or as tertiaries. But participles,
like adjectives, can stand as primaries (the living are more
valuable than the dead) and as adjuncts (the living dog). Infinitives,
according to circumstances, may belong to each of the three ranks;
in some positions they require in English to (cf. G. zu, Dan. at). I
ought strictly to have entered such combinations as to go, etc.,
under the heading “rank of word groups.”

Infinitives as Primaries: to see is to believe (cf. seeing is believing) |
she wants to rest (cf. she wants some rest, with the corresponding
substantive). Fr. espérer, c'est jouir | il est défendu de fumer ici |
sans courir | au lieu de courir. G. denken ist schwer | er verspricht
zu kommen | ohne zu laufen | anstatt zu laufen, etc.

Infinitives as Adjuncts: in times to come | there isn't a girl to
touch her | the correct thing to do | in a way not to be forgotten
the never to be forgotten look (MEG II, 14. 4 and 15. 8). Fr. la
chose à faire | du tabac à fumer. (In G. a special passive participle
has developed from the corresponding use of the infinitive: das
zu lesende buch.) Spanish: todas las academias existentes y
por existir (Galdos). This use of the infinitive in some way
makes up for the want of a complete set of participles (future,
passive, etc.).

Infinitives as Subjuncts: to see him, one would think | I shudder
to think of it | he came here to see you.

Adverbs.

Adverbs as Primaries. This use is raid; as an instance may
be mentioned “he did not stay for long | he's only just back from
100abroad.” With pronominal adverbs it is more frequent: from
here | till now. Another instance is “he left there at two o'clock”:
there is taken as the object of left. Here and there may also be
real substantives in philosophical parlance: “Motion requires a
here and a there | in the Space-field lie innumerable other theres”
(NED, see MEG II, 8. 12).

Adverbs as Adjuncts. This, too, is somewhat rare: the off
side | in after years | the few nearby trees (US) | all the well passengers
(US) | a so-so matron (Byron). In most instances the
adjunct use of an adverb is unnecessary, as there is a corresponding
adjective available. (Pronominal adverbs: the then government |
the hither shore) MEG II, 14. 9.

Adverbs as Subjuncts. No examples needed, as this is the
ordinary employment of this word-class.

When a substantive is formed from an adjective or verb, a
defining word is, as it were, lifted up to a higher plane, becoming
secondary instead of tertiary, and wherever possible, this is shown
by the use of an adjective instead of an adverb form.

It is worth noting that adjectives indicating size (great, small)
are used as shifted equivalents of adverbs of degree (much, little):
a great admirer of Tennyson, Fr. un grand admirateur de Tennyson.
On these shifted subjunct-adjuncts, cf. MEG II, 12. 2, and on nexus-words,
p. 137, below. Curme (GG 136) mentions G. die geistig
armen, etwas längst bekanntes, where geistig and längst remain
uninflected like adverbs “though modifying a substantive”:
the explanation is that armen and bekanntes are not substantives,
but merely adjective primaries, as indicated by their flexion.
Some English words may be used in two ways: “these are full
equivalents (for)” or “fully equivalent (to),” “the direct opposites
(of)” or “directly opposite (to)”; Macaulay writes: “The government
of the Tudors was the direct opposite to the government of
Augustus” (E2. 99), where to seems to fit better with the adjective
opposite than with the substantive, while direct presupposes the
latter. In Dan. people hesitate between den indbildt syge and den
indbildte syge as a translation of le malade imaginaire.101

Word groups.

Word groups consisting of two or more words, the mutual
relation of which may be of the most different character, in many
instances occupy the same rank as a single word. In some cases
it is indeed difficult to decide whether we have one word or two
words, cf. p. 93 f. To-day was originally two words, now there is
a growing tendency to spell it without the hyphen today, and as a
matter of fact the possibility of saying from today shows that to is
no longer felt to have its original signification. Tomorrow, too, is
now one word, and it is even possible to say “I look forward
to tomorrow.” For our purpose in this chapter it is, however, of
no consequence at all whether we reckon these and other doubtful
cases as one word or two words, for we see that a word group
(just as much as a single word) may be either a primary or an
adjunct or a subjunct.

Word groups of various kinds as Primaries: Sunday afternoon
was fine | I spent Sunday afternoon at home | we met the kind
old Archbishop of York | it had taken him ever since to get used to
the idea | You have till ten to-night | From infancy to manhood
is rather a tedious period (Cowper). Cf. Fr. jusqu'au roi l'a cru;
nous avons assez pour jusqu'à samedi; Sp. hasta los malvados creen
en el (Galdos).

Word groups as Adjuncts: a Sunday afternoon concert | the
Archbishop of York | the party in power | the kind old Archbishop
of York's daughter | a Saturday to Monday excursion | the time
between two and four | his after dinner pipe.

Word groups as Subjuncts (tertiaries): he slept all Sunday
afternoon | he smokes after dinner | he went to all the principal
cities of Europe | he lives next door to Captain Strong | the canal ran
north and south | he used to laugh a good deal | five feet high he
wants things his own way | tilings shall go man-of-war fashion he
ran upstairs three steps at a time. Cf. the “absolute construction”
in the chapter on Nexus (IX).

As will have been seen already by these examples, the group,
whether primary, secondary, or tertiary, may itself contain elements
standing to one another in the relation of subordination indicated
by the three ranks. The rank of the group is one thing, the rank
within the group another. In this way more or less complicated
relations may come into existence, which, however, are always
easy to analyze from the point of view developed in this chapter.
Some illustrations will make this clear “We met the kind old
Archbishop of York”: the last six words together form one group
primary, the object of met; but the group itself consists of a
102primary Archbishop and four adjuncts, the, kind, old, of York, or,
we should rather say that Archbishop of York, consisting of the
primary Archbishop and the adjunct of York, is a group primary
qualified by the three adjuncts the, kind, and old. But the
adjunct of York in its turn consists of the particle (preposition) of
and its object, the primary York. Now, the whole of this group
may be turned into a group adjunct by being put in the genitive:
We met the kind old Archbishop of York's daughter.

He lives on this side the river: here the whole group consisting
of the last five words is tertiary to lives; on this side, which consists
of the particle (preposition) on with its object this (adjunct) side
(primary), forms itself a group preposition, which here takes as an
object the group the (adjunct) river (primary). But in the sentence
the buildings on this side the river are ancient, the same five-word
group is an adjunct to buildings. In this way we may arrive at
a natural and consistent analysis even of the most complicated
combinations found in actual language. 145

Clauses.

A special case of great importance is presented by those groups
that are generally called clauses. We may define a clause as a
member of a sentence which has in itself the form of a sentence
(as a rule it contains a finite verb). A clause then, according to
circumstances, may be either primary, secondary, or tertiary.

I. Clauses as Primaries (clause primaries).

In the first three sentences the clause is the subject, in the rest
it is the object, either of the verb or of the preposition of. But
there is a kind of pseudo-grammatical analysis against which I
must specially warn the reader: it says that in sentences like the
103second the subject of steals trash is a he which is said to be implied
in who, and to which the relative clause stands in the same relation
as it does to the man in the man who steals — one of the numerous
uncalled-for fictions which have vitiated and complicated grammar
without contributing to a real understanding of the facts of
language. 146

II. Clauses as Adjuncts (clause adjuncts).

It is worth remarking that often when we have seemingly two
relative clauses belonging to the same antecedent (i.e. primary)
the second really qualifies the antecedent as already qualified by
the first, thus is adjunct to a group primary consisting of a primary
and the first relative clause as adjunct. I print this group primary
in italics in the following examples: they murdered all they met
whom they thought gentlemen | there is no one who knows him
that does not like him | it is not the hen who cackles the most that
lays the largest eggs.

III. Clauses as Subjuncts or tertiaries (clause subjuncts).

Note here especially the first example, in which the clause
introduced by whoever is neither subject nor object as the clauses
considered above were, but stands in a looser relation to it is true.

The definition of the term “clause” necessitates some remarks
on the usual terminology, according to which the clauses here
mentioned would be termed ‘dependent’ or ‘subordinate’ clauses
as opposed to ‘the principal clause’ (or ‘principal proposition’);
corresponding terms are used in other languages, e.g. G. ‘nebensatz,
hauptsatz.’ But it is not at all necessary to have a special term
for what is usually called a principal clause. It should first be
remarked that the principal idea is not always expressed in the
‘principal clause,’ for instance not in “This was because he was
ill.” The idea which is expressed in the ‘principal clause’ in
“It is true that he is very learned,” may be rendered by a simple
adverb in “Certainly he is very learned” — does that change his
being learned from a subordinate to a principal idea? Compare
also the two expressions “I tell you that he is mad” and “He is
mad, as I tell you.” Further, if the ‘principal clause’ is defined
as what remains after the subordinate clauses have been peeled
off, we often obtain curious results. It must be admitted that in
some cases the subordinate clauses may be left out without any
material detriment to the meaning, which is to some extent complete
in itself, as in “I shall go to London (if I can)” or “(When he
got back) he dined with his brother.” But even here it does not
seem necessary to have a special term for what remains after the
whole combination has been stripped of those elements, any more
than if the same result had followed from the omission of
105some synonymous expressions of another form, e.g. “I shall go to
London (in that case)” or “(After his return) he dined with his
brother.” If we take away the clause where I was born from the
three sentences quoted above, what remains is (1) I do not know,
(2) This is the land, (3) It is a custom; but there is just as little
reason for treating these as a separate grammatical category as
if they had originated by the omission of the underlined parts of
the sentences (1) I do not know my birth-place, (2) This is my
native land, (3) It is a custom at home. Worse still, what is left
after deduction of the dependent clauses very often gives no meaning
at all, as in “(Who steals my purse) steals trash” and even more
absurdly in “(What surprises me) is (that he should get angry).”
Can it really be said here that the little word is contains the principal
idea? The grammatical unit is the whole sentence including
all that the speaker or writer has brought together to express his
thought; this should be taken as a whole, and then it will be seen
to be of little importance whether the subject or some other part
of it is in the form of a sentence and can thus be termed a clause
or whether it is a single word or a word group of some other form.

Final remarks.

The grammatical terminology here advocated, by which the distinction
of the three ranks is treated as different from the distinction between substantives,
adjectives, and adverbs, is in many ways preferable to the often
confused and self-contradictory terminology found in many grammatical
works. Corresponding to my three ranks we often find the words substantival,
adjectival, and adverbial, or a word is said to be “used adverbially,”
etc. (Thus NED, for instance, in speaking of a sight too clever.) Others
will frankly call what or several in one connexion substantives, in another
adjectives, though giving both under the heading pronouns (Wendt.) Falk
and Torp call Norw. sig the substantival reflexive pronoun, and sin the
adjectival reflexive pronoun, but the latter is substantival in “hver tog
sin, så tog jeg min.” Many scholars speak of the ‘adnominal genitive’
(= adjunct) as opposed to the ‘adverbial genitive,’ but the latter expression
is by some, though not by all, restricted to the use with verbs. In “The
King's English” the term ‘adverbials’ is used for subjunct groups and
clauses, but I do not think I have seen “adjectivals” or “substantivals”
used for the corresponding adjuncts and primaries. For my own ‘adjective
primary’ the following terms are in use: substantival adjective, substantivized
adjective, absolute adjective, adjective used absolutely (but “absolute”
is also used in totally different applications, e.g. in absolute ablative), quasi-substantive
(e.g. NED the great), a free adjective (Sweet NEG § 178 on G.
die gute), an adjective partially converted into a noun (ib. § 179 about E.
the good), a substantive-equivalent, a noun-equivalent. Onions (AS § 9)
uses the last expression; he applies the term ‘adjective-equivalent’ among
other things to “a noun in apposition,” e.g. ‘Simon Lee, the old huntsman’
and ‘a noun or verb-noun forming part of a compound noun,’ e.g. “cannon
balls.” In a lunatic asylum he says that lunatic is a noun (this is correct,
as shown by the pl. lunatics), but this noun is called ‘an adjective-equivalent’;
consequently he must say that in sick room the word sick is an adjective
which is a noun-equivalent (§ 9. 3), but this noun-equivalent at the same
time must be an adjective-equivalent according to his § 10 6 1 This is an
106example of the “simplified” uniform terminology used in Sonnenschein's
series. Cf. MEG II, 12. 41. London in the London papers is called an adjective-equivalent,
and the poor, when standing by itself, a noun-equivalent;
thus in the London poor the substantive must be an adjective-equivalent,
and the adjective a noun-equivalent. Some say that in the top one the substantive
is first adjectivized and then again substantivized, and both these
conversions are effected by the word one. Cf. MEG II, 10. 86: top in my
system always remains a substantive, but is here adjunct to the primary
one. My terminology is also much simpler than that found, for instance,
in Poutsma's Gr., where we find such expressions as ‘an attributive adnominal
adjunct consisting of a (pro)noun preceded by a preposition’ for my ‘prepositional
(group) adjunct’ (Poutsma using the word adjunct in a wider
sense than mine).

We are now in a position rightly to appreciate what Sweet
said in 1876 (CP 24): “It is a curious fact, hitherto overlooked
by grammarians and logicians, that the definition of the noun
applies strictly only to the nominative case. The oblique cases
are really attribute-words, and inflexion is practically nothing
but a device for turning a noun into an adjective or adverb. This
is perfectly clear as regards the genitive. … It is also clear that
noctem in flet noctem is a pure adverb of time.” Sweet did not,
however, in his own Anglo-Saxon Grammar place the genitive
of nouns under adjectives, and he was right in not doing so, for
what he says is only half true: the oblique cases are devices for
turning the substantive, which in the nominative is a primary,
into a secondary word (adjunct) or tertiary word, but it remains
a substantive all the same. There is a certain correspondence
between the tripartition substantive, adjective, adverb, and the
three ranks, and in course of time we often see adjunct forms of
substantives pass into real adjectives, and subjunct forms into
adverbs (prepositions, etc.), but the correspondence is only partial,
not complete. The ‘part of speech’ classification and the ‘rank’
classification represent different angles from which the same word
or form may be viewed, first as it is in itself, and then as it is in
combination with other words.107

Chapter VIII
Junction and Nexus

Adjuncts. Nexus.

Adjuncts.

It will be our task now to inquire into the function of adjuncts:
for what purpose or purposes are adjuncts added to primary words!

Various classes of adjuncts may here be distinguished.

The most important of these undoubtedly is the one composed
of what may be called restrictive or qualifying adjuncts: their
function is to restrict the primary, to limit the number of objects to
which it may be applied; in other words, to specialize or define it.
Thus red in a red rose restricts the applicability of the word rose
to one particular sub-class of the whole class of roses, it specializes
and defines the rose of which I am speaking by excluding white
and yellow roses; and so in most other instances: Napoleon the
third | a new book | Icelandic peasants | a poor widow, etc.

Now it may be remembered that these identical examples
were given above as illustrations of the thesis that substantives
are more special than adjectives, and it may be asked: is not
there a contradiction between what was said there and what has
just been asserted here? But on closer inspection it will be seen
that it is really most natural that a less special term is used in
order further to specialize what is already to some extent special:
the method of attaining a high degree of specialization is analogous
to that of reaching the roof of a building by means of ladders:
if one ladder will not do, you first take the tallest ladder you have
and tie the second tallest to the top of it, and if that is not enough,
you tie on the next in length, etc. In the same way, if widow is
not special enough, you add poor, which is less special than widow,
and yet, if it is added, enables you to reach farther in specialization;
if that does not suffice, you add the subjunct very, which
in itself is much more general than poor. Widow is special, poor
widow more special, and very poor widow still more special, but
very is less special than poor, and that again than widow.

Though proper names are highly specialized, yet it is possible
to specialize them still more by adjuncts Young Burns means
108either a different person from old Burns, or if there is only one
person of that name in the mind of the actual speaker (and hearer)
it mentions him with some emphasis laid on the fact that he is
still young (in which case it falls outside the restrictive adjuncts,
see below, p. 111).

Among restrictive adjuncts, some of a pronominal character
should be noticed. This and that, in this rose, that rose differ from
most other adjuncts in not being in any way descriptive: what
they do, whether accompanied by some pointing gesture or not,
is to specify. The same is true of the so-called definite article
the, which would be better called the defining or determining
article; this is the least special of adjuncts and yet specializes
more than most other words and just as much as this or that (of
which latter it is phonetically a weakened form). In the rose, rose
is restricted to that one definite rose which is at this very moment
in my thought and must be in yours, too, because we have just
mentioned it, or because everything in the situation points towards
that particular rose. Cf. “Shut the door, please.” While king in
itself may be applied to hundreds of individuals, the king is as
definite as a proper name: if we are in the middle of a story or a
conversation about some particular king, then it is he that is meant,
otherwise it means ‘our king,’ the present king of the country
in which we are living. But the situation may change, and then
the value of the definition contained in the article changes automatically.
“The King is dead. Long live the King!” (Le
roi est mort. Vive le roi!) In the first sentence mention is made
of one king, the king whom the audience thinks to be still king
here; in the second sentence the same two words necessarily
refer to another man, the legal successor of the former. It is
exactly the same with cases like “the Doctor said that the patient
was likely to die soon,” and again with those cases in which Sweet
(NEG § 2031) finds the “unique article”; the Devil [why does he
say that a devil has a different sense?], the sun, the moon, the earth,
etc. (similarly Deutschbein SNS 245). There is, really, no reason
for singling out a class of “persons or things which are unique in
themselves.”

This, however, is not the only function of the definite article.
In cases like the English King | the King of England | the eldest
boy | the boy who stole the apples, etc., the adjuncts here printed
in italics are in themselves quite sufficient to individualize, and
the article may be said so far to be logically superfluous though
required by usage, not only in English but in other languages.
We may perhaps call this the article of supplementary determination.
The relation between the King and the English King is
parallel to that between he, they, standing alone as sufficient to
109denote the person or persons pointed out by the situation (he can
afford it | they can afford it) and the same pronouns as determined
by an adjunct relative clause (he that is rich can afford it | they that
are rich can afford it). Cf. also the two uses of the same, first by
itself, meaning ‘the identical person or thing that has just been
mentioned,’ and second supplemented with a relative clause: the
same boy as (or, that) stole the apples. But, as remarked in NED,
the definite article with same often denotes an indeterminate
object, as in “all the planets travel round the sun in the same
direction,” in which sense French may employ the indefinite article
(deux mots qui signifient une même chose) and English often says
one and the same, where one may be said to neutralize the definite
article; so in other languages, Lat. unus et idem, Gr. (ho) heis kai
ho autos, G. ein und derselbe, Dan. een og samme. (N.B. without the
definite article. 147)

An adjunct consisting of a genitive or a possessive pronoun
always restricts, though not always to the same extent as the
definite article. My father and John's head are as definite and
individualized as possible, because a man Can only have one father
and one head; but what about my brother and John's hat? I
may have several brothers, and John may possess more than one
hat, and yet in most connexions these expressions will be understood
as perfectly definite: My brother arrived yesterday | Did
you see my brother this morning? | John's hat blew off his head — the
situation and context will show in each case which of my brothers
is meant, and in the last sentence the allusion, of course, is to the
particular hat which John was wearing on the occasion mentioned.
But when these expressions are used in the predicative the same
degree of definiteness is not found: when a man is introduced
with the words “This is my brother” or when I say “That is
not John's hat,” these words may mean indefinitely ‘one of my
110brothers’ and ‘one of John's hats.’ In German a proposed genitive
renders definite (Schiller's gedichte) but a postposed genitive
does not, whence the possibility of saying einige gedichte Schiller's
and the necessity of adding the definite article (die gedichte Schiller's)
if the same degree of definiteness is wanted as in the preposed genitive.
Where a prepositional group is used instead of the genitive,
the article is similarly required: die gedichte von Schiller, so in
other languages: the poems of Schiller, les poèmes de Schiller, i
poemi dello Schiller.

In some languages it is possible to use a possessive pronoun in
the incompletely restricted sense. MHG had ein sîn bruoder,
where now ein bruder von ihm is said. In Italian, possessives are
not definite, hence the possibility of saying un mio amico | alcuni
suoi amid | con due o tre amid suoi | si comunicarono certe loro
idee di gastronomia (Serao, Cap. Sans. 304). Consequently the
article is needed to make the expression definite: il mio amico.
But there is an interesting exception to this rule: with names
indicating close relationship no article is used: mio fratello, suo
zio. If I am not mistaken this must have originated with mio
padre, mia madre, where definiteness is a natural consequence of
one's having only one father and one mother, and have been analogically
extended to the other terms of kinship. It is perfectly
natural that the article should be required with a plural: i miei
fratelli, and on the other hand that it should not be used with a
predicative: questo libro i, mio. In French the possessives are
definite, as shown through their combination with a comparative
as in mon meilleur ami ‘my best friend,’ where the pronoun has
the same effect as the article in le meilleur ami. 148 But a different
form is used in (the obsolete) un mien ami = It. un mio amico,
now usually un de mes amis (un ami à moi). In English indefiniteness
of a possessive is expressed by means of combinations with
of: a friend of mine | some friends of hers, cf. also any friend of
Brown's, a combination which is also used to avoid the collocation
of a possessive (or genitive) and some other determining pronoun:
that noble heart of hers | this great America of yours, etc. As a
partitive explanation 249 is excluded here, we may call this construction
“pseudo-partitive.”

Next we come to non-restrictive adjuncts as in my dear little
Ann! As the adjuncts here are used not to tell which among
several Anns I am speaking of (or to), but simply to characterize
111her, they may be termed ornamental (“epitheta ornantia”)
or from another point of view parenthetical adjuncts. Their
use is generally of an emotional or even sentimental, though not
always complimentary, character, while restrictive adjuncts are
purely intellectual. They are very often added to proper names:
Rare Ben Jonson | Beautiful Evelyn Hope is dead (Browning) |
poor, hearty, honest, little Miss La Creevy (Dickens) | dear, dirty
Dublin | le bon Dieu. In this extremely sagacious little man, this
alone defines, the other adjuncts merely describe parenthetically,
but in he is an extremely sagacious man the adjunct is restrictive.

It may sometimes be doubtful whether an adjunct is of one
or the other kind. His first important poem generally means ‘the
first among his important poems’ (after he had- written others of
no importance), but it may also mean the first he ever wrote and
add the information that it was important (this may be made clear
in the spoken sentence by the tone, and in the written by a comma).
The industrious Japanese will conquer in the long run: does this
mean that the J. as a nation will conquer, because they are industrious,
or that the industrious among the Japanese nation will
conquer?

I take a good illustration of the difference between the two
kinds of adjuncts from Bernhard Schmitz's French Grammar:
Arabia Felix is one part of Arabia, but the well-known epigram
about (the whole of) Austria, which extends her frontiers by marriages,
while other countries can only extend theirs by war, says:
“Tu, felix Austria, nube.” The same difference between a preposed
non-restrictive and a postposed restrictive adjunct is seen
in the well-known rules of French Grammar, according to which ses
pauvres parents comprises all his relatives in sympathetic compassion,
while ses parents pauvres means those of his relatives
that are poor — a distinction which is not, however, carried through
consistently with all adjectives.

The distinction between the two kinds of adjuncts is important
with regard to relative clauses. In English, while the pronouns
who and which may be found in both, only restrictive clauses can
be introduced by that or without any pronoun: the soldiers that were
brave ran forward | the soldiers, who were brave, ran forward |
everybody I saw there worked very hard. The difference between
the first two sentences can be made still more evident by the insertion
of all: all the soldiers that were brave… | the soldiers, who
were all of them brave… It will be noticed that there is also a
marked difference in tone, a non-restrictive clause beginning on a
deeper tone than a restrictive one; besides, a pause is permissible
before a non-restrictive, but hardly before a restrictive clause;
cf. the use of a comma in writing. In Danish the difference is
112shown by the article of the antecedent: (alle) de soldater som var
modige lab frem | soldaterne, som (alle) var modige, fab frem. But
this criterion is not always available; if the antecedent has another
adjunct the only difference is in the stress of the preposed article:
ˈde franske soldater som… | de ˈfranske soldater, som… A
so-called continuative relative clause is, of course, non-restrictive:
he gave the letter to the clerk, who then copied it, Dan. han gav brevet til
kontoristen, som så skrev det av (but: … to the clerk who was to
copy it… til den kontorist som skulde skrive det av).

The following examples will serve further to illustrate the two
kinds of relative clause adjuncts: there were few passengers that
escaped without serious injuries | there were few passengers,
who escaped without serious injuries | they divide women into two
classes: those they want to kiss, and those they want to kick,
who are those they don't want to kiss.

The distinction between restrictive and non-restrictive adjuncts
(which are both in a certain sense qualifiers) does not affect quantifying
adjuncts, such as many, much, some, few, little, more, less,
no, one and the other numerals. Whenever these are found with
adjectives as adjuncts to the same primary they are always placed
first: many small boys | much good wine | two young girls. There
is a curious relation between such quantifiers and combinations of
substantives denoting number or quantity followed by an of-group
(or in languages with a more complicated form-system, a
partitive genitive or a partitive case): hundred was originally a
substantive and in the plural is treated as such: hundreds of
soldiers, but in the singular, in spite of the preposed one or a, it
is treated like the other numerals: a hundred soldiers; thus also
three hundred soldiers; cp. dozens of bottles, a dozen bottles. Where
E. has a couple of days, a pair of lovers, G. has ein paar tage, Dan.
et par doge, even die paar tage, de par doge exactly as die zwei tage,
de to doge. To E. much wine, many bottles, no friends, corresponds
Ft. beaucoup de vin, beaucoup de bouteilles, pas d'amis; to E. a
pound of meat, a bottle of wine corresponds G. ein pfund fleisch, eine
flasche wein, Dan. et pund kød, en flaske vin, etc.

Wherever an indefinite article is developed, it seems always to
be an unemphatic form of the numeral one: uno, un, ein, en, an (a),
Chinese i, a weak form of yit (Russ. odin is often used like an indefinite
article). In English a has in some cases the value of the
numeral, as in four at a time, birds of a feather, and in some cases the
full and the weakened forms are synonymous, as in one Mr. Brown
= a Mr. Brown, where we may also say a certain Mr. Brown. This
use of the word certain reminds us that in most cases where we use
the “indefinite” article we have really something very definite
in our mind, and “indefinite” in the grammatical sense practically
113means nothing but “what shall not (not yet) be named,” as in the
beginning of a story: “In a certain town there once lived a tailor
who had a young daughter” — when we go on we use the definite
form about the same man and say: “The tailor was known in
that town under the name of, etc.” (On the “generic” use of the
indefinite article see p. 152 and Ch. XV.)

As the indefinite article is a weakened numeral, it is not used
with “uncountables” (mass-words, Ch. XIV). And as one — and consequently
a(n) — has no plural, there is no plural indefinite article,
unless you count the curious Sp. unos as one. But in a different
way French has developed what may be called an indefinite article
to be used with mass-words and plurals in its “partitive article,”
as in du vin, de l'or, des amis. This, of course, originated in a prepositional
group, but is now hardly felt as such and at any rate
can be used after another preposition: avec du vin | j'en ai parlé à
des amis. It is now just as good an adjunct as any numeral or as
the synonym quelque(s) or E. some.

Nexus.

We now proceed to what was above (p. 97) termed nexus.
The example there given was the dog barks furiously as contrasted
with the junction a furiously barking dog. The tertiary element
furiously is the same in both combinations, and may therefore
here be left out of account. The relation between the dog barks
and a barking dog is evidently the same as that between the rose
is red and a red rose. In the dog barks and the rose is red we have
complete meanings, complete sentences, in which it is usual to
speak of the dog and the rose as the subject, and of barks and
is red as the predicate, while the combination is spoken of as
predication. But what is the difference between these and the
other combinations?

Paul thinks that an adjunct is a weakened predicate (ein degradiertes
prädikat, P 140 ff.), and in the same way Sheffield says that
an adjunct “involves a latent copula” (GTh 56). If this means
that a red rose is equivalent to (or had its origin in) a rose which is red,
and that therefore red is always a kind of predicative, it should
not be overlooked that the relative pronoun is here smuggled into
the combination, but the function of the relative is precisely that
of making the whole thing into an adjunct (an attribute, an epithet).
Barking is not a degraded barks, though a barking dog is a dog who
barks. Peano is much more right when he says that the relative
pronoun and the copula are like a positive and a negative addition
of the same quantity which thus annul one another (which = — is,
or — which = + is), thus which is = 0.114

While Paul thinks that junction (attributivverhältnis) has
developed from a predicate relation, and therefore ultimately from
a sentence, Sweet does not say anything about the relative priority
of the two combinations, when he says that “assumption” (his
name for what is here called junction) is implied or latent predication,
and on the other hand, that predication is a kind of strengthened
or developed assumption (NEG § 44). But this way of looking at
the question really leads nowhere.

Wundt and Sütterlin distinguish the two kinds as open and
closed combinations (offene und geschlossene wortverbindungen).
It would probably be better to say that one is unfinished and makes
one expect a continuation (a red rose, — well, what about that rose?)
and the other is rounded off so as to form a connected whole (the
rose is red). The former is a lifeless, stiff combination, the latter
has life in it. This is generally ascribed to the presence of a finite
verb (the rose is red; the dog barks), and there is certainly much
truth in the name given to a verb by Chinese grammarians, “the
living word” as opposed to a noun which is lifeless. Still, it is
not the words themselves so much as their combinations that impart
life or are deprived of life and, as we shall see presently, we have
combinations without any finite verb which are in every respect
to be ranged with combinations like the rose is red, or the dog barks.
These form complete sentences, i.e. complete communications, and
this, of course, is very important, even from the grammarian's
point of view. But exactly the same relation between a primary
and a secondary word that is found in such complete sentences is
also found in a great many other combinations which are not
so rounded off and complete in themselves as to form real sentences.
We need not look beyond ordinary subordinate clauses to see this,
e.g. in (I see) that the rose is red, or (she is alarmed) when the dog
barks. Further, the relation between the last two words in he
painted the door red is evidently parallel to that in the door is red and
different from that in the red door, and the two ideas “the Doctor”
and “arrive” are connected in essentially the same way in the
four combinations (1) the Doctor arrived, (2) I saw that the Doctor
arrived, (3) I saw the Doctor arrive, (4) I saw the Doctor's arrival.
What is common to these, and to some more combinations to be
considered in the next chapter, is what I term a nexus, and I shall
now try to determine what constitutes the difference between
a nexus and a junction, asking the reader to bear in mind that on
the one hand the presence of a finite verb is not required in a nexus,
and that on the other hand a nexus may, but does not always, form
a complete sentence.

In a junction a secondary element (an adjunct) is joined to
a primary word as a label or distinguishing mark: a house is
115characterized by being mentioned as the next house or the Doctor's
house. Adjunct and primary together form one denomination, a
composite name for what conceivably might just as well have been
called by a single name. As a matter of fact, instead of new-born
dog we often say puppy, instead of silly person we may say fool;
compare also the composite expressions a female horse, the warm
season, an unnaturally small person, an offensive smell with the
single-word expressions a mare, the summer, a dwarf, a stench, etc.
What in one language is expressed by one word, must often in
another be rendered by means of a primary with an adjunct:
E. claret, Fr. vin rouge, and on the other hand, Fr. patrie, E. native
country. A junction is therefore a unit or single idea, expressed
more or less accidentally by means of two elements. 150

A nexus, on the contrary, always contains two ideas which
must necessarily remain separate: the secondary term adds something
new to what has already been named. Whereas the junction
is more stiff or rigid, the nexus is more pliable; it is, as it were,
animate or articulated. Comparisons, of course, are always to
some extent inadequate, still as these things are very hard to
express in a completely logical or scientific way, we may be allowed
to say that the way in which the adjunct is joined to its primary
is like the way in which the nose and the ears are fixed on the head,
while an adnex rests on its primary as the head on the trunk or
a door on a wall. A junction is like a picture, a nexus like a process
or a drama. The distinction between a composite name for
one idea and the connexion of one concept with another concept
is most easily seen if we contrast two such sentences as the blue
dress is the oldest and the oldest dress is blue; the fresh information
imparted about the dress is, in the first sentence that it is the oldest,
and in the second that it is blue; cf. also a dancing woman charms
and a charming woman dances.

We shall now consider more in detail the various grammatical
combinations characterized by nexus. Some of these are well
known to grammarians, but the collocation of them all from this
point of view, so far as I know, is new.116

Chapter IX
Various Kinds of Nexus

Finite Verb. Infinitival Nexus. Nexus without a Verb. Nexus-Object,
etc. Nexus-Subjunct. Nexus of Deprecation. Summary. Copula.
Predicative.

Finite verb.

In attempting to classify the various kinds of nexus we shall first
very briefly mention the three kinds which contain a finite verb:
first the ordinary complete sentences, as in “the dog barks” | “the
rose is red.” Second, the same combinations in subordinate
clauses, that is, as parts of a sentence, as in “she is afraid when
the dog barks | I see that the rose is red.” Third, the very interesting
phenomenon seen in “Arthur whom they say is kill'd to-night”
(Shakesp., John IV, 2. 165). The nexus whom is kill'd is the object
of they say, whence the use of the accusative whom. In the
Appendix I shall give other examples of this construction as well
as my reasons for defending the form whom, which is generally
considered as a gross error.

Infinitival Nexus.

Next we have a series of constructions containing an infinitive.

The accusative with the infinitive. Examples of this well-known
construction: I heard her sing | I made her sing | I caused
her to sing — thus in some combinations with, and in others without,
to. Similarly in other languages. Sweet, § 124, notices the
difference between I like quiet boys and I like boys to be quiet, the
latter sentence implying not even the slightest liking for boys, as
the former does, but he does not see the real reason for this difference,
as according to him “the only word that I like governs grammatically
is boys, to be quiet being only a grammatical adjunct to
boys.” It would be more correct to say that it is not boys that is
the object, but the whole nexus consisting of the primary boys and
the infinitive, exactly as it is the whole clause and not only the
subject of it that would be the object, if we were to translate it
into “I like that boys are quiet.” (This construction is rare with
this verb, though NED has a quotation from Scott; with other
verbs which also take the acc. with the inf., such as see, believe, it is
117in common use.) Sonnenschein § 487, here speaks of “two direct
objects” and places the sentence on the same footing as “he asked
me a question,” but this is misleading, for without change of sense
we may say “he asked a question,” while “I like to be quiet” is
totally different from the sentence with boys inserted. The relation
between boys and the infinitive is not at all the same as that between
me and a question, but is exactly the same as between the two parts
of any other nexus, e.g. between the subject and the predicate
of a complete sentence.

The same construction is frequently found in English in cases
where the nexus is the object not of a verb, but of a preposition, or
perhaps rather of a phrase consisting of a verb and a preposition,
which is often synonymous with a single verb (look on = consider,
prevail on = induce, etc.). Examples: I looked upon myself to be
fully settled (Swift) | she can hardly prevail upon him to eat | you
may count on him to come.

While “I long for you to come” can be analyzed in the same
way, this is not true of some other combinations with for and an
infinitive that have developed in modern English. The original
division of a sentence like “It is good for a man not to touch a
woman” was “It is good for a man | not to touch a woman,” but
it came to be apprehended as “It is good | for a man not to touch
a woman,” where for a man was taken to belong more closely to
the infinitive. This led to the possibility of placing for and the
word it governs first, as in: for a man to tell how human life began
is hard (Milton) | for you to call would be the best thing, and to the
further use after than: Nothing was more frequent than for a bailiff
to seize Jack (Swift) | nothing could be better than for you to call:
for and its object are now nothing but the primary (subject) of the
nexus, whose secondary part is the infinitive; combinations like
“it might seem disrespectful to his memory for me to be on good
terms with [his enemy]” (Miss Austen) show how far the construction
has wandered from its original use, as to his memory here serves
the same purpose as the for-phrase did at first. (See my paper on
this shifting in “Festschrift W. Viëtor.” Die neueren Sprachen,
1910.)

There is a close parallel to this English development in Slavic,
where a dative with an infinitive is frequent in places where Greek
and Latin would have an acc. with inf., see Miklosich, Synt. 619,
Vondrák, SG 2. 366 and especially C. W. Smith in Opuscula philol.
ad I. N. Madvigium, 1876, 21 ff. From such sentences as OSl.
dobro jestĭ namŭ sĭde byti ‘it is good for us to be here,’ where the
dative originally belonged to ‘is good’ it was extended to cases
like ne dobro jestĭ mnogomŭ bogomŭ byti ‘it is not good for many
gods to be, i.e. that there are many gods’; the construction is used
118even with verbs which cannot naturally take a dative. In the early
Gothonic languages there was a similar construction, and Grimm
and others speak of a dative-with-infinitive construction in Gothic
jah wairþ þairhgaggan imma þairh atisk (Mark 2. 23 ‘and it happened
for him to go through the field’) and similar instances in the
related languages; they can, however, scarcely be considered as
more than the first abortive beginnings of the development that
proved so fruitful in Slavic (see the able discussion in Morgan
Callaway, The lnfin. in Anglo-Saxon, Washington, 1913, p. 127 and
248 ff., where earlier writers on the subject are quoted).

We have seen the primary, or what is virtually the subject of
an infinitive, put in the accusative, and in the dative, and with
the preposition for; but in some languages it may also be put in
the nominative. In ME the common case of substantives, which
represents the earlier nominative and accusative alike, was used in
combinations like: Lo! swich it is a millere to be fals (Chaucer) |
And verelye one man to lyue in pleasure, whyles all other wepe…
that is the parte of a iayler (More). In pronouns we find the
nominative: Thow to lye by our moder is to muche shame for vs to
sufrre (Malory). In Spanish we have a nominative: Es causa
bastante Para tener hambre yo? ‘Is that reason enough for me to
be hungry?’ | Que importara, si esta muerto Mi honor, el quedar yo
vivo! ‘What matters it that I remain alive, if my honour is dead?’
(both from Calderon, Ale. de Zal. 1. 308 and 2. 840). In the same
way in Italian, and in Portuguese also with eu ‘I.’ 151 An Italian
combination like “prima di narrarci il poeta la favola,” in which
the infinitive has both a subject and two objects, reminds one
strongly of a subordinate clause (“before the poet tells us the
story”), from which it differs only in not having a finite verb.
Similarly in Arabic, according to Steinthal, Charakteristik, 267, I
transcribe his translation of one example: ‘es ist gemeldet-mir die
tödtung (nominat.) Mahmud (nominat.) seinen-bruder, d. h. dass
Mahmud seinen bruder getodtet hat.’

The following instances show another way in which a nominative
may be the notional subject of an infinitive. If the object of he
believes in “he believes me to be guilty” is the whole nexus consisting
of the four last words, it is necessary to say that in the
passive construction “I am believed to be guilty” the subject is
not “I” alone, but the nexus I to be guilty, although these words
119do not stand together, and though the person of the verb is determined
by the first word alone. What is believed is my guilt. In
the same way he is said (expected, supposed) to come at five (his arrival
at five is expected) | I am made (caused) to work hard (what is
caused is not “I,” but my working) and correspondingly in other
languages. 152

The same consideration holds good in active constructions, e.g.
he seems to work hard | er scheint hart zu arbeiten | il semble (paraît)
travailler durement (where Dan. has the passive form just as in the
above-mentioned sentences: han synes at arbejde hårdt): the real
subject is the whole underlined nexus. 253 This analysis must consistently
be extended to instances like E. he is sure (likely) to come |
she happened to look up, etc., though these latter constructions are
historically developed from older ones in which what is now in the
nominative was put in the .dative case.

While all the infinitive-combinations hitherto mentioned are
primary members of the main sentence, we have now to deal with
the rare cases in which similar combinations are subjuncts, e.g.
the caul was put up in a raffle to fifty members at half-a-crown a
head, the winner to spend five shillings (Dickens) | we divided it:
he to speak to the Spaniards and I to the English (Defoe). The
infinitive here has the same signification of what is destined or
enjoined as in he is to spend, and the whole nexus may be said to
be used instead of the clumsy the winner being to spend, which
would belong in a following paragraph.

A further kind of nexus is found, as already noted (p. 115),
in combinations like “I heard of the Doctor's arrival.” But these
verbal substantives will require a separate chapter (Ch. X). The
only thing to be mentioned here is that the similarity between such
combinations and sentences like “the Doctor arrived” is recognized
in the traditional term “subjective genitive” as contrasted with
the “possessive genitive” in “the Doctor's house, the Doctor's
father.”

Nexus without a Verb.

A final series of nexuses consists of those which contain neither
a finite verb nor an infinitive nor a verbal substantive.

Here we first encounter the so-called nominal sentences, containing
120a subject and a predicative, which may be either a substantive
or an adjective. These sentences are extremely frequent
both in such languages as have not developed a “copula,” i.e. a
verb meaning ‘to be,’ and in those languages which have a copula,
but do not use it as extensively as e.g. English. Among the latter
are some of the oldest languages of our family — for instance, old
Greek; see especially Meillet, La phrase nominale, en indo-européen,
MSL 14, 1906, p. 1 ff. In Russian this is the ordinary construction
where we use the present tense of be: ‘I am ill’ is ja bolen, ‘he is a
soldier’ on soldat; a difference is made in the form of an adjective
according as it is used as a predicative or as an adjunct, e.g. dom
nov ‘the house is new,’ dom novyj ‘a new house, the new house.’
The verb ‘be,’ however, has to be expressed in other tenses, as well
as in sentences meaning ‘there is, or are.’

It is generally said that such “nominal” sentences are no
longer found in our West-European languages, but as a matter
of fact there is one particular form in which they are extremely
common. Under the influence of strong feeling there seems to
be everywhere a tendency to place the predicative first, to which
the subject is added as a kind of afterthought, but without the
verb is. In this way we get sentences which are analogous in every
respect to the Greek as “Ouk agathon polukoiraniē” (Not a good
thing, government by the many), for instance: Nice goings on,
those in the Balkans! | Quite serious all this, though it reads like
a joke (Ruskin) | Amazing the things that Russians will gather
together and keep (H. Walpole) | what a beastly and pitiful wretch
that Wordsworth (Shelley; such that-phrases are frequent. 154) | Fr.
Charmante, la petite Pauline! | Dan. Et skrsekkeligt baest, den
Christensen! | Godt det samme!

This construction is frequent with expressions for “happy”:
Gr. Trismakares Danaoi kai tetrakis, hoi tot' olonto Troiēi en eureiēi
'thrice and four times happy the Danaans who perished then in
broad Troy (Odyss. 5. 306) | felix qui potuit rerum cognoscere
causas (Virg.) | Beati possidentes | Happy the man, whose wish
and care A few paternal acres bound (Pope) | Thrice blest whose
lives are faithful prayers (Tennyson) | Dan. Lykkelig den, hvis lykke
folk foragter! (Rørdam); cf. also Gothic Hails þiudans iudaie
(Joh. 19. 3) | ON. Heill þū nū Vafþrūþner | All haile Macbeth! 255
Another frequent form is: Now I am in Arden, the more fool
I! (Sh.).121

Very often the subject that follows the predicative is an infinitive
or a whole clause: Gr. Argaleon, basileia, diēnckeōs agoreusai
‘difficult, your Majesty, to speak at length’ (Od. 7. 241) | Needless
to say, his case is irrefutable | Fr. Inutile d'insister davantage |
What a pity that he should die so young | Wie schade dass er so
früh sterben sollte | Quel dommage qu'il soit mort si tôt | Skade at
han døde så ung | Small wonder that we all loved him exceedingly
How true, that there is nothing dead in this Universe (Carlyle)
true, she had not dared to stick to them.

In a special French form we have que before the subject:
Singulier homme qu'Aristote! | Mauvais prétexte que tout cela!

I have given all these examples, because grammarians generally
fail to appreciate these constructions. It is no use saying that we
have here ellipsis of is; it would only weaken the idiomatic force of
such sentences if we were to add the verb, though it would be
required if the subject were placed first.

Corresponding verbless combinations are also found in clauses:
Russian govorjat čto on bolen ‘they say that he is ill’ | However great
the loss, he is always happy | the greater his losses, the more will he
sing | his patrimony was so small that no wonder he worked now
and then for a living wage (Locke).

Nexus-Object, etc.

A nexus-object is often found: “I found the cage empty” which
is easily distinguished from “I found the empty cage” where empty
is an adjunct. It is usual here to say that the cage is the object
and that empty is used predicatively of, or with, the object, but it
is more correct to look upon the whole combination the cage empty
as the object. (Cf. “I found that the cage was empty” and
“I found the cage to be empty.”) This is particularly clear in
sentences like “I found her gone” (thus did not find her!), cf.
also the contrast between “I found Fanny not at home,” where
the negative belongs to the subordinate nexus, and “I did not find
Fanny at home,” where not negatives find.

Other examples: they made him President (him President is
the object of result) | he made (rendered) her unhappy | does that
prove me wrong? | he gets things done | she had something the
matter with her spine | what makes you in such a hurry? | she only
wishes the dinner at an end. The predicate-part of the nexus may
be any word or group that can be a predicative after the verb to be.

The most interesting thing here is that a verb may take a nexus-object
which is quite different from its usual objects, as in he drank
122himself drunk | the gentleman had drunke himselfe out of his five
senses (Sh.; he drank himself is absurd) and that verbs otherwise
intransitive may have a nexus-object of result: he slept himself
sober | A louer's eyes will gaze an eagle blind (Sh.) | Lily was nearly
screaming herself into a fit.

Other languages present similar phenomena, e.g. Dan. de drak
Jeppe fuld | de drak Jeppe under bordet | ON. þeir biðja hana
gráta Baldr ór helju ‘they ask her to weep B. out of Hades.’ Paul
P. 154 mentions combinations like: die augen rot weinen | die
füsse wund laufen | er schwatzt das blaue vom himmel herunter |
denke dich in meine lage hinein; but his remarks do not show
clearly how he apprehends this “freie verwendung des akkusativs.”
In Finnish we have here the characteristic case called “translative,”
as in: äiti makasi lapsensa kuoliaaksi ‘the mother slept her child
(into) dead (overlay it)’ | hän joi itsensä siaksi ‘he drank himself
(into) a swine’; the examples taken from Eliot FG 128, others in
Setala, Finska språkets satslära §29.

The close analogy between the accusative with infinitive and
this nexus-object makes it easy to understand that we sometimes
find the same verb taking both constructions in the same sentence:
a winning frankness of manner which made most people fond of her,
and pity her (Thackeray) | a crowd round me only made me proud,
and try to draw as well as I could (Ruskin) | he felt himself dishonored,
and his son to be an evil in the tribe (Wister).

In the passive turn corresponding to sentences with nexus-objects,
we must consistently (as in the infinitive-constructions,
p. 119) look upon the whole nexus as the (notional) subject, thus he
… President in “he was made President,” etc., though, of course,
the person of the verb is dependent on the primary part of the
nexus only: if I am made President. In Danish we have constructions
like “han blev drukket under bordet | pakken ønskes
(bedes) bragt til mit kontor,” literally, ‘the parcel is wished (asked)
brought to my office.’ Cf. ON. at biðja, at Baldr væri grátinn
ór Helju ‘to ask that Baldr should be wept out of Hades.’

Analogous constructions are sometimes found with active
verbs, as in Greek: allous men pantas elanthane dakrua leibōn
(Od. 8. 532) ‘he escaped the attention of the others shedding tears,
i.e. the fact that he shed…’ | hōs de epausato lalōn (Luk. 5. 4;
the E. translation “when he had left off speaking” is only seemingly
in accordance with the Greek text, for speaking is the verbal substantive
as object of left, not a participle in the nom. as lalon). 156

A nexus may be the object of a preposition. In English this
is particularly frequent after with as in: I sat at work in the schoolroom
123with the window open (different from: near the open window) |
you sneak back with her kisses hot on your lips (Kipl.) | he fell asleep
with his candle lit | let him dye, With euery ioynt a wound (Sh.) | he
kept standing with his hat on. The character of the construction
and the peculiar signification of with (different from that in “he
stood with his brother on the steps”) is particularly clear when
the adnex neutralizes the usual meaning of with: with both of us
absent | wailed the little Chartist, with nerve utterly gone | I hope
I'm not the same now, with all the prettiness and youth removed.

Without also is found governing a nexus: like a rose, full-blown,
but without one petal yet fallen.

In Danish med often takes a nexus: med hænderne tomme ‘with
the hands empty,’ different from med de tomme hænder ‘with the
empty hands,’ which presupposes some action by means of the
hands, while the former combination implies nothing more than a
clause (while, or as, his hands are, or were, empty). Similarly also
in other languages.

With other prepositions we have the well-known Latin constructions
post urbem conditam | ante Christum natum. When
Madvig here says that the idea is not so much of the person or thing
in a certain condition, as of the action as a substantival conception
he is thinking of the (Danish, etc.) translation by means of a substantive,
but this, of course, is of the class described below as
nexus-substantive (‘after the construction of the town, before the
birth of Christ’), which is different from ordinary substantival
conceptions, and calls for a separate elucidation, so that Madvig's
explanation leaves us just where we were. Nor do we get much
further with Allen and Greenough's comment that “a noun and a
passive participle are often so united that the participle and not
the noun contains the main idea.” Brugmann (IF 5. 145 ff.)
characterizes the explanation by means of an abbreviated clause
as “sterile linguistic philosophy” 157 and thinks himself that the
construction took its origin in a shifting of the syntactic structure
(verschiebung der syntaktischen gliederung) in combinations like
post hoc factum, which at first meant ‘after this fact’ (hoc adjunct
to the primary factum, if I may use my own terms), but was afterwards
apprehended with hoc as primary and factum as secondary,
124this being subsequently extended to other cases. The whole
explanation seems rather far-fetched. None of these grammarians
thinks of classing the phenomenon with the rest of the constructions
which I mention in this chapter (absolute ablative, etc.), though
it is only through a collective treatment that they can be fully
understood as illustrating one another.

In Italian the same construction is pretty frequent after dopo:
dopo vuotato il suo bicchiere, Fileno disse | Cercava di rilegger
posatamente, dopo fatta la correzione (Serao) | Dopo letta questa
risposta, gli esperti francesi hanno dichiarato che… (Newspaper).

Milton's “after Eve seduc'd” and Dryden's “the royal feast
for Persia won” are no doubt due to conscious imitation of Latin
syntax, but that does not account for similar constructions found
here and there in less learned writers: before one dewty done
(Heywood) | they had heard of a world ransom'd, or one destroyed
(Sh., may be adjunct) | after light and mercy received (Bunyan) |
he wished her joy on a rival gone (Anthony Hope) — to pick out only
a few of the examples I have collected.

Similar nexuses may be found also in other positions, where
they are not the object either of a verb or of a preposition, thus in
Lat.: dubitabat nemo quin violati hospites, legati necati, pacati
atque socii nefario bello lacessiti, fana vexata hanc tantam efficerent
vastitatem (Cicero, translated by Brugmann ‘dass die mishandlung
der gastfreunde, die ermorderung der gesandten, die ruchlosen
angriffe auf friedliche und verbündete volker, die schandung der
heiligtumer’).

A similar example is found in Shakespeare: Prouided that my
banishment repeal'd, And lands restor'd againe be freely graunted
(R2 III. 3. 40 = the repealing of my b. and restoration of my 1.).
But in cases like the following it may be doubtful whether we have
a participle or a verbal substantive: the Squire's portrait being
found united with ours, was a honour too great to escape envy
(Goldsmith) | And is a wench having a bastard all your news?
(Fielding).

French examples have been collected by Sandfeld Jensen
(Bisætningerne i moderne fransk, 1909, p. 120) and E. Lerch (Prädikative
partizipia für verbalsubstantiva im französ., 1912), e.g. le
verrou poussé l'avait surprise ‘the fact that the door was bolted’ |
c'était son rêve accompli ‘das war die erfüllung ihres traumes.’
The adnex need not be a participle, as is seen by some relative
clauses analyzed by Sandfeld Jensen: Deux jurys qui condamnent
un homme, ça vous impressionne, in which ça (singular) clearly
shows the character of the combination. Cf. now Brunot PL 208.

I am inclined to include here some combinations with “quantifiers,”
which are not to be taken in the usual way, e.g. the proverb
125too many cooks spoil the broth = the circumstance that the cooks are
too numerous spoils. Thus also: trop de cuisiniers gâtent la sauce |
viele köche verderben den brei | mange kokke fordærver maden |
many hands make quick work | mange hunde er harens død | no
news is good news | you must put up with no hot dinner. This is
evidently quite different from the adjuncts in “too many people
are poor” or “no news arrived on that day.”

Nexus Subjunct.

We next come to nexus subjuncts. None of the usual names
(duo ablativi, ablativi consequential, ablativi absoluti, absolute
participles) get at the essence of the phenomenon: “absolute”
must mean ‘standing out of the syntactic connexion,’ but do these
words stand more outside than other subjuncts? Participle should
not be mentioned in the name, for no participle is required, e.g.
dinner over | Scipione autore, etc. Brugmann (KG. § 815) makes
an attempt at explaining the various cases employed (gen. in Gr.,
and Sanskrit, abl. in Lat., dat. in Gothic, O.H.G., OE., ON., etc.);
he thinks that the participle to begin with was an ordinary adjunct,
which later through a “verschiebung der syntaktischen gliederung”
was felt together with some other word to be “eine art von (temporalem
oder dgl.) nebensatz.” In my view what is characteristic
of the construction is contained in two things: (1) that there are two
members standing to another in the peculiar relation here termed
nexus, thus parallel to the relation between subject and verb in
“the dog barks,” and (2) that this combination plays the part
of a subjunct in the sentence. I am not here concerned with the
question how the Latin ablative is to be explained, whether as
originally local or temporal or instrumental; in the language as
we know it the temporal Tarquinio rege only differs from hoc
tempore in this, that rege stands in another relation to its primary
Tarquinio than hoc (adjunct) to its primary tempore. The same
difference is seen in me invito as against hoc modo, both combinations
denoting manner. 158126

In the Romanic languages, the nexus-subjunct is still so common
that a few examples will suffice: It. morto mio padre, dovci andare
a Roma | sonate le cinque, non è più permesso a nessuno d'entrare |
Fr. Ces dispositions faites, il s'est retiré | Dieu aidant, nous y
parviendrons. 159 Sp. concluidos los estudios… pues no hube
clase… Examinadas imparcialmente las cualidades de aquel
nifio, era imposible desconocer su merito (Galdos, D. Perf. 83).

In English the construction is frequent, though apart from
certain restricted applications it is more literary than popular:
we shall go, weather permitting | everything considered, we may feel
quite easy | this done, he shut the window | she sat, her hands
crossed on her lap, her eyes absently bent upon them 260 | he stood, pipe
in mouth 261 | dinner over, we left the hotel. Thus very often with
one of the other words or groups that can be predicatives besides
adjectives and participles.

There is in certain cases a tendency to introduce the nexus-subjunct
by some word like once: Once the murderer found, the
rest was easy enough | Fr. Une fois l'action terminée, nous rentrâmes
chez nous (sitôt achevée cette tâche).

In German nexus-subjunets are pretty common now, though
comparatively young in the language; I select a few of Paul's
examples (Gr. 3. 278): Louise kommt zurück, einen mantel umgeworfen |
alle hände vol', wollen Sie noch immer mehr greifen |
einen kritischen freund an der seite kommt man schneller vom fleck.
Paul is not explicit as to how this “art des freien akkusativs” is
to be apprehended, but his remark (after examples with a passive
participle) “In allen diesen fallen konnte man statt des passiven
ein aktives attributives partizipium einsetzen” and his mention
(on p. 284) of the accusative as an acc. of object leave us in the
lurch with regard to those combinations that contain no participles.
Curme (GG 266, 553) also takes the participle in an active sense
and thinks that habend is understood: Dies vorausgeschickt [habend]
fahre ich in meiner erzdhlung fort | Solche hindernis alle ungeachtet
[habend], richtet gott diesen zug aus. I am very sceptical with regard
to this explanation of the origin of the construction through sub-audition;
anyhow, it does not explain how (in Curme's own words)
“the construction has become productive, so that we now find as
predicate of the clause [what I call the nexus] not only a perfect
participle of a transitive verb, but also the perfect participle of an
intransitive verb, an adjective, adverb, or a prepositional phrase.”127

As nexus-subjuncts we may also consider the genitives in
unverrichteter dinge kam er zurück | wankenden schrittes…
erscheint der alte mann (Raabe, quoted by Curme).

The “absolute dative” in the old Gothonic languages is often
explained as an imitation of the Latin construction. In Dan.
the construction plays only a subordinate rôle, apart from a few
fixed combinations like “Alt vel overvejet, rejser jeg imorgen | alt
iberegnet | dine ord i ære, tror jeg dog…” as in G. dein wort in
ehren, literally ‘your words in honour,’ i.e. with due deference
to your words.

To begin with, the subject-part of this nexus-subjunct was
everywhere put in some oblique case, though, as we have seen,
this case was different in different languages. But independently
of one another, various languages began to use the nominative
case as more conformable to the rôle as subject. This is the rule
now in Modern Greek (Thumb, Handb. 2 ed. 161), and goes far
back, as Sandfeld tells me, e.g. in the apocryphal Evang. Thomæ
10. 1 Met' oligas hēmeras skhizōn tis xula… epesen hē axinē. To
the same friend I am indebted for an early mediaeval Latin example:
Peregrinatio Silvias 16. 7 benedicens nos episcopus profecti sumus.
In Romanic languages the case is not shown in substantives, but
with pronouns we have the nominative, e.g. It. essendo egli Cristiano,
io Saracina (Ariosto), Sp. Rosario no se opondrá, queriendolo yo
(Galdós, D. Perf. 121). In English the nominative has prevailed
in the standard language: For, he being dead, with him is beautie
slaine (Sh. Ven. 1019). In G. the nominative is found now and
then, see Paul Gr. 3. 281 and 283, who gives the following
example from Grillparzer: der wurf geworfen, fliegt der stein,
and Curme GG 554, who has examples from Schiller, Auerbach,
Hauptmann, etc.

In this notwithstanding (notwithstanding this) and notwiths.anding
all our efforts we have properly a nexus-subjunct with this and all
our efforts as primaries and the negative participle as adnex, but
the construction is now practically to be considered as containing
a preposition and its object; thus also G. ungeachtet unserer bemühungen,
Dan. uagtet vore anstrengelser. In the same way Fr. pendant
ce temps, E. during that time (orig. ‘while that time dures or lasts’).
German here goes still further in metanalysis: the old genitive
nexus-subjunct währendes krieges, pl. währender kriege, is dissolved
into während des krieges, während der kriege: in this way während
has become a preposition governing the genitive.

A nexus is very often expressed by means of a genitive and an
“abstract substantive” as in I doubt the Doctor's cleverness, which
means the same thing as ‘I doubt that the Doctor is clever.’ The
parallelism with verbal substantives, as in the Doctor's arrival,
is obvious, but nevertheless traditional grammatical terminology
restricts the use of the name ‘subjective genitive’ to the latter
combination, though it might just as well be applied to cases like
the Doctor's cleverness. 162 On both kinds of substantives see the
next chapter.

Nexus of Deprecation.

In all the various kinds of nexus thus far considered the connexion
between the two members is to be taken in a direct or positive
sense. But we now come to what might be termed the nexus
of deprecation in which the connexion is as it were brushed aside
at once as impossible; the meaning is thus negative, and this
is expressed in speech by the intonation, which is the same as in
questions, often in an exaggerated form and not infrequently given to
129the two members separately: we shall see in a later chapter that
question and negation are often closely akin.

There are two forms of deprecating nexus: first with an infinitive,
e.g. What? I loue! I sue! I seeke a wife! (Sh.) | “Did you
dance with her?” “Me dance!” says Mr. Barnes (Thackeray) |
I say anything disrespectful of Dr. Kenn? Heaven forbid! (G.
Eliot). 163 In the last example, the words “Heaven forbid” show
how the idea of the nexus is rejected; the following example from
Browning shows how the construction, if continued so as to form
a whole sentence of the regular pattern, conforms to the type
mentioned above, p. 121: She to be his, were hardly less absurd
Than that he took her name into his mouth. It is not, however,
common to complete the sentence in this way, the emotion having
found sufficient vent in the subject and the infinitive in the particular
tone of voice to which I have referred.

Other languages use the same trick, e.g. Er! so was sagen! |
Han gifte sig! | Toi faire ça! | Io far questo! | Mene incepto
desistere victam? — in Latin with the accusative with infinitive
that would be required if a proper predicate were added. 264

Second, a subject and a predicative may be placed together
with the same interrogative tone and the same effect of brushing
aside the idea of their combination as real or possible: Why,
his grandfather was a tradesman! he a gentleman! (Defoe) | The
denunciation rang in his head day and night. He arrogant, uncharitable,
cruel! (Locke). — It is, of course, possible to add a negative
in the form of an answer so as to make the meaning perfectly
clear: He arrogant? No, never! or, Not he!

In the same way in other languages: Hun, utaknemlig! | Er!
in Paris! | Lui avare? etc. In G. also with und: er sagte, er
wolle landvogt werden. Der und landvogt! Aus dem ist nie was
geworden (Frenssen).

These sentences with nexus of deprecation may be added to
those mentioned above, in which we had complete (independent)
sentences without a verb in one of the finite forms. From
another point of view they may be given as instances of aposiopesis:
under the influence of a strong emotion the speaker does not
trouble to finish his sentence, and not infrequently it would
be difficult to go on so as to produce a regularly constructed
sentence.130

Summary.

We may end this chapter by giving a tabulated survey of the
principal instances of nexus, using characteristic examples instead
of descriptive class-names. In the first column I place instances
in which a verb (finite or infinitive) or a verbal substantive is found,
in the second instances without such a form.

In 1 and 10 the nexus forms a complete sentence, in all the
other instances it forms only part of a sentence, either the subject,
the object or a subjunct.

Appendix to Chapter IX.
Copula. Predicative.

This may be the proper place to insert a few remarks on what is often
termed the copula, i.e. the verb is as the sign of a completed combination
(nexus) of two ideas which stand in the relation of subject and predicate.
Logicians are fond of analyzing all sentences into the three elements, subject,
copula, and predicate; the man walks is taken to contain the subject the man,
the copula is, and the predicate walking. A linguist must find this analysis
unsatisfactory, not only from the point of view of English grammar, where
is walking means something different from walks, but also from a general
point of view. The analysis presents some difficulties when the present
tense is not used; the man walked cannot be dissolved into anything containing
the form is, but only into the man was walking — but then logicians
move always in the present of eternal truths! The copula is so far from
being the typical verb, that many languages have never developed any
copula at all, and others dispense with it in many cases, as we have seen
above. The verb be has become what it is through a long process of wearing
down a more concrete signification (‘grow’); it took a predicative in
exactly the same way as many other verbs with a fuller signification still
do: he grows old | goes mad | the dream will come true | my blood runs
cold | he fell silent | he looks healthy | it looms large [it seems important |
she blushed red | it tastes delicious | this sounds correct, etc. It may be
remarked also that a predicative is found not only after verbs, but also after
some particles, in English especially for, to, into, as: I take it for granted |
you will be hanged for a pirate (Defoe) | ho set himself down for an ass | he
took her to wife (obsolete) | she grew into a tall, handsome girl | I look upon
131him as a fool, etc. This is particularly interesting in the combinations
mentioned above (p. 124): with his brother as protector | the Committee,
with the Bishop and the Mayor for its presidents, had already held.several
meetings. Similarly in other languages: Goth, ei tawidedeina ina du þiudana
‘that they might make him (to) king’ | G. das wasser wurde zu wein |
Dan. blive til nar, holde een for nar. Note the nominative in G. Was für
ein mensch, so also in Dutch wat voor een and Russian after čto za (cf. Shakespeare's
What is he for a foole?). It is interesting that in this way the
preposition for may govern an adjective (participle), which is not otherwise
possible: I gave myself over for lost; cp. Lat. sublatus pro occiso | quum
pro damnato mortuoque esset | pro certo habere aliquid; It. Giovanni non
si diede por vinto; Fr. Ainsi vous n'êtes pas assassiné, car pur volé nous
savons que vous l'êtes. — The parallel with a predicative after a verb is also
seen in the E. rules for the use of the indefinite article, which are the same
in both cases: in his capacity as a Bishop | in his capacity as Bishop of
Durham.132

Chapter X
Nexus-Substantives. Final Words
on Nexus

“Abstracts.” Infinitives and Gerunds. Final Words on Nexus.

“Abstracts.”

Those who define substantives as names of substances or things
encounter difficulties with such words as beauty, wisdom, whiteness,
which evidently are substantives and in all languages are treated
as such, yet cannot be said to be names of substances or things.
On the strength of this consideration it is habitual to distinguish
two classes of substantives, concrete and abstract. The former are
also called reality nouns (dingnamen, substanzbezeichnende substantiva),
they comprise names of persons and of “objects,”to
which are also reckoned such more or less “intangible” phenomena
as sound, echo, poem, lightning, month, etc. “Abstracts” are also
called thought-names (begriffsnamen, verdinglichungen). The distinction
of the two classes seems easy enough, for we hardly ever
hesitate to which class we are to assign any given noun; yet it
is by no means easy to find a satisfactory definition of “abstract
substantives.”

Let us first look at the question as treated by a distinguished
logician.

J. N. Keynes (FL, p. 16) expands the definition that a concrete
name is the name of a thing, whilst an abstract name is the name
of an attribute, by saying that “a concrete name is the name of
anything which is regarded as possessing attributes, i.e. as a subject
of attributes; while an abstract name is the name of anything which
is regarded as an attribute of something else, i.e. as an attribute of
subjects.” But on p. 18 he mentions that attributes may themselves
be the subjects of attributes, as in the sentence “unpunctuality is
irritating,” and says that “Unpunctuality, therefore, although
primarily an abstract name, can also be used in such a way that it
is, according to our definition, concrete.” But when “names which
are primarily formed as abstracts and continue to be used as such
are apt also to be used as concretes, that is to say, they are names
of attributes which can themselves be regarded as possessing
133attributes,” Keynes has to admit that “this result is paradoxical”
He sees two ways of avoiding this difficulty, but rejects the first
as logically of no value. This consists in defining an abstract name
as the name of anything which can be regarded as an attribute of
something else, and a concrete name as the name of that which
cannot be regarded as an attribute of something else. He therefore
prefers the second way out, that is, he gives up for logical purposes
the distinction between concrete and abstract names, and substitutes
for it a distinction between the concrete and the abstract
use of names, adding that “as logicians we have very little to do
with the abstract use of names,” for “when a name appears either
as the subject or as the predicate of a non-verbal proposition 165
its use is always concrete.”

This is really tantamount to brushing away the whole distinction,
and yet there is no denying that such a word as hardness is on a
different plane altogether from stone, etc. I think Dr. Keynes's
result has been arrived at on account of the unhappy term
“abstract” and especially of its contrast “concrete,” because
these words in ordinary language are often applied to differences
which have no connexion with the distinction occupying us here.
This is seen with particular clearness in V. Dahlerup's article
“Abstrakter og konkreter” (Dania 10. 65 ff.), in which he says
that the distinction between abstract and concrete is a relative
one and applies not only to substantives, but to all other word-classes
as well. Hard is concrete in “a hard stone,” but abstract
in “hard work,” towards in concrete in “he moved towards the
town,” but abstract in “his behaviour towards her,” turn is concrete
in “he turned round,” but abstract in “he turned pale,” etc.
This usage, according to which “concrete” stands chiefly for what
is found in the exterior world as something palpable, space-filling,
perceptible to the senses, and “abstract” refers to something only
found in the mind, evidently agrees with popular language, but
it does not assist us in understanding what is peculiar to such
words as “whiteness” in contradistinction to other substantives.

W. Hazlitt (New and Improved Grammar, 1810, Preface viii)
says: “a substantive is neither the name of a thing, nor the name
of a substance, but the name of a substance or of any other thing
or idea, considered as it is in itself, or as a distinct individual.
That is, it is not the name of a thing really subsisting by itself
(according to the old definition), but of a thing considered as subsisting
by itself. So if we speak of white as a circumstance or
quality of snow, it is an adjective; but if we abstract the idea of
134white from the substance to which it belongs, and consider this
colour as it really is in itself, or as a distinct subject of discourse,
it then becomes a substantive, as in the sentence, White or whiteness
is hurtful to the sight.”

Essentially the same idea is found in many recent writers,
who define substantives like “whiteness” with slight variations
as “fictitiously substantival words,” “names of only imaginary
substances,” “vorstellungen, welche als selbständige gegenstände
gedacht werden,” “gegenstandlich gedachte begriffe,” etc., “mere
names, thought of, and consequently grammatically treated as if
they were independent things” (Noreen VS 5. 256 f. 166). In spite
of this consensus I must confess that when I speak of a young girl's
beauty or of an old man's wisdom, I do not think of these qualities
as “things” or “real objects”; these are to me only other ways
of expressing the thought that she is beautiful and he is wise.
When Wundt says that humanity (menschlichkeit) denotes a quality
just as much as human does, he is perfectly right, but not so when
he adds that the substantival form makes it easier to treat this
quality in our thoughts as an object (gegenstand). Misteli avoids
this fiction and lays stress exclusively on the grammatical treatment,
but no one really explains how and why all languages come
to have such substantives for adjectival notions.

Sweet long before Wundt and Misteli had expressed similar
ideas (1876, CP 18, cf. NEG §80, 99): “The change of white into
whiteness is a purely formal device to enable us to place an attribute-word
as the subject of a proposition… Whiteness is correctly
described as an “abstract” name, as signifying an attribute without
reference to the things that possess the attribute. White, however,
is held to be connotative… The truth is, of course, that white is as
much an abstract name as whiteness is, the two being absolutely
identical in meaning.” To Sweet, therefore, “the only satisfactory
definition of a part of speech must be a purely formal one:
snow for instance, is not a noun because it stands for a thing, but
because it can stand as the subject of a proposition, because it
can form its plural by adding s, because it has a definite prefix [i.e.
the definite article], etc., and whiteness is a noun for precisely
the same reasons.” 267

Sweet is right in saying that white and whiteness are equally
abstract (in the sense ‘separated from individual things’), but not
in maintaining that the two are absolutely identical in meaning.
135The difference may be slight, but it is nevertheless a real one, else
why should all nations have separate words for the two ideas?
Observe that we use different verbs in the two cases: being white
= having whiteness; the minister is (becomes) wise, he possesses
(acquires) wisdom. In Ido Couturat ingeniously created the
ending -eso for these nouns, which is the root of the verb es-ar ‘to
be’ with the substantive ending -o: blind-es-o ‘the being blind,’
i.e. ‘blindness,’ superbeso ‘pride,’ etc. Here we might perhaps
say that the idea of ‘being’ is smuggled into the word, exactly
as our linguistic habits incline us to smuggle a (neither expressed
nor necessary) ‘is’ into such Russian sentences as dom nov ‘the
house (is) new’; but Couturat rightly perceived the cardinal truth
that in such substantives the adjectival element enters as a predicative.
This then is what is really characteristic of these formations:
they are predicative-substantives. 168

There is evidently great similarity between the substantives
here considered, which are formed from adjectives, and verbal
substantives (nouns of action, nomina actionis) like coining, arrival,
movement, change, existence, repose, sleep, love, etc. 269 But the examples
show that the name “noun of action” is not adequate, unless we
count such states as rest and sleep as actions. My own view has
already been indicated: starting from the fact that “I saw the
Doctor's arrival” = “I saw the Doctor arrive, I saw that the
Doctor arrived” and that “I doubt the Doctor's cleverness”
= “I doubt that the Doctor is clever” we have to recognize a
separate class of words which we shall term nexus-substantives
and subdivide into verbal nexus-words (arrival) and predicative
nexus-words (cleverness).

The task then remains of investigating the use of this class, or
the purpose for which these words are employed in actual speech.
So far as I can see, their use lies in the power they afford us of
avoiding many clumsy expressions, because subordinate clauses
would otherwise be necessary to render the same idea. Try, for
instance, to express without the italicized substantives the following
passage from a recent novel: “His display of anger was equivalent
to an admission of belief in the other's boasted power of divination.”136

The value of this power of creating handy expressions for
complex thoughts is greatly increased by the fact that when a
verb or a predicative is thus raised into a substantive, subordinate
members are also in consequence raised to a higher plane: tertiary
members are made secondary, and quaternary, tertiary. In other
words, subjuncts become adjuncts, and sub-subjuncts become
subjuncts, and we are able to construct sentences with a facility
which more than makes up for the concomitant change of a primary
member (the subject or object) into a secondary member (an
adjunct, “subjective” or “objective” genitive).

This must be illustrated by a few examples. “The Doctor's
extremely quick arrival and uncommonly careful examination of
the patient brought about her very speedy recovery” — if we compare
this with the sentences “the Doctor arrived extremely quickly and
examined the patient uncommonly carefully; she recovered very
speedily,” we shall see that (giving the rank of the word in Roman
numbers) the verbs arrived, examined, recovered (II) have been
turned into the substantives arrival, examination, recovery (I), the
subjuncts (adverbs) quickly, carefully, speedily (III) have become
the adjuncts (adjectives) quick, careful, speedy (II), while the change
from sub-subjuncts (IV) into subjuncts (III) has entailed no formal
change in extremely, uncommonly, very. On the other hand, the
primary words (subject and object) the Doctor, the patient, she (I)
have been turned into the secondary members (adjuncts) the Doctor's,
of the patient, her (II).

Similar shiftings are observed in the sentence “we noticed the
Doctor's (II) really (III) astonishing (II) cleverness (I),” as compared
with “the Doctor (I) was really (IV) astonishingly (III) clever (II).”
(If really is here referred to the verb was, it has the rank III.)

Predicative-nouns are also very handy in the frequent combinations
in which they are made the object of the preposition with,
as they enable us to get rid of long-winded subjunct combinations:
“He worked with positively surprising rapidity” (instead of
“positively surprisingly rapidly”), “with absolute freedom,”
“with approximate accuracy,” etc. Cf. the shiftings mentioned
above, p. 91.

We are now in a position to get a clearer view of a grammatical
phenomenon which is generally termed “the cognate object.” 170
137Its purpose cannot be fully understood if we start from such
examples as “I dreamed a dream” (Onions, AS 35) or “servitutem
servire,” for such combinations are, to say the least, extremely
rare in actual speech, for the simple reason that such an object is
inane and adds nothing to the verbal notion. In actual speech we
meet with such sentences as: I would faine dye a dry death (Sh.) | I
never saw a man die a violent death (Ruskin) | she smiled a little
smile and bowed a little bow (Trollope) | Mowgli laughed a little
short ugly laugh (Kipling) | he laughed his usual careless laugh
(Locke) | he lived the life, and died the death of a Christian
(Cowper), etc.

These examples make it clear that the nexus-substantive is
simply introduced to give us an easy means of adding some descriptive
trait in the form of an adjunct which it would be difficult or
impossible to tack on to the verb in the form of a subjunct (cf. also
“fight the good fight,” which is different from “fight well”).
Sometimes this extra description is added as a kind of “appositum,”
marked off by means of a comma or dash, as in: The dog sighed,
the insincere and pity-seeking sigh of a spoilt animal (Bennett) J
Kitty laughed — a laugh musical but malicious (Mrs. H. Ward). We
see the same device employed in other cases, where some special
addition to a secondary word cannot conveniently be expressed by
means of a subjunct; a predicative-word is consequently loosely
attached to the sentence as the bearer of the specialization in the
form of an adjunct, thus in: her face was very pale, a greyish pallor
(Mrs. Ward) | he had been too proud to ask — the terrible pride of
the benefactor (Bennett). Not infrequently the addition is introduced
by the preposition with: she was pretty, with the prettiness
of twenty | I am sick with a sickness more than of body, a sickness
of mind and my own shame (Carlyle).

If I add that nexus-substantives are also often convenient
in cases where idiomatic usage does not allow a dependent clause,
as after upon in “Close upon his resignation followed his last
illness and death,” I hope I have accounted sufficiently for the
rôle played in the economy of speech by these formations. 171 But
like most good things in this world substantives of this type can
138be abused. This is well brought out in an interesting paper by
Hermann Jacobi on the Nominal Style in Sanskrit (IF 14. 236 ff.).
When languages begin to grow old (alternde sprachen!!) they
tend, he says, to nominal expressions, especially when they have
for a long time served as vehicles for scientific thinking. It seems
possible to express ideas with greater precision and adequacy by
means of nouns than by means of the more pictorial verbs (die
mehr der sphäre der anschauung sich nähernden verba). “Sanskrit
had become the privileged vehicle for the higher education in
India; it had become unintelligible to the lower classes of the
people and had ceased to be used for all purposes of human life.
While Sanskrit was increasingly diverted from the practical details
of everyday life and was simultaneously used more and more to
serve the interests of the higher life of the intellect, abstract methods
of diction were more and more needed as the sphere of ideas to be
expressed became narrower and narrower,” and that led naturally
to the preference for substantives, i.e. our nexus-substantives.

I think the difference between the two kinds of style can be
illustrated by comparing my English translation of the last sentence
with the German original: “Mit der zunehmenden abkehr von der
gemeinen alltäglichkeit des daseins und der damit hand in hand
gehenden zuwendung zum höheren geistigen leben stieg in dem sich
also einengenden ideenkreise, welchem das Sanskrit als ausdrucksmittel
diente, das bedürfnis begrifflicher darstellung.” German
scientific prose sometimes approaches the Sanskrit style described
by Jacobi. When we express by means of nouns what is generally
expressed by finite verbs, our language becomes not only more
abstract, but more abstruse, owing among other things to the
fact that in the verbal substantive some of the life-giving elements
of the verb (time, mood, person) disappear. While the nominal
style may therefore serve the purposes of philosophy, where, however,
it now and then does nothing but disguise simple thoughts
in the garb of profound wisdom, it does not lend itself so well to
the purposes of everyday life.

Infinitives and Gerunds.

It is interesting to note in the history of language how verbal
substantives sometimes tend to discard some of the characteristics
of substantives and to assume some of those verbal characteristics
which were above alluded to as “life-giving,” or in other words
how speakers have here and there treated them as they were accustomed
to treat finite verbs.

This is the case with our infinitives, which are now universally
admitted to be fossilized case-forms of old verbal substantives
139They have approached the finite verbs morphologically and syntactically,
though not to the same extent in all languages: they
can take their object in the same case as the ordinary verb (accusative,
dative, etc.), they admit the usual combinations with negatives
and other subjuncts, they develop tense-distinctions (perfect
infinitive like Lat. amavisse, E. to have loved, in some languages also
future infinitive), and the distinction between active and passive
(the latter in Lat. amari, E. to be loved, etc.). All these traits are
alien to such words as movement, construction, or belief. A further
assimilation of the infinitive to finite verbs is seen in those
languages which admit of its being combined with a subject in the
nominative; see p. 119.

In some languages the infinitive can be used with the definite
article. This substantival trait has the advantage that the case-form
of the article shows the function of the infinitive in the sentence.
Where this can be applied to a combination like the Greek
accusative with the infinitive, it is of greater value than where it is
only the “naked” infinitive that can take the article, as in German. 172

A development corresponding to what we have here observed
in the infinitive is found in some other verbal substantives. An
object in the accusative is seen in rare cases in Sanskrit, Greek and
Latin as in the often-quoted Plautine sentence “Quid tibi hanc
curatiost rem?” (Delbrück, Synt. 1. 386). In some Slavic languages,
for instance Bulgarian, it has become quite a common thing to add
an object in the accusative to the verbal substantive in -anije and
corresponding endings. In Danish the verbal substantive in -en can
take an object, though only if verb and object enter into a close
semantic union which is shown by unity-stress on the latter:
denne skiften tilstand, tagen del i lykken, etc., examples in my
Fonetik, 565.

The most interesting case in point is the English form in -ing,
where we witness a long historical development by which what was
originally a pure substantive formed only from some particular
verbs comes to be formed from any verb and acquires more and more
of the characteristics of the finite verb (GS § 197 ff.). It can
take an object in the accusative (on seeing him) and an adverb
(he proposed our immediately drinking a bottle together), it develops a
perfect (happy in having found a friend) and a passive (for fear of
140being killed). As for the subject, which originally had always to be
put in the genitive and is still often found so, it is now often put in
the common case (he insisted on the Chamber carrying out his policy |
without one blow being struck) and may even exceptionally in colloquial
speech be put in the nominative (Instead of he converting the Zulus,
the Zulu chief converted him, with strong stress on he). When an
Englishman now says “There is some possibility of the place having
never been inspected by the police,” he deviates in four grammatical
points from the construction that would have been possible to one
of his ancestors six hundred years ago (common case, perfect,
passive, adverb).

Here we may mention also the Latin Gerund. The development
of this form is rather interesting. Latin had a passive participle
in -ndus (the “gerundive”) which might be used in the same
way as other participles and adjectives so as to imply a nexus
(cf. above, p. 125), thus in “elegantia augetur legendis oratoribus
et poetis,” ‘elegance is increased through read orators and poets,’
i.e. through the fact that they are read, through reading them.
By the side of cupiditas libri legendi, which is to be interpreted in
this way, it became possible to say cupiditas legendi without any
substantive as primary; this further led to legendi being felt as a
kind of genitive of the infinitive and admitting an object in the
accusative. Thus was created what is now given as a separate
form of the verbs, inflected in the various cases (except the
nominative) of the singular like an ordinary neuter substantive
and termed the “gerund” (see, e.g., Sommer, Handb. d. lat.
laut- u. formenlehre 631). The original and the derived constructions
are found side by side in Caesar's “neque consilii
habendi neque arma capiendi spatio dato.” 173

Final Words on Nexus.

As I have emphasized the existence of two no Sons in a nexus
(as opposed to junctions, where the two members together formed
one notion), the reader may be surprised to find that I am here
putting the question whether it is not possible to have a nexus
consisting of only one member, and still more to find that I am
answering that question in the affirmative. We do find cases in
which we have either a primary alone or a secondary alone, and
which nevertheless offer so close an analogy to an ordinary nexus
that it is impossible to separate them from undoubted instances
of nexus. But an accurate analysis will show that the usual two
141members are everywhere present to the mind, and that it is only
in the linguistic expression that one of them may now and then
be absent.

First we may have a primary alone or, in other words, a nexus
without an adnex. This is seen in such an English sentence as
(Did they run?) Yes, I made them: this means the same thing as
I made them run, and thus, however paradoxical it may sound, it is
an accusative-with-infinitive without the infinitive; them implies
a real nexus and is different from the object in (Who made these
frames?) I made them. In the same way in colloquial English we
may have an isolated to standing as a representative of an infinitive
with to: I told them to (= I told them to run). Psychologically
these are cases of aposiopesis (‘stop-short sentences’ or ‘pull-up
sentences,’ as I have called them, Language, 251): the infinitive
is left out as in (Will you play?) Yes, I will, or Yes, I am going to
(I am willing to, anxious to).

Next we have the secondary part of a nexus alone, without
any primary. This is extremely frequent in exclamations, where
it is not necessary to tell the hearer what one is speaking about;
they form complete pieces of communication and should unhesitatingly
be termed “sentences.” Thus, for instance, Beautiful |
How nice! | What an extraordinary piece of good luck! These are
really predicatives, cf. This is beautiful, etc.: the predicative
comes first to the mind of the speaker; if afterwards he thinks of
adding the subject, the result is a sentence of the form considered
above, p. 121: Beautiful this view! Or he may choose another
form by adding a question: Beautiful, isn't it? (just as in This view
is beautiful, isn't it? 174)

I think we may speak also of a nexus with the primary unexpressed
in all those cases in which a finite verbal form is sufficient
in itself without a noun or pronoun as subject, e.g. Lat. dico, dicis,
dicunt, etc. In many cases a verb in the third person in various
languages is expressive of the “generic person” (Fr. on); see the
interesting collections by H. Pedersen and J. Zubatý in KZ 40.
134 and 478 ff.

In our modern languages, the subject must generally be
expressed, and those few cases in which it is omitted, may
be explained through prosiopesis, which sometimes becomes
habitual in certain stock exclamations like Thank you | G. danke |
G. bitte | Bless you | Confound it! Cf. also Hope I'm not boring
you.142

In all the cases so far considered a one-member nexus has
been an independent sentence. It may also be merely a part of a
sentence. There is no primary in the nexus which forms the object
of makes in the E. proverb “practice makes perfect,” i.e. makes
one perfect; this is very frequent in Danish, e.g. “penge alene
gør ikke lykkelig” (money alone does not make [a man] happy) | jeg
skal gøre opmærksom på at…, G. ich mache darauf aufmerksam,
dass…

An accusative-with-infinitive without the accusative is not at
all rare, e.g. live and let live | make believe | I have heard say | Lat
see now who shal telle another tale (Chaucer; this is obsolete).
In Dan. frequent: han lod lyse til brylluppet | jeg har hørt sige at…,
etc. Thus also in German and Fr. The unexpressed primary
is the ‘generic person.’ In G. ich bitte zu bedenken it may be the
second person.

Nor are these the only instances in which the primary of a nexus
is left unexpressed, for in the great majority of cases in which we
use either an infinitive or a nexus-substantive there is no necessity
expressly to indicate who or what is the subject of the nexus. This
may be either definite, as shown by the actual context, as in:
I like to travel, or I like travelling (the unexpressed primary is I) | it
amused her to tease him (the primary is she) | he found happiness
in activity and temperance (the primary is he), etc. Or else it may
be the indefinite ‘generic person’ (Fr. on): to travel (travelling) is
not easy nowadays | activity leads to happiness | poverty is no
disgrace, etc. That the primary, though not expressed, is present
to the mind is shown by the possibility of using a “reflexive”
pronoun, i.e. one indicating identity of subject and object, etc.,
with infinitives and nexus-substantives: to deceive oneself |
control of oneself (self-control) | contentment with oneself | Dan.
at elske sin næste som sig selv er vanskeligt | glæde over sit
eget hjem | G. sich mitzuteilen ist natur | Lat. contentum rebus
suis esse maximæ sunt divitiæ (Cic.), and similarly in other
languages.

I think that by laying stress on the notion of nexus and the
inherent necessity of a “primary” or subject-part I have attained
a better understanding of “abstracts,” of “nomina actionis,”
and of infinitives, and especially of the rôle these forms play in
the economy of speech than by the usual definitions. Nothing is
really gained by defining the infinitive as “that form of a verb
which expresses simply the notion of the verb without predicating
it of any subject” (NED) or as “the form that expresses the notion
of a verb in general without indicating it as predicated of any definite
subject, with which it might form a sentence” (Madvig) — to which
it might be objected that as a matter of fact there is very often a
143definite subject, sometimes expressed and sometimes to be gathered
from the context, and that on the other hand the subject of a finite
verb is very often just as indefinite as that of an isolated infinitive.
I venture to hope that the reader will find that the numerous
phenomena brought together in this and the preceding chapter
throw so much light on one another that it warrants my grouping
of these constructions in a separate class, for which the term
“nexus” may not be found inappropriate.144

Chapter XI
Subject and Predicate

Various Definitions. Psychological and Logical Subject. Grammatical
Subject. There is.

Various Definitions.

The discussion of the two members of a nexus has already to
some extent anticipated the question of the relation of subject
and predicate, for in those nexuses which constitute complete
sentences, the “primary” has been shown to be identical with
the subject, and the adnex (secondary member) identical with
the predicate; in other forms of nexus, we might also use the
terms “subject-part” and “predicate-part” instead of “primary”
and “adnex.”

We have now to discuss various definitions given of the terms
“subject” and “predicate” by previous writers, who have not
as a rule taken into consideration anything but “sentences” or
even the more restricted class called “judgments.” An exhaustive
critical examination of everything that has been said by grammarians
and logicians on this question would require a whole
volume, but I hope the following remarks will be found comprehensive
enough.

The subject is sometimes said to be the relatively familiar
element, to which the predicate is added as something new.
“The utterer throws into his subject all that he knows the
receiver is already willing to grant him, and to this he adds in
the predicate what constitutes the new information to be conveyed
by the sentence … In ‘A is B’ we say, ‘I know that you
know who A is, perhaps you don't know also know that he is the
same person as B’” (Baldwin's Dict. of Philosophy and Psychol.
1902, vol. 2. 364). This may be true of most sentences, but not
of all, for if in answer to the question “Who said that?” we
say “Peter said it,” Peter is the new element, and yet it is undoubtedly
the subject. The “new information” is not always
contained in the predicate, but it is always inherent in the connexion
of the two elements, — in the fact that these two elements
are put together, i.e. in the “nexus,” cf. what was said about the
difference between junction and nexus on pp. 114-117.145

Others say that the rôle of the predicate is to specify or determine
what was at the outset indefinite and indeterminate, that
the subject is thus a determinandum which only by means of the
predicate becomes a determinatum (Keynes FL 96, Noreen VS
6. 153, Stout AP 2. 213). But this description is far more
true of an adjunct as blushing in the blushing girl than of blushes
in the girl blushes. What is here made determinate is not the girl
but the whole situation.

Another definition that is frequently given is that the subject
is what you talk about, and the predicate is what is said about
this subject. This is true about many, perhaps most, sentences,
though the man in the street would probably be inclined to say
that it does not help him very much, for in such a sentence as
“John promised Mary a gold ring” he would say that there are
four things of which something is said, and which might therefore
all of them be said to be “subjects,” namely (1) John, (2) a
promise, (3) Mary, and (4) a ring. This popular definition, according
to which subject is identified with subject-matter or topic, is
really unsatisfactory, as may perhaps be best appreciated if we
see where it leads a distinguished psychologist like Stout, who
in a famous passage (AP 2. 212 ff.) starts from it and then lands
us at a point which is admittedly very far from the grammarian's
conception of subject and predicate: “The predicate of a sentence
is the determination of what was previously indeterminate.
The subject is the previous qualification of the general topic to
which the new qualification is attached. The subject is that
product of previous thinking which forms the immediate basis
and starting-point of further development. The further development
is the predicate. Sentences are in the process of thinking
what steps are in the process of walking. The foot on which the
weight of the body rests corresponds to the subject. The foot
which is moved forward in order to occupy new ground corresponds
to the predicate.… All answers to questions are, as
such, predicates, and all predicates may be regarded as answers
to possible questions. If the statement “I am hungry” be a
reply to the question, “Who is hungry?” then “I” is the predicate.
If it be an answer to the question, “Is there anything
amiss with you?” then “hungry” is the predicate. If the question
is, “Are you really hungry?” then “am” is the predicate.
Every fresh step in a train of thought may be regarded as an
answer to a question. The subject is, so to speak, the formulation
of the question; the predication is the answer.”

If this is the logical consequence of the popular definition of
‘subject,’ then the grammarian cannot use that definition, for it
does not assist him in the least. It is, indeed, unfortunate that
146the grammarian has to use the word “subject,” which in ordinary
language means, among other things, also ‘topic’ (‘subject-matter’).

Psychological and Logical Subject

The confusion arising from the ambiguity of the word “subject”
is also responsible for much of what linguists and logicians
have written on the so-called psychological and logical subject and
predicate. As a matter of fact, these terms are by various
writers used of totally different concepts, as will be seen
from the following survey, which is probably not by any means
exhaustive.

(1) Sequence in time. Thus G. v. d. Gabelentz (Zeitschr. f.
völkerpsychologie u. sprachwissensch. VI and VIII and shorter in
Spr. 348 ff.): the hearer first apprehends a word A and asks full of
expectation: What about this A? Then he receives the next word
or idea B, adds together these two and asks: Now, what about
this (A + B)? The answer is the next idea C, and so forth. Each
successive word is the predicate of the subject contained in what
he has already heard. It is as with the two rolls of paper in a
telegraphic apparatus, on the one side there is the roll filled with
writing, which is continually expanding, on the other side the
blank roll, which is continually gliding over and swelling the other.
The speaker knows beforehand both what is contained in one
roll and what is to fill the empty paper. What now makes him
mention A first, and then B, etc.? Evidently he will place first
what makes him think: his ‘psychological subject,’ and next
what he thinks about it; his ‘psychological predicate’; after
that both together may be made the subject of further thinking
and speech. (Similarly, Mauthner, Kritik der sprache, 3. 217 ff.)

This is interesting, and Gabelentz's clever analysis from this
point of view of the sentence “Habemus senatusconsultum in te
vehemens et grave” might be quoted in any study of the psychological
effect of word-order; but the analogy between this and
the subject-predicate relation is far too loose for the same name
to be applied to both. Wegener's name “exposition” for what
Gabelentz calls psychological subject is much more to the point.
But it should always be remembered that word-order in actual
language is not exclusively determined by psychological reasons,
but is often purely conventional and determined by idiomatic
rules peculiar to the language in question and independent of the
will of the individual speaker.

(2) Novelty and importance. Paul (Gr. 3. 12) seems first to
agree with Gabelentz when defining the psychological subject as
the idea or group of ideas that is first present in the mind of the
147speaker, and the psychological predicate as what is then joined
(neu angeknüpft) to it. But he neutralizes that definition when
he adds that even if the subject-idea is the first in the mind of the
speaker, it is sometimes placed later, because in the moment when
he begins to speak, the predicate-idea presses for utterance as
the new and more important one, especially under the influence
of strong emotion. In his former work (P 283) he says that the
psychological predicate is the most important element, that which
it is the aim of the sentence to communicate and which therefore
carries the strongest tone. If in “Karl fahrt morgen nach Berlin”
everything is equally new to the hearer, then Karl is the
subject to which the predicate fahrt is added; to the latter as
subject comes as a first predicate morgen, and as a second predicate
nach Berlin. If on the other hand the hearer knows about
Karl's trip to-morrow but is ignorant of his destination, then
nach Berlin is the predicate; if he knows that he is going to Berlin,
but does not know when, then morgen is the predicate, etc. Paul
even goes so far as to say that if the only thing he is ignorant of
is the manner of getting there (whether on horseback, or in a
carriage, or on foot), then fährt “ist gewissermassen in zwei
bestandteile zu zerlegen, ein allgemeines verbum der bewegung
und eine bestimmung dazu, welche die art der bewegung bezeichnet,
und nur die letzere ist prädikat.” It would be difficult to imagine
greater or more unnecessary subtlety. Why not avoid the terms
subject and predicate in this sense and simply say that what is
new to the hearer in any piece of communication may be found
according to circumstances in any part of the sentence?

(3) Stress (or tone). This view is hardly to be kept distinct
from the former. Høffding (Den menneskelige tanke, 88) says that
the logical predicate is often the grammatical subject or an adjective
belonging to it: “You are the man” | “All the guests have
arrived.” It is recognized everywhere by the stress: “The king
will not come” | “He has gone.” In sentences of descriptive
contents nearly every word may express a logical predicate because
it may receive stress as containing new information. What is
here termed logical predicate is nearly identical with what Paul
calls psychological predicate, but it would be better to recognize
that it has very little to do with logic proper: in the same writer's
textbook of formal logic he continually uses the words subject
and predicate, for instance in the rules he gives for syllogisms,
but there the words will be always found to be taken not in their
logical, but in their grammatical signification without any regard
to stress. As this is generally determined less by strictly logical
considerations than by emotion (the interest felt in an idea or the
value ascribed to it at the moment), Bloomfield (SL 114) rightly
148prefers the term the emotionally dominant element 175 for what Paul
calls the logical and Høffding the psychological predicate.

(4) Any primary word in a sentence is the logical subject.
Thus according to Couturat (Revue de Métaphysique, Janvier 1912,
5) in the sentence “Pierre donne un livre à Paul,” which means
the same thing as “Paul reçoit un livre de Pierre,” the three words
Pierre, livre, Paul (by him called termes) are all of them “les sujets
du verbe qui exprime leur relation.”

(5) “In guter vater ist gut, logisch betrachtet, eben so wohl
prädicat zum subject vater, wie in der vater ist gut; in einen brief
schreiben, schon schrieben, hat, logisch genommen, das subject
schreiben sein prädicat einen brief, schön” (Steinthal, Charakteristik
101).

(6) Wegener (U 138) analyzes the G. verb satteln as consisting
of sattel + the suffix which makes it into a verb, and says that
the two elements are respectively the logical predicate (sattel)
and the logical subject (-n).

(7) Sweet (NEG, p. 48) says that in a sentence like “I came
home yesterday morning” the word came by itself is the grammatical
predicate, but came-home-yesterday-morning the logical
predicate. And in another place (HL 49) he says that in gold is
a metal, the strictly grammatical predicate is is, but the logical
predicate is metal.

(8) Many grammarians use the term “logical subject” for
that part of a passive sentence which would be the subject if the
same idea had been expressed in the active turn, thus his father
in “he was loved by his father” (called ‘converted subject’ below,
Ch. XII).

(9) Others will say that in “It is difficult to find one's way in
London,” “it cannot be denied that Newton was a genius,” it is the
formal subject, and the infinitive or the clause the logical subject.

(10) Still other grammarians will say that in such a “subjectless”
sentence as G. mich friert the logical subject is “I.” 276

(11) A final use of the same term (closely related to 10) is seen
when the transition from the old construction “Me dreamed a
strange dream” to the modern “I dreamed a strange dream”
is described by saying that the psychological (or logical) subject
has become also the grammatical subject.

It is no wonder that after all this purposeless talking about
logical and psychological subjects some writers have tried to avoid
149the term subject altogether. Thus Schuchardt (Br 243) would
substitute the word agens, but that does not seem appropriate in
he suffers, he broke his leg, etc., and in A loves B we should rather
say that B acts on A than inversely. The only two linguists, so
far as I know, who have seriously tried to dispense with the term
subject in their grammatical analysis are the Swedes Svedelius
and Noreen. Nothing, however, is gained by this. It is much
better to retain the traditional terms, but to restrict them to
domains where everybody knows what they import, i.e. to use
subject and predicate exclusively in the sense of grammatical
subject and predicate, and to discountenance any proposals to
attach to these words the adjuncts ‘logical’ and ‘psychological.’

Grammatical Subject.

Clearly to understand what the word subject means in its
grammatical application, it will be well to recur to what was said
in the chapter on the three ranks. In every sentence there are
some elements (secondary words) which are comparatively fluid
or liquid, and others (primary words) that are more firmly fixed
and resemble rocks rising out of the sea. The subject is always
a primary, though not necessarily the only primary in the sentence;
this amounts to saying that the subject is comparatively definite
and special, while the predicate is less definite, and thus applicable
to a greater number of things.

Doubt as to which word is the subject may sometimes arise
when the colourless verb be is followed by a predicative, 177 though
even here there is generally no difficulty in seeing which is the
subject if we keep in mind what has been said about the more
specialized nature of a subject as contrasted with a predicate.

After the results attained by our inquiry in Chapter V we are
prepared to find that adjectives are extremely frequent as predicatives,
because they are less special than substantives and applicable
to a greater number of different things; thus in my father
is old | the dress was blue, no one doubts that the words printed
in italics are the subjects, and the two adjectives the predicatives.

Where two substantives are connected by means of is, we
can formulate some rules in accordance with our principle.

It is worth noticing how the value of the indefinite article
shifts automatically. Take a conversation like the following:
A says: “The sailor shot an albatross,” i.e. one individual of
that species. B asks: “What is an albatross?” The question
is not about that one albatross, but about the whole species, and
accordingly A's reply “An albatross is a big sea-bird” relates
to the whole species, and says that all albatrosses belong to the
wider class of sea-birds.

In the former the words my brother are more definite (my only
brother, or the brother whom we are talking about) than in the
second (one of my brothers, or leaving the question open whether
I have more than one). Cf. on the meaning of possessives, p. 110
above.

The question is not very important, and if we look at it from
the point of view here advocated, we may say that one term is
just as special as the other. Yet it seems natural in such cases
to take the proper name as the more special and therefore as the
subject. We see this if we formulate the corresponding questions,
for the neuter what always takes the place of the predicative;
now both sentences are natural answers to either of the questions:
What was Miss C.? and Who was the prettiest girl? 178 but What
was the prettiest girl at the ball? would be a question about something
else. We obtain the same result by noticing that it is possible
to say “I look on Miss C. as the prettiest girl at the ball,” but
not “I look on the prettiest girl at the ball as Miss C.” 279

Where there is perfect identity (coextension) of the two terms
connected by is, they may change places as subject and predicative;
this is what Keats implied in his line: “Beauty is truth;
truth, beauty.” But as we have seen, perfect identity is rare,
and it is important to remark that the linguistic “copula” is
153does not mean or imply identity, but subsumption in the sense
of the old Aristotelian logic, which is thus in closer accordance
with grammar than the so-called logic of identity (Leibniz, Jevons,
Hoffding). According to the latter the sentence “Peter is stupid”
should be analyzed as “Peter is a stupid Peter,” or, as it is also
maintained that the substance of the predicate influences that of
the subject, we obtain perfect identity only by saying “Stupid
Peter is stupid Peter.” In this way, however, the character of
communication from speaker to hearer is lost; by the words “is
stupid Peter” the hearer is told nothing more than he had heard
at the beginning, and the sentence has no value whatever.
Ordinary mortals, therefore, will always prefer the formula “Peter
is stupid,” by which Peter is ranged among those beings (and
things) that can be called “stupid.”

In the mathematical formula A = B we should not take the
sign = as the copula and B as predicative, but insert the copula
is before the predicative equal to B, and thus read it as meaning:
A is comprised among the (possibly several) objects that are equal
to B (whether ‘equal’ connotes only quantitative equality or
perfect identity).

In some idiomatic uses we may be inclined to take is as implying
identity, e.g. “to see her is to love her.” “Seeing is believing.”
But the identity is more apparent than real. It would be
impossible to invert the terms, and the logical purport of the
saying is merely this: seeing immediately leads to, or causes,
love, or belief. Thus also; “To raise this question is to answer
it,” etc. 180

There is.

In connexion with what has been said about the subject of
a sentence being more special and more definite than the predicative,
we may mention the disinclination to take as subject a
word with the indefinite article, except when this is meant as
the “generic” article designating the whole species, which is
really a definite idea. Instead of beginning a story in this way:
“A tailor was once living in a small house,” etc., it is much more
natural to begin: “Once upon a time there was a tailor,” etc.
By putting the weak there in the place usually occupied by the
subject we as it were hide away the subject and reduce it to an
inferior position, because it is indefinite.

The word there, which is used to introduce such a sentence,
though spelt in the same way as the local there, has really become
154as different from it as the indefinite is from the definite article;
it has no stress and is generally pronounced with the neutral (mid-mixed)
vowel [ðə] instead of [ðɛˈə]; its indefinite signification is
shown by the possibility of combining it in the same sentence
with the local (stressed) there or with here. It is followed by an
indefinite subject: there was a time when… | there were many
people present | there was no moon | there came a beggar, etc.
The weak there also takes the place of the subject in combinations
like “Let there be light” and “on account of there being no
money in the box.” Cf. also from a modern novel: No other
little girl ever fell in love with you, did there?

The indefiniteness here spoken of is not always formally indicated,
thus those is notionally indefinite in “there are those who
believe it” (= there are some who; sunt qui credunt) and thus
different from the definite those with which we begin a sentence:
“Those who believe it are very stupid.” “In Brown's room
there was the greatest disorder” = a very great disorder, different
from “The greatest disorder was in Brown's room,” i.e. greater
than in the other rooms. Note also the different word-order in
“There [ðə] was found the greatest disorder” and “There [ðɛˈə]
the greatest disorder was found,” though the former sentence may
also be read with stressed there.

Sentences corresponding to English sentences with there is or
there are, in which the existence of something is asserted or denied
— if we want a term for them, we may call them existential sentences —
present some striking peculiarities in many languages.
Whether or not a word like there is used to introduce them, the
verb precedes the subject, and the latter is hardly treated grammatically
like a real subject. In Danish it has the same form as
an object, though the verb is is: der er dem som tror, even with
the passive der gives dem. In Danish the verb was here put
in the singular before a plural word, even at a time when
the distinction between sg. er and pl. ere was generally observed;
in English there is the same tendency to use there's before
plurals, though in the literary language it is not now quite so
strong as it was formerly; in Italian, too, one finds v'è instead
of vi sono.

In Russian the verb ‘is’ is in most other sentences unexpressed,
but in these sentences we have a preposed verb, e.g. byl
mal'čik ‘there was a boy,’ žila vdova ‘there lived a widow.’ The
form jest' ‘there is,’ originally a third person singular, is used even
before a plural word, and even before pronouns of the other persons
(Vondrák SG 2. 267), and finally we may mention the curious
form naěxalo gostej ‘there came driving (neuter sg.) some guests’
(gen. pl., Berneker, Russ. Gramm. 156).155

In Ancient Greek the verb is was not necessarily expressed in
ordinary sentences, but in these sentences we find a preposed esti,
as in II. 3. 45 all' ouk esti biē phresin, oude tis alkē; cf. Meillet
MSL 14. 9.

In German we have the well-known es gibt, which, of course,
precedes the indication of that which is said to exist; this latter
is the object of the verb, though some West German dialects use
it in the nominative and say es geben viele äpfel — Grimm, Wörterbuch
IV, 1. 1704, Paul Gr 3. 28.

Many languages have expressions containing the word ‘has,’
followed by what was originally its object, but is now not always
distinct in form from the subject-case, thus Fr. il y a, Sp. hay
(from ha ‘it has’ y ‘there’), It. v'ha (in v'hanno molti ‘there are
many’ molti is treated as subject), South German es hat, Serbian
and Bulgarian ima, Mod. Gr. ekhei. (Cf. also H. Pedersen, KZ
40. 137.) Chinese has the otherwise invariable rule that the subject
is placed before the verb, but these sentences begin with yeù,
originally ‘have’; see Gabelentz, Chin. Gramm. 144. Finck (KZ
41. 226) transcribes the same word yu³, e.g. yu³ ko lang² ‘there
once was a wolf,’ orig. ‘has piece wolf.’

I may here mention some peculiarities of Finnish grammar.
The nominative is used only with definite subjects, among which
are also reckoned generic expressions; if, on the other hand,
something indefinite is denoted, the partitive is used; cp. thus viini
(nom.) on pöydällä ‘the wine is on the table,’ viini on hyvää ‘wine
(the species) is good,’ viiniä (partitive) on poydällä ‘there is wine
on the table.’ Just as in English and Dan. we do not as a rule
use there, der, when the verb has an object, because this seems
to imply a kind of definiteness, Finnish in such cases has the
nom., even if ‘some’ are implied: varkaat (or jotkut varkaat, nom.)
varastivat tavarani ‘thieves (some thieves) stole my things,’ but
varkaita (part.) tuli talooni ‘there came some thieves into my
house’ (Eliot FG 121 f.).156

Chapter XII
Object. Active and Passive

What is an Object? Object of Result. Subject and Object. Reciprocity.
Two Objects. Adjectives and Adverbs with Objects. Passive. Use
of the Passive. Middle Voice. Active and Passive Adjectives. Active
and Passive Substantives. Nexus-Substantives. Infinitives.

What is an Object?

It is easy enough to see what is the subject of a sentence when
this contains only one primary, as in John slept | the door opened
slowly; and we have seen that in sentences containing two terms
connected by means of is or a similar verb (and also in those sentences
without a verb mentioned in Ch. IX.) the member which is
most special is the subject (primary) and the less special member
the predicative. But many sentences contain two (or three)
primaries: here one is the subject and the other (or the two
others) the object (or objects); thus in John beats Paul | John shows
Paul the way, John is the subject, and Paul and the way are objects.
In sentences containing a verb it is nearly always easy to find
the subject, for it is that primary that has the most immediate
relation to the verb in the form in which the latter actually occurs
in the sentence: this applies to sentences like those just mentioned
as well as to sentences of the form Peter is beaten by John,
where we might according to other definitions feel inclined to
regard John as the subject because he is the agent.

Various definitions have been given of object; the most popular
one is that the object denotes the person or thing on which the
action of the verb is performed. This covers a great many instances,
such as John beats Paul | John frightened the children | John
burns the papers, but it is difficult to apply the definition to countless
other sentences in which, however, grammarians never hesitate
to use the term object, e.g. John burns his fingers (i.e. he suffers
in his fingers from burning) | John suffers pain, etc.

Sweet long ago saw this difficulty and said (CP 25): “With
such verbs as beat, carry, etc., the accusative unmistakably denotes
the object of the action expressed by the verb, but with such
verbs as see, hear, it is clearly a mere metaphor to talk of an
‘object.’ A man cannot be beaten without feeling it, but he
157can be seen without knowing anything about it, and in many
cases there is no action or volition at all involved in seeing. And
in such a sentence as he fears the man, the relations are exactly
reversed, the grammatical nominative being really the object
affected, while the grammatical accusative represents the cause.” 181
Sweet concludes that in many cases the accusative has no meaning
at all — it would be better to say that it has not the meaning
implied in the narrow definition usually given, but varies according
to the infinitely varying meanings of the verbs themselves, as
seen in such instances as: kill the calf | kill time | the picture
represents the king | he represented the University | it represents
the best British tradition | run a risk | run a business | answer a
letter, a question, a person | he answered not a word | pay the
bill | pay six shillings | pay the cabman | I shall miss the train | I
shall miss you | entertain guests | entertain the idea | fill a pipe | fill
an office, etc., etc. (Cf. Spr. L. 83.)

In some cases it may be difficult to tell whether a word is to
be called a predicative or an object. The object can in many
cases be recognized by the possibility of turning it into the subject
of a passive sentence. The object is more closely connected
with the verb of the sentence, and the predicative with the subject
(to which it might under altered circumstances be joined as an
adjunct). Thus it is natural that the predicative adjective in
those languages which inflect it is made to agree with the subject
in number and gender, and that the predicative, whether substantive
or adjective, is in many languages put in the same case
as the subject (nominative). Something between an object and
a predicative is seen in English after make (she will make a good
wife) and in German dialects after geben (see examples in Grimm's
Wörterbuch, 1702: welche nit gern spinnen, die geben gute wirtin |
wöttu en bildhauer gäwen = willst du ein steinmetzer werden).

Subjuncts (“substantives used adverbially”) often resemble
objects, and it is not always easy to draw the line between the
two categories, e.g. in he walks three miles. We do not hesitate
to regard stones in throw stones as the object of the verb, but many
languages here use the instrumental case (which in old Gothonio
was merged into the dative); in OE. the word for ‘throw’
weorpan may take a dative (teoselum weorpeþ ‘throws dice’),
though it more often takes an accusative; ON has kasta (verpa)
steinum ‘throw (with) stones’; in Russian, brosat' ‘throw’ takes
either the acc. or the instrumental. English has, of course, no
longer any instrumental case, but we might speak of an “object
of instrument” in cases like: she nods her head | claps her hands |
shrugs her shoulders | pointed her forefinger at me | it rained fire
and brimstone.

Object of Result.

A subdivision of ‘objects of result’ comprises those ‘inner
objects’ which I mentioned under the head of nexus-substantives
(dream a strange dream | fight the good fight, etc., p. 137 f.).
Another is seen in grope one's way | force an entrance | he smiled
his acquiescence, etc.

Subject and Object.

The relation between subject and object cannot be determined
once and for all by pure logic or by definition, but must in each
case be determined according to the special nature of the verb
employed. Both subject and object are primary members, and
we may to some extent accept Madvig's dictum that the object
is as it were a hidden subject, or Schuchardt's that “jedes objekt
ist ein in den schatten gerücktes subjekt” (Sitzungsber. d. preuss.
Akad. d. wiss. 1920, 462). In many ways we see that there is some
kinship between subject and object.

If this were not so, we should be at a loss to understand the
frequency of shiftings from one to the other in course of time,
as in ME him (O = object) dreams a strange dream (S = subject),
which has become he (S) dreams a strange dream (O), a transition
which, of course, was facilitated by the great number of sentences
in which the form did not show the first word to be an object,
as the king dreamed.… This transition causes a semantic change
in the verb like, which from the meaning ‘please, be agreeable
to’ (him like oysters) came to mean ‘feel pleasure in’ (he likes
oysters). By this change the name of the person, which had
always been placed first because of its emotional importance,
now by becoming the subject became the foremost word of the
sentence from a grammatical point of view as well.

While, then, in English and Danish a certain number of verbs
ceased in this way to be “impersonal” and became “personal,”
a corresponding change in Italian led to the development of a
160kind of pronoun for the “generic person” (see on this term the
chapter on Person). Si dice cosi means literally ‘(it) says itself
thus,’ G. ‘es sagt sich so,’ but that is equivalent to G. ‘man sagt
so,’ and what was at first the object came to be regarded as the
subject, and vice versa, as in si pud vederlo ‘you can see him’;
this is shown in the change of number from si vendono biglietti,
where biglietti is subject, into si vende biglietti, where it is object.
Both constructions are now found side by side, thus in Fogazzaro,
Santo, p. 291, Pregò che si togliessero le candele, but p. 290 disse
che si aspettava solamente loro. 183

The logical kinship between subject and object also accounts
for the fact that there are here and there sentences without a
formal subject but with an object, as G. mich friert, mich hungert.
In the vast majority of cases, however, where a verb has only one
primary, this will be felt as the subject and accordingly is, or in
course of time comes to be, put in the nominative as the proper
subject-case.

Reciprocity.

Some verbs by virtue of their meaning make it possible to
reverse the relation between subject and object. If A meets B,
B also meets A (note that where we say I met an old man, the
Germans usually, though having the same word-order, will make
an old man into the subject: mir begegnete ein alter mann). When
in geometry one line cuts (intersects) another line, the second
line also cuts the former. If Mary resembles Ann, Ann also
resembles Mary; and if Jack marries Jill, Jill also marries Jack.
In such cases we often make the two words into one connected
subject and use each other as object; the old man and I met each
other | the two lines cut one another | Mary and Ann resemble
each other | Jack and Jill marry one another. Reciprocity may,
of course, also occur without being necessarily implied in the meaning
of the verb itself: A may hate B without B hating A, but
if B does hate him back, we may express it in the same way:
A and B hate one another. In English the verb in itself often
suffices to express reciprocity: A and B meet (marry, kiss, fight)
= A meets (marries, kisses, fights) B, and B meets (marries, kisses,
fights) A. In some of these cases Danish has the form in -s (old
reflexive): A og B mødes, kysses, slåss.

Two Objects.

There may be two objects in the same sentence, e.g. He gave
his daughter a watch | he showed his daughter the way | he taught
161his daughter arithmetic, etc. (But it should be noted that in “they
made Brown President” we have only one object, namely the
whole nexus, as in “they made Brown laugh”.) In languages
with separate forms for the accusative and the dative, the person
is generally put in the dative, and the thing in the accusative;
the former is called the indirect, and the latter the direct object.
But sometimes we find the dative where there is only one object,
and in some cases both objects are in the accusative — which shows
that the difference between the dative and the accusative is not
a notional one, but purely syntactic, dependent in each language
on idiomatic rules; on this, and on the use of other cases for the
object, see the chapter on Case (XIII).

Instead of a case-form for the indirect object we often find
a preposition, which loses its original local meaning, thus E. to,
Romanic a. This originally indicated direction and would be
appropriately used with such verbs as give, but its use was extended
to cases in which any idea of direction would be out of the question,
e.g. with deny. In Spanish à is used even with the direct
object, if this denotes a person. In English the preposition on
is sometimes used idiomatically: bestow something on a person,
confer a degree on him.

The point of view which determines whether something is the
direct or the indirect object may sometimes vary, even within
one and the same language, as in E. present something to a person
or present a person with something (Fr. présenter quelque chose à
quelqu'un). Where French has fournir qch à qqn, English says
furnish someone with something. Only the briefest mention can
here be made of the French inclination to treat a verb and a
dependent infinitive as one verb, and therefore to turn the person
into the indirect object: il lui fit voir le cheval (as il lui montra
le cheval), but il le fit chanter; 184 and then further: je lui ai entendu
dire que…

Where the active verb has two objects, one of them may be
made the subject in the corresponding passive turn. 285 In most
cases it is the direct object which is treated in this way, and many
languages are strict in not allowing what in the active is in the
162dative case to be made a subject in the passive. Cf., however,
Fr. je veux être obéi. In English there is a growing tendency to
make the person into the subject of the passive verb; this is
quite natural because there is now no formal difference between
dative and accusative, and because for emotional reasons one
always tends to place the name of the person first. Thus people
will naturally say: the girl was promised an apple | he was awarded
a gold medal, etc. Grammarians have opposed this tendency,
chiefly because they have had in their heads the rules of Latin
grammar, but the native speech-instinct cannot be put down by
pedantic schoolmasters. Curiously enough the pedants seem to
have had fewer objections to constructions like: he was taken
no notice of, which find their explanation in a following paragraph.

Adjectives and Adverbs with Objects.

Verbs are not the only words that can take an object. In
English there are a few adjectives which can do the same: he
is not worth his salt | he is like his father; Dan. han er det franske
sprog mægtig, G. (with gen.) er ist der französischen sprache mächtig;
Lat. avidus laudis | plenus timoris. We have also English combinations
like conscious that something had happened | anxious to
avoid a scandal, where the clause and the infinitive are objects.
These adjectives, however, cannot take a substantive as their
object except with a preposition: conscious of evil | anxious for
our safety, where we may say that the whole groups of evil, for our
safety are notional objects, even if we do not acknowledge them
as grammatical objects. The same remark applies to of-groups
after such adjectives as suggestive, indicative, etc. In Latin we
have the rule that participles in -ns take their object in the accusative
when the verbal feeling is strong: amans patriam, but in the
genitive (like adjectives such as tenax) when they denote a more
constant characteristic: amans patriæ.

If an adverb takes an object, the adverb becomes what is
commonly termed a preposition; see Ch. VI. Observe that the
German preposition nach is nothing but a phonetic variant of the
adverb nah.

When a verb is followed by an adverb (preposition) with its
object, the latter may often be looked upon as the object of the
whole combination verb + adverb; hence we find vacillations,
e.g. G. er läuft ihr nach (um ihr nachzulaufen): er läuft nach ihr
(um nach ihr zu laufen), Fr. il lui court après — il court après elle.
In OE. he him æfter rād (æfterrād) ‘he rode after him,’ æfter may
be taken as a postpositive preposition; notice also that the inseparable
Dan. (at) efterfølge, (at) efterstræbe = the separable G.
163nach(zu)folgen, nach(zu)streben. Hence come the passive constructions
found in E. he was laughed at | he is to be depended on, etc.

Passive.

It is customary in English to speak about the active and
passive voice (Fr. voix). William James, in his Talks to Teachers,
p. 152, relates how one of his relatives was trying to explain to
a little girl what was meant by the passive voice. “Suppose
that you kill me: you who do the killing are in the active voice,
and I, who am killed, am in the passive voice.” “But how can
you speak if you're killed?” said the child. “Oh, well, you may
suppose that I am not yet quite dead!” The next day the child
was asked, in class, to explain the passive voice, and said, “It's
the kind of voice you speak with when you ain't quite dead.”
The anecdote shows not only the bad blunders that may be committed
in the teaching of grammar (absurd examples, stupid
explanations), but also the drawback of the traditional term voice.
Some grammarians in Germany and elsewhere use the word genus
(genus verbi), which has the inconvenience that it is also used of
gender (genus substantivi). It would be best, probably, to use
164the word turn: and say ‘active and passive turn.’ The words
active and passive cannot very well be dispensed with, though
they, too, may lead to misconceptions: even in works by good
scholars one may occasionally find words to the effect that such
verbs as suffer, sleep, die should be called passive rather than
active, or that Lat. vapulo ‘I am thrashed’ is a passive in spite
of its active form, or that there is nothing active in A sees B, A
loves B. These ideas start from the erroneous conception that
the distinction between active and passive in the linguistic sense
is congruent with the distinction between bodily or mental activity
and passivity — an error which is connected with the similar one
we saw above where we were speaking of the definition of the
object.

It is important here as elsewhere to distinguish between syntactic
and notional categories. Whether a verb is syntactically
active or passive depends on its form alone; but the same idea
may be expressed sometimes by an active, sometimes by a passive
form: A precedes B = A is followed by B; A likes B = A is
attracted by B. The passive Lat. nascitur has given way to the
active Fr. naît in the same sense and is rendered in English sometimes
by the passive is born, sometimes by the active, originates,
comes into existence; the circumstance that Lat. vapulo in other
languages is translated by a passive does not alter its grammatical
character as an active; and Gr. apothnēskei is just as active when
we render it ‘is killed’ (thus when it is followed by hupo ‘by’)
as when we simply say ‘dies.’ There is thus nothing in the ideas
themselves to stamp verbs as active or passive. And yet we may
speak of ‘active’ and ‘passive’ as notional as well as syntactic
categories, but only as applied to the meaning of each verb separately,
and — what is very important — only in case of a transposition
of the relation of the subject (and object if there is one) to the verb
itself. “Jill is loved by Jack” and “es wird getanzt” are
notionally as well as syntactically in the passive, because the
subjects are different from those in “Jack loves Jill” and “sie
tanzen.” In other cases there is disagreement between the syntactic
and the notional active or passive.

Other English examples: he opened the door; the door
opened | he moved heaven and earth; the earth moves round
the sun | roll a stone; the stone rolls | turn the leaf; the tide
turns | burst the boiler; the boiler bursts | burn the wood; the
wood burns, etc.

It is rarer to find verbs with passive forms that may be used
in both these ways. The Dan. mindes has a passive form; it
generally means ‘remember’ and may then be said to be a notional
active, but when it is used, as occasionally happens, in the sense
‘be remembered’ (“det skal mindes længe”) it is a notional
passive; similarly we have “vi må omgås ham med varsomhed”
‘we must deal cautiously with him,’ and “han må omgås med
varsomhed” ‘he must be dealt with cautiously.’ We shall see
other instances of notional passive unexpressed in form in verbal-substantives
and infinitives.

Cp. the use in the genitive: nanˑup niaqua Peˑlip takuva.
‘Pele saw the bear's head’ | nanˑup niaqua angivoq ‘the bear's
head was large’ | Peˑlip niaqua nanˑup takuvaˑ ‘the bear saw
Pele's head.’

Similar rules are found in Basque, in some languages of the
Caucasus, and in some Amerindian languages. On this basis it
has been conjectured that the primitive Aryan language had one
form, characterized by -s, and used as an active (energetic, subjective
or possessive) case, thus only with names of animate beings
(masculine and feminine), and on the other side a form with no
ending or with -m, which was used as a passive or objective case,
serving also as the subject of intransitive verbs and coming naturally
to be used as a ‘nominative’ of names of inanimate things
(neuter). The -s-case later was differentiated into a nominative
and a genitive, the latter being characterized in some instances
by a different accent, in others by the addition of a second suffix.
166But originally it denoted not so much possession proper as some
intimate natural union or connexion. 187 It will be seen that these
speculations help to account for some peculiarities of our gender-system
as well as of our case-system, and they should be remembered
when we come to speak of the “subjective” genitive, though
there we shall see that this is used not only with nouns from transitive
verbs, but also with intransitives and passives and cannot
be distinguished from the “objective” genitive.

Use of the Passive.

We use the active or passive turn according as we shift our
point of view from one to the other of the primaries contained
in the sentence. “Jack loves Jill” and “Jill is loved by Jack”
mean essentially the same thing, and yet they are not in every
respect exactly synonymous, and it is therefore not superfluous
for a language to have both turns. As a rule the person or thing
that is the centre of the interest at the moment is made the subject
of the sentence, and therefore the verb must in some cases be
put in the active, in others in the passive. If we go through all
the passives found in some connected text we shall find that
in the vast majority of cases the choice of this turn is due to one
of the following reasons.

(1) The active subject is unknown or cannot easily be stated,
e.g. He was killed in the Boer war | the city is well supplied with
water | I was tempted to go on | the murderer was caught yesterday:
here the fact of his capture is more important than the
statement what policeman it was who caught him. Very often
the active subject is the ‘generic person’: it is known = ‘on
sait.’ In “the doctor was sent for” neither the sender nor the
person sent is mentioned.

(2) The active subject is self-evident from the context: His
memory of these events was lost beyond recovery | She told me
that her master had dismissed her. No reason had been assigned;
no objection had been made to her conduct. She had been forbidden
to appeal to her mistress, etc.

(3) There may be a special reason (tact or delicacy of sentiment)
for not mentioning the active subject; thus the mention
of the first person is often avoided, in writing more frequently
than in speaking: “Enough has been said here of a subject which
will be treated more fully in a subsequent chapter.” In Swedish
167the passive turn is rather frequent to avoid the clumsy substitutes
for the second personal pronouns: Önskas en tändstick ‘do you
want a match?’ | Finns inte en tändstick? ‘Haven't you got
a match?’

In none of these cases is the active subject mentioned, and
it has often been pointed out that this is the general rule with
passive sentences in many languages (Arabic, Lettish, old Latin,
Wackernagel VS 143). Statistical investigations made by some
of my pupils showed me many years ago that between 70 and
94 per cent, of passive sentences in various English writers contained
no mention of the active subject.

(4) Even if the active subject is indicated (“converted subject”)
the passive turn is preferred if one takes naturally a greater
interest in the passive than in the active subject: the house was
struck by lightning | his son was run over by a motor car.

(5) The passive turn may facilitate the connexion of one sentence
with another: he rose to speak and was listened to with
enthusiasm by the great crowd present.

In most languages there are certain restrictions on the use of
the passive turn, which are not always easy to account for. The
verb have (have got) in its proper sense is seldom used in the passive
(though it may be used, e.g. in “This may be had for twopence
at any grocer's”). Pedants sometimes object to sentences like:
“this word ought to be pronounced differently” (because a word
can have no duty!) or “her name will have to be mentioned.”
Intransitive verbs in the passive are common in some languages:
Lat. itur, itum est, curritur, G. es wird getanzt, even “Was nützte
es auch, gereist musste werden; man musste eben vorwärts, solange
es ging” (Ch. Bischoff), Dan. der danses, her må arbejdes — but
not in English or French.

Middle Voice.

On the “middle voice” as found, for instance, in Greek there
is no necessity to say much here, as it has no separate notional
character of its own: sometimes it is purely reflexive, i.e. denotes
identity of subject and (unexpressed) object, sometimes a vaguer
reference to the subject, sometimes it is purely passive and sometimes
scarcely to be distinguished from the ordinary active; in
some verbs it has developed special semantic values not easily
classified.

Active and Passive Adjectives.

The notional distinction between active and passive also applies
to some adjectives derived from or connected with verbs. We
168have active and passive participles (E. knowing, known, etc.,
though the latter is not purely passive). It is also a common
conviction among comparative linguists that the old Aryan participles
in -to and -no, which are at the bottom of our weak and
strong second participles, were at first neither active nor passive
in character. 188 Besides these we have adjectives with such endings
as some (troublesome, wearisome), -ive (suggestive, talkative), -ous
(murderous, laborious), which are all of them active, and adjectives
in -ble, which are generally passive (respectable, eatable, credible,
visible), but occasionally active (perishable, serviceable, forcible)
-less is active in sleepless, passive in tireless. Sometimes there are
two correlated forms for active and passive: contemptuous:
contemptible, desirous: desirable; sometimes the same word may
have now an active, and now a passive meaning: suspicious, curious.
It is the same in other languages. Some of the active adjectives
may take a notional object by means of the preposition of: suggestive
of treason, oblivious of our presence, etc.

Active and Passive Substantives.

If we ask whether substantives can be active and passive,
and whether they can take objects, we first encounter the so-called
agent-nouns, which are active, e.g. fisher, liar, conqueror, saviour,
creator, recipient. What would be the object of the corresponding
verb, is put in the genitive (Ann's lover) or more often, follows
the preposition of (the owner of this house, the saviour of the world).
We may here as above speak of notional or shifted objects. — Substantives
of the form pickpocket, breakwater contain an active verb
with its object; a pickpocket may be defined as ‘a picker of
pockets.’

In English we have a curious class of passive substantives in
-ee: lessee, referee, etc., ‘one to whom a lease is given, to whom
a question is referred,’ examinee ‘person examined’ (but with the
same ending we have the active substantives refugee, absentee).

Nexus-substantives.

Next we come to nexus-substantives. These are originally
neither active nor passive, but may according to circumstances
be looked upon as one or the other. To take first a familiar Latin
example: amor dei may mean either the love that God feels, or
the love that someone else feels with God as its object. In the
first case we call dei a subjective genitive (which by some is
taken simply as a possessive genitive, inasmuch as God ‘has’ or
169‘possesses’ the feeling); in the second we call it an objective genitive.
In the first dei is, in the symbols used above, Sa, in the second Oa,
but as we have seen that Oa = Sp, we may just as well say that
dei in both cases is a subjective genitive, but that amor in the
first case is an active, and in the second case a passive word. In
both cases we have a nexus, in which the genitive indicates the
primary, and amor the secondary element; the nexus in itself
is neither active nor passive, the only thing expressed being a
connexion between the two elements God and love, in which it
is left to the hearer whether he will take it as meaning the fact
that God loves, or the fact that God is loved. In the same way
odium Cæsaris, timor hostium are ambiguous. So also in Greek,
e.g. 2 Cor. 5. 14 hē gar agapē tou Khristou sunekhei hēmas (in
A.V.: the loue of Christ constreineth vs).

English sometimes presents the same ambiguity. Hodgson
(Errors in the Use of Engl. 91) has the following anecdote: An
attorney, not celebrated for his probity, was robbed one night
on his way from Wicklow to Dublin. His father, meeting Baron
O'Grady the next day, said: “My lord, have you heard of my
son's robbery?” “No, indeed,” replied the Baron, “pray whom
did he rob?”

Memory is used in two ways in Hamlet: 'Tis in my memory
locked — this is the common usage, S^a — and: a great mans memory
may outliue his life half a year — this is the rarer S^p. Formerly
the objective genitive (S^p) was more common than now, e.g. from
Shakespeare: Reuenge his foule and most vnnaturall murther (the
fact that he has been murdered) | thou didst denie the golds receit.
There are, however, certain definite rules for the use of the genitive
(and of possessive pronouns) though they have not been recognized
by grammarians. The chief ones are the following.

(1) It is obvious that with intransitive verbs there can be no
question of any passive sense; the genitive therefore is always
S^a: the doctor's arrival, existence, life, death, etc.

The following rules apply to transitive verbs, but rules (2)
to (5) concern only the combination of genitive and substantive,
when this is not followed by a prepositional group.

(2) Substantives formed from such transitive verbs as cannot
on account of their meaning have a person as object are taken
in the active sense: his (S^a) suggestion, decision, supposition, etc.

(3) Where the meaning of the verb is such that its subject
generally is a person and that it may take a person as object, the
genitive or possessive is generally taken as S^a: his attack, discovery,
admiration, love, respect, approbation interruption, etc. Here,
however, we notice a curious difference, according as the nexus-substantive
is the subject of the sentence or is used after a
170preposition: His assistance (S^a) is required | come to his assistance
(S^p). Thus also: his service (support, defence) is valuable | at his
service (in his support, defence). Cf. also the somewhat archaic:
in order to his humiliation. The substantive has the same passive
sense without a genitive after verbs like need; want: he needs
support, asks for approbation (but my is S^a in: he asks for my
approbation).

(4) The genitive or possessive will, however, be understood in
its objective sense when more interest is taken in the person who
is the object of an action than in the person who is the agent in
the case. Thus in a recent number of an English paper I found,
at a few lines' distance, De Valera's capture and De Valera's arrest
mentioned as possibilities: it is of no importance who captures
or arrests the Irish leader. Other examples: a man's trial (the
fact that he is brought before a judge) | his defeat | his overthrow |
his deliverance | his release | his education. The passive sense is
also found in: her reception was unique | he escaped recognition.
In “he is full of your praises” the person who praises naturally
is he, and your therefore represents S^p = O^a.

(5) Where the subject of a verb is as often, or more often, a
thing than a person, and where, on the other hand, the object is
a person, the nexus-substantive is taken in a passive sense: his
(S^p) astonishment, surprise, amazement, amusement, irritation, etc.

Next we have to consider the use of prepositions with nexus-substantives.
Of in itself is just as ambiguous as the genitive,
the love of God, S^a or S^p. But it is unambiguous if it is combined
with a genitive, for then the latter always means S°, and the
of-group S^p: my trials of thy loue (Sh.) | his instinctive avoidance
of my brother, etc. When the genitive combinations mentioned
under (4) are thus followed by of, they immediately change their
meaning: Luther's (S^a) deliverance of Germany from priestcraft | he
won praise by his release of his prisoners | her reception of her guests.

In the nineteenth century the construction with by began to
be common as an unambiguous means of denoting S^a; it is the
same by that is used with the passive verb, but curiously enough
this recent use is not mentioned in the NED: the purchase, by
the rich, of power to tax the poor (Ruskin) | a plea for the education
by the State of neglected country girls | the massacre of
Christians by Chinese. If by is used, the genitive may be used for
S^p: his expulsion from power by the Tories (Thackeray).

For S^p there is also a growing tendency to use other prepositions
than the ambiguous of, thus: your love for my daughter | the
love of Browning for Italy | his dislike to (for) that officer | there
would have been no hatred of Protestant to Catholic | contempt,
fear for, attack on. With certain substantives similar prepositions
171are common in other languages as well, Dan. for, til, Lat. odium
in Antonium, It. la sua ammirazione per le dieci dame più belle
(Serao). 189

The English verbal substantive in -ing had also originally the
same double character, though it has generally an active sense:
His (S^a) throwing, etc. In former times S^p was frequent, cf.:
Shall we excuse his throwing into the water (Sh. = his having been
thrown). The passive sense is also seen in “Vse euerie man after
his desart, and who should scape whipping?” (Sh.), and is still
found in combinations like: the roads want mending, but the
creation in comparatively recent times of the passive combination
being thrown (having been thrown) restricts the simple form in the
vast majority of cases to the active use. On the case of the
notional subject see p. 141.

Infinitives.

We must here also say something about that early form of
verbal substantive which developed into our infinitive. This, too,
at first was neither active nor passive, but in course of time passive
simple forms or combinations developed: amari, be loved, etc.
Traces of the (active or indifferent) form as a notional passive
are still found, in English for instance in “they were not to blame”
(cf. they were not to be seen) | the reason is not far to seek | the
reason is not difficult to see, where the reason is the subject of is,
but at the same time may be considered a kind of object for to
see, or subject for to see if this is taken in the passive sense. 290 Cf.
further: there is a lot to see in Rome | there is a lot to be seen in
Rome (the two sentences are not exactly synonymous). In the
following quotation we have the three possibilities in close succession:
There was no one to ask (active form, passive sense),
no one to guide him (the same in active sense); there was nothing
to be relied upon.

Other well-known instances of this double-sided character of
the infinitive: G. er liess ihn (S^a) kommen | er liess ihn (S^p)
strafen | Dan. han lod ham komme | han lod ham straffe | Fr. je
l'ai vu jouer | je l'ai vu battre. In Engl., where the passive
form is now extensively used in such cases, the active form was
formerly used in a passive sense, e.g. (he) leet anon his deere doghter
calle (Chaucer: ‘let her be called, caused her to be called’) | he
made cast her in to the riuer (NED make 53 d).172

Chapter XIII
Case

Number of English Cases. Genitive. Nominative and Oblique. Vocative
Final Words about Cases. Prepositional Groups.

Number of English Cases.

The subject of this chapter, which has already to some extent
been touched upon in the previous chapter, is a most difficult
one, because languages differ very much on this point, and because
the underlying ideas expressed by the various cases are not as
palpable as, e.g., the difference between one and more, or between
past, present and future, which are to form the subjects of some
other chapters. It will, perhaps, be best to start from a concrete
example, which illustrates the fundamental difference between the
two originally related languages, Latin and English.

If we are to speak of separate datives and accusatives in
English, I for one do not know where in this list the dative goes
out and the accusative comes in, and I find no guidance in those
grammars that speak of these two cases.

Someone might suggest that we have a criterion in the possibility
of a word's being made the subject of a passive sentence,
as this is allowable with accusatives only. This would be a purely
linguistic test — but it is not applicable. In the first place it is
not every “accusative” that can be made the subject of a passive
sentence; witness the second “accusatives” in “they made Brown
Mayor,” “they appointed Kirkman professor.” Secondly, a
“dative” is made the subject of the passive sentences “he was
awarded a medal” | “she was refused admittance,” as has been
already mentioned (p. 163). Until other more infallible tests are
forthcoming, we may therefore safely assert that there is no
separate dative, and no separate accusative, in modern English.

This conclusion is strengthened when we see the way in
which the ablest advocate of the distinction, Professor Sonnenschein,
carries it out in his grammar, where it will be difficult to
find any consistent system that will guide us in other cases than
those that are mentioned. Sometimes historical reasons are
174invoked, thus when the rule is given that the case after any preposition
is the accusative (§ 169, 489): “In OE. some prepositions
took the dative… but a change passed over the language, so
that in late Old English there was a strong tendency to use the
accusative after all prepositions.” This is at any rate not the
whole truth, for the dative was kept very late in some instances;
Bee, e.g., Chaucer's of towne, yeer by yere, by weste, etc., with the e
sounded. We have traces of this to this day in some forms, thus
the dat. sg. in alive (on life), Atterbury (æt Þære byrig), the dat. pl.
in (by) inchmeal, on foot, which may be looked on as a continuation
of OE. on fotum, ME. on foten, on fote, at any rate when used of
more than one person, as in “they are on foot.” Apart from
such isolated survivals the plain historical truth is that in most
pronouns it was only the dative that survived, in the plurals of
substantives the accusative (= nom.), and in the singulars of
substantives a form in which nominative, accusative, and dative
are indistinguishably mingled — but whatever their origin, from an
early period these forms (him, kings, king) were used indiscriminately
both where formerly a dative, and where an accusative
was required. 191

To return to the way in which Professor Sonnenschein distributes
the two cases in modern English. In “he asked me a
question” both me and question are said to be direct objects,
probably because OE. ascian took two accusatives; in teach him
French we are left at liberty to call him an accusative or a dative,
though the former seems to be preferred, in spite of the fact that
teach is OE. tæcan, which takes a dative and an accusative. We
should probably never have heard of two accusatives with this
verb, had it not been for the fact that Lat. doceo and G. lehren
have this construction 292 — but that surely is quite irrelevant to
English grammar, otherwise we may expect some day to hear
that use takes the ablative like Lat. utor.

Sometimes the rules given are evidently incomplete. In § 173
the dative as indirect object seems to be recognized only where
the same sentence also contains an object in the accusative, as
in “Forgive us our trespasses,” but if we have simply “Forgive
us,” are we to say that us is in the accusative? Is him in “I
175paid him” in the accusative, because it is the only object, or is
it in the dative, because it is the indirect object in “I paid him
a shilling”? Such questions arise by the score as soon as you
begin to put asunder what nature has joined together into one
case, and while in German it is possible to answer them because
the form actually used guides us, we have nothing to go by in
English. In hit him a blow who is to say whether him is the
indirect object (dative) and a blow the direct object (“acc.”), or
else him the direct object (“acc.”) and a blow a subjunct (“instrumental”
or “adverbial”)? Most people when asked about the
simple sentence hit him (without the addition a blow) would
probably say that him was the direct object, and thus in the
“accusative.”

Sonnenschein recognizes “adverbial” uses of both cases, but
it is not possible to discover any reasons for the distribution.
“Near him” — dative, why? If because of OE. syntax, then him
in to him, from him should also be a dative; here, however, it is
said to be an accusative because of the fiction that all prepositions
take the accusative, but why is it not the same with near, which
is recognized as a preposition by the NED? “He blew his pipe
three times” — accusative, why? (In OE. it would be a dative.)
And thus we might go on, for there is nothing to justify the perfectly
arbitrary assignation of words to one or the other case.
The rules have to be learned by rote by the pupils, for they cannot
be understood.

Professor Sonnenschein says that a study of the history of
English grammars has led him emphatically to deny the view held
by many scholars that progress in English grammar has actually
been due to its gradual emancipation from Latin grammar. In
Modern Language Teaching, March 1915, he said that a straight
line led from the earliest grammarians, who did not see any analogy
between English and Latin grammar, to a gradually increasing
recognition of the same cases as in Latin, a full understanding of
the agreement of the two languages having only been made
possible after comparative grammar had cleared up the relationship
between them. But this view of a steady ‘progress’ towards
the Sonnenscheinian system is far from representing the whole
truth, for it has been overlooked that Sonnenschein's system is
found full-fledged as early as 1586, when Bullokar said that English
has five cases, and that in the sentence “How, John, Robert gives
Richard a shirt,” John is vocative, Robert nominative, shirt accusative,
and Richard dative (or, as it is quaintly called, gainative) —
four cases being thus recognized besides the genitive. In 1920
Professor Sonnenschein himself, in the Preface to the second volume
of his Grammar, mentions some early grammarians (Gil 1619,
176Mason 1622), who based English grammar on Latin grammar,
but though there seem thus at all times to have been two conflicting
ways of viewing this part of English grammar, Sonnenschein
thinks that “in the main” the line of direction and progress
has been as indicated by him. He does not mention such
excellent grammarians as William Hazlitt, 193 William Cobbett, and
Henry Sweet, who were opposed to his view of the cases, but
mentions with special praise Lindley Murray, who took “the
momentous step of recognizing an ‘objective’ case of nouns”
and thus “rendered English grammar the service of liberating it
from the false definition of case” and “opened the door” to
the next momentous step, Sonnenschein's recognition of a dative
case. What is the next step to be in this progressive series, one
wonders? Probably someone will thank Sonnenschein for thus
opening the door to the admission of an ablative case, and why
not proceed with an instrumental, locative, etc.? All the Professor's
arguments for admitting a dative apply to these cases
with exactly the same force.

He says that cases denote categories of meaning, not categories
of form, and that this is just as true of Latin grammar as it is of
English grammar. The different cases of a Latin noun do not
always differ from one another in form: the accusative of neuter
nouns has always the same form as the nominative, all ablative
plurals are the same in form as dative plurals, in some nouns the
dative singular does not differ in form from the genitive singular,
in others from the ablative singular. All this is perfectly true,
but it does not invalidate the view that the case distinctions of
Latin grammar are primarily based on formal distinctions, to
which different functions are attached. No one would have dreamt
of postulating a Latin ablative case if it had not in many instances
been different in form from the dative. And where the two cases
are identical in form, we are still justified in saying that we have
now one, and now the other case, because other words in the same
position show us which is used. We say that Julio is the dative
in do Julio librum, but the ablative in cum Julio, because in the
corresponding sentences with Julia we have different forms: do
Juliæ librum, cum Julia. Templum in some sentences is in the
nominative, in others in the accusative, because in the first we
177should have used the form domus, and in the others the form
domum. And thus in all the other instances, exactly as above
(p. 51) we recognized cut as a preterit in I cut my finger yesterday,
though there is nothing in the form of that particular verb to
show that it is not the present. But with English nouns it is
impossible to argue in the same way: there is a fundamental
incongruity between the Latin system where the case-distinctions
are generally, though not always, expressed in form, and the
English System where they are never thus expressed. To put
the English accusative and dative, which are always identical in
form, on the same footing as these two cases in Latin, which are
different in more than ninety instances out of a hundred, is simply
turning all scientific principles upside down.

It is quite true that we should base our grammatical treatment
of English on the established facts of comparative and historical
grammar, but one of the most important truths of that science
is the differentiation which in course of time has torn asunder
languages that were at first closely akin, thereby rendering it
impossible to apply everywhere exactly the same categories. We
do not speak of a dual number in English grammar as we do in
Greek, although here the notional category is clear enough; why
then speak of a dative case, when there is just as little foundation
from a formal point of view, and when the meaning of the dative
in those languages that possess it is vague and indistinct from a
notional point of view?

Professor Sonnenschein says that cases “denote categories of
meaning.” But he does not, and cannot, specify what the particular
meaning of the dative is. 194 If we look through the rules
of any German, Latin, or Greek grammar, we shall find in each
a great variety of uses, or functions, i.e. meanings assigned to the
dative, but many of them differ from one language to another.
Nor is this strange, if we consider the way these languages have
developed out of the Proto-Aryan language which is the common
“ancestor” of all of them. As Paul says, it is really perfectly
gratuitous (es ist im grunde reine willkür) to call the case we have
in German (and Old English) a dative, for besides the functions
of the dative it fulfils the functions of the old locative, ablative,
and instrumental. Formally it corresponds to the old dative
only in the singular of part of the words, in some words it represents
the old locative, while in all words the dative plural is an
old instrumental. The Greek dative in the third declension in
178the singular is an old locative, and the dative of all words has taken
over the functions of the locative and instrumental as well as
those of the old dative proper. However far back we go, we
nowhere find a case with only one well-defined function: in every
language every case served different purposes, and the boundaries
between these are far from being clear-cut. This, in connexion
with irregularities and inconsistencies in the formal elements characterizing
the cases, serves to explain the numerous coalescences
we witness in linguistic history (“syncretism”) and the chaotic
rules found in individual languages — rules which even thus are
to a great extent historically inexplicable. If the English language
has gone farther than the others in simplifying these rules, we
should be devoutly thankful and not go out of our way to force
it back into the disorder and complexity of centuries ago.

But if no clear-cut meaning can be attached to the dative as
actually found in any of the old languages of our family, the same
is true of the accusative. Some scholars have maintained a
“localistic” case-theory and have seen in the accusative primarily
a case denoting movement to or towards, from which the other
uses have gradually developed: Romam ire ‘go to Rome’ led
to Romam petere, and this to the other accusatives of the object,
thus finally even to Romam linquere ‘leave Rome.’ Others consider
the objective use the original function, and others again
think that the accusative was the maid of all work who stepped
in where neither the nominative nor any of the special cases was
required. The only thing certain is that the accusative combined
the connotation of a (direct) object with that of movement towards
a place and that of spatial and temporal extension. It may even
originally have had further uses which are now lost to us.

That the meanings of the accusative and dative cannot be
kept strictly distinct, is shown also by the fact that the same verb
may in the same language take sometimes one case and sometimes
the other. Thus in German we find vacillation between them
after rufen, gelten, nachahmen, helfen, kleiden, liebkosen, versichern
and others (many examples in Andresen, Sprachgebrauch, 267 ff).
In OE., folgian and scildan vacillated in the same way. The
object after onfon ‘take, receive’ is now in the accusative, now
in the dative, and now in the genitive. If we were to go by linguistic
history, we should say that of the three synonyms in
English, help governs the dative, and aid and assist the accusative.
There is, of course, no foundation in the history of language for
what seems to be at the root of Sonnenschein's rule, that (apart
from his “adverbial” uses) a dative is found only when the verb
has also another object (which then is said to be in the accusative):
that rule is found in no language and in Sonnenschein's grammar it
179is due to a decree that is just as arbitrary as the Professor's ruling
that all prepositions govern the accusative.

Professor Sonnenschein tries to prop up his views by a pedagogical
argument (Part III, Preface): the pupil who has mastered
the uses of the English cases, as set forth in his book, will have
little to learn when he comes to Latin, except that Latin has an
extra case — the ablative. This means that part of the difficultly
of Latin grammar is shifted on to the English lessons; the subject
in itself is not made easier even for those pupils who are going on
with Latin afterwards, the only difference is that they have to
learn part of it now at an earlier stage, and in connexion with
a language where it is perhaps more difficult to understand because
the memory has no support in tangible forms on which to fasten
the functions. And what of all those pupils who are never to take
up Latin? Is it really justifiable to burden every boy and girl
of them with learning distinctions which will be of no earthly use
to them in later life?

Genitive.

Not a single one of the old Aryan cases is so well-defined in
its meaning that we can say that it has some single function or
application that marks it off from all the rest. The genitive combines
two functions which are kept separate in two Finnish cases,
the genitive and the partitive. But what the former function is
cannot be indicated except in the vaguest way as belonging to,
or belonging together, appertaining to, connexion with, relation
to or association with: 195 in English the use of this case is greatly
restricted, yet we find such different relations indicated by means
of the genitive as are seen in Peter's house, Peter's father, Peter's
son, Peter's work, Peter's books (those he owns, and those he has
written), Peter's servants, Peter's master, Peter's enemies, an hour's
rest, out of harm's way, etc. Some grammarians try to classify
these various uses of the genitive, but in many cases the special
meaning depends not on the use of the genitive in itself, but on
the intrinsic meaning of each of the two words connected, and
is therefore in each case readily understood by the hearer. Here
we must also mention the “subjective” and “objective” genitives
considered above (p. 169 ff.).

English has preserved only those uses in which the genitive
serves to connect two nouns, one of which is in this way made
an adjunct to the other (“adnominal genitive”), and the derived
use in which the genitive stands by itself as a primary, e.g. at the
grocer's. In the older languages the genitive was also used in
other ways, thus with certain verbs, where it formed a kind of
180object, with some adjectives, etc. The relation between this
genitive and an ordinary object is seen clearly in German, where
some verbs, e.g. vergessen, wahrnehmen, schonen, which used to
take the genitive, are now followed by an accusative; es in ich
kann es nicht los werden, ich bin es zufrieden was originally a genitive,
but is now apprehended as an accusative.

We next come to the second value of the old Aryan genitive,
the partitive, which cannot be separated from the so-called
genitivus generis. In Latin it is chiefly used with primaries (substantives,
etc.), e.g. magna pars militum, major fratrum, multum
temporis. This in so far agrees with the other value of the case,
as the genitive is an adjunct either way; but there are other
applications of the partitive genitive in which it comes to fulfil
more independent functions in the sentence. The genitive is often
used as the object of a verb, and so comes into competition with
the accusative, as in OE. bruceÞ fodres ‘partakes of food,’ Gr.
phagein ton artou ‘eat (some part) of the bread,’ earlier German,
e.g. Luther's wer des wassers trincken wird, Russian, e.g. daite mně
xlěba ‘give me of bread, some bread.’ In Russian this use of
a genitive as the object has been extended (with loss of the partitive
idea) to all masculines and plurals denoting living beings.
The partitive may also be used as the subject of a sentence, and
so come into competition with the nominative. This is frequent
with the partitive in Finnish, and the same use is found here and
there in our own family of languages, thus in negative sentences
in Russian, e.g. nět xlěba ‘there is no bread,’ ne stalo našego druga
‘there was no more of our friend, i.e. he died.’ We see corresponding
phenomena in the Romanic languages, in which the
preposition de has taken the place of the old genitive even in its
use as a partitive, in which it is now often called the “partitive
article”; it is noteworthy that the noun with this partitive
article may be used not only as an object of a verb (j'y ai vu des
amis), but also as the subject of a sentence (ce soir des amis vont
arriver | il tombe de la pluie), as a predicative (ceci est du vin),
and after prepositions (avec du vin | après des detours | je le
donnerai à des amis). If the subject-use is comparatively rare,
this is explained by the general disinclination that speakers have
to indefinite subjects (see p. 154; in voici du vin, il y a du vin, il
faut du vin we originally had objects).

The expression of the partitive idea ‘some (indefinite) quantity
of…’ thus as it were comes athwart the ordinary case-system,
because it comes to be used in the same functions for which many
languages have separate cases (nominative, accusative); this is
true whether this partitive idea is expressed by means of a
separate case, as in Finnish, or by means of the genitive, as
181in Greek, or finally by the French combination with the preposition
de.

If the distinction between the different cases was really one
of meaning, that is, if each case had its own distinctive notional
value, it would be quite unthinkable to have for one and the
same construction, namely the so-called “absolute” construction
(nexus-subjunct, as I call it) such complete divergence in usage
as we actually find: ablative (Latin), dative (Old English), genitive
(Greek), accusative (German), nominative (Modern English).
It may be possible to account for this historically, but it can never
be explained logically on the ground of some supposed intrinsic
meaning of these cases.

The irrationality of the old case distinctions may perhaps also
be brought out by the following consideration. The dative and
the genitive seem to be in some way opposites, as indicated by
the fact that when the old cases are replaced by prepositional
groups, the preposition chosen in the former case is to, ad, and
in the latter one which from the first denoted the opposite movement,
of (a weak form of off), de. And yet the dative (or its substitute)
often comes to mean the same thing as a genitive, as in
the popular G. dem kerl seine mutter ‘that fellow's mother,’ Fr.
ce n'est pas ma faute à moi, sa mère à lui, and the popular la mère
à Jean (OFr. je te donrai le file a un roi u a un conte, Aucass).
C'est à moi means ‘it is mine.’ In Norwegian dialects, combinations
with til and åt (‘to, at’) and in Faeroese, combinations with
hjá (‘with, chez’) have largely supplanted the obsolete genitive. 196

Nominative and Oblique.

If the reader will recur to the question put at the beginning
of this chapter, how many cases we are to recognize in the English
sentence “Peter gives Paul's son a book,” he will, I hope, now
agree with me that it is impossible to say that son and book are
in different cases (dative and accusative); but so far nothing
has been said against the second possibility that we have in both
an oblique case to be kept distinct from the nominative, of which
in our sentence Peter is an example. Old French had such a
system in its nouns, for there ‘Peter’ and ‘son’ in the nominative
would be Pierres and fils, and in the oblique case Pierre and fil.
Though there is no such formal distinction in the English substantives,
I can imagine someone saying that on the strength of
my own principles I should recognize the distinction, for it is
found in pronouns like I — me, he — him, etc., and just as I say
182that sheep in many sheep, though not distinct in form from the
singular, is a plural, because lambs in many lambs is distinct from
the singular lamb, and that cut in some sentences must be similarly
recognized as a preterit, so I ought to say that Peter and son are
in the nominative in those combinations in which we should use
the form he, and in the oblique case wherever we should use the
form him. This looks like a strong argument; yet I do not think
it is decisive. In the case of sheep and cut the parallel was with
words belonging to the same word-class, where the conditions are
practically the same, but here the argument is drawn from another
word-class, the pronouns, which present a great many peculiarities
of their own and keep up distinctions found nowhere else. If
we were to distinguish cases on the strength of their being distinct
in some pronouns, we might just as well distinguish gender in
English substantives on account of the distinctions seen in he,
she, it, and who, what, and split up adjectives and genitives into
two “states” or whatever you would call them, according as they
corresponded to my (adjunct) or to mine (non-adjunct). But as
a matter of fact, no grammarian thinks of making such distinctions,
any more than Old English grammars speak of a dual
number in substantives, while naturally recognizing it in the
personal pronouns, where it has distinct forms. Thus we see that
distinctions which are appropriate and unavoidable in one word-class
cannot always be transferred to other parts of speech.

With regard to the meaning of the nominative as distinct from
the other cases, we are accustomed from the grammar of Latin
and other languages to look upon it as self-evident that not only
the subject of a sentence, but also the predicative, is put in the
nominative. From a logical point of view this, however, is not
the only natural thing, for subject and predicative are not to be
regarded as notionally identical or even necessarily closely akin.
Here as elsewhere it serves to broaden one's view to see how
the same ideas are expressed in other languages. In Finnish the
predicative is (1) in the nominative, e.g. pojat ovat iloiset ‘the
boys are glad,’ (2) in the partitive “if the subject is regarded as
referred to a class in common with which the subject shares the
quality in question” (Eliot), “to denote qualities which are found
always or habitually in the subject” (Setala), e.g. pojat ovat iloisia
‘boys are (naturally) glad,’ (3) in the essive to denote the state
in which the subject is at a given time, e.g. isäni on kipeänä ‘my
father is. (now) ill,’ 197 and (4) in the translative after verbs signifying
to become (change into a state), e.g. isäni on jo tullut vanhaksi
‘my father has grown old.’ 298183

Even in our West-European languages the predicative does
not always stand in the nominative. In Danish for a couple of
centuries it has been recognized as good grammar to use the accusative
(or rather oblique case) and thus to look upon the predicative
as a kind of object: det er mig. And in English we have colloquially
the same use: it's me. The habitual omission of the
relative pronoun in such sentences as this: “Swinburne could
not have been the great poet he was without his study of the
Elizabethans” (thus also in Danish) also seems to show that
popular instinct classes the predicative with the object. 199

In English and Danish this cannot be separated from the
tendency to restrict the use of the nominative to its use in immediate
connexion with a (finite) verb to which it serves as subject
(I do | do I), and to use the oblique form everywhere else, thus
e.g. after than and as (he is older than me | not so old as me) and
when the pronoun stands by itself (Who is that? — Me!). This
tendency has prevailed in French, where we have moi when the
word is isolated, and the nom. je, acc. me in connexion with a
verbal form, and similarly with the other personal pronouns;
cf. also the isolated lui, lei, loro in Italian. 2100 (Cf. on this development
in English Progr. in Language, Ch. VII, reprinted ChE
Ch. II.)

Vocative.

On the so-called Vocative very little need be said here. In
some languages, e.g. Latin, it has a separate form, and must consequently
be reckoned a separate case. In most languages, however,
it is identical with the nominative, and therefore does not
require a separate name. The vocative, where it is found, may
be said to indicate that a noun is used as a second person and
placed outside a sentence, or as a sentence in itself. It has
points of contact with the imperative, and might like this
be said to express a request to the hearer, viz. ‘hear’ or ‘be
attentive.’

The close relation between the vocative and the nominative
is seen with an imperative, when “You, take that chair!” with
you outside the sentence (exactly as in “John, take that chair”)
by rapid enunciation becomes “You take that chair!” with you
as the subject of the imperative.184

Final Words about Cases.

Another classification, which in some ways would be better,
would be according to the three ranks considered in Ch. VII.

Many of the notions, however, are ill-defined and pass imperceptibly
into one another. No wonder, therefore, that languages
vary enormously, even those which go back ultimately to the
same ‘parent-language.’ Cases form one of the most irrational
part of language in general. 1101

Prepositional groups.

The reader will have observed that in this chapter I speak
only of the so-called synthetic cases, not of the “analytic cases,”
which consist of a preposition and its object; these, as I maintain,
should not be separated from any other prepositional group. In
English, to a man is no more a dative case than by a man is an
instrumental case, or in a man a locative case, etc. Deutschbein
is an extreme representative of the opposite view, for in his SNS,
p. 278 ff., he gives as examples of the English dative, among
others: he came to London | this happened to him | complain to
the magistrate | adhere to someone | the ancient Trojans were fools
to your father | he behaved respectfully to her | you are like
daughters to me | bring the book to me | I have bought a villa
for my son | What's Hecuba to him? | it is not easy for a foreigner
to apprehend — thus both with to and for, probably because German
has a dative in most of these cases. It is much sounder to
recognize these combinations as what they really are, prepositional
groups, and to avoid the name “dative” except where we
have something analogous to the Latin, or Old English, or German
dative. It is curious to observe that Deutschbein with his
emphasizing of “Der räumliche dativ” (“he came to London”)
is in direct opposition to the old theory which deduced all cases
from local relations, for according to that the dative was thought
of as the case of ‘rest,’ the accusative as the case of ‘movement
to,’ and the genitive the case of ‘movement from’; if Deutschbein
calls to London a dative, why not also into the house? But then
186the German in das haus would be a dative in spite of the actual
use of the accusative, which here means something different from
the dative in in dem haus. Even if the two expressions “I gave
a shilling to the boy” and “I gave the boy a shilling” are synonymous,
it does not follow that we should apply the same grammatical
term to both constructions: man-made institutions and
institutions made by man mean the same thing, but are not grammatically
identical.

The local meaning of the preposition to is often more or less
effaced, but that should not make us speak of a dative even where
to is wholly non-local. Thus also in French, where j'irai an ministre
and je dirai au ministre are analogous, though with a pronoun
the dative case is used in one, but not in the other construction:
j'irai à lui and je lui dirai.

With the genitive the same considerations hold good. Deutschbein
speaks of a genitive, not only in the works of Shakespeare,
but also in: participate of the nature of satire | smell of brandy |
proud of his country, and, if I am not mistaken, the man from
Birmingham | free from opposition (SNS 286 ff.). Some grammarians
speak of “die trennung eines genitivs von seinem regierenden
worte durch andere satzteile” and mean instances like “the
arrival at Cowes of the German Emperor,” where we have simply
two parallel prepositional group-adjuncts; some will even use
such a term as “split genitive” (Anglia, Beibl. 1922, 207) with
examples like “the celebrated picture by Gainsborough of the
Duchess of Devonshire,” where it would be just as reasonable to
call by Gainsborough a genitive as to use that name of the of-combination.
Both are prepositional groups and nothing else.

I may perhaps take this opportunity of entering a protest against a
certain kind of ‘national psychology’ which is becoming the fashion in
some German university circles, but which seems to me fundamentally
unsound and unnatural. It affects case-syntax in the following passage:
“Wenn nun der sächs. gen. bei zeitbestimmungen im lebendigen gebrauch
ist, so deutet dies darauf hin, dass der zeit im englischen sprachbewusstsein
eine bevorzugte rolle eingeräumt wird, was namentlich in gewissen berufskreisen
wie bei verlegern, herausgebern, zeitungsschreibern der fall sein
wird” (Deutschbein SNS 289). In the same work, p. 269, the dative in
G. ich helfe meinen freunden is taken as a sign of “ein persönliches vertrauensverhältnis
statischen charakters zwischen mir und meinen freunden,” but
“wenn im ne. to help (I help my friend) mit dem akk. konstruiert wird, so
verzichtet es darauf, das persönliche verhältnis von mir zu meinen freunde
auszudrücken… das ne. besitzt demnach einen dynamischen grundcharakter,
der auch in anderen zahlreichen erscheinungen der sprache
bemerkbar ist.” What does dynamic mean in that connexion? And how
does Deutschbein know that the case after help is not a dative still? In
give my friend a book he acknowledges friend as a dative, why not here?
The form is the same. The function is exactly the same as in the corresponding
OE. sentence ic helpe minum freonde, of which it forms an uninterrupted
continuation, and which in its turn corresponds in every respect
to G. ich helfe meinem freunde. Why not simply say that in Modern English
it is neither accusative nor dative, and then leave out all conclusions about
“personal,” “dynamic,” and “static” national characters?187

Chapter XIV
Number 1102

Counting. The Normal Plural. Plural of Approximation. Higher Units.
Common Number. Mass-Words.

Counting.

Number might appear to be one of the simplest natural categories,
as simple as ‘two and two are four.’ Yet on closer inspection it
presents a great many difficulties, both logical and linguistic.

From a logical point of view the obvious distinction is between
one and more than one, the latter class being subdivided into
2, 3, 4, etc.; as a separate class may be recognized ‘all’; while
beyond all these there is a class of ‘things’ to which words like one,
two are inapplicable; we may call them uncountables, though
dictionaries do not recognize this use of the word uncountable, which
is known to them only in the relative sense ‘too numerous to be
(easily) counted’ (like innumerable, numberless, countless).

The corresponding syntactic distinctions are singular and plural,
which are found in most languages, while some besides the ordinary
plural have a dual, and very few a trial.

We can only speak of “more than one” in regard to things
which without being identical belong to the same kind. Plurality
thus presupposes difference, but on the other hand if the difference
188is too great, it is impossible to use words like two or three. A pear
and an apple are two fruits; a brick and a castle can barely be
called two things; a brick and a musical sound are not two, a
man and a truth and the taste of an apple do not make three, and
so on.

What objects can be counted together, generally depends on
the linguistic expression. In the majority of cases the classification
is so natural that it is practically identical in most languages; but
in some cases there are differences called forth by varieties of
linguistic structure. Thus in English there is no difficulty in
saying “Tom and Mary are cousins,” as cousin means both a male
and a female cousin; Danish (like German and other languages)
has different words, and therefore must say “T. og M. er fætter og
kusine,” and E. five cousins cannot be translated exactly into
Danish. On the other hand, English has no comprehensive term
for what the Germans call geschwister, Dan. søskende. Sometimes,
however, a numeral is placed before such a collocation as brothers
and sisters: “they have ten brothers and sisters,” which may be
= 2 brothers + 8 sisters or any other combinations; “we have
twenty cocks and hens” (= Dan. tyve høns). The natural need for a
linguistic term which will cover male and female beings of the
same kind has in some languages led to the syntactical rule that
the masculine plural serves for both sexes: Italian gli zii, Span.
los padres (see p. 233).

In some cases it is not possible to tell beforehand what to reckon
as one object: with regard to some composite things different languages
have different points of view; compare un pantalon — a
pair of trousers, et par buxer, ein paar hosen; eine brille — a pair
of spectacles, une paire de lunettes, et par briller; en sax, eine
schere — a pair of scissors, une paire de ciseaux.

English sometimes tends to use the plural form in such cases
as a singular, thus a scissors, a tongs, a tweezers.

In modern Icelandic we have the curious plural of einn ‘one’
in einir sokkar ‘one pair of socks’ (to denote more than one pair
the ‘distributive’ numerals are used: tvennir vetlingar ‘two pairs
of gloves’).

With parts of the body there can generally be no doubt what
to consider as one and what as two; yet in English there is (or
rather was) some vacillation with regard to moustache, which is
in the NED defined as (a) the hair on both sides of the upper lip,
(b) the hair covering either side of the upper lip, so that what to
one is a pair of moustaches, to another is a moustache: “he twirled
first one moustache and then the other.”

In Magyar it is a fixed rule that those parts of the body which
occur in pairs are looked upon as wholes; where the English
189say “my eyes are weak” or “his hands tremble” the Hungarian
will use the singular: a szemem (sg.) gyenge, reszket a keze (sg.).
The natural consequence, which to us appears very unnatural,
is that when one eye or hand or foot is spoken of, the
word fél ‘half’ is used: fél szemmel ‘with one eye,’ literally
‘with half eye(s),’ fel lábára sánta ‘lame of one foot.’ This
applies also to words for gloves, boots, etc.: keztyű (pair of)
gloves, fel keztyű (a half… i.e.) one glove, csizma (sg.)
‘boots,’ fel csizma ‘a boot.’ The plural forms of such words
(keztyűk, czizmák) are used to denote several pairs or different
kinds of gloves, boots.

The Normal Plural.

The simplest and easiest use of the plural is that seen, e.g., in
horses = (one) horse + (another) horse + (a third) horse.…
(We might use the formula: Apl. = Aa + Ab + Ac…) This
may be called the normal plural and calls for very few remarks, as
in most languages grammar and logic here agree in the vast majority
of cases.

There are, however, instances in which different languages
do not agree, chiefly on account of formal peculiarities. English
and French have the plural of the substantive in the eighteenth and
nineteenth centuries, les siècles dix-huitième et dix-neuvième, while
German and Danish have the singular, the reason being not that
the English and French are in themselves more logical than other
nations, but a purely formal one: in French the article, which
shows the number, is placed before the substantive and is not in
immediate contact with the adjectives; in English the article is
the same in both numbers, and can therefore be placed before the
(singular) adjective as if it were in the singular itself without
hindering the use of the natural plural in centuries. In German,
on the other hand, you have to choose at once between the singular
and the plural form of the article, but the latter form, die, would
be felt as incongruous before the adjective achzehnte, which is in
the neuter singular; if, on the other hand, you begin with the
(singular) article das, it would be equally odd to end with the plural
of the substantive (das 18te und 19te jahrhunderte), whence the
grammatically consistent, if logically reprehensible use of the
singular throughout. It is the same in Danish. In English, too,
when the indefinite article is used, the singular is preferred for the
same reason: an upper and a lower shelf. Sometimes the singular
may be used to avoid misunderstandings, as when Thackeray
writes “The elder and younger son of the house of Crawley were
never at home together”: the form sons might have implied the
190existence of more than one son in each class. (See other special
cases in MEG II, p. 73 ff. 1103)

The English difference between the two synonymous expressions
more weeks than one and more than one week shows clearly the psychological
influence of proximity (attraction). The force of this is not
equally strong in all languages: where Italian has the singular in
ventun anno on account of un, English says twenty-one years exactly
as it says one and twenty years; thus also a thousand and one nights.
But German and Danish here show the influence of attraction with
peculiar clearness because each language has the plural when the
word for ‘one’ is removed from the substantive, and the singular
when it immediately precedes it: ein und zwanzig tage, tausend und
eine nacht; een og tyve dage, tusend og een nat.

With fractions there are some difficulties: should one and a
half be connected with a substantive in the singular or in the
plural? Of course one can get out of the difficulty by saying one
mile and a half, but this will not do in languages which have an
indivisible expression like G. anderthalb, Dan. halvanden; German
seems to have the plural (anderthalb ellen), but Danish has the singular
(halvanden krone) though with a curious tendency to put a
preposed adjective in the plural though the substantive is in the
singular: med mine stakkels halvanden lunge (Karl Larsen), i disse
halvandet år (Pontoppidan). Where English has two and a half
hours (pl.), Danish has attraction: to og en halv time (sg.).

Where each of several persons has only one thing, sometimes
the singular, and sometimes the plural is preferred: Danish says
hjertet sad os i halsen (sg.), while English has our hearts leaped to
our mouths, though not always consistently (three men came marching
along, pipe in mouth and sword in hand; see for details MEG II,
p. 76 ff.). Wackernagel (VS 1. 92) gives an example from Euripides
where the mother asks the children to give her the right hand:
dot' ō tekna, det' aspasasthai mētri dexian khera.

Plural of Approximation.

I next come to speak of what I have termed the plural of approximation,
where several objects or individuals are comprised in the
191same form though not belonging exactly to the same kind. Sixties
(a man in the sixties; the sixties of the last century) means, not
(one) sixty + (another) sixty…, but sixty + sixty-one + sixty-two
and so forth till sixty-nine. The corresponding usage is found
in Danish (treserne), but not, for instance, in French.

The most important instance of the plural of approximation is
we, which means I + one or more not-I's. It follows from the
definition of the first person that it is only thinkable in the singular,
as it means the speaker in this particular instance. Even when a
body of men, in response to “Who will join me?” answer “We all
will,” it means in the mouth of each speaker nothing but “I will
and all the others will (I presume).”

The word we is essentially vague and gives no indication whom
the speaker wants to include besides himself. It has often, therefore,
to be supplemented by some addition: we doctors, we gentlemen, we
Yorkshiremen, we of this city. Numerous languages, in Africa and
elsewhere, have a distinction between an “exclusive” and an
“inclusive” plural, as shown by the well-known anecdote of the
missionary who told the negroes “We are all of us sinners, and we
all need conversion,” but unhappily used the form for “we” that
meant “I and mine, to the exclusion of you whom I am addressing,”
instead of the inclusive plural (Friedrich Müller). In several
languages it is possible after we to add the name of the person or
persons who together with “I” make up the plural, either without
any connective or with “and” or “with”: OE. wit Scilling I
and Scilling, unc Adame ‘for me and Adam,’ ON. vit Gunnarr ‘I and
Gunnarr’ (cf. þeir Sigurðr ‘S. and his people,’ þau Hjalti ‘H. and
his wife’), Frisian wat en Ellen ‘we two, I and E,’ G. pop. wir sind
heute mit ihm spazieren gegangen, ‘I and he…,’ Fr. pop. nous
chantions avec lui ‘I and he sang,’ Ital. quando siamo giunti con mia
cugina ‘when my cousin and I arrived,’ Russian my s bratom
pridëm ‘we with brother, i.e. I and my brother, will come,’ etc. 1104

The plural of the second person may be, according to circumstances,
the normal plural (ye = thou + a different thou + a third
thou, etc.), or else a plural of approximation (ye = thou + one or
more other people not addressed at the moment). Hence we find
in some languages similar combinations to those mentioned above
with we: OE. git Iohannis ‘ye two (thou and) John,’ ON. it Egill
‘thou and E.’, Russ. vy s sestroj ‘ye, (thou) with thy sister.’

The idea that “we” and “ye” imply some other person(s)
besides “I” and “thou” is at the root of the Fr. combination
192nous (or vous) autres Français, i.e. ‘I (or thou) and the other Frenchmen.’
In Spanish nosotros, vosotros have been generalized and are
used instead of nos, vos, when isolated or emphatic.

In most grammars the rule is given that if the words composing
the subject are of different persons, then the plural verb is of the
first person rather than the second or third, and of the second
person rather than the third. It will be seen that this rule when
given in a Latin grammar (with examples like “si tu et Tullia
valetis, ego et Cicero valemus”) is really superfluous, as the first
person plural by definition is nothing else than the first person
singular plus someone else, and the second person plural correspondingly.
In an English grammar (with examples like “he and
I are friends; you and they would agree on that point; he and his
brother were to have come,” Onions, AS 21) it is even more superfluous,
as no English verb ever distinguishes persons in the plural.

A third instance of the plural of approximation is seen in the
Vincent Crummleses, meaning Vincent Crummies and his family,
Fr. les Paul = Paul et sa femme; “Et Mme de Rosen les signalait:
Tiens… les un tel” (Daudet, L'Immortel 160). 1105

When a person speaks of himself as “we” instead of “I” it
may in some cases be due to a modest reluctance to obtrude his
own person on his hearers or readers; he hides his own opinion or
action behind that of others. But the practice may even more
frequently be due to a sense of superiority, as in the “plural of
majesty.” This was particularly influential in the case of the
Roman emperors who spoke of themselves as nos 2106 and required to
be addressed as vos. This in course of time led to the French way
of addressing all superiors (and later through courtesy also equals,
especially strangers) with the plural pronoun vous. In the Middle
Ages this fashion spread to many countries; in English it eventually
led to the old singular thou being practically superseded by you,
which is now the sole pronoun of the second person and no longer a
sign of deference or respect. You now is a common-number form,
and the same is true to some extent of It. voi, Russian vy, etc. The
use of the “plural of social inequality” entails several anomalies,
as the German Sie (and in imitation of that, Dan. De) in speaking to
one person, Russian oni, one (‘they,’ m. and f.) in speaking of one
person of superior standing; grammatical irregularities are seen,
e.g., in the singular self in the royal ourself, Fr. vous-même, and in
the singular of the predicative in Dan. De er så god, Russ. vy
segodnja ne takaja kak včera (Pedersen RG 90) ‘you are not the
193same (sg. fem.) to-day as yesterday.’ Mention should also be
made of the use of the plural of deference in German verbs, when
no pronoun is used: Was wünschen der herr general? ‘What do
you want, General?’ Politeness and servility are not always free
from a comic tinge. 1107

Higher Units.

It is very often necessary or at any rate convenient to have a
linguistic expression in which several beings or things are comprehended
into a unit of a higher order. We must here distinguish
between various ways in which this fusion may be effected.

In the first place the plural form may be used in itself. English
has a facility in this respect which appears to be unknown to the
same extent in other languages; the indefinite article or another
pronoun in the singular number may be simply put before a plural
combination: that delightful three weeks | another five pounds | a
second United States | every three days | a Zoological Gardens,
etc. There can be no doubt that this is chiefly rendered possible
by the fact that the preposed adjective does not show whether it is
singular or plural, for a combination like that delightful three weeks
would be felt as incongruous in a language in which delightful was
either definitely singular or plural in form; but the English uninflected
form can easily be connected both with the singular that
and the plural three weeks.

A slightly different case is seen in a sixpence (a threepence), which
has been made a new singular substantive with a new plural:
sixpences (threepences). In the corresponding Danish name for
the coin worth two kroner the analogy of the singular en krone, en
eenkrone has prevailed and the form is en tokrone, pl. mange tokroner.
This reminds one of the E. a fortnight, a sennight (fourteen nights,
seven nights), in which, however, the latter element is the OE.
plural niht (the ending s in nights is a later analogical formation);
thus also a twelvemonth (OE. pl. monaþ).

In the second place the unification of a plural may be effected
through the separate formation of a singular substantive. Thus in
Greek we have from deka ‘ten’ the sb. dekas, L. decas, whence E.
decade; in French we have the words in -aine: une douzaine,
vingtaine, trentaine, etc., the first of which has passed into several
other languages: dozen, dutzend, dusin. Corresponding to dekas
the old Gothonic languages had a substantive (Goth, tigus), which
as is well known, enters into the compounds E. twenty, thirty, etc.,
G. zwanzig, dreissig, etc. These were therefore originally substantives,
194though now they have become adjectives. Lat. centum,
mille, E. (Gothonic) hundred, thousand were also substantives of
this kind, and reminiscences of this usage are still found, e.g. in
Fr. deux cents and in the E. use of a, one: a hundred, one thousand;
cf. also a million, a billion. A peculiar type of half-disguised compounds
may be seen in Lat. biduum, triduum, biennium, triennium
for periods of two or three days or years.

With these must be classed words like a pair (of gloves), a couple
(of friends), and this leads up to words denoting an assemblage of
things as a set (of tools, of volumes), a pack (of hounds, of cards),
a bunch (of flowers, of keys), a herd (of oxen, of goats), a flock, a
bevy, etc.

Such words are rightly termed collectives, and I think this
term should not be used in the loose way often found in grammatical
works, but only in the strict sense of words which denote a unit
made up of several things or beings which may be counted
separately; a collective, then, is logically from one point of view
“one” and from another point of view “more than one,” and this
accounts for the linguistic properties of such words which take
sometimes a singular and sometimes a plural construction. (On
the difference between collectives and mass-words see below.)

Some collectives are derivatives from the words denoting the
smaller units: brotherhood, from brother, cp. also nobility, peasantry,
soldiery, mankind. There is an interesting class in Gothonic languages
with the prefix ga-, ge- and the neuter suffix -ja; Gothic
had gaskohi ‘pair of shoes’; these formations became especially
numerous in OHG, where we have, e.g., gidermi ‘bowels,’ giknihti
‘body of servants,’ gibirgi ‘mountainous district,’ gifildi ‘fields,
plain.’ In modern G. we have gebirge, gepäck, gewitter, ungeziefer,
and others, partly with changed signification or construction.
Geschwister at first meant ‘sisters’ (“zwei bruder und drei
geschwister”), later it came to mean ‘brothers and sisters’ and
even sometimes may be used in the singular of a single brother
or sister, when it is desirable not to specify the sex. But in ordinary
speech it is now no longer used as a collective, but as an ordinary
plural.

Latin familia meant at first a collection of famuli, i.e. ‘housemates,’
later ‘servants’; when the word famulus went out of use,
familia acquired its present European meaning, and as an unanalyzable
collective must be classed with such words as crew, crowd,
swarm, company, army, tribe, nation, mob.

Some words may develop a collective signification by metonymy,
as when the parish is said for the inhabitants of the parish, all
the world = ‘all men,’ the sex ‘women,’ the Church, the bench,
society, etc.195

The double-sidedness of collectives is shown grammatically;
they are units, and as such can be used not only with a or one preposed,
but also in the plural in the same way as other countables:
two flocks, many nations, etc. On the other hand, they denote
plurality, and therefore may take the verb and the predicative
in the plural (my family are early risers; la plupart disent, thus in
many other languages as well) and may be referred to by such a
pronoun as they. It is, however, worthy of note that this plural
construction is found with such collectives only as denote living
beings, and never with others, like library or train, though they
mean ‘collection of books, of railway-carriages.’ Sometimes a
collective may show the two sides of its nature in the same sentence:
this (sg.) family are (pl.) unanimous in condemning him. This
should be thought neither illogical nor “antigrammatical” (as
Sweet calls it, NEG § 116), but only a natural consequence of the
twofold nature of such words.

In some instances languages go farther than this and admit
combinations in which the same form which is really a singular
is treated as if it were the plural of the word denoting the smaller
unit: those people (= those men), many people (as distinct from
many peoples = many nations), a few police, twenty clergy. In
Danish we have this with folk (as in E. with the word spelt in the
same way), which is a true collective in et folk (a nation, with the
separate pl. mange folkeslag), but is now also treated as a plural:
defolk, mange folk, though we cannot say tyve folk ‘twenty people’;
there is a curious mixture in de godtfolk ‘those brave people,’
godt is sg. neuter. (Quotations for E. 80,000 cattle, six clergy, five
hundred infantry, six hundred troops, etc., are found in MEG II,
p. 100 ff. 1108)

The transition from a collective to a plural is also seen in the
Aryan substantives in -a. Originally they were collectives in the
feminine singular; we have seen an instance in Lat. familia. In
many cases these collectives corresponded to neuters, as in opera,
gen. operæ ‘work’: opus ‘piece of work’; hence -a came finally
to be used as the regular way of forming the plural of neuters,
though a survival of the old value of the ending is found in the
Greek rule that neuters in the plural take the verb in the singular
(see the full and learned treatment in J. Schmidt, Die Pluralbildungen
der indogerm. Neutra, 1889, a short summary in my book Language,
p. 395). It is interesting to see the development in the Romanic
languages, where the same ending still serves to form a plural in
196many Italian words (frutta, uova, paja), but has generally again
become a fem. sg., though not in a collective sense; cp. It. foglia,
Fr. feuille from Lat. folia.

Wherever we have a plural of any of the words mentioned in
this section, we may speak of a “plural raised to the second power,”
e.g. decads, hundreds, two elevens (two teams of eleven each), sixpences,
crowds, etc. But the same term, a plural raised to the second
power, may be applied to other cases as well, e.g. E. children, where
the plural ending -en is added to the original pl. childer, possibly
at first with the idea that several sets (families) of children were
meant, as in the Sc. dialectal shuins mentioned by Murray as
meaning the shoes of several people, while shuin means one pair only
(Dial. of the Southern Counties, 161; see also MEG II, 5. 793).
This logical meaning of a double plural (a plural of a plural) cannot,
however, be supposed to have been in all cases present to the minds
of those who created double plurals: often they were probably
from the very first simple redundancies, and at any rate they are
now felt as simple plurals in such cases as children, kine, breeches,
etc. Breton has plurals of plurals: bugel child, pl. bugale, but
bugale-ou ‘plusieurs bandes d'enfants,’ loer ‘stocking,’ pl. lerou ‘pair
of stockings,’ but lereier ‘several pairs of stockings,’ daou-lagad-ou,
‘eyes of several persons’ (H. Pedersen, GKS 2. 71). We have a
double plural in form, but not in sense, in G. tränen, zähren ‘tears.’
Here the old plural form träne (trehene), zähre (zähere) has now
become a singular.

In Latin the use of a separate set of numerals serves to indicate
the plural of a plural. Litera is a letter (buchstabe), pl. literæ may
stand for ‘letters (buchstaben)’ or for the composite unit ‘a letter
(epistle)’ or the logical plural of this ‘letters (epistles)’; now
quinque literæ means ‘fünf buchstaben,’ but quinæ literæ ‘fünf
briefe.’ Castra ‘a camp’ is originally the pl. of castrum ‘a fort’;
duo castra ‘two forts,’ bina castra ‘two camps.’ Similarly, in
Russian the word for ‘a watch’ or ‘clock’ is časy, formally the pl.
of čas ‘hour’; two hours is dva časa, but ‘two watches’ is dvoe
časov; with higher numerals štuk ‘pieces’ is inserted: dvadtsat' pjat'
štuk časov, sto štuk časov ‘25, 100 watches or clocks.’

In this connexion we may also notice that when we say my
spectacles, his trousers, her scissors, no one can tell whether one pair
or more pairs are meant, thus whether the correct translation
into other languages would be meine brille, son pantalon, ihre schere,
or meine brillen, ses pantalons, ihre scheren. (But when we say “he
deals in spectacles; the soldiers wore khaki trousers,” etc., the
meaning is obviously plural.) The plural forms spectacles, trousers,
scissors, in themselves thus from a notional point of view denote a
‘common number.’197

Common Number.

The want of a common number form (i.e. a form that disregards
the distinction between singular and plural) is sometimes felt, but
usually the only way to satisfy it is through such clumsy devices
as “a star or two,” “one or more stars,” “some word or words
missing here,” “the property was left to her child or children.” 1109
In “Who came?” and “Who can tell?” we have the common
number, but in “Who has come?” we are obliged to use a definite
number-form in the verb even if the question is meant to be quite
indefinite. Note also “Nobody prevents you, do they?” where
the idea would have been expressed more clearly if it had been
possible to avoid the singular in one, and the plural in the other
sentence (cf. under Gender, p. 233).

Mass-Words.

In an ideal language constructed on purely logical principles
a form which implied neither singular nor plural would be even
more called for when we left the world of countables (such as
houses, horses; days, miles; sounds, words, crimes, plans, mistakes,
etc.) and got to the world of uncountables. There are a
great many words which do not call up the idea of some definite
thing with a certain shape or precise limits. I call these “mass-words”;
they may be either material, in which case they denote
some substance in itself independent of form, such as silver, quicksilver,
water, butter, gas, air, etc., or else immaterial, such as leisure,
music, traffic, success, tact, commonsense, and especially many
“nexus-substantives” (see Ch. X) like satisfaction, admiration,
refinement, from verbs, or like restlessness, justice, safety, constancy,
from adjectives.

While countables are “quantified” by means of such words
as one, two, many, few, mass-words are quantified by means of such
words as much, little, less. If some and more may be applied to both
classes, a translation into other languages shows that the idea is
really different: some horse, some horses, more horses — some quicksilver,
more quicksilver, more admiration: G. irgend ein pferd,
einige pferde, mehr (mehrere) pferde (Dan. flere heste) — etwas quecksilber,
mehr puecksilber, mehr bewunderung (Dan. mere beundring).198

As there is no separate grammatical “common-number,”
languages must in the case of mass-words choose one of the two
existing formal numbers; either the singular, as in the examples
hitherto adduced, or the plural, e.g. victuals, dregs, lees — proceeds,
belongings, sweepings — measles, rickets, throes and such colloquial
names of unpleasant states of mind as the blues, creeps, sulks, etc.
In many cases there is some vacillation between the two numbers
(coal(s), brain(s), and others), and where one language has a singular,
another may have a plural. It is curious that while Southern
English and Standard Danish looks upon porridge and grød as
singulars, the same words are in Scotland and Jutland treated as
plurals. Corresponding to the E. plurals lees, dregs, German and
other languages have singular mass-words: hefe. With immaterial
mass-words it is the same: much knowledge must be rendered in
German viele kenntnisse, in Danish mange kundskaber.

The delimitation of mass-words offers some difficult problems,
because many words have several meanings. Some things adapt
themselves naturally to different points of view, as seen, for instance,
in fruit, hair (much fruit, many fruits: “shee hath more hair then
wit, and more faults then haires,” Shakespeare); cf. also a little
more cake, a few more cakes. In a Latin edictum dry vegetables
and meat are given as singulars, i.e. as mass-words, while fresh
ones are given in the plural, because they are counted (Wackernagel,
VS 1. 88). Note also verse: “He writes both prose and verse.”
“I like his verses to Lesbia.”

Sometimes the original signification may belong to one, sometimes
to the other of these two classes. Sometimes a word is
differentiated, thus shade and shadow are derived from different
case-forms of the same word (OE. sceadu, sceadowe). As a rule,
shade is used as a mass-word, and shadow as a countable, but
in some connexions shade is just as much a thing-word as shadow,
e.g. when we speak of different shades (= nuances) of colour.
Cloth in one sense is a mass-word as denoting one particular
kind of material, but as denoting one particular thing (as a
199table-cloth, or a covering for a horse) it is a thing-word and has
developed the new plural cloths, while the old plural clothes is now
separated from cloth and must be termed a distinct word: a mass-word
with plural form.

A name of a tree, e.g. oak, may be made a mass-word, not
only to denote the wood or timber obtained from the tree, but also
to denote a mass of growing trees (cf. barley, wheat): “oak and beech
began to take the place of willow and elm.” A corresponding usage
is also found in other languages. A related case is seen in the use
offish, not only to denote the “flesh” of fish which we eat, but also
the living animals as an object for fishing; this is found in other
languages besides English, thus in Danish (fish), Russian (ryba,
Asboth, Gramm. 68), Magyar (Simonyi US. 259). In English and
Danish this has been one of the causes that have led to the use
of the unchanged plural as in many fish, mange fish.

Mass-words are often made into names for countables, though
languages differ considerably in this respect. Thus in E., but not
in Danish, tin is used for a receptacle made of tin (for sardines, etc.).
In English, bread is only a mass-word, but the corresponding word
in many languages is used for what in E. is called a loaf: un peu de
pain, un petit pain = a little bread, a small loaf.

Immaterial mass-words undergo a similar change of signification
when they come to stand for a single act or instance of the quality,
as when we talk of a stupidity — a stupid act, many follies or kindnesses,
etc. This usage, however, is not so universal in English
as in many other languages, and the best rendering of eine unerhörte
unverschämtheit is a piece of monstrous impudence, cf. also an insufferable
piece of injustice, another piece of scandal, an act of perfidy,
etc. (examples MEG II, 5. 33 ff.). This construction is strictly
analogous with a piece of wood, two lumps of sugar, etc.

In one more way mass-words may become thing-words, when a
nexus-substantive like beauty comes to stand for a thing (or a
person) possessing the quality indicated. And finally we must
mention the use of a mass-word to denote one kind of the mass:
this tea is better than the one we had last week; and then naturally in
the plural: various sauces; the best Italian wines come from Tuscany.

Through the term “mass-word” and through the restriction of the term
“collective” to a well-defined class of words, so that the two terms are
consistently opposed to one another (the notion of number being logically
inapplicable to mass-words, while it is doubly applicable to collectives)
I hope to have contributed something towards clarifying a difficult subject.
The necessity of a term like mass-word is seen in many places in dictionaries;
in the NED, for example, we often read definitions like the
following: “claptrap (1) with pl.: A trick… (2) without a or pl.:
Language designed to catch applause” — i.e. (1) as a thing-word, (2) as a
mass-word. My own division seems preferable to the two best thought-out
divisions I know, those of Sweet and Noreen.200

Sweet does not see the essential similarity between his ‘material nouns’
and ‘abstract nouns’; nor is his name ‘material nouns’ a fortunate one,
because many names of immaterial phenomena present the same characteristics
as iron or glass. Neither can I see the value of the distinction he
makes between singular class-nouns (like sun in popular as contrasted with
scientific language) and plural nouns (like tree): both represent ‘countables,’
even if there is more occasion in one case than in the other to use the word
in the plural.

Chapter XV
Number — concluded

Various Anomalies. The Generic Singular and Plural. Dual. Number in
Secondary Words. Plural of the Verbal Idea.

Various Anomalies.

In all languages there are words which serve the purpose of singling
out the individual members of a plurality and thus in the form of
a singular expressing what is common to all: every, each. There
is only a shade of difference between “everybody was glad” and
“all were glad” (cf. the neuter “everything” and all in “all is
well that ends well” = all things). Note also Lat. uterque vir,
utraque lingua, utrumque ‘each (either) of the two men, both men,
both languages, both things.’ A closely related case is that seen
in many a man, which individualizes, where many men generalizes;
thus also in many other languages: manch ein mann, mangen en
mand, mucha palabra loca (Hanssen, Sp. gr. § 56. 6), Fr. obsolete
maint homme.

Here and there we find anomalies in the use of number-forms
which are difficult to explain, but which at any rate show that people
are not absolutely rational beings, thus in OE. the use of the singular
with the tens, as in Beowulf 3042 se wæs fiftiges fotgemearces lang
‘it was 50 feet long,’ ib. 379 pritiges manna mægencræft ‘the strength
of 30 men,’ thus with some inconsistency, as fotgemearces is sg. and
manna is pl. — In Middle English we find the singular a before a
numeral, a forty men, meaning ‘about forty,’ thus very frequently
in Dan. en tyve stykher ‘about twenty (pieces),’ and this may be compared
with E. a few (in Jutland dialects æn lile fo); the sg. article
here turns the plural words from a quasi-negative quantity (he has
few friends) into a positive (he has a few friends). But a few may
have been induced by a many, where many may be the collective
substantive and not the adjective — the forms of these, which were
at first separated, have been confounded together. Fr. vers les une
heures (as well as vers les midi) with its numerical incongruity
is evidently due to the analogy of other indications of time such as
vers les deux heures; it is as if vers-les had become one amalgamated
preposition with denominations of the hour. The G. interrogative
pronoun wer, like E. who, above 198, is independent of number,
202but when one wants expressly to indicate that the question refers
to more than one person this may be achieved through the addition
of alles, in the singular neuter! “Wer kommt denn alles?” (‘Who
are coming?’ — “Wer kommt?” ‘Who is coming?’) “Wen hast
du alles gesehen?” — implying that he has seen several people.
Cf. what is said below under Sex on beides and mehreres as neuters
to the personal beide, mehrere (p. 237).

The Generic Singular and Plural.

One and the same generic truth is differently expressed in the
G. proverb “Ein unglück kommt nie allein” and E. “Misfortunes
never come singly” (cf. Shakespeare: “When sorrows come, they
come not single spies, But in battalions”). — Compare also twice a
week with deux fois la semaine.

The difference between this “indefinite person” and the generic
use of man (in “man is mortal”) is not easy to define, and seems
often to be emotional rather than intellectual. Hence also the
frequent use of man, one, si as a disguised “I” when one wishes
to avoid mentioning oneself, and therefore generalizes what one
wants to say: a similar motive leads to the use of you in the same
sense. But it is worth mentioning as something connected with
the “generic” character of the “indefinite person” that man or on
is not unfrequently followed by a plural word. Dan. “man blev
enige” | Fr. “la femme qui vient de vous jouer un mauvais tour
mais voudrait qu'on reste amis quand même” (Daudet, L'Immortel
151). 1113 Thus also in It. with si: Serao, l.c. 223 si resta liberi per
tre mesi | Rovetta, Moglie S. Ecc. 49 Si diventa ministri, ma si
nasce poeti, pittori!

Dual.

In languages possessing a dual, two different conceptions are
found. One is represented in Greenlandic, where nuna ‘land’
forms its dual nunak and its plural nunat; here “the dual is chiefly
used when the speaker wants expressly to point out that the question
is about a duality; if, on the other hand, the duality is obvious
as a matter of course, as in the case of those parts of the body
which are found in pairs, the plural form is nearly always employed.
Thus it is customary to say issai, his eyes, siutai his ears, talê his
arms, etc., not issik, siutik, tatdlik, his two eyes, etc. Even with
the numeral mardluk (two), which is in itself a dual, the plural is
often used, e.g. inuit mardluk two men” (Kleinschmidt, Gramm. d.
grönländ. spr. 13).

The other conception, according to which the dual is preferably
used in names of objects naturally found in pairs, as in Gr. osse
‘the eyes,’ is represented in Aryan. In many of the older languages
of this family duals were found; they tended to disappear as time
went on, and now survive only in a few isolated dialects (Lithuanian,
Sorb, Slovene; a few Bavarian dialects in the personal pronouns).
The gradual disappearance of dual forms in the Aryan languages 2114
presents many interesting features which cannot be here detailed.
205The existence of a dual is generally (Lévy-Bruhl, Meillet) looked
upon as a mark of primitive mentality; its disappearance is
therefore considered as a consequence or accompaniment of progress
in civilization, (In my own view of linguistic development any
simplification, any discarding of old superfluous distinctions is
progressive, though a causal nexus between civilization in general
and particular grammatical phenomena cannot be demonstrated
in detail.)

The Greek dual was lost at an early period in the colonies,
where the civilization was relatively advanced, while it was kept
more tenaciously in continental Greece, e.g. in Lacedæmon, Bœotia,
and Attica. In Homer duals are frequent, but they appear to be
an artificial archaism used for poetical purposes, especially for the
sake of the metre, while the plural is often used in speaking of two
even in the same breath as the dual (cp. collocations like amphō
kheiras, Od. 8. 135). In Gothic dual forms are found only in the
pronouns of the first and second persons and in the corresponding
forms of verbs, but these latter are few in number; and in the other
old Gothonic languages only the pronouns ‘we’ and ‘ye’ keep
the old distinction, which was later generally given up. (Inversely
the duals við, þið have ousted the old plurals vér, þér, in modern
Icelandic, and possibly also in Dan. vi, l.) Isolated traces of the
old dual have been found in the forms of a few substantives, such
as door (originally the two leaves) and breast, but even in these
cases from the oldest times the forms were understood not as duals,
but as singulars. The only words which may now be said to be
in the dual are two and both, but it should be noted that the
latter when used as a “conjunction” is often applied to more
than two, as in “both London, Paris, and Amsterdam”; though
this is found in many good writers, some grammarians object
to it. 1115

According to Gauthiot, the dual forms Sanskrit akṣī, Gr. osse,
Lithuanian akí do not properly mean ‘the two eyes,’ not even
‘the eye and the other eye,’ but ‘the eye in so far as it is double,’
thus mitrā is ‘Mitra, in so far as he is double,’ i.e. Mitra and Varuna,
for Varuna is the double of Mitra. Similarly we have Sanskr.
áhanī ‘the day and (the night),’ pitárāu ‘the father and (the
mother),’ mātárāu ‘the mother and (the father),’ and then also
pitárāu mātárāu ‘father and mother’ (both in the dual), and,
somewhat differently, Gr. Aiante Teukron te ‘Aias (dual) and
Teukros.’ Ugro-Finnic has parallels to most of these constructions,
thus both words are put in the plural in combinations like
206imeŋen igeŋen ‘the old man and the old woman,’ teteŋen tuŋgen
‘winter and summer.’

In some cases the lost dual left some traces behind it, the true
character of which has been forgotten. Thus in Old Norse, the
pronoun þau ‘they two’ is an old dual form, but as it happens to
be also the neuter plural, it leads to the syntactic rule that the neuter
plural is used when persons of the two sexes are spoken of together.

In Russian the old dual in some words happened to have the
same form as the genitive singular; cases like dva mužika ‘two
peasants’ then led to the use of the gen. sg. in other words, and,
curiously enough, after the notion of a dual had been entirely
forgotten, even after the words for 3 and 4, tri, četyre: četyre
goda ‘four years,’ etc.

Number in Secondary Words.

When Sweet (NEG, § 269) says that the only grammatical
category that verbs have in common with nouns is that of number,
he is right so far as actual (English) grammar is concerned; but
it should be remembered that the plural does not mean the same
thing in verbs as in substantives. In the latter it means plurality
of that which is denoted by the word itself, while in the verb the
number refers not to the action or state denoted by the verb, but to
the subject: compare (two) sticks or (two) walks with (they) walk,
which is in the plural, but implies not more walks than one, but
more walkers than one. In the same way, when in Latin and other
languages adjunct adjectives are put in the plural, as in urbes magnæ,
G. grosse städte, this does not indicate any plurality of the adjectival
idea, the plurality referring to ‘towns’ and to nothing else. In
both cases we have the purely grammatical phenomenon termed
“concord” which has nothing to do with logic, but pervaded all
the older stages of languages of the Aryan family; it affected not
only the number forms, but also the case forms of adjectival words,
which were “made to agree” with the primaries they belonged to.
But this rule of concord is really superfluous (cf. Language, 335 ff.),
and as the notion of plurality belongs logically to the primary
word alone, it is no wonder that many languages more or less
consistently have given up the indication of number in secondary
words.

In the adjectives, Danish, like German, still keeps up the distinction
between en stor mand (ein grosser mann) and store mænd
(grosse männer), while English is here more progressive and makes
no distinction between the singular and the plural in adjectives
(a great man, great men), the only survivals of the old rule of concord
being that man, those men, this man, these men. — In an ideal language
207neither adjuncts nor verbs would have any separate plural
forms. 1116

In Magyar there is the inverse rule that number is indicated
in a secondary and not in a primary word, but only when a substantive
is accompanied by a numeral. It is, then, put in the singular,
as if we were to say “three house.” This is termed “illogical”
by the eminent native linguist Simonyi: I should rather call it an
instance of wise economy, as in this case any express indication of
the plurality of the substantive would be superfluous. The same
rule is found in other languages; in Finnic with the curious addition
that in the subject not the nominative singular, but the partitive
singular is used; in the other cases there is agreement between
the numeral and the substantive. There is some approximation
to the same rule in Danish (tyve mand stærk, fern daler ‘five dollars,’
the value, different from fern dalere ‘five dollar pieces,’ to fod), in
German (zwei fuss, drei mark, 400 mann), and even in English (five
dozen, three score, five foot nine, five stone; details in MEG II, 57 ff.).

The first part of a compound substantive is in many respects
like an adjunct of the second. It is well known that in the ancient
type of Aryan compounds the stem itself is used, thus number is
not shown: Gr. hippo-damos may be one who curbs one horse, or
several horses. In E. the singular form is usually employed, even
when the idea is manifestly plural; as in the printed book section |
a three-volume novel. But in many, chiefly recent, formations the
plural is found in the first part: a savings-bank | the Contagious
Diseases Act. In Danish there is a curious instance of both parts
being inflected: bondegård, pl. bøndergårde ‘peasants' farms’;
generally the singular form of the first part is kept in the plural:
tandlæger, etc.

In verbs, English has discarded the distinction between singular
and plural in all preterits (gave, ended, drank, etc., with the sole
exception of was, were) and in some present tenses as well (can,
shall, must and others, which were originally preterits); where it
has been preserved, it is only in the third person (he comes, they
come), while in the first and second persons no difference is now
made (I come, we come, you come). In Danish the numerical distinction
has been totally given up in verbs, where the old singular
form has become a “common number”; it is always so in spoken
Danish, and now nearly always so in the literary language.

There seems to be a strong tendency everywhere to use the
singular form of the verb instead of the plural (rather than inversely)
208when the verb precedes the subject; the reason may often be that
at the moment of his uttering the verb the speaker has not made
up his mind what words are to follow. From OE I may quote
“Eac wæs gesewen on ðæm wage atifred ealle da heargas,” from
Shakespeare “that spirit upon whose weal depends and rests The
lives of many.” This is particularly frequent with there is (Thackeray:
there's some things I can't resist). It is the same in other
languages. In literary Danish it was the rule to have der er with a
plural subject at the time when ere was the form otherwise always
required when the subject was in the plural. Similarly very often
in Italian (“in teatro c'era quattro o sei persone”). The same
tendency to use the singular when the verb precedes is seen in the
same language when Evviva is used with a plural subject (Rovetta:
Evviva le bionde al potere!)

Those languages which have kept the old rule of concord in
secondary words are very often thereby involved in difficulties,
and grammars have to give more or less intricate rules which are
not always observed in ordinary life — even by the “best writers.”
A few English quotations (taken from MEG II, Ch. VI) will show
the nature of such difficulties with verbs: not one in ten of them
write it so badly | ten is one and nine | none are wretched but by
their own fault | none has more keenly felt them | neither of your
heads are safe | much care and patience were needed | if the death
of neither man nor gnat are designed | father and mother is man
and wife; man and wife is one flesh | his hair as well as his eyebrows
was now white | the fine lady, or fine gentleman, who show me their
teeth | one or two of his things are still worth your reading | his
meat was locusts and wild honey | fools are my theme | both death
and I am found eternal. All these sentences are taken from well-known
writers, the last, for instance, from Milton. Corresponding
difficulties are experienced in adjectival forms in those languages
which make their adjuncts agree in number (gender and case)
with primary words, and a simple comparison of Fr. ma femme et mes
enfants or la presse locale et les comités locaux with E. my wife and
children, the local press and committees shows the advantage to a
language of throwing overboard such superfluous distinctions in
secondary words. 1117209

Plural of the Verbal Idea.

The idea of “one or more than one” is not incompatible with
the idea expressed by the verb itself. I am not thinking here of
what R. M. Meyer (IF 24. 279 ff.) terms “verba pluralia tantum,”
for he speaks of such verbs as G. wimmeln, sich anhäufen, sich
zusammenrotten, umzingeln (English examples would be swarm,
teem, crowd, assemble, conspire), where the necessary plural idea is
not in the verb as such, but in the subject, 1118 but I am thinking of
those cases in which it is really the verbal idea itself that is made
plural. What that means is easily seen if we look first at the
corresponding verbal substantives nexus-substantives (see Ch. X).
If the plural of one walk or one action is (several) walks, actions, the
plural idea of the verb must be ‘to undertake several walks, to
perform more than one action.’ But in English and in most
languages there is no separate form of the verb to indicate this;
when I say he walks (shoots), they walk (shoot), it is impossible to
tell whether one or more than one walk (shot) is meant. If
we say “they often kissed” we see that the adverb expresses
exactly the same plural idea as the plural form (and the adjective)
in (many) kisses. In other words, the real plural of the verb is
what in some languages is expressed by the so-called frequentative
or iterative — sometimes a separate “form” of the verb which is
often classed with the tense 2119 or aspect system of the language in
question, as when repetition (as well as duration, etc.) is in Semitic
languages expressed by a strengthening (doubling, lengthening)
of the middle consonant, or in Chamorro by a reduplication of the
stressed syllable of the verbal root (K. Wulff, Festschrift Vilh.
Thomsen 49). Sometimes a separate verb is formed to express
repeated or habitual action, thus in some cases in Latin by means
of the ending -ito: cantilo, ventito ‘sing frequently, come often’;
visito is from a formal point of view a double frequentative, as
it is formed from viso, which is in itself a frequentative of video,
but the plural idea tends to disappear, and Fr. visiter, E. visit
may be used of a single coming. In Slav this category of plural or
frequentative verbs is well developed, e.g. Russ. strělivat' ‘to fire
several shots,’ from strěljat' ‘to fire one shot.’ In English several
verbs in -er, -le imply repeated or habitual action: stutter, patter,
chatter, cackle, babble. Otherwise repeated action must be rendered
in various other ways: he talked and talked | he used to talk
of his mother | he was in the habit of talking | he would talk of
210his mother for hours | he talked of his mother over and over
again, etc.

Having mentioned the plural of such verbal substantives as
walk shot, kiss, we may remind the reader of the other kind of
“nexus substantives,” those containing a predicative, such as
stupidity, kindness, folly. These also may be put in the plural,
though, as remarked above, they are then changed from mass-words
into countables (as they are indeed when the singular is
used with the indefinite article: a stupidity = ‘a stupid act, an
instance of being stupid’).

Adverbs, of course, have no distinct number, the only exceptions
being such adverbs as twice, thrice, often, which may be said to
be plurals of once because logically these adverbs are equivalent
to ‘two times, three times, many times’; the plural idea thus
refers to the substantival idea contained in the subjunct, just as in
group subjuncts like “at two (three, many) places.” Similarly the
groups now and then, here and there may be said to contain a plural
idea, as they signify the same thing as ‘at various times, at various
places,’ but this, of course, does not affect the truth of the general
assertion that the notion of number is inapplicable to adverbs.

Appendix to the Chapters on Number.

To indicate place in a series most (all?) languages have words derived
from (cardinal) numerals; these are called ordinals. Very often the first
ordinals are not formed from the corresponding cardinals in the usual way:
primus, first, erst bear no relation to unus, one, ein, but from the beginning
denote foremost in point of place or time. Lat. secundus originally means
‘following’ and leaves it to the imagination to infer how many precede;
frequently we have a word for 2nd which at the same time has the vague
meaning ‘different,’ thus OE oðer (preserved in the indefinite sense in MnE
other, while the cardinal has been taken from French), G. ander, Dan. anden.
In French there is a new regular formation from deux: deuxième (at first
probably used in combinations like vingt-deuxième, cf. vingt-et-unième).

In many cases cardinals are used where a stricter logic would require
ordinals; this is due to considerations of convenience, especially where
high numbers are concerned, thus in 1922= the 1922nd year after Christ's
birth (Russian here uses the ordinal); further in reading such indications
as “line 725,” “page 32,” “Chapter XVIII,” etc., in French also in
“Louis XIV,” “le 14 septembre,” etc.

After the word for “number” (numero, etc.) this use of the cardinal
instead of the ordinal is universal: “number seven” means the seventh
of a series. Cf. also the indication of the “hour”: at two o'clock, at three-fifty.

Note the use of the ordinal in G. drittehalb, Dan. halvtredie ‘two and a
half’ (the third is only half), and the somewhat different usage in Scotch
at half three, Dan. klokken halv tre, G. um halb drei uhr ‘at half-past two.’

In many languages ordinals (with or without the word for ‘part’ added)
have to do duty to express fractions: five-sevenths, cinq septièmes, fünf
siebentel, fern syvendedel, etc. For 1/2, however, there is a separate word
half, demi, etc.211

Chapter XVI
Person

Definitions. Common and Generic Person. Notional and Grammatical
Person. Indirect Speech. Fourth Person. Reflexive and Reciprocal
Pronouns.

Definitions.

In the NED “person” as used in grammar is defined as follows:
“Each of the three classes of personal pronouns, and corresponding
distinctions in verbs, denoting or indicating respectively the person
speaking (first person), the person spoken to (second person), and
the person or thing spoken of (third person).” But though the
same definition is found in other good dictionaries and in most
grammars, it is evidently wrong, for when I say “I am ill” or
“you must go” it is undoubtedly “I” and “you” that are
spoken of; the real contrast thus is between (1) the speaker,
(2) spoken to, and (3) neither speaker nor spoken to. In the first
person one speaks of oneself, in the second of the person to whom
the speech is addressed, and in the third of neither.

Further, it is important to remember that in this use the word
“person” qualified with one of the first three ordinals means
something quite different from the ordinary signification of
“person” and does not imply “personality” as a human or rational
being; “the horse runs” and “the sun shines” are in the third
person; and if in a fable we make the horse say “I run” or the
sun say “I shine,” both sentences are in the first person. This
use of the word “person,” which goes back to Latin grammarians
and through them to Greek (prosōpon) is one of the many inconveniences
of traditional grammatical terminology which are too
firmly rooted to be now abolished, however strange it may be to
an unsophisticated mind to be taught that “impersonal verbs”
are always put in the “third person”: pluit, it rains. Some people
have objected to the inclusion of a pronoun like it among “personal
pronouns,” but the inclusion is justified if we take the expression
“personal pronoun” to mean pronoun indicating person in
the sense here mentioned. But when we come to speak of the
distinction between the two interrogative pronouns who and what,
and find that the former refers to persons and the latter to anything
212that is not a person, we might feel inclined to call who a personal
pronoun, — which would be decidedly awkward.

It is a simple consequence of the definition that the first person,
strictly speaking, is found in the singular only; 1120 in a preceding
chapter (p. 192) mention has already been made of the fact that
the so-called first person plural “we” is really “I + someone else
or some others,” and in some works dealing with Amerindian
languages the figures 1/2 and 1/3 are conveniently used to designate
“we” according as the others that are added to “I” are of the
second or third persons respectively.

For the curiosity of the matter I may quote here a sentence
to illustrate the emotional value of the three persons. “With
Ruskin the people are always ‘You’; with Carlyle they are even
farther away, they are ‘they’; but with Morris the people are
always ‘We’” (William Morris, by Bruce Glacier).

In many languages the distinction between the three persons
is found not only in pronouns, but in verbs as well, thus in Latin
(amo, amas, amat), Italian, Hebrew, Finnish, etc. In such languages
many sentences have no explicit indication of the subject,
and ego amo, tu amas is at first said only when it is necessary or
desirable to lay special stress on the idea “I, thou.” In course
of time, however, it became more and more usual to add the pronoun
even when no special emphasis was intended, and this paved
the way for the gradual obscuration of the sound of the personal
endings in the verbs, as these became more and more superfluous
for the right understanding of the sentences. Thus in Fr. j'aime,
tu aimes, il aime, je veux, tu veux, il veut, je vis, tu vis, il vit are
identical in sound. In English we have the same form in I can,
you can, he can, I saw, you saw, he saw, and even in the plural we
can, you can, they can, we saw, you saw, they saw — phonetic and
analogical levellings have gone hand in hand to wipe out old personal
distinctions. These, however, have not disappeared entirely,
survivals being found in Fr. j'ai, tu as, il a, nous avons, vous avez,
ils ont, and in E. I go, he goes, and generally in the third person
singular of the present tense. In modern Danish all these distinctions
have disappeared: jeg ser, du ser, han ser, vi ser, I ser,
de ser, and so in all verbs and all tenses, exactly as in Chinese and
some other languages. This must be considered the ideal or logical
state of language, as the distinction rightly belongs to the primary
idea only and need not be repeated in secondary words.213

In English a distinction has developed in the auxiliary verbs
used to express futurity: I shall go, you will go, he will go, and
correspondingly to express conditional unreality: I should go, you
would go, he would go.

Any imperative (and we might add, any vocative) is virtually
in the second person, even in such cases as “Oh, please, someone
go in and tell her” or “Go one and cal the Iew into the court”
(Sh.), as seen clearly, for instance, by the addition in “And bring
out my hat, somebody, will you” (Dickens). In English the form
of the verb does not show which person is used, but other languages
have a third person of the imperative, in which case we must say
that there is a conflict between the grammatical third person and
the notional second person. Sometimes, however, the latter
prevails, even in form, as when in Greek we find “sigan nun
hapas ekhe sigan” where ekhe (2nd p.) according to Wackernagel
(VS 106) stands instead of ekhetō (3rd p.): ‘everyone now hold
silence.’ Where we have a first person plural in the imperative,
as in It. diamo, Fr. donnons, the virtual meaning is ‘you give,
and I will give, too,’ and so the imperative here as always refers
to the second person. In English the old give we has been supplanted
by let us give (as in Danish and, to some extent, also in German);
here let, of course, is, grammatically as well as notionally, in the
second person, and the first person pl. is only shown in the dependent
nexus us give.

The local adverb corresponding to the first person is here, and
where we have two adverbs for ‘not-here,’ as in northern English
dialects there and yonder (yon, yond), we might say that there corresponds
to the second, and yonder to the third person; 1121 but very
often there is only one adverb for both ideas, as in Standard English
there (yonder being obsolete). The connexion between the first
person and ‘here’ is seen in Italian, where the adverb ci ‘here’
is used very extensively as a pronoun of the first person plural in
the oblique cases instead of ni ‘us.’ In German we have the two
adverbs of movement, hin for a movement towards, and her for
a movement away from, the speaker.

I mention this as an interesting view, though in this volume
I generally keep aloof from speculations about primitive grammar
and the origin of grammatical elements.

Common and Generic Person.

We have seen above (p. 198) that it is, or would be, convenient
in some cases to have a form for a “common number”; in the
same way the want of a “common person” is also sometimes felt.
As already remarked, we is really a case in point, as it stands for
“I and you” or “I and someone else,” and the plural you, ye
also often stands for “thou and someone else” and thus combines
the second and third persons. But this does not cover the instances
in which the two persons are not joined by means of and,
but separated, for instance, by a disjunctive conjunction. Here
we have considerable difficulties in those languages which distinguish
persons in their verbs: “either you or I are (or am or is?) wrong”;
see the examples given in Language, p. 335 f. Note also the use of
our in “Clive and I went each to our habitation” (Thackeray,
Newc. 297), where it would also be possible to say: “… each to
his home,” and where Danish certainly would use its reflexive
pronoun of the third person: “C. og jeg gik hver til sit hjem”
(cp. “vi tog hver sin hat”), but a common-person form would be
more logical.

A curious case in which a common-person form would have
solved the difficulty is mentioned by Wackernagel (VS 107):
Uter meruistis culpam (Plautus) ‘which of you two has deserved
blame?’ — uter would require the third person singular, but the
verb is put in the second person plural because two men are
addressed.

The choice between these several expressions depends on a more
or less emotional element: sometimes one wants to emphasize
the fact that one is included oneself in the general assertion, sometimes
one wants to make a kind of special appeal to the person
addressed at the moment, 1122 and sometimes one wants to keep one's
own person in the background, though what is meant is really the
first person more than anything else (one, a fellow). But the name
“generic person” covers the notion underlying all these uses of
various grammatical persons.

It is interesting to notice that in some languages the pronoun
for ‘we’ is disappearing and is being replaced by the generic
expression (‘one’). Thus in French “Je suis prêt, est-ce qu'on
part?” for … nous partons (Bally, LV 59); from Benjamin,
Gaspard, I quote “Nous, on va s'batte, nous on va s'tuer” (with
strong emphasis of contrast on nous, p. 13), and “Moi, j'attends
le ballet, et c'est nous qu'on dansera avec les petites Allemandes”
and it is we who will dance, p. 18). In Italian this is quite common:
Verga Eros 27 la piazzetta dove noi si giocava a volano | Fogazzaro
Dan. Cortis 31 noi si potrebbe anche partire da un momento all'
altro | id. Santo 139 la signora Dessalle e io si va stamani a visitare
i Conventi | 216 Noi si sa che lui non vole andare. 2123 The frequency
of this phenomenon in Italian seems to show that the reason for
it cannot be that suggested by Bally, I.e., that in the first person
plural nous chantons the verb has preserved a special ending which
is useless and does not harmonize with those of je chante, tu chantes,
il chante, ils chantent, which have become alike in pronunciation
(but then what about vous chantez?). But Bally is probably right
when he says that while the forms moi je chante, toi tu chantes, lui
il chante, eux ils chantent are perfectly natural, the combination
with emphatic first person pl. nous nous chantons is obscure and
216inharmonious, and that therefore the form nous on has been preferred
as more satisfactory to the ear and to the mind.

Notional and Grammatical Person.

In the vast majority of cases there is complete agreement
between notional and grammatical person, i.e. the pronoun “I”
and the corresponding verbal forms are used where the speaker
really speaks of himself, and so with the other persons. Still,
deviations are by no means rare; servility, deference, or simply
politeness, may make the speaker avoid the direct mention of
his own personality, and thus we may have such third-person
substitutes for “I” as your humble servant; cf. Spanish “Disponga
V., caballero, de este su servidor.” In languages of the east this
is carried to an extreme, and words meaning originally ‘slave’
or ‘subject’ or ‘servant’ have become the normal expressions for
“I” (see, e.g., Fr. Müller, Gr. II, 2. 121). In Western Europe,
with its greater self-assertion, such expressions are chiefly used
in jocular speech, thus E. yours truly (from the subscription in
letters), this child (vulgarly this baby). A distinctively self-assertive
jocular substitute for “I” is number one. Some writers avoid
the mention of “I” as much as possible by using passive constructions,
etc., and when such devices are not possible, they say the
author, the (present) writer, or the reviewer. A famous example
of self-effacement in order to produce the impression of absolute
objectivity is Cæsar, who in his commentaries throughout uses
Cæsar instead of the first pronoun. But it is, of course, different
when the same trick of using one's own name instead of the personal
pronoun is used by Marlowe's Faustus or Shakespeare's Julius
Cæsar or Cordelia or Richard II, or Lessing's Saladin, or Oehlenschläger's
Hakon (many examples from German, Old Norse, Greek,
etc., in Grimm's Personenwechsel, 7 ff.). In some cases this may
be a kind of introduction of oneself to the audience, but generally
it is the outcome of pride or haughtiness. Still another case is
found when grown-up people in talking to small children say
“papa” or “Aunt Mary” instead of “I” in order to be more
easily understood. 1124

Present company may sometimes be used instead of “we,”
“us”: “You fancy yourself above present company.”

Among substitutes for notional second person I shall first
mention the paternal we, often used by teachers and doctors
217(“Well, and how are we to-day?”) and denoting kindness through
identifying the interests of speaker and hearer. This seems to be
common in many countries, e.g. in Denmark, in Germany (Grimm,
Personenwechsel, 19), in France (Bourget, Disc. 94 “Hé bien, nous
deviendrons un grand savant comme le père?” | Maupassant,
Fort c. l. m. 224 “Oui, nous avons de l'anémie, des troubles nerveux”
— immediately followed by vous). The usual tinge of
protection in this we is absent from the frequent Danish “Jeg
skal sige os” (Let me tell you).

Next we have the deferential substitutes consisting of a possessive
pronoun and the name of a quality: your highness (= you
that are so high), your excellency, your Majesty, your Lordship,
etc. It is well known that in Spanish vuestra merced ‘your grace,’
shortened into usted, has become the usual polite word for ‘you.’
In French, Monsieur, Madame, Mademoiselle may be used instead
of vous (Monsieur désire? etc.). In countries in which great stress
is laid upon titles the simple and natural personal pronouns have
often to give way to such expressions as abound in German and
Swedish: “Was wünscht (wünschen) der herr lieutenant?”
“Darf ich dem gnädigen fräulein etwas wein einschenken?” etc.
In Sweden it is not easy to carry on a polite conversation
with a person whose title one is ignorant of or happens to
have forgotten; and I am sorry to say that my own countrymen
of late years have begun more and more to imitate
their neighbours to the South and to the East in this respect,
and to ask “Hvad mener professoren?” instead of “Hvad
mener De?”

In German it was formerly usual to say er, sie with the verb
in the third person singular instead of du, especially in speaking
to inferiors, and the corresponding practice (han, hun) prevailed
in Denmark until well into the nineteenth century. The third
person plural Sie has now become the usual polite word for notional
second person (sg. and pl.) in German, and this usage, which Grimm
rightly calls an indelible stain on the German language, 1125 has been
servilely imitated in Denmark: De.

There is a different use of the third person for a notional second
person which may be illustrated from Shaw's play, where Candida
says to her husband: “My boy is not looking well. Has he been
overworking?” Similarly a lover may say my darling or my own
girl instead of you. There is also a petting way of addressing a
child as it, which may have originated in the habit of half mentioning,
half addressing an infant that is too small to understand what
is being said to it. This, too, may be exemplified from Candida,
218who says to Marchbanks: “Poor boy! have I been cruel? Did
I make it slice nasty little red onions?”

With the English possessive compounds with self (myself,
yourself) we have a conflict between the grammatical person (third)
and the notional person (first, second); the verb is generally made
to agree with the notional person (myself am, yourself are), though
occasionally the third person is used (Shakespeare sometimes has
my self hath, thy self is, etc.).

Indirect speech.

It should be remarked, however, that in the cases 2, 5, 8, and
9 clearness would certainly gain by the use of the name instead
of one or more of the ambiguous he's.

It is a simple consequence of the nature of the plural we, that
it frequently remains unshifted, as in: “He said that he still
believed in our glorious future as a nation.”

In English the auxiliary shall (should) is often used in reported
speech to show that the second or third person is a shifted first
person: “Do you think you shall soon recover?” “He thought
219he should soon recover” — contrast with this the continuation
“but the Doctor knew that he would die.”

There is a rather unusual case of a shifted personal (possessive)
pronoun in the Merchant of Venice (II. 8. 23): Shylock exclaims
“My stones, my daughter, my ducats,” and when the street-boys
mimic him, this is reported: “Why all the boyes in Venice follow
him, Crying his stones, his daughter, and his ducats.” Here the
direct speech would be more natural. In Icelandic sagas it is
quite usual to find that the beginning of a reported speech only is
shifted, and that after one sentence the rest is given in the exact
form in which the speech had been made.

Fourth person.

Should we recognize a fourth person by the side of the third?
This was the opinion of Rask (Vejledning 1811, 96, Prisskr. 1818,
241), who said that in “he beats him” him is in the fourth, while in
“he beats himself” himself is in the third person like the subject.
(Inversely, Thalbitzer, in Handbook of American Ind. Lang. 1021,
denotes by “fourth person” the reflexive.) Yet it is easy to see
that if we accept the definition of “person” given above, both
these are in the third person, and that no fourth “person” is
thinkable, however true it is that the same pronoun or verbal
form (in the third person) may refer to different beings or things,
in the same or in successive sentences.

Some Amerindian languages have very subtle distinctions, see
Uhlenbeck, Grammatische onderscheidingen in het Algonkinsch
(Akad. van Wetensch., Amsterdam, 1909): in Chippeway the first
time a third person is mentioned this is not especially marked,
but the subordinate second tertia persona, also called obviativus,
is marked by a suffix -n, and the third tertia persona (called superobviativus,
by Uhlenbeck subobviativus) by the suffix -ini. In
“Joseph took the boy and his mother” the boy is the second, and
his mother the third tertia persona, and it is exactly indicated whether
his refers to Joseph or to the boy. This makes Brinton (Essays
of an Americanist, Philadelphia, 1890, 324) regret the poverty
of English, where the sentence “John told Robert's son that he
must help him” is capable of six different meanings which in
Chippeway would be carefully distinguished. Nevertheless, it
must be said that nearly always the meaning of such pronouns
as he and his will be made sufficiently clear by the situation and
context, even in such sentences as these (Alford): “Jack was very
respectful to Tom, and always took off his hat when he met him.”
“Jack was very rude to Tom, and always knocked off his hat
when he met him.” Sully relates how a little girl of five was much
220puzzled by the old hymn: “And Satan trembles when he sees
The weakest saint upon his knees.” — “Whatever, she asked, did
they want to sit on Satan's knees for?”

Note also the fun that was made of the Kaiser's telegram (1914)
to the Crown Princess: “Freue mich mit dir über Wilhelm's
ersten sieg. Wie herrlich hat Gott ihm zu seite gestanden. Ihm
sei dank und ehre. Ich habe ihm eisernes kreuz zweiter und
erster klasse verliehen.”

In the spoken language extra stress serves in many cases to
remove any ambiguity and to show who is meant. In John Stuart
Mill's Essay on Poetry we read: “Shelley is the very reverse of
all this. Where Wordsworth is strong, he is weak; where Wordsworth
is weak, he is strong.” This makes nonsense if read with
unstressed he, for that would mean Wordsworth, but it gives perfect
sense if read with stressed he, which then comes to mean Shelley;
it might even be readily understood if after stressing the first he
we substitute a weak he for the second Wordsworth. This clarifying
stress is indicated by the italicizing of they in Lamb's sentence:
“Children love to listen to stories about their elders, when they
were children.” In Somersetshire dialect Bill cut's vinger means
‘his own,’ Bill cut ees vinger means ‘the other person's.’

Reflexive and Reciprocal Pronouns.

Many languages have developed reflexive pronouns, by means
of which many ambiguities are obviated. Their function is to
indicate identity with what has been mentioned before, in most
cases with the subject, whence it comes that these pronouns generally
have no nominative.

In the Aryan languages we have the pronouns originally beginning
with sw-, but their sphere of application is not everywhere
the same, so it may be of some interest to give a short survey of
their employment in the languages best known to us.

(1) Originally the reflexive pronoun was used in all three persons
and without any regard to number, e.g. in Sanskrit and in the oldest
Greek. This use is still preserved in Lithuanian and Slav, e.g. Russian
ty vrediš' sebě ‘you hurt yourself,’ my dovol'ny soboju ‘we are pleased
with ourselves’ (examples taken from H. Pedersen's grammar).

(2) In many languages the reflexive pronoun has been restricted
to the third person, whether singular or plural; thus Lat. se and
the forms derived from this in Romanic languages; further G
sich, ON. sik, Dan. sig, though, as we shall see immediately, with
some restrictions.

(3) In the dialects of Jutland this pronoun sig is used only
when referring to a singular subject; when referring to a plural
221subject dem is used. This use of dem instead of the received
sig is not at all rare in literary Danish, even in writers who were
not born in Jutland; thus Kierkegaard writes, Enten Ell. 1. 294
naar de ikke kede dem,

(4) While in German the polite pronoun Sie (notional second
person) takes the reflexive sich: Wollen Sie sich setzen, the Danish
imitation De is always now followed by Dem: Vil De ikke sætte
Dem (in the eighteenth century sometimes sig).

(5) Though the Fr. unstressed form se is used of any third
person subject in both numbers, the stressed form soi is restricted
to the singular and is generally used only when referring to an
indefinite subject: ce qu'on laisse derrière soi, but of a definite
subject: ce qu'il laisse derrière lui, ce qu'elle laisse derrière elle
(ce qu'ils laissent derrière eux). Exceptions to this rule are found
now and then, thus pretty frequently in Holland, e.g. J. Chr. 7. 81
Il était trop peu sûr de soi pour ce rôle (also ib. 3. 213, 4. 6).

(6) English very early went further than any of the related
languages, as the only remnant of the reflexive pronouns — and
that only in the oldest period — was the possessive sin (see below).
The old expressions, therefore, were “I wash me, thou washest
thee, he washes him, she washes her, we wash us, ye wash you,
they wash them.” Survivals of this are found in prepositional
combinations like “I have no money about me, he has no money
about him,” etc. In many cases the simple verb besides its transitive
function has now also a reflexive meaning: “I wash, dress,
shave,” etc. But in most cases the reflexive meaning is expressly
indicated by the combinations with self: “I defend myself, you
defend yourself (yourselves), he defends himself,” etc. In this
way reflexive pronouns have developed which differ from the original
Aryan ones in distinguishing the three persons and the two numbers,
and thus resemble those of Finnish, which are formed by means of
itse, to which are appended the usual possessive suffixes: itseni
myself, itsemme ourselves, itsesi yourself, itsensä himself (herself),
etc. Compare also the later Greek emauton, seauton, heauton, etc.,
and especially the curious Modern Greek formations ton emauto
mou myself, ton emauto sou yourself, ton emauto sas yourselves,
ton emauto tou, tēs himself, herself, ton emauto mas ourselves, etc.

The development of the reflexive possessive has followed the
same lines, though it has not been completely parallel with that
of se, etc.

(1) To begin with, it referred to all persons in all numbers.
This is still the Russian usage, e.g. ja vzjal svoj platok ‘I took my
pocket-handkerchief.’

(2) It is restricted to the third person, but may refer to plurals
as well as to singulars. This stage is found in Lat. suus and in
222the old Gothonic languages, e.g. Gothic Lk 6. 18 qemun hailjan
sik sauhte seinaizo ‘they came to be healed of their diseases’ |
Mk 15. 29 wiþondans haubida seina ‘shaking their heads.’ The
OE. poetical sin is found corresponding to ‘his’ and ‘her,’ but
only rarely referring to a plural subject, and the pronoun seems
to have disappeared pretty early from ordinary conversational
language. ON sinn may refer to plural as well as to singular subjects;
this use is still found in Norwegian: de vasker sine hander
‘they wash their hands,’ and in Swedish.

(3) But in Danish sin is used only with a subject in the singular:
han (hun) vasker sine hænder; de vasker deres hænder.

(4) In the dialects of Jutland we have the further restriction
that sin refers to an indefinite subject only: enhver (en) vasker
sine hænder, but han vasker hans hænder, hun vasker hender hænder.

(5) In some languages this pronoun has lost its reflexive power
and is used as a general possessive of the third person singular,
thus in French, where ses mains can be used in any position, meaning
‘his or her hands.’

(6) Thus also in German, only with the restriction that it
means only ‘his’ (or ‘its’): seine hände ‘his hands,’ but in the
fern, ihre hände ‘her hands.’ 1126

Considerations of space prevent me from dealing here with
the question of the range of reflexive pronouns, which differs
widely in the languages possessing them, especially in participal
and infinitival constructions and dependent clauses. 2127

Where reference is possible to two different persons in complicated
combinations the existence of a reflexive pronoun is in some
cases no security against ambiguity, as in Lat. “Publius dicit
223Gaium se occidere voluisse,” or in Dan. “han fandt Peter liggende
i sin seng,” which is no more clear than the E. “he found Peter
lying in his bed.” Cf. the German use of dessen, where sein would
be ambiguous: “Der graf hat diesem manne und dessen sohne
alles anvertraut” (Curme GG 168).

Closely related to the reflexive pronouns are the reciprocal
pronouns, meaning ‘each other’: each part of those mentioned
as the subject acting upon (or with regard to) and being in turn
acted upon by all the other parts. This meaning is often expressed
by the simple reflexive pronoun, either alone as in Fr. ils se haïssent
or with some addition, as in ils se haïssent entr'eux, Lat. inter se
confligunt, Goth. Mk 1. 27 sokidedun miþ sis misso, cf. G. sie half en
sich gegenseitig, or in Fr. ils se sont tués l'un l'autre (as ils se sont
tués might be taken to mean ‘they have committed suicide’).
Combinations like l'un l'autre are also used without any reflexive
pronouns in various languages, where they always tend to become
one inseparable whole, as they have done in Gr. allēlous, Dan.
hinanden, hverandre, Dutch elkaar, mekaar, G. einander. On
the development of the German word see the interesting article
in Grimm's Wörterbuch, which also gives corresponding expressions
from various other languages (Romanic, Slav, Lithuanian, Keltic).
In English the elements formerly separated, as in Shakespeare's
gazed each on other or what we speak one to another, have now in
ordinary language been fused together: gaze on each other, speak
to one another. In Russian drug druga is separated by a preposition
(drug s drugom with one another), but the tendency to look upon
the combination as a unit is shown by the fact that it is used uniformly
without regard to gender and number (Boyer and Speranski,
M 273). Magyar egy-mas seems to be simply a translation of G.
einander. 1128

Reciprocal pronouns are sometimes found as the subject of
a dependent clause, thus in a recent English novel: “Miss C. and
I are going to find out what each other are like.” Similar sentences
may be heard in Danish.

Many grammars deal with the theory of reflexives in a chapter
about various kinds of verbs, giving “reflexive verbs” as one kind
(and “reciprocal verbs” as another). But surely the verb is exactly
the same in “we hurt him,” “we hurt ourselves,” “we hurt one
224another,” the only difference being the identity or non-identity
of subject and object. Thus also G. “ich schmeichele mir,” “ich
spotte meiner” contain the same verb as “ich schmeichele dir,”
“ich spotte semer.” The only cases in which one might fairly
speak of a reflexive verb would be those in which a verb is found
idiomatically with no other object than a reflexive pronoun, as
in E. I pride myself, Dan. jeg forsnakker mig, G. ich schäme mich.
The identity of subject and object (direct or indirect) influences the
choice of the auxiliary in Fr. il s'est tué ‘he has killed himself,’
nous nous sommes demandé ‘we have asked ourselves (or one
another)’. It is a different thing that what is expressed in our
languages with a reflexive pronoun may in some languages be
expressed by a separate form of the verb, as in the Greek “middle
voice”: louomai ‘I wash myself,’ etc. (the same form having also
a passive signification, see Ch. XII, p. 168). In Scandinavian the
reflexive pronoun sik has in a reduced form been fused with many
verbal forms, which then generally have acquired a purely passive
meaning: han kaldes, originally ‘he calls himself,’ now ‘he is
called.’ Sometimes the meaning is reciprocal: de slås (with a
short vowel) ‘they fight (strike one another)’; in this verb there
is another form with a long vowel (and glottal catch) for the passive
slå(e)s ‘is struck.’ In Russian the reflexive pronoun tends in a similar
way to be fused with verbs in the two forms sja and s' (in spite
of the spelling pronounced with a non-palatalized s); on the various
meanings (distinctly reflexive, vaguely reflexive, reciprocal, approximately
passive) see H. Pedersen RG 190, Boyer and Speranski M 247.225

Chapter XVII
Sex and gender

Various Languages. Aryan Gender. Sex. Common Sex. Animate and
Inanimate. Conceptional Neuter.

Various languages.

By the term gender is here meant any grammatical class-division
presenting some analogy to the distinction in the Aryan languages
between masculine, feminine, and neuter, whether the division be
based on the natural division into the two sexes, 1129 or on that between
animate and inanimate, or on something else. While a great many,
probably the vast majority, of languages, have no gender in this
sense, there are some languages which divide nouns into gender
classes. Only the briefest mention of some of these class-distinctions
can here be given, just enough to show, on the one hand the similarities,
and on the other hand the dissimilarities with our own system.

In the Bantu languages of South Africa every substantive
belongs to one of several classes, each of these being characterized
by its own prefix, which is repeated in a more or less weakened
form as a “reminder” in all subordinate words referring to the
substantive in question, whether adjuncts or verbs. Some of
these classes imply the singular, others the plural number, but
none of them has any reference to sex, though some are used mainly
of living beings and others of things. The number of the classes
varies in different languages belonging to the group, the maximum
being sixteen, but some of the classes are apt to be confounded,
and it is not possible to indicate the ultimate reason for the division.
(See Lang. 352 ff. and the works there quoted.)

‘Heavy’ when said of a man is watshi, of a woman jatshi, of
a thing batshi, and heaviness correspondingly is watshol, jatshol,
batshol. Wašo is brother, jašo sister, woh boy, joh girl.

In the related Tshetshensian ‘I am’ is suo wu when spoken by
a man, suo ju by a woman, suo du by a child (Fr. Müller, Grundriss
III, 2. 162).

In Andaman one class comprises inanimate things, another
animate beings, which are subdivided into human and non-human.
There is a sevenfold division of parts of the human body, but this
division is transferred to inanimate things that have some relation
to these several parts of the human body (P. W. Schmidt, Stellung
der Pygmäervölker, 121).

Algonkin languages have a distinction between animate and
inanimate, though the distribution presents many points that to
us appear strange, as when parts of the human body are generally
looked upon as inanimate, while various parts of the bodies of
animals are reckoned among animate things. (See J. P. B. Josselin
de Jong, De Waardeeringsonderscheiding van Levend en Levenloos,
Leiden, 1913, which compares this system and the Aryan genders,
and discusses the theories advanced about the origin of the latter.)

In Hamitic languages we have a partition into two classes,
one comprising names of persons, of big or important things, and
of males, and the other those of things, small things, and females,
sometimes with the curious rule that words of the first class in
the plural belong to the second class, and vice versa. By interchange
of the same prefixes we thus turn man into small man,
brother into sister, and he-dog into bitch or small dog; in Bedauyo
ando ‘excrement’ is masculine of a horse, ox, or camel, feminine
of smaller animals. A woman's breast is masculine, a man's
(because smaller), feminine. (Meinhof, Spr. der Hamiten, 23, and
passim; Die mod. sprachforsch. in Afrika, 134 ff.)

The genders of the Semitic languages are generally considered
as most similar to the Aryan genders, though there is no neuter,
and though in Semitic even verbal forms are made to agree with
the gender (sex) of the subject. Thus Arabic katabta ‘thou (m.)
hast written,’ katabti ‘thou (f.) hast written,’ kataba ‘he has written,’
katabat ‘she has written,’ plural 2. pers. katabtum (m.), katabtunna
(f.), 3. pers. katabū (m.), katabna (f.); in the first person no such
distinction is found: katabtu ‘I have written,’ katabnā ‘we have
written.’

Aryan gender.

Our own family of Aryan languages in the earliest historically
accessible forms distinguishes three genders, masculine, feminine,
and neuter, the last of which may to some extent be considered
227a subdivision of masculine, characterized chiefly by making no
distinction between the nominative and the accusative. The distribution
of words into these three classes is partly rational, partly
irrational. It is rational in so far as many names of male beings
are of the masculine gender, many names of females being feminine,
and many names of sexless things neuter. But by the side of this
we find in some cases names of male beings as feminines or neuters,
names of female beings as masculines or neuters, and names of things
or ideas without a natural sex as either feminines or masculines. 1130
I have spoken about various attempts to explain the origin of this
singular system or want of system in Language, p. 391 ff., 2131 and of
the practical disadvantages of it, ibid. 346 ff. It may be possible
to assign reasons why some words have a certain gender; thus
Handel Jakób has recently pointed out (Bulletin de l'Acad. polonaise
des Sciences, 1919-20, p. 17 ff.) that words meaning ‘earth’ (Gr
khthōn, khōra, Lat. terra, Slav, ziemia, G. erde) are made f., because
the earth is thought of as a mother producing plants, etc.; similarly
names of trees, because these bring forth fruits; he adduces
some Semitic parallels. But the main problem remains, why is
this classification extended to all words, even where it is not possible
to see any connexion with natural sex? Why, to take only
one instance, is the common Aryan word for ‘foot’ (pous, pes,
fot, etc.) m., while the various unconnected words for ‘hand’
are f. (kheir, manus, handus, ruka)? Words for ‘table, thought,
fruit, thunder,’ etc., are in one language m., in another f. It is
certainly impossible to find any single governing-principle in this
chaos.

Gender is shown partly by form, as when in Latin the nom. and
acc. are distinguished in rex regem m., lex legem f., while the two
cases are identical in regnum n., but it is chiefly a syntactic phenomenon,
different forms of adjeotives and pronouns being required
with the different genders: ille rex bonus est, illa lex bona est, illud
regnum bonum est. 3132

In the vast majority of cases the gender of words is handed down
traditionally from generation to generation without any change;
but sometimes changes occur. In not a few cases these are due
to purely formal accidents; thus it has been noted that, in French,
228words beginning with a vowel are particularly liable to changes in
gender, because there the form of the definite article is the same
in all cases, viz. l' (the indefinite article un, une, too, was formerly
pronounced [yn] before a word beginning with a vowel). Words
ending in the feminine -e (or, we might say in conformity with
actual pronunciation, words ending in a consonant sound) tend
to become feminine. Both these causes operate together in making
énigme, épigramme, épithète f. instead of m. In other cases the
change of gender is due to the meaning of the words. There is a
natural tendency to have the same gender in words of related
meaning (such words being, moreover, often mentioned in close
succession), thus Fr. été from f. becomes m. on account of the other
names of the seasons, hiver, printemps, automne (the last of these
in former times vacillating between the original m. and f.); la
minuit under the influence of le midi becomes le minuit. In the
same way G. die mittwoche ‘Wednesday’ has become der mittwoch
after der tag and the names of the other days of the week.

Similarly the gender of new words (or newly adopted foreign
words) is in many cases determined by formal considerations, as
when etage in German is fem. (in Fr. it is m.), but in others by sense-analogies,
as when in G. beefsteak becomes neuter (after rindfleisch),
and lift masculine (after aufzug) or when in Danish we say et vita
(after et liv), en examen (after en prøve), etc., the same word being
even sometimes treated differently in different senses, e.g. fotografien
‘photography’ (after kunsten), fotografiet ‘photograph’
(after billedet), imperativen (måden), det kategoriske imperativ
(buddet). When the metrical system was introduced, gram and
kilogram (kilo) were made neuter after et pund, et lod, but we
say en liter after en pot, en pægl, and en meter after en alen, en fod.

We see the influence of accidents of form on a broader scale
in the way in which the original trinity of Aryan gender has been
reduced to a duality in some languages. In the Romanic languages
the distinctive features of masculine and neuter were obliterated,
chiefly through the loss of any distinction in the sounds of the
endings, while the ending of the feminine with its full vowel -a
was kept apart, the consequence being that there are two genders
only, masculine and feminine (on the remains of the old neuter
see below). In Danish, on the other hand, the distinction between
the masculine and feminine articles (ON. enn, en or inn, in, einn,
ein, etc.), was lost, and thus the old m. and f. were fused together
in one “common gender” as in hesten, bogen, den gamle hest, den
gamle bog, as distinct from the neuter as in dyret, det gamle dyr.
But in those Danish dialects in which the old final -nn and -n are
kept phonetically apart (the former having a palatalized form of
the nasal) the old trinity of m., f. and n. is preserved.229

In the following remarks I am chiefly concerned with the relation
between notional (that is, in this case, natural) and grammatical
categories, and shall try to show how here and there languages
have in course of time developed other and more rational groupings
than the old traditional ones.

Sex.

Though, as has been remarked above, there are many examples
of incongruity, still the correspondence between male and masculine
on the one hand, and female and feminine on the other hand, is
strong enough to be very actively felt, and combinations which
are sometimes necessary, like G. eine männliche maus, ein weiblicher
hase, will always be felt as inharmonious and as containing a contradiction
between the form of the article and the meaning of the
adjective. In a comic paper I find the following illustration:
“L'instituteur. Comment donc? Vous êtes incapable de faire
l'analyse grammatical de cette simple phrase: ‘L'alouette
chante.’ Vous avez écrit dans votre devoir: Alouette, substantif
masculin singulier. — L'élève. Sans doute. Et je maintiens
energiquement ‘masculin’: chez les alouettes, il n'y a que le mâle
qui chante.” — Cf. also from Sweden: “Hvad heter den här apan?
— Hon heter Kalle, för det är en hanne” (Noreen VS 5. 314, i.e.
What is the name of that ape? She is called Charles, for it is
a he. In Swedish apa is feminine). And from North Jutland
i honkat nöwne wi åse me haj (Grönborg Optegnelser 72, i.e. we
say he of a she-cat; kat is m., as shown by the article i).

There is therefore a natural tendency to bring about conformity
between gender and sex. 1133 This may be achieved in the first place
by a change in form, as when Lat. lupa was formed instead of the
earlier lupus which had been used, for instance of Romulus's she-wolf
(Havet), or when much later Sp. leona, Fr. lionne and It.
signora, Sp. senora were formed from Lat. leo, senior, which did
not distinguish sex. In Greek the old neania ‘youth’ adopted
the masculine ending -s to become neanias ‘young man.’ Or else
the form is retained, but the syntactic construction is changed, as
when Lat. nauta, auriga when applied to men (a ‘sailor, charioteer’)
become masculine (i.e. take adjectives in m.): originally they
were abstracts and meant ‘sailoring, driving’; or when the Spanish
say el justicia ‘the judge,’ el cura ‘the curate,’ el gallina ‘the coward,’
el figura ‘the ridiculous fellow’ (la justicia ‘justice,’ la cura ‘curacy,’
la gallina ‘hen,’ la figura ‘figure’). Thus also Fr. le trompette
‘the trumpeter’ (la trompette ‘the trumpet’); cp. also la jument
230‘the mare.’ In Sw. statsråd ‘councillor of State,’ orig. ‘council,’
is still neuter, but an adjective predicative is generally put in the
form common to masculine and feminine: statsrådet är sjuk (not
sjukt); in Danish the word in this sense has definitely given up
its neuter gender: statsråden er syg. Thus also Dan. viv, which
formerly was n. (like G. das weib, OE. þæt wif, Sw. vivet) is now of
the common gender, and instead of the old gudet, troldet ‘the god,
the troll’ we say now guden, trolden.

Common Sex.

French homme is just as ambiguous as E. man, and one is therefore
sometimes obliged to say un être humain; in scientific books
one finds even the long-winded un être humain, sans acception de
sexe, where other languages have simple words like mensch, by
the side of mann, Greek anthropōs, by the side of anēr, etc. (Cf.
Meillet LH 273 ff.)

While a great many special names for human beings are applicable
to both sexes, e.g. liar, possessor, inhabitant, Christian, aristocrat,
fool, stranger, neighbour, etc., others, though possessing no
distinctive mark, are as a matter of fact chiefly or even exclusively
applied to one sex only, because the corresponding social functions
have been restricted either to men or to women. This is true of
minister, bishop, lawyer, baker, shoemaker and many others on the
one hand, nurse, dressmaker, milliner on the other. It is curious
that some words have in course of time been restricted to women,
though originally applicable to men as well, thus leman (OE. leofman
‘dear man,’ in Chaucer and even in Shakespeare of a man,
later only of a woman, now obsolete), bawd, witch, girl.

Where it is desired to restrict common-sex words to one sex,
this may be done in various ways, thus man-servant or servant-man,
maid-servant, servant-girl, a he-devil, a she-devil, her girlfriends,
a poetess (but it is a higher praise to say that Mrs. Browning
was a great poet, than to call her a great poetess). Author is still
to a great extent a common-sex word, though the word authoress
exists, but there is no corresponding formation to denote the female
teacher or singer. Most languages present similar inconsistencies,
and in many cases linguistic difficulties have been created through
the recent extension of the activities of women to spheres that used
to be reserved for men. 1134 Of the artificial languages there is only
one that has successfully tackled the problem of having on the
one hand common-sex words and on the other hand special-sex
words, namely Ido, where all denominations without any special
ending are applicable to both sexes, while male is denoted by the
ending -ulo and female by -ino, e.g. frato brother or sister, fratulo
brother, fratino sister, frati G. geschwister, homo mensch, homulo
mann, homino woman, sposo spouse, spozulo husband, spozino
wife, and thus dentisto, dentistulo, dentistino, etc. 2135

In the plural there is naturally even greater need for common-sex
words than in the singular, but it is only few languages that
232can use the plural masculine in the same way as It. gli zii ‘uncle
and aunt’ (lo zio e la zia), i fratelli (il fratello e la sorella), i suoceri
(but not i padri instead of i genitrici) or Spanish los padres ‘father
and mother,’ los hermanos ‘brother(s) and sister(s)’ sus dos hijos,
Juan y Perfecta (Galdós, D. Perf. 29).

With regard to animals, only those few that have the greatest
importance to men have separate common-sex and special-sex
words or forms (as horse, stallion, mare); from these we have several
gradations (e.g. dog, he-dog or simply dog for the male, bitch or
she-dog; sparrow, cock-sparrow, hen-sparrow) down to animals
whose sex has no interest to ordinary speakers (fly, worm).

In pronouns and adjectives, where a common-sex form is
not available, as it is in somebody, everybody, each, the masculine
is most often used, as in Fr. quelqu'un, chacun, Jean et Marie
étaient très contents d'eux-mêmes; some incongruity is inevitable
in sentences like “Was Maria und Fritz so zueinander zog, war,
dass jeder von ihnen am anderen sah, wie er unglücklich war”
or “Dona Perfecta… su hermano… pasaron unos pocos
afios sin que uno y otro se vieran” (Galdós, D. Perf. 32).

It seems to be of special importance to have a common-sex
interrogative pronoun, because in asking “Who did it?” one does
not know beforehand whether it is a he or a she; hence most
languages have only one form here (not infrequently a form which
has a masculine ending), thus Gr. tis, Goth, hwas (the fern, form
hwo given in grammars, probably never occurs as an interrogative
primary), OE. hwa, E. who, G. wer, Du. wie, Dan. (hvo), hvem,
Russ. kto, etc. Exceptions are ON. m. hverr, f. hver, m. hvárr,
f. hvár and Lat. m. quis, f. quæ, but in modern Icelandic the difference
has disappeared, at any rate in the nominative (hver, hvor), and
in the Romanic languages only the masculine form survives as a
common-sex form: It. chi, Fr. qui, Sp. quién.

In the personal pronouns for the third person he and she are
distinguished in English as in the other languages of our family;
when a common-sex pronoun is wanted, he may be used instead
of he or she, but colloquially the pl. they is often used (“Nobody
prevents you, do they?” etc., Lang. 347, MEG II, 5. 56). In
the plural most Gothonic languages have now generalized one form
for both sexes (E. they, G. sie, Dan. de, etc.), which is very natural
as one has very often to talk of groups of persons of different sex.
Thus also in Russian except in the nom., where oni, one are kept
apart. In the Romanic languages the two sexes are kept apart:
eglino, elleno; ellos, ellas; ils (eux), elles, except in the dative:
loro, les, leur, and in the Fr. acc. with verbs: les. ON. has separate
forms in the nom. and acc.; þeir, þær; þá, þær, but not in the
dat.: þeim; in the nom. and acc. it has also a separate form for
233the neuter: þau, and this is also used as a common-sex plural, a
phenomenon which is generally accounted for from the accidental
fact that the old dual (which would often be used for ‘he and she’)
came to be phonetically identical with the neuter plural. If that
is so, the use of the neuter singular as a common-sex form may be
transferred from the dual-plural; an example of both is found in
Laxd. S. 69. 20 Eptir þetta skilja þau Gutðrún talit, ok bað hvárt
þeira annat vel fara ‘after this G. and he (Snorri) stop talking,
and bade each other farewell’ (þau n. pl., hvárt and annat n. sg.).
On the corresponding rule in Gothic and OHG see Willmanns DG 3.
768, Streitberg GE 166. Old Dan. Jysk 1. 4. 3. hwat lengær liuær
mothær æthe barn ‘which lives longer, mother or child.’

Animate and Inanimate.

A distinction between living and lifeless, or animate and inanimate,
or sometimes between human and extra-human, personal
and non-personal (things which are not always easy to keep apart),
pervades many parts of the grammars of many languages, sometimes
in close connexion with sex-gender, sometimes independent of
sex-gender. This distinction may be shown grammatically in
the most different ways, and I cannot claim that the following
survey is complete even for the languages with which I am most
familiar.

From the oldest times there has been a strong tendency to
use the pronoun it (OE. hit) to represent things. It was so even
when the old threefold gender, m., f., n., was still living and showed
itself in the forms of adjuncts (articles, pronouns, adjectives).
Thus (to give some of the examples adduced in the interesting
article “Grammatical and Natural Gender in Middle English,”
by S. Moore, Publ. Mod. L. Ass. 1921) hlæw… beorhtne (acc.
m.) … hit | anne arc… hit | ænne colic… hit | þisne calic,
234hit | þeos race… hit. From the Ancrene Riwle: þene kinedom
… hit | þeo ilke scheadewe… hit | þene drunch… hit. (In
Moore's article this phenomenon is mixed up with the use of heo
(she) when referred to such words as the neuter wif, mægden or
the masculine wifman, or of he referring to the neuter did; it
would have been better to treat these things separately: the latter,
but not the former usage is pretty frequent in Modern German.)
This use of it quite naturally became even more predominant after
the old distinctions of case and gender in adjunct pronouns and
adjectives had disappeared, and about 1600 it led to the creation
of a new genitive case its, where formerly his was in use both for
the masculine and the neuter; its also superseded the dialectal
gen. it, which had begun to be used in Standard English.

It is, however, impossible to draw a hard and fast line of demarcation
in English between an animate gender, represented by he
or she, and an inanimate gender, represented by it. For it may be
used in speaking of a small child or an animal if its sex is unknown
to the speaker or if his interest in the child or animal is not great:
the greater personal interest one takes in the child or animal, the
less inclined one will be to use it, and he or she is even used in many
cases of an animal independently of any knowledge of the actual
sex of the individual referred to (a hare… she, a canary-bird…
he, a crocodile… he, an ant… she, etc.). On the other
hand, things may, in more or less jocular style, be mentioned as
he or she, by way of indicating a kind of personal interest. The
best-known and most universal example of this is the sailor's
she of a ship; in Dickens a coach is she, and this is nowadays the
fashion among motorists in talking of their cars.

A country may from different points of view be treated either as
inanimate or animate. On the one hand, in speaking of France, we
may say “it certainly is smaller than Spain, but then it is much
more fertile,” and on the other hand, “I do not approve of her
policy in the reparations question”: in the latter case France is
viewed as a personal agent, hence the sex-indicating pronoun is
chosen, and if this is in the feminine in spite of the fact that the
political leaders are (still!) men, this is due to literary tradition
from French and Latin, where the names of countries happened
to be feminine. In German and Danish, where this influence is
not so strong, states even as political agents are mentioned in the
neuter, es, det (though we may sometimes substitute the personal
name Franskmanden ‘the Frenchman’ and say “Ja, Franskmanden,
han veed nok hvad han vil” without having any individual Frenchman
in view).

A somewhat similar case in seen with heaven, which may be
referred to as he, when it is a veiled expression for God. Nature
235when viewed as an agent is she from the Latin (and Fr.) gender,
and this is transferred to Fate by Browning (“Let fate reach me how
she likes”) in spite of the Latin gender. 1136 When the sun is mentioned
as he, and the moon as she, this has very little to do with a real
feeling of them as animate, but is purely artificial literary tradition
from Latin: it is well known that in OE. as in the other Gothonic
languages the sun was f. and the moon m.

There can be no doubt that the poetic tendency to personify
lifeless things or abstract notions, for instance to apostrophize
Death as if it were a living being, and the related representation
in plastic art of such notions, are largely due to the influence of
languages with sex-gender, chiefly, of course, Latin. But it has been
justly remarked (among others by Jenisch, 1796) that such personification
is more vivid in English than it can be, for instance, in
German, because the pronoun he or she, where everyday language
has it, at once draws attention to the idealization, which in German
is not so noticeable because every chair and every stone is er, and
every plant and every nose is sie. English poets have also greater
freedom to choose which sex they will attribute to such notions. 2137
Thum compares Shakespeare's passage “See how the morning
opes her golden gates, And takes her farewell of the glorious sun,”
in which the morning is the mistress who takes leave of her lover,
with Schlegel's translation “Sieh, wie sein tor der goldene morgen
öffnet, Und abschied von der lieben sonne nimmt,” where the relation
has been inverted on account of the gender of morgen and sonne.
In Milton, Sin is talking to Satan who has begotten on her his son
Death; this is rendered impossible in a French translation, because
le péché cannot be the mother, and la mort cannot be the son. Note
also Brunot's remark (PL 87) “le hasard des genres a créé aux
artistes de grands embarras. La Grâce, la Beauté, la Science,
prenaient facilement figure de femme, mais la Force? On a eu
recours à Hercule!”

Some of the distinctions tabulated on p. 234 are comparatively
recent; thus the relative which down to the beginning of the seventeenth
century might be used of a person. When this and that are
236used as primaries, they are inanimate; note also the difference in
such dictionary definitions as “Rubber — one who, or that which
rubs.” When the prop-word one is anaphorical (i.e. refers to a
word mentioned already) it may be either animate or inanimate
(this cake… the only one I care for), but when it does not in
that way refer to a word just mentioned, it is always personal
(‘the great ones of the earth’). All these things are dealt with in
greater detail in MEG, Vol. II, passim.

It is also worth mentioning that collectives can take the verb
in the plural only if they denote living beings (family, police),
but otherwise always take it in the singular (library, forest). It
is also noteworthy that the genitive (in -s) is extinct except in the
case of names of living beings (the man's foot, but the foot of a mountain) — apart
from some survivals of set phrases (out of harm's
way | a boat's length from the ship). 1138

In German the distinction between animate and inanimate
is not so marked as in English: many things are referred to as
er, sie, dieser, jene, etc., that is, by the same pronouns as are used
for persons. Yet there are some indications of the difference
besides the obvious instance wer and was: the datives ihm, ihr,
are not often used of things, and instead of mit ihm, mit ihr, in
ihm, in ihr, etc., the compounds damit, darin, etc., are used. There
is a greater inclination to use derselbe, dieselbe of inanimates than
of living beings; the possessive pronoun sein is generally reserved
for living beings: sie legte die hand auf den stein und empfand
dessen wärme, or die wärme desselben (Curme GG 168). The old
dative has disappeared from the neuters was, etwas, nichts, and
the compounds with wo- (womit, wovon) are used where with
animates we have mit wem, von wem.

How important the neuter conception is in some cases is shown
by the curious fact that it has been allowed to override the idea
of plurality in beides, which means ‘both things’ as distinct from
beide ‘both persons’; thus also mehreres ‘several things,’ but
mehrere ‘several persons,’ and in pretty much the same way alles
(cf. Lat. omnia pl. n.), to which we have, of course, parallels in
other languages: E. all sg. n. (which tends to be superseded by
everything, all being reserved when used alone for persons in the
pl.), Dan. alt, etc. Dan. alting was originally a pl. ‘all things,’
but is now used as a neuter sg.: alting er muligt. Cf. also much
(viel, vieles) = many things (viele dinge).

In Danish the distinction between animate and inanimate
is not well-defined grammatically. But we have the interrogative
237pronoun hvem of human beings and hvad of things corresponding
to who and what, and instead of using begge ‘both’ alone as a
primary there is a tendency to use begge to of two persons and
begge dele of things, corresponding to alle (allesammen) ‘all’ (pl.)
and alt (alting) ‘all, everything.’ The sex-indicating pronouns
han, hun ‘he, she’ are used of human beings and of such of the
higher animals as the speaker takes a personal interest in; other
animals are referred to as den or det according to the gender of the
word: lammet, svinet… det, hesten, musen… den ‘the lamb,
swine, horse, mouse’ — exactly as the same pronouns refer to
things, e.g. huset… det, muren… den ‘the house, the wall.’
As in English, though not to the same extent, there is some
disinclination to use the genitive in -s with names of inanimates:
we say taget på huset, træerne i haven more often than
husets tag, havens træer ‘the roof of the house, the trees of the
garden.’

Swedish literary language has retained much more of the old
gender system than Danish, but the tendency is towards the same
use as in Danish of den instead of the older m. and f. han, hon,
in speaking of things, see the extremely able discussion in Tegnér,
Om genus i svenskan, 1892.

In French we have, of course, qui (qui est-ce qui) over against
que (qu'est-ce que) and quoi; further en refers to something inanimate,
where with animates the possessive pronoun is used: j'en
connais la precision in speaking of a watch, je connais sa precision
in speaking of a man (but there are instances in which son is
necessary even of a thing, and the relative corresponding to en,
viz. dont, is used of both classes).

In Spanish we have the rule that the object takes the preposition
á before it if it denotes a living being: he visto al ministro
‘I have seen the minister,’ but he visto Madrid. In Russian and
the other Slav languages the rule prevails that with names of
living beings the genitive is used instead of the accusative. In
some of the modern languages of India, such as Hindustani, the
object form with living beings is marked by the ending -ho, while
in names of inanimate things the object has the same form as
the nominative (S. Konow in Festskrift til A. Torp, 99). In various
languages, therefore, a distinction between these two classes is
seen reflected in their manner of indicating the object, but as the
means by which this is achieved are entirely different, we seem
here to have a trait that has its root in the psychological sameness of
men all over the world. (Cf. also the Aryan nominative ending
-e if that was originally characteristic of the names of living beings
— which, however, is more than doubtful, as on the one hand -s
is found in inanimates like Lit. naktìs, L. nox, and on the other
238hand many animates seem never to have had -s, e.g. pater, G.
kuōn.)

The distinction between animate (or personal) and inanimate
(or impersonal) is sometimes shown indirectly in the way in which
some case-forms are allowed to survive while others disappear
The dative is more often used in words denoting living beings
than with inanimates; hence the acc. forms found in the oldest
English, mec, þec, usic, eowic were early ousted by the dat. me,
þe, us, eow (now me, thee, us, you), and somewhat later the old datives
hire (her), him, hem (mod. 'em), hwam (whom) displace the old accusatives
heo, hine, hie, hwane; them also is a dative. On the other
hand, in the neuter it is the old accusatives hit (it), that, what that
are preserved at the cost of the datives. Similarly in Dan. the
old datives ham, hende, dem, hvem have ousted the accusatives
(though it is true that in mig, dig the acc. has outlived the dative);
in North German wem instead of wen, in Fr. lui, It. lui, lei, loro
(when not used with a verb) we see the same tendency, while the acc.
has carried the day in G. was, Fr. quoi, etc.

In substantives the old nominative has sometimes prevailed
over the oblique cases in names of living beings, while the inverse
is the case in names of inanimates. Thus it has been remarked
by Behaghel, Bojunga and Tegnér that in the G. n-declension the
old nom. without -n has held its own in names of living beings
only: bote, erbe, knabe, while inanimates have generalized the
oblique cases: bogen, magen, tropfen. In Swedish similarly the
acc. has prevailed over the nom. in words like maga, båga, strupa,
aga, vana, while names of persons have retained and generalized
the nom. in -e: gubbe, granne, bonde (Tegnér G. 221). Another
nom. ending has likewise been preserved in names of persons only:
slarver, spjuver, luver (ibid. 225). Old French had a distinction
between a nominative and an oblique case; generally the latter
has been generalized, but it has been remarked by Bréal (MSL 6.
170) that all the old nominatives that have been preserved denote
human beings, e.g. traître, sœur, fils, maire.

As lifeless things are naturally reputed inferior in value to
living beings, and as the neuter gender in those languages that
have one is preferably used of things, this gender comes to have a
certain depreciatory tinge when applied to human beings and animals:
in Dan. it is noteworthy that many terms of abuse are neuter:
etfjols, pjok, fæ, bæst, drog; some words for animals that are chiefly
used in a depreciatory sense, have in historical times changed their
gender and have become neuter: øg, asen, æsel, kreatur. This
may be compared with the well-known fact that diminutives in
various languages are often neuter, even if the words from which
they are derived have another gender: Gr. paidion ‘little boy’
239from pais, G. fischlein, fräulein, bübchen, mädchen, etc. 1139 I suppose
that when Italian has so many diminutives in -ino from feminines
they were originally not real masculines but neuters: casino,
tavolino, ombrellino from casa, tavola, ombrella, also donnino, manino
by the side of donnina, manina, and I venture the conjecture that
it is the same depreciatory neuter that is behind the curious
occurrence of some forms in -o for smaller things by the side of
words in -a for bigger things: buco ‘a small hole,’ coltello ‘a small
knife,’ by the side of buca, coltella, etc. In the dialects of South-Eastern
Jutland some names for young animals, which otherwise
in Danish are of the common gender, have become neuter: et
kalv, hvalp, gris, kylling (M. Kristensen, Nydansk, 1906, 57). In
Swedish individ is always en if used of a human being, often also
of higher animals, but in speaking of a lower animal ett individ
is said (Tegnér G. 39); in Danish it is always neuter as Lat.
and G.

Here and there we find a tendency to establish a grammatical
distinction between thing-words (countables) and mass-words
(uncountables) apart from the difference dealt with in the chapter
on Number (XIV, p. 198 f.). In the south-western dialects of England
“full shapen things” are referred to as he, acc. en (from OE.
hine) and take the pronominal adjuncts theäse, thik, while “unshapen
quantities” are referred to as it and take this, that: Come
under theäse tree by this water | goo under thik tree, an zit on that
grass (Barnes, Dorset Gr. 20, Ellis EEP. 5. 85, Wright, Dial. Gr.
§ 393, 416 ff.). In other languages there is a tendency to use the
neuter gender preferably with mass-words, thus G. das gift, das
kies ‘poison, gravel’ has taken or is taking the place of the older
die gift, der hies. In the same way we have now in Danish støvet
for older støven ‘dust.’ But in Danish this is carried further.
Neuter forms of adjuncts are used to indicate quantity with mass-words
even where these in other respects are of the common gender.
Thus we say mælken, osten ‘the milk, the cheese,’ but alt det mælk,
noget andet ost ‘all that milk, some other cheese’ (as mass, — ‘another
cheese’ as thing-word is en anden ost); jeg kan ikke nøjes med det
te ‘I cannot rest content with that (much) tea,’ but… med den
tea if the kind or quality is meant. Many dialects in Jutland go
still further, all mass-words being made neuter without regard
to the original gender, and in Hanherred a complementary change
has taken place, all thing-names having been made of the common
gender: iset, jordet, skiben, husen ‘the ice, earth, ship, house,’
where Standard Danish has isen, jorden, skibet, huset.240

Conceptional Neuter.

Before concluding this chapter on gender we still have to consider
something which for want of a better name I propose to
term “the conceptional neuter.” It might be said to be the real
or notional or universal neuter in opposition to the specified or
concrete neuter which we have when in English we refer to a previously
mentioned house or worm, etc., as it, and to the arbitrary
neuter which we have when in German we refer to a previously
mentioned haus or mädchen as es because the word happens to be
of the neuter gender. It will appear from the following paragraphs
that there are certain natural or notional functions for a neuter
gender to fulfil, even though in many languages, which have not
otherwise a neuter gender, there is nothing but a few pronominal
forms to show the existence of this neuter in their grammatical
system.

The first application of this unspecified or conceptional neuter
is seen in such sentences as E. it rains, G. es regnet, Dan. det regner,
Fr. il pleut (colloquially ça pleut), further it snows, thunders, etc.,
where it is difficult or impossible to define what it stands for: the
whole situation of the atmosphere, if you like, but at any rate
something thought of as definite in the same way as we use the
definite article in “the weather is fine” or “the day is bright.”
Many languages here have no pronoun, Lat. pluit, It. piove, etc.,
and Brugmann and others see in the use of it a purely grammatical
device, called forth by the habit of always having an express subject
(he comes, il vient, where Lat. or It. has often merely the verbal
form venit, viene). There is undoubtedly much truth in this consideration,
but it does not give the whole truth, and Grimm
(Wörterbuch) is not wholly wrong when he speaks of “das geisterhafte,
gespenstige, unsichtbare, ungeheure” as expressed in the
“impersonals”; Spitzer uses the expression “das grosse neutrum
der natur,” and thinks that this it is just as much an outcome of
man's mythopoetic imagination as Juppiter tonat. 1140 I may here
adduce on the one hand the following bit of conversation from one
of Bennett's novels: “It only began to rain in earnest just as
we got to the gate. Very thoughtful of it, I'm sure!” and on the
other hand, from a totally different sphere, Brownings use of That
with a capital letter as a synonym for God: “Rejoice we are
allied To That which doth provide And not partake, effect and
not receive!” (Rabbi Ben Ezra) and Hardy's similar use of It:
241“Why doth It so and so, and ever so, This viewless, voiceless
Turner of the Wheel?” which he justifies by saying that “the
abandonment of the masculine pronoun in allusion to the First
or Fundamental Energy seemed a necessary and logical consequence
of the long abandonment by thinkers of the anthropomorphic
conception of the same” (The Dynasts).

I find the same unspecified or conceptional it (though not the
great neuter of Nature) as an object in idiomatic combinations
like to lord it | you are going it! | we can walk it perfectly well | let
us make a day of it, etc. In the following sentence a comic effect
is produced by the ambiguity of it as specified and unspecified:
He never opens his mouth but he puts his foot in it.

Corresponding uses are found idiomatically in other languages,
for instance G. sie hat es eilig | er treibt's arg | Dan. han har det
godt, sidder godt i det | han skal nok drive det vidt | Fr. l'emporter, le
prendre sur un certain ton. In Dan. the n. det curiously interchanges
with the common-gender form den: ta den med ro ‘take
it easy’ during recent times has supplanted ta det med ro, and den
is found in many idiomatic phrases: brænde den a, holde den gående,
etc.

Note here also G. es klopft an der tür, Dan. det banker på døren,
corresponding to E. someone is knocking at the door (there is a
knock at the door) and Fr. on frappe à la porte.

Next we have a conceptional neuter in words like what, nothing,
everything, something, and it is interesting to notice that in Danish,
where ting is of the common gender, ingenting and alting ‘nothing,
everything’ take the predicative in the neuter gender: den ting
er sikker, but ingenting er sikkert, etc. We see the same in the
Romanic languages where the Lat. neuter has been merged in the
masculine, but where these words, even those which were originally
feminine, are treated as masculines, i.e. neuters. Thus Fr. rien
from the Lat. f. rem: rien n'est certain, further quelquechose de
bon. In It. qualche cosa, ogni cosa, che cosa (and the abbreviated
interrogative cosa = che cosa) take the predicative in the masculine,
i.e. neuter: che cosa fu detto? Thus also nulla fu pubblicato | una
visione, un nulla che fosse femminile (Serao, Cap. Sansone 87, 123).

A conceptional neuter is also found in connexion with adjectives
in the generic the beautiful, i.e. ‘everything beautiful,’ the good,
etc. Note that Spanish here has retained the Lat. neuter in the
form of the article: lo bueno, different from the masculine el bueno
‘the good one.’

A further function of the conceptional neuter is to represent
a predicative as in: All men my brothers? Nay, thank Heaven,
that they are not (Gissing, cf. MEG II. 16. 377) | you make him
into a smith, a carpenter, a mason: he is then and thenceforth
242that and nothing else (Carlyle) | Marian grew up everything that her
father desired (Gissing) | his former friends or masters, whichever
they had been (Stevenson) | She had now become what she had
always desired to be, Amy's intimate friend (Gissing) | she treated
him like a tame cat, which is what he was (McKenna) | What is
he? Just nothing at all as yet. Sweet NEG § 212 has not understood
this function of what when he speaks of it as “used in a personal
sense”; note that the answer to the question “What is
he?” may contain any predicative: “a shoemaker” or “kindhearted,”
etc.

We have exactly the same neuter in other languages. Dan.
Er de modige? Ja, det er de. Hvad er han? G. Sind sie mutig?
Ja, das sind sie. Vom papst ist es bekannt, dass er, als er es noch
nicht war, seine verhältnisse geregelt hatte. Was ist er? Er
ist noch nichts. Fr. Si elles sont belles, et si elles ne le sont pas.
It. Pensare ch'egli era libero e che anche lei lo era! (Fogazzaro).
Sp. Personas que parecen buenas y no lo son (Galdós). Cf. also
Gr. Ouk agathon polukoiraniē, and the G. n. sg. Welches sind Ihre
bedingungen? 1141

A notional neuter is also found where a pronoun represents
a verb or a nexus: Can you forgive me? Yes, that is easy enough
| The Duke hath banished me. That he hath not (Sh.) | I'll write
or, what is better, telegraph at once. Infinitives and whole clauses
also always take articles, adjectives, etc., in the neuter gender in
those languages which have one: Gr. to pinein, G. das trinken;
Lat. humanum est errare, etc.243

Chapter XVIII
Comparison

Comparative and Superlative. Equality and Inequality. Weakened Superlatives
and Comparatives. Latent Comparisons. Formal Comparatives.
Indication of Distance. Secondaries and Tertiaries.

Comparative and Superlative.

This tripartition no doubt corresponds with the actual forms
found in the best-known languages, in which the “positive” is
the fundamental form from which the two others are derived
either by means of endings or by the addition of adverbs (subjuncts)
like more and most. In some well-known instances the two
higher degrees are taken from other stems than the positive:
good, better, best | bonus, melior, optimus, etc. 1142

Let us now look a little more closely into this system from a
logical point of view. In the first place, it does not require much
thought to discover that the “positive” cannot strictly be called
a “degree of comparison,” for when we speak of a horse or a book
as old, we do not compare it with any other horse or book; the
form, then, is rather “negative of comparison” than “positive,”
as the old grammarians termed it with their curious scorn of a
good or consistent terminology. The term does not, however,
do much harm, as it cannot very well be confounded with positive
in the sense ‘not negative.’

In both cases A is compared with B, C, and D; but the result
is in the former case given with regard to these three (the other
boys), in the latter with regard to all the boys, himself included.
The comparative must thus be supplemented by a member
(expressed or understood), added by means of than and different
from the object compared, hence the frequent use of the word
other. This kind of supplement is not possible in the case of a
superlative, which, on the other hand, is often followed by of or
among all. But as both forms really express the same idea, we
should not be surprised to find a rather frequent confusion, resulting
in such blendings as the best of all others; see, e.g., a king,
whose memory of all others we most adore (Bacon) | parents are
the last of all others to be trusted with the education of their own
children (Swift).

Now we can see how easy it was for languages that formerly
possessed a real superlative, to give up this form and content
themselves with the comparative. In the Romanic languages the
only expression for the superlative idea is the comparative rendered
definite either by the article or by some other defining word:
le plus grand malheur | mon meilleur ami, etc. (Sometimes no
defining word is required, as in “la vie, dans tout ce qu'elle a de
plus intensif.”) In Russian, the comparative similarly is often
used as a kind of superlative, which is facilitated by the fact that
the second member of comparison is added in the genitive, and
that the same case is used as a partitive and thus corresponds
both to Eng. than and of: lúčše vsegó = ‘better than all’ or ‘best
of all’ | bogáče vsěx = ‘richer than all’ or ‘richest of all.’ (Besides,
the superlative may be expressed by nai- placed before the comparative
or by sámyj ‘self’ (H. Pedersen RG, p. 89; cf. Vondrák
SG 1. 494 and 2. 71 ff.)

We have what might be called a limited superlative meaning
‘better’ (etc.) than all the others with the exception of one (two,
245etc.) in the next best, the largest but one (two, etc.), the third best, etc.
Similarly in Danish and German, where, however, no expressions
exist corresponding to the English ones with but. There are many
languages, on the contrary, which have no such easy ways of
expressing this kind of superlative.

In German a curious confusion arises when a superlative is
qualified by ‘possible,’ this word being put in the superlative
form instead of the other adjective (adverb); both expressions
are combined in a speech by Professor Jodl “das problem der
gröstmöglichen glücksbefriedigung für die möglichst grosse zahl”;
in English it would be “the greatest happiness possible for the
greatest number possible.”

Equality and Inequality.

Obviously 1 and 3 are closely connected, as both denote inequality.
English uses than with 1 and 3, and as with 2, while
other languages use the same word in all three cases, thus Fr.
meilleur que, aussi bon que. Danish distinguishes end and som
as E., but some parts of Denmark (Fyn) use som even after comparatives.
In the same way some parts of Germany use wie in
all three kinds of comparison, while other parts of Germany use
wie for equality only, and als with the comparative. Hence it
is possible in Fr. to say, for instance, “il a autant ou peut-être
plus d'argent que moi,” where other languages have no such easy
expression, for the sentence “he could box as well or better than
I” (Wells) is felt as somewhat slipshod English.

But in practice the expressions with less are naturally little used;
besides the two forms sub 2 are not exact synonyms: it would
obviously be impossible to say as young as the hills instead of as old
246as the hills. This is a natural consequence of the fact, that old,
besides having the neutral signification (as vox media) of ‘having
(this or that) age’ as in “baby is only two hours old” also signifies
‘having a great age, advanced in years’; it is, indeed, in the
latter sense that it forms a contrast to young. In some languages
the two senses are kept distinct, as in Fr. âgé de deux heures |
vieux, in Ido evanta du hori | olda.

Similarly, though more unkind than = less kind than, the terms
as unkind as and as kind as are not synonyms, because the former
implies that both persons compared are unkind, and the latter
that both are kind. Comparison by means of as is therefore
generally by no means neutral or indifferent, though it may occasionally
be, as in “I don't think man has much capacity for
development. He has got as far as he can, and that is not far,
is it?” (Wilde).

On the other hand, comparisons with than are as a rule
indifferent or neutral; “Peter is older than John” does not
imply that Peter is old, and the comparative may really therefore
indicate a lesser degree than the positive would in “Peter is old.”
Nor does the sentence “Peter is older than John” say anything
about John's being old; but that is implied if we add the subjunct
still: “Peter is still older than John” (thus also: Pierre
est encore plus vieux que Jean | Peter er endnu ældre end Jens |
Peter ist noch älter als Hans — by the way an interesting parallel
development in different languages, for this use of still is not at
all self-evident; it is also found in Russian.

If we negative stage 1 (Peter is not older than John), the
meaning may be either stage 2 (equality) or 3 (inferiority); in
English a curious distinction is made between not more than, which
is indistinct and may mean either 2 or 3, and no more than, which
implies stage 2, equality. A negative stage 2 takes the form not
so old as and practically always means stage 3 ‘less old than,
younger than’; a negative with as is not so frequent and may
sometimes mean stage 1 if it has extra emphasis on as, as when
the assertion “A is as old as B” is contradicted: “Oh no, not
as old as B, but much older.”

Weakened Superlatives and Comparatives.

There is a natural tendency to exaggerate by using the superlative
for a very high, instead of the highest, degree. This is
sometimes termed the “absolute superlative,” sometimes the
“elative.” Thus “with the greatest pleasure,” “a most learned
man,” etc. This has become the rule in Italian and Spanish to
such an extent that the old Latin superlative form is never used
247as a real superlative; It. bellissimo ‘very fine,’ Sp. doctisimo ‘very
learned,’ etc. 1144 In colloquial Norwegian we have the same with
a negative: ikke så væst ‘not so very bad.’ In Danish a difference
is made between the uninflected and the inflected superlative
form, the former alone (without the article) meaning the real
superlative, the latter the elative: med störst veltalenhed (more
eloquently than anyone else) | med störste veltalenhed (very eloquently
indeed).

Sometimes the comparative form is similarly used without
implying a comparison, as Dan. “en bedre middag” (a good, or
a pretty good, dinner). Thus also E. rather, e.g. “Does it rain? —
Rather!”

A similarly weakened comparative is found in Dan. flere, as
in “ved flere lejligheder,” where E. generally says more explicitly
more than one, a plural of one. Curiously enough in this case, in
which there is no comparison, some languages have a double comparative
ending, G. mehrere (this could formerly take als, which
is now impossible), late Lat. plusiores, whence Fr. plusieurs —
which, in spite of its form, is really weaker than the ‘positive’
viele, beaucoup.

Latent comparisons.

In some linguistic expressions the comparative idea is latent.
Thus in the verb prefer: I prefer A to B = I like A better than
B (je préfère A à B | ich ziehe A dem B vor); in Ido the ordinary
comparative connective is in this case used: me preferas A kam
B = me prizas A plu kam B. This may be found very rarely in
English, too, as in Thackeray Sk 138 preferring a solitude, and to
be a bachelor, than to put up with one of these for a companion. —
Further we have a latent comparative in too (trop, Dan. for, G. zu),
which means ‘more than enough,’ or ‘more than decent, or proper,
or good.’ Here, also, the distance may be indicated: an hour
too late | en time for sent | eine stunde zu spät | trop tard d'une
heure. — Cf. also outlast = ‘last longer than,’ outlive (survive),
Dan. overleve, G. überleben; exceed.

As latent comparatives must also be considered before and its
opposite, Fr. avant, après, G. vor, nach, etc.; note that E. after
and Dan. efter are also formal comparatives; the indication of
distance is seen in “an hour before sunrise | une heure avant le
lever du soleil | eine stunde vor dem sonnenaufgang,” etc. But
when we say “after an hour he came back” and similarly “après
une heure il rentra,” etc., we have really a confusion of the indication
of distance and the object of the preposition, as it means
248‘an hour after (his departure, or whatever was mentioned).’ This
may be compared to what has taken place in the mathematical
use of plus and minus = augmented (lessened) by, cf. the translations
“four less two,” “quatre moins deux,” “vier weniger
zwei.”

Fr cadet and aîné are also latent comparatives, il est mon cadet
de deux ans = ‘he is two years younger than I (me).’ Cf. also
“il avait un frère cadet, de dix ans moins âgé, ingénieur comme
lui” (Rolland). A similar syntax is seen in English with some
words taken over from Latin comparatives, though from a formal
point of view they cannot in English be considered as comparatives;
thus “he is my senior by two years,” etc.

The irrationality of grammatical expressions is seen in the
following facts. While Lat. post and ante are, as we have seen,
virtual comparatives, they take quam only when the second member
of comparison is a whole clause; this is expressed in ordinary
grammatical terminology by saying that post and ante are prepositions,
but postquam and antequam are conjunctions; but it
is easy to see that this is not the usual function of quam, which
here corresponds to E. that rather than to than. E. after and
before can take both words and clauses (are both prepositions and
conjunctions), cf. “he came after (before) the war” and “he
came after (before) the war was over.” In Danish the two words
are treated differently, for efter requires the addition of at in order
to be made a conjunction: “han kom efter krigen” | “han
kom efterat krigen var forbi,” while no at is required with för:
“han kom för krigen” | “han kom för krigen var forbi”; in
both cases forend may be substituted (end means “than,” the
connective after comparatives), but vulgar speech inclines to add
at to make it into a conjunction; “han kom forend at krigen
var forbi.” In G. the dative case of the demonstrative-relative
pronoun dem is required to change the preposition nach into the
conjunction nachdem, while vor (früher als) is the preposition corresponding
to the conjunction ehe. In Fr. we have après and avant
as prepositions, après que and avant que as conjunctions, where
it is impossible to tell whether que is ‘than’ or ‘that’; cf. also
It. poscia che. (With an infinitive, French has, or had, the following
constructions: avant que de partir, avant de partir, avant
que partir, avant partir.)

Formal comparatives.

On the other hand, we have a class of words which are, formally
considered, comparatives, but are not notional comparatives
in so far as they cannot take than: upper, outer and its doublet
249utter, former, etc. These have probably at no time had the true
comparative functions; but latter and elder, which now share
the same inability to take than, were formerly true comparatives
of late and old, and we still in Shakespeare find elder than. These,
then, may be called ex-comparatives.

Other is a formal comparative, though there is no corresponding
positive; it can take than (thus also in other languages autre que,
etc.). In English other sometimes infects its synonym different,
which then takes than instead of the regular from, for instance:
things will be made different for me than for others (Wilde);
inversely one may find from after another: I hope to be another
man from what I was (Dickens).

There are other well-known words in our languages formed
with the same ending and still less to be considered as comparatives,
namely pronominal words relating to the number of two like Lat.
uter, neuter, OE. ægðer, hwæðer, E. either, neither, whether, etc.

It may be doubtful, perhaps, whether this Aryan suffix -ter-
belonged originally to these pronominal words referring to two
or was from the first a comparative ending. 1145 But however that
may be, we find in many languages the rule that when there is
no direct comparison (with than) the comparative is used if two,
and the superlative if more than two are referred to; cf. Latin
major pars if something is divided into two parts, maxima pars
if there are three or more parts. In English we have, correspondingly,
e.g. “If Hercules and Lychas plaie at dice Which is the
better man, the greater throw May turne by fortune from the weaker
hand” (Sh.), but apart from some set phrases like the lower lip
and the upper end the natural tendency in modern English is to
use the superlative everywhere, as in “whose blood is reddest,
his or mine” (Sh.), see MEG II, 7. 77. This tendency has completely
prevailed in Danish. It is curious to note that German here has
a form composed of an old superlative with the comparative ending
superadded: ersterer, and that the English equivalent the former
is similarly formed from the OE. superlative forma (= primus)
and the comparative ending -er.

Indication of Distance.

With comparisons of inequality the degree of difference (the
distance) is often indicated, e.g. “he is two years older than his
brother”; also with by; in Latin the ablative is here used, in
G. frequently um, etc.250

It is, accordingly, possible to combine the two kinds of comparison
as in the sentence “She is as much better than her husband
as champagne is better than beer” (cf. she is as superior to her
husband as champagne is to beer; the distance between her and
her husband is like that between, etc.).

The distance with a comparative is in some instances indicated
by means of the form the from the OE. instrumental case þy. This
is a demonstrative pronoun in such combinations as “I like him
all the better on account of his shyness” | “that makes it all
the worse” | “so much the better” (in the two last examples all
and so much also indicate the distance in addition to the, which is
hardly felt to be more than an unmeaning expletive). But in
“the more, the merrier” and similar collocations of two members,
the first the is relative, while the second the is demonstrative; the
first member may be called the determinant, and the second
the determined. In ordinary E. the two members have exactly
the same construction, and there is nothing to show which is the
dependent and which the principal clause in “the more he gets,
the more he wants”; but in Dan. and G. (and formerly also in
E.) the word-order in such cases shows that the first is the determinant,
and the second the determined; cf. “jo mere han får,
des mere ønsker han” and “je mehr er bekommt, desto mehr
wünscht er.” The same relation between the two is sometimes
indicated by the addition of that after the former the, e.g. The
nearer that he came, the more she fled (Marlowe).

In the Russian construction with čěm… těm, the former is
shown by the form to be a relative, and the latter a demonstrative
pronoun in the instrumental denoting difference. But in French
there is as little formal difference between the two as in English,
and there is not even a word like the: “plus on est de fous, plus
on rit.” The two parts are therefore, even more than in English,
felt to be grammatically on an equality, and this often manifests
itself in the insertion of et as between two independent sentences:
“plus il a, et plus il desire.” 1146

In most cases the determinant is placed first, and it is this
nearly fixed custom which allows of the grammatical conformity
between the two members in English and French. If the order
is reversed, other more explicit or more clumsy formulas than
the usual ones must be used in F. “la figure est d'autant plus
admirable qu'elle est mieux proportionnée” (= mieux la figure
est proportionnée, plus elle est admirable) | “Si la vie réalise un
plan, elle devra manifester une harmonie plus haute à mesure
qu'elle avance plus loin” (Bergson). In English a change in the
word-order generally is all that is required to make the sense
clear: they liked the book the better, the more it made them cry
(Goldsmith).

There is an interesting sub-class of these expressions of proportional
correlation, in which the determinant is the length of
time, but is not explicitly expressed as such. Different languages
have different ways of indicating this: the usual English
way is by means of a repeated comparative, as in “it grew darker
and darker” (= the longer it lasted, the darker it grew) | he
became “more and more impatient,” etc. Similarly in Danish and
other languages. Poets often substitute the positive for the first
comparative, as “and swift and swifter grew the vessel's motion”
(Shelley); another expression is seen in “her position was
becoming daily more insecure.” A third expression is by means
of ever: he spoke ever more indistinctly. This is rare in English,
but the corresponding formula is the usual one in German: es
wurde immer dunkler | er sprach immer weniger. The usual French
equivalent is de plus en plus (de plus en plus obscur | il parla
de moins en moins, etc.). The idea here is that it was already
at the starting point darker (than previously) and that it then
became darker still (but ‘still’ is not expressed).

Secondaries and Tertiaries.

The comparison is in the vast majority of cases between two
primaries as in “John is older than Tom | this house is bigger
than ours | I like claret better than beer.” But sometimes two
secondary or tertiary notions (‘qualities’) may be compared as
in “his speech was more eloquent than convincing | he spoke
more eloquently than convincingly.” Here English requires the
periphrasis with more 1147 (similarly in Danish and German), while
Latin has the well-known illogical expression with the comparative
252in the second adjective (adverb) as well as in the first: verior
quam gratior.

Two verbs may also be compared: he felt rather than saw
her presence in the room. This really implies a stylistic rather
than a real comparison, and means something like “felt would
be a more correct expression than saw.” A similar idea is at the
bottom of such expressions as “this rather frightened him,” where
the second term of comparison is left unexpressed, but where the
original idea is “frightened is a more adequate expression than
any other verb.” This then leads us to such expressions as “there
are some things which I more than dislike” where the first term is
omitted: dislike is too weak an expression.253

Chapter XIX
Time and Tense 1148

The Nine-Tense System. Seven Tenses. Main Divisions of Time. Subordinate
Divisions of Time. Economy of Speech. Non-temporal Use
of Tenses.

The Nine-Tense System.

In this chapter we shall deal with the linguistic expressions for
the natural (or notional) concept “time” and its subdivisions.
In many languages we find time-indications expressed in verbal
forms, the so-called “tenses,” and this has appeared to many
grammarians so natural that they have considered tense-distinction
the chief characteristic of verbs (hence G. zeitwort). But
there are languages whose verbs do not distinguish tenses, and
even in English, which ordinarily distinguishes tenses, we find
such verbs as must, ought, which in the modern language have
only one “tense”; on the other hand, time is often indicated
by means of other words than verbs, and this way of indicating
time is often much more precise than that effected by means of
verbal forms can ever be, as when we say “on the third of February,
1923, at 11.23 p.m.”

Let us, however, confine ourselves in the first place to those
time-distinctions that find expression in the verbs of the best-known
languages. The first question then is, can we establish
a scheme of “tenses” of universal application?

The first line has no special designation; parallel to the others
it should be “in praesenti.”

Closely connected systems with three times three tenses are
found in other works (by Matzen, Kroman, Noreen, see details
and criticism in Tid og tempus, 374) and are there given as purely
logical systems without any regard to the way in which those
nine categories are represented in actual language. Madvig probably
meant his system as an empirical one for Latin exclusively
(in his Greek Syntax he does not give the scheme and would have
had difficulties in finding a place for the aorist in it), but even as a
description of the Latin tenses the system has certain drawbacks.
Scribam is found in two places, as præsens in futuro (I 3) and as
futurum in praesenti (III 1) while other forms are given only once.
In the III series it would be natural to expect 1. scripturus sum,
parallel with the other forms, and the reason for the discrepancy
evidently is that scripturus sum implies a near future, and Madvig
did not want to have the element of distance in time mixed up
with his system. It is, however, difficult to keep this element,
of nearness apart from the other composite forms with scripturus,
and in his Greek Syntax, § 116, Madvig applies the terms futurum
in præsenti and futurum in præterito to the combinations with
mellō and emellon, which admittedly imply nearness in time, and
the same element is also present in the III-series as given by
Kroman and Noreen. If, on the other hand, this element is discarded,
there is no necessity for having both a præsens in futuro
and a futurum in præsenti. These must be regarded as one, represented
by scribam, but then analogy would require us to identify
also I 2 præsens in præterito with II 1 præteritum in præsenti:
the difference between scribebam and scripsi is not indicated with
sufficient precision by their places in the system, as shown incidentally
by Madvig's placing scripturus eram and scripturus fui
at one and the same place (III 2). These two are not synonymous,
being distinguished exactly in the same way as scribebam and
scripsi, but the distinction, to which we shall have to revert, has
really nothing directly to do with the other time-distinctions contained
in the scheme. It would be best, therefore, to reduce the
scheme from nine to seven places, merging into one I 2 and II 1
and in the same way I 3 and III 1.

Seven Tenses.

If now we want to arrange these seven tenses in a consistent
scheme we encounter first the difficulty of terminology. It would
be best to have two separate sets of terms, one for the notional
or natural divisions of time and one for the grammatical (syntactic)
255tense-distinctions. In Danish, and also in German, it is very
convenient to use native terms for the former, and Latin terms
for the latter; thus nutid, fortid, fremtid (jetztzeit, vorzeit, zukunft)
of the three chief divisions of time, and præsens, præteritum, futurum
for the three verbal tenses. But in English we cannot do exactly
the same thing, because there are no native (Anglo-Saxon) words
corresponding to present and future, which thus must be used
both for natural time and for grammatical tense (for it would
hardly do to distinguish between present and præsens, between
future and futurum). We may, however, reserve the word past
(past time) for the notional past and use preterit about the
corresponding tense. Wherever it is required for the sake of
clearness, I shall say present time or present tense, future time
or future tense respectively. For subdivisions I would propose
the employment of the prefixes before and after as notional and
the prefixes ante and post as syntactic designations (e.g. before-past,
ante-preterit).

But this arrangement is not satisfactory, and it is much better
to arrange the seven “times” in one straight line. Before-past
is evidently “past in past,” and in the same way after-past
becomes “future in past,” and analogously before-future is “past
in future,” and after-future is “future in future,” to use clumsy
terms reminding one of Madvig's system.

We thus get a system which avoids Madvig's two serious
logical errors, (1) that of a tripartition of “now,” which as a
point has no dimensions and cannot be divided, and (2) the even
more serious mistake of arranging time in a two-dimensional
scheme with three times three compartments. For there can be
no doubt that we are obliged (by the essence of time itself, or at
any rate by a necessity of our thinking) to figure to ourselves time
as something having one dimension only, thus capable of being
represented by one straight line.256

This figure, and the letters indicating the various divisions,
show the relative value of the seven points, the subordinate “times”
being orientated with regard to some point in the past (Ab) and
in the future (Cb) exactly as the main times (A and C) are orientated
with regard to the present moment (B).

The system thus attained seems to be logically impregnable,
but, as we shall see, it does not claim to comprise all possible time-categories
nor all those tenses that are actually found in languages. 1149
It will now be our task to go through these seven divisions, taking
first the main ones and then the subordinate ones, and to examine
how they are actually expressed in various languages.

Main Divisions of Time.

(A) Simple past time.

(A) Simple past time. — For this there is in English one tense,
the preterit, e.g. wrote. Other languages have two tenses, e.g.
Lat. scripsi, scribebam; on the difference see below, p. 275. While
in these languages the distance of time from the present moment
is quite immaterial, some languages have separate preterits for
the distant and for the near past. The latter is expressed in
French by means of the periphrasis je viens d'écrire.257

Among expressions for the simple past we must here also
mention the so-called historic present, which it would be better
to call the unhistoric present, or, taking a hint thrown out by
Brugmann, the dramatic present. The speaker in using it steps
outside the frame of history, visualizing and representing what
happened in the past as if it were present before his eyes. As
Noreen has it, it serves to produce an artistic illusion. But however
artistic this trick is, it must not be imagined that it is not
popular in its origin; one need only listen to the way in which
people of the humblest ranks relate incidents that they have witnessed
themselves to see how natural, nay inevitable, this form is.
Yet Sweet thinks that in English it is due to literary influence
from French and Latin, and that in the Icelandic sagas, where it
is extremely frequent, it was borrowed from Irish (Philol. Soc.
Proceedings, 1885-87, p. xlv, NEG § 2228). Einenkel and others
think that its use in Middle English is due to Old French. But
in Middle English it is especially frequent in popular poetry, where
foreign influence of a syntactic character is highly improbable.
The non-occurrence or rare occurrence of this present in Old
English must, I think, be explained by the fact that Old English
literature gives us none of those vivid narratives in natural prose
for which Iceland is justly famous. On the whole the dramatic
present belongs to that class of everyday expressions which crop
up comparatively late in writing, because they were looked upon
as being below the dignity of literature. It is never found in
Homer, but is frequent in Herodotus. Delbrück is no doubt
right when he says that it is “gewiss uraltvolkstümlich”
(Synt. 2. 261).

(B) Simple present time.

(B) Simple present time. — For this those languages that have
tense distinctions in their verbs generally use the present tense.

But what is the present time? Theoretically it is a point,
which has no duration, any more than a point in theoretic geometry
has dimension. The present moment, “now,” is nothing but the
ever-fleeting boundary between the past and the future, it is continually
moving “to the right” along the line figured above.
But in practice “now” means a time with an appreciable duration,
the length of which varies greatly according to circumstances;
cf. such sentences as “he is hungry | he is ill | he is dead.” This
is exactly what happens with the corresponding spatial word
“here,” which according to circumstances means very different
things (in this room, in this house, in this town, in this country,
in Europe, in this world), and with the word “we,” which may
embrace a varying number of individuals beside the speaker, the
only thing required being (with here) that the spot where the
present speaker is at the moment, and (with we) that the present
258speaker, is included. With regard to the present tense all languages
seem to agree in having the rule that the only thing required
is that the theoretical zero-point, “now” in its strictest sense,
falls within the period alluded to. This definition applies to cases
like: he lives at number 7 | knives are sharp | lead is heavy |
water boils at 100 degrees Celsius | twice four is eight. With
regard to such “eternal truths” it has sometimes been (wrongly)
said that our languages are faulty because they state them only
in reference to present time without having means to express
that they were equally valid in the past and will be so in the future.
The remark loses its sting when we take into consideration that
most or all of our pronouncements about present time necessarily
concern some part of what belongs strictly to the past and to
the future. If “present time” is defined as is done here, it is
applicable even to intermittent occurrences like the following:
I get up every morning at seven (even when spoken in the evening) 1150 |
the train starts at 8.32 | the steamer leaves every Tuesday
in winter, but in summer both on Tuesdays and Fridays. In the
last sentence the present moment falls within the limits of what
is spoken about, for the saying concerns the present arrangement,
valid for the present year as well as for the last few years and
presumably for the next few years as well.

This manner of viewing things seems to me preferable to that
adopted by Sweet, who writes (NEG, § 289) that “for the purpose
of such statements (as the sun rises in the east, platinum is the
heaviest metal) the present is best suited, as being in itself the
most indefinite of the tenses” — why indefinite? Still less can
we call such sentences “timeless” (zeitlos), as is often done. 2151
It would be better to speak of “generic time” in the same way
as we have spoken of “generic number” and “generic person.”
If for such statements the present tense is generally used, it is in
order to affirm that they are valid now. But other tenses may
occasionally be used: we have the so-called “gnomic preterit”
as in Shakespeare's “Men were deceivers ever” (cf. the Greek
gnomic aorist) — a sort of stylistic trick to make the hearer himself
draw the conclusion that what has hitherto been true is so
still and will remain so to the end of time. On the other hand, the
future tense is used “gnomically” in Fr. “rira bien qui rira le
dernier,” where the corresponding proverbs in other languages
259use the present tense: the reason for the French tense is that the
proverb is most often quoted when somebody else is laughing and
the speaker wants to say that he will laugh later and that that
will be better. 1152

(C) Simple future time.

(C) Simple future time. — It is easy to understand that expressions
for times to come are less definite and less explicit in our
languages than those for the past: we do not know so much about
the future as about the past and are therefore obliged to talk
about it in a more vague way. Many languages have no future
tense proper or have even given up forms which they had once
and replaced them by circuitous substitutes. I shall here give
a survey of the principal ways in which languages have come to
possess expressions for future time.

(1) The present tense is used in a future sense. This is particularly
easy when the sentence contains a precise indication of
time in the form of a subjunct and when the distance in time from
the present moment is not very great: I dine with my uncle
to-night. The extent to which the present tense is thus used is
different in different languages; the tendency is strongest with
verbs denoting ‘go’: I start to-morrow | ich reise morgen ab |
jeg rejser imorgen | je pars demain | parto domani, etc. Gr. eîmi
‘I go’ nearly always means ‘I shall go.’ The present tense is
also extensively used in clauses beginning with when and if: “I
shall mention it when I see him (if I see him)”; in French with
si: “Je le dirai si je le vois,” but not with quand: “quand je le
verrai.”

(2) Volition. Both E. will and Dan. vil to a certain degree
retain traces of the original meaning of real volition, and therefore
E. will go cannot be given as a pure ‘future tense,’ though it
approaches that function, as seen especially when it is applied to
natural phenomena as it will certainly rain before night. There is
also an increasing tendency to use (wi)ll in the first person instead
of shall, as in I'm afraid I'll die soon (especially in Sc. and Amr.),
which makes will even more the common auxiliary of the future.
In German wollen has to be used in “es scheint regnen zu wollen,”
because the usual auxiliary werden cannot be used in the infinitive.
The future is, expressed by volition also in Rumanian
voiu canta ‘I will (shall) sing’; cp. also occasional It. vuol piovere
(Rovetta, Moglie di Sua Eccel. 155). In Modern Greek the idea
of volition seems to have been completely obliterated from the
combinations with tha: tha graphō and iha grapsō ‘I shall write’
(regularly, or once); tha, formerly thena, is derived from the
260third person the = thelei + na ‘that’ from hina and has now
become a pure temporal particle. 1153

(3) Thought, intention. ON mun. This cannot easily be
kept apart from volition.

(4) Obligation. This is the original meaning of OE. sceal,
now shall, Dutch zal. In English the meaning of obligation is
nearly effaced, but the use of the auxiliary is restricted to the first
person in assertions and to the second person in questions, though
in some classes of subordinate clauses it is used in all three persons. 2154
The meaning of obligation also clung at first to the Romanic form
from scribere-habeo ‘I have to write,’ which has now become a
pure future tense, It. scriverò, Fr. écrirai, etc. Under this head
we may also place E. is to as in “he is to start to-morrow.”

(5) Motion. Verbs meaning ‘go’ and ‘come’ are frequently
used to indicate futurity, as in Fr. je vais écrire, used of the near
future, E. I am going to write, which sometimes, though by no
means always, has the same nuance of nearness, and finally without
that nuance Swed. jag kommer att skriva, Fr. quand je viendrai
à mourir, E. I wish that you may come to be ashamed of what you
have done | they may get to know it. (But Dan. jeg kommer til at
skrive denotes either the accidental or the necessary, either ‘I
happen to write’ or ‘I (shall) have to write’.)

(6) Possibility. E. may frequently denotes a somewhat vague
futurity: this may end in disaster. Here we may mention those
cases in which an original present subjunctive has become a future
tense, as Lat. scribam.

(7) There are other ways in which expressions for futurity
may develop. G. ich werde schreiben according to some is derived
from a participial construction ich werde schreibend, but this is
not always recognized; it is not mentioned in Paul Gr 4. 127 and
148, where the treatment of the future is very unsatisfactory.
The Gr. future in -sō (leipsō, etc.) is said to have been originally
a desiderative.

A notional imperative necessarily has relation to the future
time. Where, as in Latin, there are two tenses in the imperative,
both really refer to the future, the so-called present imperative
referring either to the immediate future or to some indefinite time
261in the future, and the so-called future imperative being used chiefly
with regard to some specially indicated time. A “perfect imperative”
also refers to future time, the use of the perfect being a
stylistic trick to indicate how rapidly the speaker wants his command
executed: be gone! When we say Have done! we mean
the same thing as “Stop at once!” or “Don't go on!” but this
is expressed circuitously: ‘let that which you have already done
(said) be enough.’

Subordinate Divisions of Time.

Next we come to consider the subordinate divisions of time,
i.e. points in time anterior or posterior to some other point (past
or future) mentioned or implied in the sentence concerned.

(Aa). Before-past time. This requires to be mentioned so
frequently that many languages have developed special tenses
for it: ante-preterit (pluperfect, past perfect), either simple as
Lat. scripseram or periphrastic, as E. had written and the corresponding
forms in the other Gothonic and in the Romanic languages.
In OE. before-past was often indicated by means of the
simple preterit with the adverb ær: þæt þe he ær sæde ‘what he
had said,’ literally ‘that which he before said.’

(Ac). After-past time. I know of no language which possesses
a simple tense (post-preterit) for this notion. A usual expression
is by a verb denoting destiny or obligation, in E. most often was
to: Next year she gave birth to a son who was to cause her great
anxiety | It was Monday night. On Wednesday morning Monmouth
was to die (Macaulay) | he was not destined to arrive there
as soon as he had hoped to do (Kingsley). Similarly in other languages.
Dan.: Næste år fødte hun en søn som skulde volde
hende store bekymringer | G. Im nächsten jahre gebahr sie einen
sohn, der ihr grosse bekümmernis verursachen sollte | Fr. Quand
Jacques donna à l'électeur Frédéric sa fille qui devait être la tige
des rois actuels d'Angleterre (Jusserand) | Je ne prévoyais point
262tous les malheurs qui allaient nous frapper coup sur coup
(Sarcey). Sometimes in Fr. the future is used, which corresponds
to the dramatic present: Irrité de l'obstination de Biron et voulant
donner à la noblesse un de ces exemples que Richelieu multipliera,
Henri IV laissa exécuter la sentence. Gr.: tēn hodon hēi de
emellen enoi kaka kēde' esesthai (Od. 6. 165 ‘the expedition that
was to bring about sufferings’; cf. ibid. 7. 270, 8. 510). 1155

(Ca). Before-future time. The corresponding tense (the ante-future)
is usually termed futurum exactum or the future perfect.
Lat. scripsero, in our modern languages periphrastic: I shall have
written (he will have written), er wird geschrieben haben, il aura
écrit, etc. In Dan. the element of futurity is generally left unexpressed:
Hvis du kommer klokken 7, har han skrevet brevet
(… har vi spist, … er solen gået ned). Thus also in E. and
G. after conjunctions of time: I shall be glad when her marriage
has taken place | ich werde froh sein wenn die hochzeit
stattgefunden hat.

(Cc). After-future. This has chiefly a theoretic interest, and
I doubt very much whether forms like I shall be going to write
(which implies nearness in time to the chief future time) or scripturus
ero are of very frequent occurrence. Madvig has an example
from Cicero: “Orator eorum, apud quos aliquid aget aut acturus
erit, mentes sensusque degustet oportet,” but it will be seen that
here the future aget, which drags the after-future along with it,
is really a generic present, put in the future tense on account of
oportet: it is (now and always) the duty of the orator to consider
those before whom he is talking or will talk (is going to talk).
Otherwise it must be said that the natural expression for what
at some future time is still to come is a negative sentence: If
you come at seven, we shall not yet have dined (… the sun will
not yet have set) | si tu viens à sept heures, nous n'aurons pas
encore diné (… le soleil ne se sera pas encore couché). In Dan.
generally with the element of futurity unexpressed: hvis du
263kommer kl. 7, har vi ikke spist endnu (… er solen ikke gået ned
endnu). 1156

Economy of Speech.

Languages differ very much in their economy in the use of
tenses as well as in other respects. Those languages which admit
sentences like “I start to-morrow” use one sign for the future
time (the adverb) where other languages force their speakers to
use two, as in “eras ibo” (I shall start to-morrow). This is
parallel to the economical expression in “my old friend's father”
with only one genitive mark as compared with “pater veteris
mei amici,” or to “ten trout” as against “ten men” or “decem
viri.” Latin is often praised for its logic in such things, as when
Weise writes: “Der gesunde menschenverstand befähigte den
römer besonders zu genauer scheidung der begriffe, schärfe der
darstellung, klarheit und durchsichtigkeit der rede… . Der
gebildete römer ist peinlich sorgfältig in der tempusbezeichnung:
‘Ich werde kommen, wenn ich kann’ heisst bei ihm: veniam, si
potero; ‘wie du sähest, so wirst du ernten’: ut sememtem feceris,
ita metes; ‘so oft er fiel, stand er auf’: cum ceciderat, surgebat.”
English and Danish in these matters generally agree with German.
But it must be remembered that it cannot be called illogical to
omit the designation of what goes without saying: situation and
context make many things clear which a strict logician in a pedantic
analysis would prefer to see stated. Nor should it be forgotten
that Latin in other cases is economical enough. Postquam urbem
liquit: here the before-past time is expressed by the combination
of postquam (before) and liquit (past); English allows both the
shorter and the more explicit expression: after he left the town,
after he had left; Danish and German requires the double expression:
efterat han havde forladt byen | nachdem er die stadt verlassen
hatte. Latin is also economical in omitting the mark of past time
in hoc dum narrat, forte audivi ‘while she was telling this tale I
happened to overhear it.’ There are really two (relative) time-indications
saved in Shakespeare's “our vizards wee will change
after we leaue them” (after we shall have left them, Ca), and in
“you must leave the house before more harm is done” (= shall
have been done). Such savings of time-indications in the tense
of the verb are particularly frequent after conjunctions of time
and of condition; note thus the difference between the two when-clauses:
“We do not know when he will come, but when he comes
he will not find us ungrateful” — the first when is interrogative,
264and the second a relative adverb or a conjunction. In French
with quand we should have il viendra in both clauses, but if we
substitute if, we see the same difference as in English: “Nous
ne savons pas s'il viendra, mais s'il vient il ne nous trouvera pas
ingrats.”

Non-temporal Use of Tenses.

What is usually a grammatical sign for a time relation may
sometimes be used for other notional purposes. Thus a future
tense is often used to express a mere supposition or surmise with
regard to the present time: il dormira déjà = he will already be
asleep = er wird schon schlafen (I suppose that he is asleep) and
in the same way il l'aura vu = he will have seen it = er wird es
gesehen haben (he has probably seen it). It is true that we can
assert nothing with regard to a future time but mere suppositions
and surmises, and this truth is here linguistically reversed as if
futurity and supposition were identical. Or it may be that the
idea is this: ‘it will (some time in the future) appear that he is
already (at the present moment) asleep,’ in the same way as we
may use hope, which implies the future, with a subordinate clause
in the present or perfect: “I hope he is already asleep,” “I hope
he has paid his bill,” i.e. that it will turn out later that he is now
asleep or has now paid.

The most important non-temporal use of preterit forms is to
indicate unreality or impossibility. This is found in wishes and
in conditional sentences. If we want to find a logical connexion
between this use and the normal temporal use of the preterit,
we may say that the common link is that something is in all these
cases denied with regard to the present time. “At that time
he had money enough,” “I wish he had money enough,” and
“If he had money enough” — each of these sentences is in its
own way a contrast to “he has money enough.”

“I wish he had money enough” expresses by its preterit a
wish with regard to the present time, and at the same time its
impossibility or unreality (unfortunately he has not money
enough); in the same way the ante-preterit in “I wish he had
had money enough” expresses a wish with regard to some past
period and at the same time denies that he had money enough
then. But with regard to future time it is not as a rule possible
to deny anything so categorically, and the corresponding tense-shifting
(would instead of will) therefore merely serves to express
uncertainty of fulfilment: “I wish he would send the money to-morrow,”
whereas “I hope he will send the money to-morrow”
expresses the wish without saying anything about the probability
of its fulfilment.265

In conditional clauses we see the same shiftings. “If he had
money enough” has reference to the present time and denies
that he has; “if he had had money enough” has reference to
the past and denies that he had money enough; “if he should
have money enough” has reference to the future, but instead of
denying it only leaves it uncertain whether he will get it or no.
But the last form may be used also to express a doubt with regard
to the present time: “if he should be innocent” — meaning perhaps
in most cases “if it turns out (fut. time) that he is (now)
innocent,” etc. — In speaking of the future the simple preterit
(without should) may also be used: “It would be a pity if he
missed the boat to-morrow.” 1157

We may sometimes, chiefly in colloquial speech, meet with
a further shifting, the ante-preterit being used not only of the
past, but also of the present time, simply to intensify the unreality
irrespective of time. Thus we may say: “If I had had money
enough (at the present moment), I would have paid you,” and
“I wish I had had money enough (now) to pay you.”

It is also interesting to observe that the use of the preterit
to denote unreality at the present time leads to the consequence
that it may be used in speaking of the future, as in “It is high time
the boy went to bed.”

In wishes and conditions the unreality or impossibility was
not originally denoted by the tense-shifting in itself, but required
also the shifting from the indicative to the subjunctive, as still
in German. But in Danish there is now in the preterit (and ante-preterit)
no formal distinction between the two moods, and the
modification of meaning is thus made contingent on the tense only.
It is the same in English in more than 99 per cent, of the cases,
as the old preterit subjunctive is identical with the indicative,
except in the singular of the one verb be, where was and were are
still distinct. It is easy to understand, therefore, that the instinctive
feeling for the difference between these two forms cannot
be vivid enough to prevent the use of was, where were would have
been required some centuries ago. Since ab. 1700 was has been
increasingly frequent in these positions: I wish he was present
to hear you (Defoe) | a murder behind the scenes will affect the
audience with greater terror than if it was acted before their eyes
(Fielding). In literary language there has recently been a reaction
in favour of were, which is preferred by most teachers;
but in colloquial speech were is comparatively rare, except in the
phrase “if I were you,” and it is worth remarking that was is
266decidedly more emphatic than were, and thus may be said to
mark the impossibility better than the old subjunctive form:
“I'm rot rich. I wish I was” | “I am ill. If I wasn't, I should
come with you” — thus often in the negative form. In this way
we get a distinction between “If he were to call” with weak
were, denoting vaguely a future possibility, and “If he was to
call” with strong was, denying that he is to call (now),
with the use of is to which is nearly synonymous with has to,
is bound to: “If I was to open my heart to you, I could show
you strange sights” (Cowper) | “If I was to be shot for it I
couldn't” (Shaw). 1158

In French we have the corresponding use of the preterit and
ante-preterit in conditional sentences, and here too the indicative
has prevailed over the subjunctive, though the forms were more
different than was the case in English and Danish: “s'il avait
assez d'argent, il payerait,” formerly “s'il eût…”

I have here spoken of the tense in the conditional (subordinate)
clause only, but originally the same rules applied to the conditioned
(principal) clause as well. Thus we have: “But if my father
had not scanted me…. Yourselfe, renowned Prince, than stood
as faire As any commer (Sh.) | She were an excellent wife for Benedick”
(Sh.). Correspondingly in the ante-preterit: “If thou hadst
bene here, my brother had not died” (A.V.). But just as there is
a strong tendency to express the future more clearly in principal
sentences than in subordinate clauses (which in English is effected
by the use of will or shall), in the same way the shorter expression
has in these conditioned sentences been supplanted by a fuller
one with should or would: you would stand | she would be | my
brother would not have died, etc. Could and might are still used
in the old way in principal sentences because these verbs have
no infinitives and thus cannot be combined with should or would;
e.g. How could I be angry with you? | He might stay if he liked.
In French we witness a similar development, il vînt (venait) in
a conditioned sentence having been ousted by il viendrait, which
originally denoted an obligation in the past (‘he had to come’),
but is now chiefly used as what is generally termed “le conditionnel,”
e.g. in “s'il pouvait, il viendrait.” Similarly in the past: mon
frère ne serait pas mort, s'il l'avait su.

Special applications of the preterit of unreality are seen in
the use of should and ought to indicate an obligation or duty, etc.,
267in the present time, and in the “modest” use of could for can.
(Could you tell me the right time), of would for will (Would you
kindly tell me…) and of might for may (Might I ask…)
It has finally led to the change of must from a preterit into a present
tense; cf. also Swed. måste. Further details must be left to
special grammars.268

Chapter XX
Time And Tense — concluded

The Perfect. Inclusive Time. Passive Tenses. Aorist and Imperfect.
The English Expanded Tenses. Terms for the Tenses. Time-Relations
in Nouns (including Infinitives). Aspect.

The Perfect.

The system of tenses given above will probably have to meet
the objection that it assigns no place to the perfect, have written,
habe geschrieben, ai écrit, etc., one of the two sides of Lat. scripsi,
and in Latin often called perfectum absolutum or “perfect
definite.” This, however, is really no defect in the system, for
the perfect cannot be fitted into the simple series, because besides
the purely temporal element it contains the element of result.
It is a present, but a permansive present: it represents the present
state as the outcome of past events, and may therefore be called
a retrospective variety of the present. That it is a variety of the
present and not of the past is seen by the fact that the adverb
now can stand with it: “Now I have eaten enough.” “He has
become mad” means that he is mad now, while “he became
mad” says nothing about his present state. “Have you written
the letter?” is a question about the present time, “Did you
write the letter?” is a question about some definite time in the
past. Note also the difference of tense in the dependent clause
in “He has given orders that all spies are to be shot at once”
and “He gave orders that all spies were to be shot at once.” We
may perhaps figure this by means of the letters BA or B(A) —
the letters A and B being taken in the sense shown on p. 257 above.

It is highly probable that the old Aryan perfect was at first
an intensive present or “permansive”; this view is advocated
very cogently by Sarauw (Festschrift Vilh. Thomsen, 1912, p. 60):
“The perfect originally denoted the state: odi I hate, memini
I remember, hestēka I stand, kektēmai I possess, kekeutha I contain
hidden within me, heimai I wear, oida I have before my eyes.
The meaning of perfect was gained by an inference: he who
possesses has acquired; he who wears a garment has put it on.”

The two sides of the perfect-notion cannot easily be maintained
in a stable equilibrium. Some of the old perfects are used
269exclusively as real presents, e.g. Lat. odi, memini; in the Gothonic
languages the so-called præteritopræsentia, which would be better
called perfectopræsentia, 1159 e.g. E. can, may, Gothic wait, corresponding
to Gr. oida, ON. veil, OE. wat, obsolete E. wot, etc. But
apart from these what were perfects in the Gothonic languages
have lost the present-element and have become pure preterits, as
E. drove, sang, held, etc. To express the perfect-meaning compounds
with have were then formed: I have driven, sung, held,
etc. In quite recent times one of these combinations has become
a pure present (thus a new perfectopresent verb): I have got (I've
got): the retrospective element is quite absent in I've got no time |
you've got to do it 2160

The Latin perfect, which originated in an amalgamation of
old preterits (aorists) and perfects, 3161 combines the syntactic functions
of those two tenses. In Romanic verbs, however, we witness
the same development as in the majority of the Gothonic verbs,
the old perfect forms having lost their perfect-function and having
become pure preterits, though with this difference from the
Gothonic verbs, that they are aorists (now termed passé défini,
passé historique, past historic), because side by side with them
there are imperfects (see below). The real perfect as in Gothonic
is expressed periphrastically: ho scritto, ai écrit, etc. (On have
as an element in the perfect of many languages see Meillet LH 189.)

Now, in spite of the employment of the present-tense form
have in these new perfects, it appears difficult to keep up the sharp
distinction between the idea of the present result of past events
and that of these past events themselves: the perfect tends to
become a mere preterit, though the tendency is not equally strong
in all languages. English is more strict than most languages, and
does not allow the use of the perfect if a definite point in the past
is meant, whether this be expressly mentioned or not. Sentences
containing words like yesterday or in 1879 require the simple preterit,
so also sentences about people who are dead, except when
something is stated as the present effect of their doings, e.g. in
Newton has explained the movements of the moon (the movements
of the moon have been explained — namely by Newton). On the
other hand: Newton believed in an omnipotent God. “We can
say ‘England has had many able rulers,’ but if we substitute
Assyria for England the tense must be changed” (Bradley ME. 67).

German is much more lax in this respect, and South German
tends to use the compound perfect everywhere: ich habe ihn
gestern gesehen. On the other hand, Germans (North Germans?)
270will often say: Waren Sie in Berlin? where an Englishman would
have to say “Have you been in Berlin?” When an Englishman
hears a German ask “Were you in Berlin?” his natural inclination
is to retort: “When?” Danish steers a middle course
between the strictness of English and the laxity of German; a
Dane, for instance, will always ask “Har De været i Berlin?”
but has no objection to combinations like “jeg har set ham
igår” (I have seen him yesterday). If, however, the indication
of time precedes, the preterit is required: “igår såe jeg ham”
— the psychological reason being that in the former case the sentence
was at first framed as it would be without any time-indication,
and the indication is as it were an afterthought, added to
sentence when virtually completed “jeg har set ham,” whereas
if we begin with “ yesterday ” the tense naturally follows suit.

In Spanish the distinction seems to be accurately observed;
Hanssen (Sp. gr. 95) has examples corresponding to the English
ones given above: Roma se hizo señora del mundo | La Inglaterra
se ha hecho señora del mar. But in French the feeling for the distinction
is lost, at any rate in present-day colloquial Parisian and
North French, where the passé défini is entirely disused: Je l'ai
vu hier | ils se sont mariés en 1910. The transition from a perfect
to a preterit seems to be due to a universal tendency; at any
rate we meet with it in so remote a language as Magyar, where
írt ‘has written’ in the ordinary language has supplanted íra
‘wrote’ (Simonyi US 365).

A retrospective past time, bearing the same relation to some period
in the past as the perfect does to the present, cannot be kept distinct
from the before-past (ante-preterit) mentioned above: had written 1162

In the same way what was above called before-future (ante-future)
cannot be kept apart from a retrospective future: will
have written. The periphrasis with forms of the verb have seems
to indicate that people are inclined to look upon these two tenses
as parallel with the perfect rather than with the simple preterit;
hence also the terms “past perfect” and “future perfect.”

Inclusive time.

Not infrequently one may need to speak of something belonging
at once to the past and to the present time. Two tenses may be
271combined: I was (then) and am (still) an admirer of Mozart | I
have been and am an admirer of Mozart. But if an indication of
duration is added, we can combine the two into what might be
called an inclusive past-and-present. On account of the composite
character of this idea some languages use the perfect, like
English and Danish, and others the present tense, like German
and French: I have known him for two years | jeg har kendt
ham i to år || ich kenne ihn seit zwei jahren | je le connais depuis
deux ans. Note the difference in the preposition used in the
different cases. In Latin we have the same rule as in French,
only without a preposition: annum jam audis Cratippum. It is
evident that this time relation renders it impossible to find a place
for it in our time-series above; but it might be expressed by
means of the letters B&A.

Corresponding expressions are found with reference to the
past and to the future time: in 1912 I had known him for two
years | i 1912 havde jeg kendt ham i to ar || in 1912 kannte ich
ihn seit zwei jahren | en 1912 je le connaissais depuis deux ans ||
next month I shall have known him for two years | næste måned
har jeg (vil jeg ha) kendt ham i to ar || im nächsten monat werde
ich ihn seit zwei jahren kennen | le mois prochain je le connaîtrai
depuis deux ans. It goes without saying that these latter expressions
are not very frequent.

Passive Tenses.

It will be well to keep in mind the double-sided character of the
perfect when we come to treat of the tenses in the periphrastic
passive of the Romanic and Gothonic verbs. In classical Latin,
where we had the real present passive in -r: scribitur, the composite
form scriptum est is a perfect ‘it is written, i.e. has been
written, exists now after having been written.’ But in the
Romanic languages the r-passive has disappeared, and the meaning
of the periphrasis has been partly modified. This subject
has been treated by Diez (GRS 3. 202) better than by anybody
else. He quotes from early documents examples like quæ ibi
sunt aspecta for aspiciuntur, est possessum for possidetur, and then
goes on to divide verbs into two classes. In the first the action
is either confined to one single moment, e.g. catch, surprise, awake,
leave, end, kill, or imply a final aim (endzweck), e.g. make, bring
about, adorn, construct, beat; here the passive participle denotes
the action as accomplished and finished, and the combination
with sum in Romanic as in Latin is a perfect. Ex. il nemico è
battuto, l'ennemi est battu = hostis victus est; era battuto, io sono
abandonato, sorpreso; la cosa è tolta via. Diez calls these verbs
272perfective. The second class (imperfective) comprises verbs
denoting an activity which is not begun in order to be finished,
e.g. love, hate, praise, blame, admire, see, hear, etc. Here the
participle combined with sum denotes present time: egli è amato
da tutti, il est aimé de tout le monde = amatur ab omnibus; è biasimato,
lodato, odiato, riverito, temuto, veduto. In Romanic as in
Latin the participles of the first class by losing their temporal
signification tend to become adjectives (eruditus est, terra ornata
est floribus). If now the notion of past time has to be attached
to those participles which tend to become adjectives, the new
participle of esse is used for that purpose: il nemico è stato battuto,
l'ennemi a été battu. For the present time the active construction
is preferred: batton il nemico, on bat l'ennemi. In It. and Sp.
venire may also be used as an auxiliary of the passive for present
time.

The distinction between two classes, which Diez thus saw very
clearly, has been developed by H. Lindroth in two excellent papers
(PBB 31. 238 and Om adjektivering of particip, Lund 1906).
Lindroth for the first class uses the term ‘successive’ (with the
subdivisions ‘terminative’ and ‘resultative’), and for the second
the term ‘cursive.’ Even at the risk of seeming needlessly to
multiply existing terms I venture to propose the names conclusive
and non-conclusive.

In German and Danish, where there are two auxiliaries,
werden, blive on the one hand, and sein, være on the other, it does
not matter very much whether one or the other is chosen with
verbs of the second class (non-conclusive): er wird geliebt (ist
geliebt) von jedermann, han bliver elsket (er elsket) av alle = jedermann
liebt ihn, alle elsker ham. 1163 But with verbs of the first
class (conclusive) the auxiliaries denote different tenses: er wird
überwunden, han bliver overvundet = man überwindet ihn, man
overvinder ham; but er ist überwunden, han er overvundet = man
hat ihn überwunden, man har overvundet ham. In the latter
case it is possible to denote the perfect passive more explicitly
by means of the composite er ist überwunden worden, han er blevet
overvundet.

In English the old auxiliary weorðan, corresponding to G.
werden, has disappeared, and matters are now pretty much as
in French. If first we consider non-conclusive verbs (Diez's second
class), we see that when participles like honoured, admired, despised
273are used as adjuncts as in an honoured colleague, they say nothing
about time and may according to circumstances be used about
any time (an honoured colleague of Bacon). The combination
is honoured, is admired, etc., therefore belongs to the same (present)
tense as the simple is.

It is different with conclusive participles like paid, conquered,
lost, etc. In adjunct-combinations they denote the result of past
action: a paid bill | conquered towns | a lost battle. Combinations
with the auxiliary is may have two different meanings, according
as the perfect-signification inherent in the participle or the present-signification
of is comes to predominate; cf. the two sentences:
his bills are paid, so he owes nothing now (sind bezahlt; he has
paid) | his bills are paid regularly on the first of every month
(werden bezahlt, he pays). The preterit “his bills were paid”
may, of course, have the two corresponding meanings. Cf. the
following instances: he was dressed in the latest fashion | the
children were dressed every morning by their mother | at that
time they were not yet married, but they were married yesterday.
I take a final example from a paper by Curme, only modifying it
slightly: When I came at five, the door was shut (war geschlossen),
but I do not know when it was shut (geschlossen wurde). I think
the best way to make the distinction clear is to point out how
the opposite statement would run: When I came at five, the door
was open (thus the adj.), but I do not know when it was opened.

There is evidently a source of ambiguity here, 1164 but it must
be recognized that some correctives have been developed in the
course of the last few centuries. In the first place the combinations
has been, had been with a participle, which were rare in
Elizabethan English, have become increasingly frequent. Shakespeare
very often has is, where a modern writer would undoubtedly
use has been, e.g. Sonn. 76 Spending againe what is already spent…
So is my loue still telling what is told | John IV. 2. 165 Arthur,
whom they say is kill'd to-night on your suggestion. Thus also
in the Authorized Version, e.g. Matt. 5. 10 Blessed are they which
are persecuted for righteousness sake, in the Revised Version:
Blessed are they that have been persecuted 2165 In the second place
the verbs become and, especially in colloquial speech, get, are more
and more used where be would be ambiguous, e.g. taking it into
his head rather late in life that he must get married (Dickens) |
“I am engaged to Mr. W.” — “ You are not engaged to anyone.
274When you do become engaged to anyone, I or your father will
inform you of the fact” (Wilde). 1166 Finally the comparatively
recent combination is being is in some cases available to make
the meaning unmistakable. Thus we see that present-day English
has no less than three new expressions by the side of the old the
book is read, namely the book has been read, gets read, is being read.
This specialization has been an evident gain to the language.

Aorist and Imperfect.

We saw above that Lat. scripsi besides being a perfect (‘have
written’) was a preterit (‘wrote’), but that in the latter capacity
it had beside it another preterit scribebam. We shall now discuss
the difference between these two kinds of preterit, using the names
found in Greek grammars, aorist and imperfect. In French grammars,
as we have also seen, the aorist is variously termed le passé
défini or le passé historique; the latter name (past historic) has
been adopted by the Committee on Grammatical Terminology,
though the historian seems to require not only that kind of preterit,
but also the imperfect.

Now, as to the meaning of the aorist and the imperfect. Both
denote past time and they cannot be placed at different points
of the time-line drawn, p. 257, for they bear the same relation to
the present moment and have no relation to the subdivisions
denoted by the prefixes before and after. Nor have they any
reference in themselves to the duration of the action concerned,
and we cannot say that one is momentary or punctual, and the
275other durative. An indication of length of duration may be added
to both, e.g. in: ebasileuse tessera kai pentēkonta etea ‘he reigned
fifty-four years’ | Lucullus multos annos Asias præfuit | Louis XIV
régna soixante-douze ans et mourut en 1715 | De retour de ces
campagnes il fut longtemps malade; il languit pendant des années
entières.

The two tenses correspond to the two meanings of E. then,
(1) next, after that, as in “then he went to France” (Dan. dærpå),
and (2) ‘at that time’ as in “then he lived in France” (Dan.
dengang). The aorist carries the narrative on, it tells us what
happened next, while the imperfect lingers over the conditions as
they were at that time and expatiates on them with more or less
of prolixity. One tense gives movement, the other a pause. One
Latin grammarian, whom I have seen quoted I forget where,
expresses this tersely: Perfecto procedit, Imperfecto insistit
oratio. Krüger similarly says that the aorist grips (zusammenfasst)
and concentrates, the imperfect discloses (entfaltet). Sarauw
expands this (KZ 38. 151), saying that in the former “abstraction
is made from what is inessential, from the circumstances under
which the action took place and from interruptions that may
have occurred, and what was really a whole series of actions is
condensed into one action, the duration of which is not, however,
abbreviated.” It is noteworthy that, as Sarauw emphasizes, an
aorist was formed from the imperfective as well as from the perfective
verbs in Old Slavic. In the same way French uses its
aorist (passé historique) with any verb, no matter what its meaning
is. We may perhaps be allowed with some exaggeration to
say in the biblical phrase that the imperfect is used by him
to whom one day is as a thousand years, and the aorist by him to
whom a thousand years are as one day. At any rate we see that
terms like the G. “aktionsart” are very wide of the mark: the
distinction has no reference to the action itself, and we get much
nearer the truth of the matter if we say that it is a difference in
the speed of the narrative; if the speaker wants in his presentation
of the facts to hurry on towards the present moment, he
will choose the aorist; if, on the other hand, he lingers and takes
a look round, he will use the imperfect. This tense-distinction is
really, therefore, a tempo-distinction: the imperfect is lento and
the aorist allegro, or perhaps we should say ritardando and
accelerando respectively.

This will make us understand also that there is often a distinctive
emotional colouring in the imperfect which is wanting in
the aorist tense.

In the composite before-past the corresponding distinction
exists in Fr. j'avais écrit and j'eus écrit. Here too ai eu has been
276substituted in popular language for eus, as in “Quand ma femme
a eu trouvé une place, elle a donné son enfant à une vieille pour le
ramener au pays” (Daudet).

In the same way as the Latin perfect had two functions, the
imperfect in Latin, Romanic, Greek, etc., has two functions, for
besides the lingering action we have just been discussing it denotes
an habitual action in some past period. Here, therefore, the
time-notion is bound up with the idea of repetition, which is really
a numerical idea (cf. under Number, p. 210): the plural idea with
regard to the verbal action which is expressed in this use of the
imperfect is of the same order as that which finds a stronger
expression in iterative or frequentative formations.

We are now in a position to give the following comparative
scheme of tenses in some well-known language, line 1 denoting
the real perfect, line 2 the aorist, line 3 the habitual imperfect,
and line 4 the descriptive imperfect. This survey shows clearly
how some languages confuse time distinctions which in others are
kept strictly apart.

The English Expanded Tenses.

In the survey just given we found two renderings of Lat.
scribebam in English, wrote for the habitual action, and was writing
for the descriptive imperfect. Corresponding expressions are found
in the present, etc., as English possesses a whole set of composite
tense-forms: is writing, was writing, has been writing, will (shall)
be writing, will (shall) have been writing, would (should) be writing,
would (should) have been writing, and in the passive is being written,
was being written — Sweet in his tense system even gives I have
been being seen, I had been being seen, I shall be being seen, I should
be being seen, I shall have been being seen, though it would certainly
be possible to read the whole of English literature without being
able to collect half a dozen examples of some of these “forms.”
Very much has been written by grammarians about these combinations,
which have been called by various names, definite tenses,
progressive tenses, continuous tenses. I prefer to call them
expanded tenses, because this name is sufficiently descriptive of
277the formation without prejudging anything with regard to its
employment.

With regard to the historical development of these forms
I have given a preliminary account of my researches in Tid og
Tempus, pp. 406-120 with criticism of earlier views, and shall here
give only a very short summary. My main result is that the
modern construction owes very little to the OE. construction
wæs feohtende, which in ME. plays no important rôle, but that it
arose chiefly through aphesis from the construction of the verbal
substantive with the preposition on: is on huntinge, is a-hunting,
is hunting (as burst out on weeping, a weeping, weeping; set the
clock on going, a going, going). This explains the fact that these
forms become more common just when aphesis (in back from on
bæc, aback, etc., etc.) became particularly frequent, while it also
explains the use of the prep, of before an object (still heard in
vu gar speech), and the passive signification in the house was
building, and — last, not least — it helps us to understand the exact
meaning of the expanded tenses in Modern English, which is much
more precise than was that of the OE. and ME. participial combinations.
We must remember that the preposition on was often
used where now we say in: he is on hunting means ‘he is in (the
middle of) the action of hunting,’ and thus contains two elements,
first ‘being,’ with which is connected the time-indication, and
second ‘hunting,’ which forms as it were a frame round ‘is.’ The
action described by the word hunting has begun before the moment
denoted by is (was), but has not yet ceased; cf. Fr. il était à se
raser, quand est venu son beau-frère.

Verbs denoting psychological states, feelings, etc., cannot as
a rule be used in the expanded tenses; this is easily explained
278if we start from the combination is on -ing, for we can hardly
say: he is on (engaged in, occupied in) liking fish, etc. Nevertheless
it is possible in speaking of a passing state to say “I am feeling
cold.”

The expanded forms of verbs denoting movement, like go,
come, must be specially mentioned. They are first used in the
ordinary way wherever the verbs have some special signification
which does not in itself call up the idea of a beginning movement:
My watch has stopped, but the clock is going | things are coming
my way now | you are going it, I must say. In the second place
they may be used where a single action of coming or going is out
of the question: the real hardships are now coming fast upon
us | She turned to the window. Her breath was coming quickly |
cigarettes were then coming into fashion. But in most cases is
coming, is going are used of the future, exactly as the corresponding
verbs in many languages acquire the meaning of future time in
their present tense (Gr. eîmi, etc., see p. 261). The auctioneer will
say: Going, going, gone. Thus also: I am going to Birmingham
next week | Christmas is coining, the geese are getting fat. Thus
we get the expression for a near future: he is going to give up
business; even: he is going to go.

Most of the uses of the expanded tenses in Modern English
will be covered by the rules given here, and what has been said
about the longer time as a frame for something else will be found
particularly helpful. Yet it cannot be denied that there are
applications which cannot easily be explained in this way, thus
many combinations with subjuncts like always, ever, constantly,
all day long, all the afternoon. But it is worth mentioning that
these were especially frequent in ME., before the great influx of
cases arising from the aphesis in a-hunting, etc., changed the whole
character of the construction.

It is a natural consequence of the use of the expanded tenses
to form a time-frame round something else that they often denote
a transitory as contrasted with a permanent state which for its
expression requires the corresponding unexpanded tense. The
expanded form makes us think of the time-limits, within which
something happens, while the simple form indicates no time-limit.
Compare then “he is staying at the Savoy Hotel” with “he lives
in London,” or “What are you doing for a living? I am writing
for the papers” with “What do you do for a living? I write
for the papers.” Habits must generally be expressed by the
unexpanded tenses; see, e.g., the following sentences: A great awe
seemed to have fallen upon her, and she was behaving as she behaved
in church I Now he dines at seven, but last year he dined at half-past
| Thanks, I don't smoke (cp. I am not smoking).279

But if the habitual action is viewed as a frame for something
else, the expanded tense is required: I realize my own stupidity
when I am playing chess with him | Every morning when he was
having his breakfast his wife asked him for money (while complete
coextension in time may be expressed by expanded preterit in
both sentences: “Every morning when he was having his breakfast
his dog was staring at him”).

The use of the expanded form to express the transitory in
contrast to the permanent state has in quite recent times been
extended to the simple verb be, though the distinction between
“he is being polite ” of the present moment and “he is polite”
of a permanent trait of his character is only now beginning to be
observed. But it is curious to see how in other languages the
same distinction is sometimes expressed by means which have
nothing to do with the tense system of the verb. In Danish av
sig in some cases serves to mark the quality as a characteristic
trait (han er bange av sig ‘he is naturally timid’), while han er
bange means that he is afraid at the present moment; but the
addition has a very limited sphere of application. In Spanish we
have the distinction between the two verbs meaning ‘to be,’ ser
for the generic, and estar for the individual time: mi hermano
es muy activo ‘my brother is very active’ | mi hermano esta
enfermo ‘my brother is ill’; I find a good example in Calderon,
Ale. de Zal. 3. 275 Tu hija soy, sin honra estoy ‘I am your
daughter, but am dishonoured.’ With other verbs we have the
expansion nearly as in English: el esta comiendo ‘he is dining’ 1168 |
el come a las siete ‘he dines at seven.’ In Russian the predicative
is put in the nominative if generic time is meant: on byl kupec
‘he was a merchant’ (permanently), but in the instrumental if
an individual time is meant: on byl kupcom ‘he was (for the time
being) a merchant’; this distinction, however, applies to substantives
only, adjective predicatives being always put in the
nominative. On a similar distinction in modern Irish see H.
Pedersen, GKS 2. 76. In Finnish the predicative is put in the
nominative if a generic time is meant: isäni on kipeä ‘my father
is ill’ (permanently, is an invalid), but otherwise in the essive:
isäni on kipeänä ‘my father is ill’ (at the moment). (See also
the chapter on Case, p. 183.)

Finally we have to consider the passive construction in the
obsolete the house is building, and in the still usual “while the
tea was brewing | my MS. is now copying.” In my previous paper
I have stated my reasons for disbelief in the early occurrence of
this construction, as well as in the theory that these constructions
have their origin in the notionally passive use of English verbs (his
280prose reads like poetry | it lookes ill, it eates drily, marry 'tis a
wither'd pear (Sh.)). This latter use may assist in explaining
some examples of is -ing (preparing, brewing, maturing), but not
all, and in particular not the one which is perhaps of most frequent
occurrence: the house is building, for it is impossible to say the
house builds in a passive sense. The chief source of the construction
is in my view the combination on with the verbal substantive
in -ing, which as other verbal substantives is in itself neither active
nor passive (see above, p. 172) and therefore admits the passive interpretation
(cp. the house is in construction). Combinations with the
preposition a were not at all rare in former times in the passive
signification: as this was a doynq (Malory) | there is some ill
a-brewing towards my rest (Sh.) | while my mittimus was a making
(Bunyan). This naturally explains the construction in: while
grace is saying | while meat was bringing in. There is decidedly
a difference between “my periwigg that was mending there”
(Pepys) and “he is now mending rapidly,” for in the latter, but
not in the former case, the unexpanded forms mends, mended,
may be used. Compare also “while something is dressing for
our dinner” (Pepys) and “while George was dressing for dinner”
cf. George dresses for dinner).

Just as the ambiguity of some other combinations with the
substantive in -ing in its original use as neither active nor passive
gave rise to the comparatively recent construction with being
(foxes enjoy hunting, but do not enjoy being hunted), it was quite
natural that the older construction is building should be restricted
to the active sense, and that a new is being built should come into
existence. It is well known that this clumsy, but unambiguous
construction began to appear towards the end of the eighteenth
century, and that it met with violent opposition in the nineteenth
century before it was finally acknowledged as a legitimate part
of the English language.

Terms for the Tenses.

A final word about terminology. With the extensive use of
various auxiliaries in modern languages it becomes impossible or at
any rate impracticable to have a special term for all possible combinations,
the more so as many of them have more than one
function (he would go in “He would go if he could” is different
from the shifted I will go in “He said he would go to-morrow”).
Why should the combinations would go and would have gone have
special terms rather than might go and might have gone, or dared
go, etc.? The only reason is that these forms serve to translate
simple tense-forms of certain other languages. There is really no
281necessity for such terms as the “Future Perfect in the Past” for
would have written, which, as we have seen, in its chief employment
has nothing whatever to do with future time, and which
still retains some trace of the original meaning of volition in its
first element. If we give / shall write, you will write, he will write
as a paradigm of the future tense, we meet with difficulties when
we come to consider he shall write in “he says that he shall write”
as a shifted (indirect) “I shall write.” It is really easier to make
our pupils understand all these things if we take each auxiliary
by itself and see its original and its later weakened meaning, and
then on the other hand show how futurity (future time) is expressed
by various devices in English, sometimes by a weakened
will (volition), sometimes by a weakened shall or is to (obligation),
sometimes by other means (is coming), and how very often it
is implied in the context without any formal indication. Thus
we shall say, not that I shall go and he will go are “a future
tense,” but that they contain an auxiliary in the present tense
and the infinitive. The only instance in which there is perhaps
some ground for a special tense-name is have written (had
written), because the ordinary meaning of have is here totally
lost and because the combination serves exclusively to mark
one very special time-relation. But even here it might be
questioned whether it would not be better to do without the
term “perfect.”

Time-Relations in Nouns (including Infinitives).

After thus dealing in detail with time-relations as expressed
by means of tenses in finite verbs, it remains to examine whether
similar grammatical phenomena may not be found outside this
domain. It is, of course, possible to imagine a language so constructed
that we might see from the form of the word whether
the sunset we are speaking about belongs to the past, to the
present, or to the future. In such a language the words for ‘bride,
wife, widow’ would be three tense-forms of the same root. We
may find a first feeble approximation to this in the prefix ex-,
which in recent times has come into common use in several
European languages: ex-king, ex-roi, etc. Otherwise we must
have recourse to adjuncts of various kinds: the late Lord Mayor;
a future Prime Minister; an owner, present or prospective, of property;
he dreamt of home, or of what was home once; the life
to come; she was already the expectant mother of his child, etc.
In a novel I find the combination “governors and ex-governors
and prospective governors.” 1169282

In some far-off languages tense-distinctions of substantives are
better represented. Thus, in the Alaska Eskimo we find that
ningla ‘cold, frost,’ has a preterit ninglithluk and a future ninglikak,
and from puyok ‘smoke’ is formed a preterit puyuthluk ‘what
has been smoke,’ and a future puyoqkak ‘what will become smoke,’
an ingenious name for gunpowder (Barnum, Grammatical Fundamentals
of the Innuit Language of Alaska, Boston, 1901, p. 17).
Similarly in other American languages. Thus the prefix -neen in
Athapascan (Hupa) denotes past time both in substantives and
verbs, e.g. xontaneen ‘a house in ruins,’ xoutneen ‘his deceased
wife’ (Boas, Handbook of American Indian Languages, Washington,
1911, pp. 105, 111; cf. also Uhlenbeck, Grammatische onderscheidingen
in het Algonkinsch, Amsterdam Ac. 1909).

It would seem natural to have tense-indications in those nouns
that are derived from, and closely connected with, verbs. Yet
agent-nouns generally are as indifferent to time as other substantives
: though creator most often means ‘he who has created’
this is by no means necessary, and baker, liar, beggar, reader, etc.,
tell us nothing of the time when the action takes place. 1170 In most
cases habitual action is implied, but there are exceptions (in
English more often than in Danish), e.g. the speaker, the sitter = the
person who sits for his portrait.

With active participles some languages have developed tense-distinctions,
e.g. Gr. graphōn, grapsōn, grapsas, gegraphōs, Lat.
scribens, scripturus. The Gothonic languages have only one active
participle, G. schreibend, E. writing, cf. also in Romanic languages
It. scrivendo, Fr. écrivant, which is generally called the present
participle, though it is really no more present than any other
tense, the time-notion being dependent on the tense of the main
verb; cf. “I saw a man sitting on a stone | I see a man sitting on
a stone | you will see a man sitting on a stone.” Note also the
phrase “for the time being.” The composite form having written,
ayant écrit better deserves its name of perfect participle.

With regard to the participle found, for instance, in It. scritto,
Fr. écrit, E. written, G. geschrieben, etc., some remarks on the time-relation
indicated by it have already been given above, p. 272. The
usual term, the past participle, or the perfect participle, may be
suitable in some cases, e.g. printed books, but is inadequate, for
instance, in “Judged by this standard, the system is perfect | He
can say a few words in broken English | My beloved brethren | he
is expected every moment | many books are printed every year in
England,” etc. Some grammarians, seeing this terminological
283difficulty, use the words active and passive participle for writing
and written, and this is correct, so far as the former is concerned
(apart from the old-fashioned the house building now = a-building);
but the other participle is not always passive in its character. It
is distinctively active in “a well-read man | a well-spoken lad |
mounted soldiers | he is possessed of landed propriety,” and even
if the participle was passive in the original construction underlying
the composite perfect (/ have caught a fish, originally ‘I
have a fish (as) caught’), this has long ago ceased to be true, as
we see in “I have lost it” and especially with intransitive verbs
“I have slept, come, fallen, been,” where the whole combination
is undoubtedly active. Bréal (S 224) goes so far as to say that
the participle itself has (par contagion) become active, which he
proves by the fact that one writes in telegraphic style: “Reçu
de mauvaises nouvelles. Pris la ligne directe.” As there is therefore
no really descriptive name possible for the two participles
as used in actual language, I see no other way out of the terminological
difficulty than the not very satisfactory method of numbering
the forms, calling the -ing- participle the first and the other
the second participle. 1171

Nexus-substantives do not as a rule any more than other substantives
admit of any indication of time-relations; his movement
may according to circumstances correspond in meaning to he
moves, he moved, he will move. Similarly on account of his coming
may be equivalent to ‘because he comes’ or ‘came’ or ‘will
come.’ I intend seeing the doctor refers to the future, I remember
seeing the doctor to the past. But from ab. 1600 the composite
form with having has been in use, as in “He thought himself happy
in having found a man who knew the world ” (Johnson).

The infinitive, as we have mentioned above, p. 139 f., is an old
verbal substantive, and it still has something of the old indifference
to time-distinctions: I am glad to see her refers to present time,
I was glad to see her to past, and I am anxious to see her to future
time. 2172 But in some languages, for instance Greek, tense-forms
have developed in the infinitive; cf. also Lat. scripsisse by the
side of scribere. This perfect infinitive has been given up in the
Romanic languages, in which we have now the composite perfect
284infinitive Ft. avoir écrit, etc.; the corresponding composite form
is also found in the Gothonic languages, E. (to) have written,
G. geschrieben (zu) haben.

The English perfect infinitive corresponds not only to the
perfect ('Tis better to have loved and lost Than never to have loved
at all), but also to an ordinary preterit (You meant that? I suppose
I must have meant that) and to an ante-future (future perfect:
This day week I hope to have finished my work). It was formerly
used fairly often to indicate an intention which was not carried
into effect (With that Leander stoopt to haue imbrac'd her, But
from his spreading armes away she cast her. — Marlowe); this cannot
be separated from its use corresponding to the preterit of
unreality, a use which is generally overlooked by grammarians,
but winch presents more features of interest than I can here point
out; I must content myself with giving a few of my examples
without classification and without any comment: To have fallen
into the hands of the savages, had been as bad (Defoe, = it would
have been as bad if I had fallen) | it would have been wiser to
have left us (Ruskin) | it would have been extremely interesting
to have heard Milton's opinion (Saintsbury) | a lew would haue
wept to haue seene our parting (Sh.) | she would haue made Hercules
haue turnd spit (Sh.) | she was old enough to have made it
herself (Lamb) | it seems likely to have been a desirable match for
Jane (Miss Austin, = that it would have been) | We were to have
gone and seen Coleridge to-morrow (Carlyle). The form of the
infinitive in the phrase it would have been better for him to have
stayed outside implies (in the same way as if he had stayed) that
he did not stay outside, which the simple to stay in it would have
been better for him to stay outside does not; the latter infinitive is
just as “neutral” with regard to the question of reality or
unreality as staying outside would have been better; similarly he
ought to have come here implies that he has not come, as compared
with he ought to come here.

Hence we find as synonymous expressions I should like to have
seen and I should have liked to have seen 1173 by the side of I should
have liked to see. In some composite verbal expressions the indication
of the past might in itself with equal reason be added to
either verb: to E. he could have done it and Dan. han kunde ha
gjort det corresponds Fr. il aurait pu le faire, G. er hätte es tun
können 2174 In Dan. we may also say han havde kunnet gore det,
but this is not possible in English, as can has no participle; for
285the same reason the perfect infinitive has to be used in he might
(must, should, would, ought to) have done it.

Instead of saying you needed not say that (cf. G. “das brauchten
Sie nicht zu sagen”), which denies the necessity in the past time,
it is now customary to shift the time-indication on to the infinitive:
you needn't have said that.

The opposite shifting is found in I shall hope to see you to-morrow,
which really means a present hope of a future visit; as there is
no future infinitive in English, the sign of the future is added to
hope instead. 1175

Aspect.

I must here very briefly deal with a subject which has already
been touched upon and which has been very warmly discussed
in recent decades, namely what has generally in English been
called the aspect of the verb, and in German aktionsart, though
some writers would use the two terms for two different things.
It is generally assumed that our Aryan languages had at first no
real forms in their verbs for tense-distinctions, but denoted various
aspects, perfective, imperfective, punctual, durative, inceptive, or
others, and that out of these distinctions were gradually evolved
the tense-systems which we find in the oldest Aryan languages
and which are the foundation of the systems existing to-day.
Scholars took this idea of aspect from Slavic verbs, where it is
fundamental and comparatively clear and clean-cut, but when
they began to find something similar to this in other languages,
each of them as a rule partially or wholly rejected the systems
of his predecessors and set up a terminology of his own, so that
nowadays it would be possible, had one the time and inclination,
to give a very long list of terms, many of them with two or three
or even more definitions, some of which are not at all easy to
understand. 2176 Nor have these writers always distinguished the
four possible expressions for ‘aspects,’ (1) the ordinary meaning
of the verb itself, (2) the occasional meaning of the verb as
occasioned by context or situation, (3) a derivative suffix, and
(4) a tense-form. In thus criticizing my predecessors, I may
seem to some to live in a glass-house, for I am now going to give
286my own classification, which after all may not be much better
than previous attempts. Still I venture to hope that it may be
taken as a distinctively progressive step, that I do not give the
following system as representing various “aspects” or “aktionsarten”
of the verb, but expressly say that the different phenomena
which others have brought together under this one class (or these
two classes) should not from a purely notional point of view be
classed together, but should rather be distributed into totally
different pigeonholes. This, then, is how I should divide and
describe these things.

(1) The tempo-distinction between the aorist and the imperfect;
this affects (independently of the signification of the verb
itself) the tense-form in some languages; see above, p. 276.

(2) The distinction between conclusive and non-conclusive
verbs. Here the meaning of the verb affects the meaning of the
second participle in Romanic and Gothonic languages, and thus
has influence on the time-meaning of passive combinations; see
above, p. 272.

(3) The distinction between durative or permanent and punctual
or transitory. We have seen above that this is one of the
functions of the English distinction between unexpanded and
expanded tenses, and that the same distinction is in other languages
expressed by totally different means.

(4) The distinction between finished and unfinished. This
latter is one of the functions of the expanded forms in English:
he was writing a letter, as compared with he wrote a letter; in Dan.
it is often expressed by means of the preposition på: han skrev
på et brev; cf. G. an etwas arbeiten.

(5) The distinction between what takes place only once, and
repeated or habitual action or happening. As already remarked,
this really belongs to the chapter about “number.” Habitual
action is very frequently not expressed separately (“he doesn't
drink”); in some languages we have suffixes to express it, in
which case we speak of iterative or frequentative verbs. Many
E. verbs in -er and -le belong here: totter, chatter, babble, etc.

(6) The distinction between stability and change. Sometimes
we have a pair of corresponding verbs, such as have: get, be: become
(and its synonyms: get, turn, grow). 1177 Hence the two kinds
of passive mentioned p. 274 above (be married, get married). Most
verbs derived from adjectives denote a change (becoming): ripen,
slow (down), and a change is also implied in the transitive verbs
of corresponding formation: flatten, weaken, etc. (causatives). 2178
287But a state is expressed by the verb halt = ‘be lame’ (from the
obsolete adj. halt). Many verbs denote both state and change;
in lie down the latter meaning is denoted by the adverb. There
are other ways of expressing similar changes: jail asleep, go to
sleep, get to know, begin to look, cp. the states: sleep, know, look.
Some languages have special derivative endings to express change
into a state, or beginning (inchoative, inceptive, ingressive verbs). 1179
But it is interesting to notice how this signification of beginning
has often in course of time been weakened or lost; thus in the
Romanic verbs derived from the Latin inchoatives in -isco, e.g.
Fr. je finis, je punis, whence E. finish, punish. Similarly ME. gan
lost its original force, and he gan look came to mean simply ‘he
did look, he looked.’ To is used with a predicative at first only
when a change is implied (take her to wife), but later also without
this meaning (he had her to wife); similarly in Dan. til.

The opposite kind of change, where some state ceases, is sometimes
expressed by a separate formation, as in G. verblühen, Dan.
avblomstre ‘cease blooming,’ but generally by means of such verbs
as cease, stop.

Note the three expressions for (a) change into a state: (b) being
in the state: (c) change from the state, in fall in love with (begin
to love): be in love with (love): fall out of love with (cease to love) |
fall asleep: sleep: wake (wake up). But wake in that sense may
also be considered as ‘change into a state,’ the corresponding
stability-verb being to be awake, or sometimes wake (cp. Danish
vågne: våge = Fr. s'éveiller: veiller).

(7) The distinction according to the implication or non-implication
of a result. The G. compounds with er- frequently are
resultative, e.g. ersteigen, and this is generally given as one of
the chief examples of “perfektivierung durch zusammensetzung”;
but it is difficult to see why, for instance, ergreifen should be more
perfective than the simple greifen.

I think it would be better to do without the terms perfective
and imperfective except in dealing with the Slavic verb, where
they have a definite sense and have long been in universal use.
In other languages it will be well in each separate instance to
examine carefully what is the meaning of the verbal expression
concerned, and whether it is due to the verb itself, to its prefix
or suffix, to its tense-form, or to the context. Different things
are comprised under the term perfective. If, thus, we analyze
the interesting collection of Gothic instances with the prefix ga-
which is given by Streitberg, Gotisches elementarbuch, 5th ed. 1920,
p. 196, we shall see that “perfectivation” here means, first,
288finishing: swalt lay a dying, gaswalt was dead, sagq was setting,
gasagq set (above, No. 4) — second, change: slepan be asleep,
gaslepan fall asleep, þahan be silent, gaþahan become silent, and
others (above, No. 6) — third, obtaining through the action:
fraihnan ask, gafraihnan learn by asking, rinnan run, garinnan
durch das laufen erreichen, erringen. 1180 This is akin to No. 7 above,
though it is not exactly the same thing, for he who ersteigt a mountain
does not gain the mountain. On the other hand, it has some
connexion with what was above, p. 159, termed object of result,
as in dig a hole (cp. dig the garden), but has evidently nothing to
do with time- or tense-distinctions.289

Chapter XXI
Direct and Indirect Speech

Two Kinds. Shifting of Tenses. Shifting of Mood. Questions in Indirect
Speech. Indirect Requests. Final Remarks.

Two Kinds.

When one wishes to report what someone else says or has said
(thinks or has thought) — or what one has said or thought oneself
on some previous occasion — two ways are open to one.

Either one gives, or purports to give, the exact words of the
speaker (or writer): direct speech (oratio recta).

Or else one adapts the words according to the circumstances
in which they are now quoted: indirect speech (oratio obliqua).

The direct speech (direct discourse) may be preceded by some
sentence like “He said” or “She asked,” etc., but very frequent y
the reference to the speaker is inserted after some part of the
reported speech: “I wonder, she said (or, said she), what will
become of us?” Latin has a separate word for ‘say’ which is
used only in such insertions, inquam, inquit.

The direct quotation is an outcome of the same psychological
state with its vivid imagination of the past that calls forth the
“dramatic present tense” (p. 258). Hence we often find that
tense employed in the inserted “says he, say(s) I” instead of
“said.”

There are two kinds of indirect speech (indirect discourse),
Which I .shall call dependent and represented speech. The former 1181
is generally made dependent on an immediately preceding verb,
“he said (thought, hoped, etc.)” or “he asked (wondered, wanted
to know, had no idea, etc.),” while in the second class this is as
a rule understood from the whole connexion,

What is meant by the second kind of indirect speech may
perhaps be best shown by an example. After Pendennis has
been “plucked” at the University, Thackeray writes (p. 238):
“I don't envy Pen's feelings as he thought of what he had done.
He had slept, and the tortoise had won the race. He had marred
at its outset what might have been a brilliant career. He had
290dipped ungenerously into a generous mother's purse; basely and
recklessly spilt her little cruse. Oh! it was a coward hand that
could strike and rob a creature so tender… Poor Arthur
Pendennia felt perfectly convinced that all England would remark
the absence of his name from the examination lists, and talk about
his misfortune. His wounded tutor, his many duns, the undergraduates
of his own time and the years below him, whom he had
patronised and scorned — how could he bear to look any of them
in the face now?” A few pages farther on we read of his mother:
“All that the Rector could say could not bring Helen to feel any
indignation or particular unhappiness, except that the boy should
be unhappy. What was this degree that they made such an
outcry about, and what good would it do Pen? Why did Doctor
Portman and his uncle insist upon sending the boy to a place
where there was so much temptation to be risked, and so little
good to be won? Why didn't they leave him at home with his
mother? As for his debts, of course they must be paid; — his
debts! — wasn't his father's money all his, and hadn't he a
right to spend it? In this way the widow met the virtuous
Doctor,” etc.

It is not easy to find an adequate descriptive name for indirect
discourse of this kind. Lorck rightly rejects Tobler's term (mingling
of direct and indirect discourse), Kalepky's (veiled speech, verschleierte
rede) and Bally's (style indirect libre), but his own
term “erlebte rede,” which might perhaps be rendered “experienced
speech,” does not seem much better. I have found no
better term than “represented speech.” (In German I should say
“vorgestellte rede” and in Danish “forestillet tale.”) 1182

Bally thought that this phenomenon was peculiar to French,
but Lerch and Lorck give a great many German instances, though
thinking that in German it may be due to French influence,
especially to that of Zola (!). But it is very frequent in England
(where it is found long before Zola's time, for instance in Jane
Austen) and in Denmark, probably also in other countries (I have
recently found Spanish examples), and it seems on the whole so
natural that it may easily have come into existence independently
in different places. It is chiefly used in long connected narratives
where the relation of happenings in the exterior world is interrupted — very
often without any transition like “he said” or
“he thought” — by a report of what the person mentioned was
saying or thinking at the time, as if these sayings or thought were
the immediate continuation of the outward happenings. The
writer does not experience or “live” (erleben) these thoughts or
291speeches, but represents them to us, hence the name I have
chosen.

Represented speech is more vivid on the whole than the first
class of indirect speech. As it is nearer to direct speech, it retains
some of its elements, especially those of an emotional nature,
whether the emotion is expressed in intonation or in separate
words like “Oh!”, “Alas!”, “Thank God!”, etc.

It is chiefly in the last two kinds of changes that the difference
between dependent and represented speech is seen. The shifting
of person has already been considered in Ch. XVI; here we
shall deal with the others.

Shifting of Tenses.

The ante-preterit cannot be further shifted: “I had already
seen her before she nodded” becomes “He said that he had already
seen her before she nodded.” The preterit of unreality is often
left unshifted, “He said that he would pay if he could” may thus
be a rendering of “I would pay if I could” as well as of “I will
pay if I can.” As must has now only one form, it is unchanged
in indirect discourse: “He said that he must leave at once”
= “He said: I must leave at once.” This is practically the only
292way in which must can be used as a preterit in modern colloquial
speech.

It will be been that the indirect preterit and the indirect perfect
are formally identical with the ante-preterit (before-past); and
the indirect future is formally identical with the conditional;
thus also in French j'écrirais fulfils the two functions of conditional
(j'écrirais si je savais son adresse) and of indirect simple future
(il disait qu'il écrirait le plus tôt possible = the direct: j'écrirai
le plus tôt possible).

If we now ask what is the relation between these indirect
tenses and the series of tenses established above (p. 257), the
answer is that they should not be placed in that series, where they
have nothing to do, being orientated with another zero-point
(“then”) than that of the original series (“now”). A sentence
like “(He said that) he should come as soon as he could” tells us
nothing about the moment of his coming in its relation to the
present time, but only in its relation to the time when he spoke.
He may already have come, or he may be coming just now, or at
some future time — all this is left undecided, and the only thing
we are now told is that when he spoke he mentioned his coming
as due to happen at some time which then belonged to the future.

Nor is it necessary to have special terms for the tenses arising
from this shifting. The NED (shall 14b) speaks of the “anterior
future” or “future in the past” in “he had expected that he
should be able to push forward” — this is simply a shifted (or
indirect) future, and of the “anterior future perfect,” no example
is given, but the reference must be to cases like “he said that he
should have dined by eight,” which is = the direct: “I shall have
dined by eight,” thus a shifted (or indirect) before-future time
(or, if it is to be designated as a tense: a shifted or indirect ante-future
tense).

The shifting of tenses in indirect speech is very natural and
in many cases even inevitable: He told me that he was ill, but now
he is all right — here the use of the preterit was is motived by the
actual facts of the matter, and was is at the same time the direct
past and the indirect present. But this is not always the case,
and very often the verb is put in the preterit for no other reason
than that the main verb is in that tense and that the speaker
does not stop the current of his speech to deliberate whether the
thing mentioned belongs to this or that period of time, measured
from the present moment. Van Ginneken mentions this: “Je
ne savais pas qui il était. Est-ce que je veux dire par-là qu'il est
quelque autre maintenant? Nullement. Etait se trouve là par
inertie, et par savait seul on comprend qu'il faut entendre la chose
ainsi: était et est encore” (LP 499). Or rather, we might say.
293it is left unsaid whether things are now as they were. “I told
you he was ill” — he may still be ill, or he may have recovered.
In the following instances it is the nature of the thing signified
more than the words that shows that the present time is meant,
but the shifting is perfectly natural: What did you say your
name was? | I didn't know you knew Bright | How did you know
I was here? The last example is particularly interesting on account
of the contradictio in adjecto between his presence here and the
form was: I am here now, but how did you know that?

It requires some mental effort to leave the preterit and use
the more logical present tense, even where one has to enounce
some universal truth. We cannot, therefore, expect that speakers
will always be consistent in their practice with regard to the
consecutio temporum. We may hesitate in a case like this: “He
told us that an unmarried man was (or, is) only half a man,” but
we should probably prefer the unshifted in: “It was he who
taught me that twice two is four.”

The use of the unshifted present here implies that the actual
speaker is himself convinced of the truth of the assertion, whereas
the shifting of the tense also shifts the responsibility for the saying
on to the original speaker; hence the difference in “He told us
that it was sometimes lawful to kill” (but he may have been
wrong) and “I did not know then that it is sometimes lawful to
kill” (but it is). Note the preterit in Falstaff's “Did I say you
were an honest man?” with the continuation: “Setting my
knighthood and my souldiership aside, I had lyed in my throat,
if I had said so.” Sometimes the tone of the sentence is decisive:
“I thought he was married” with one intonation means ‘I now
find that I was mistaken in thinking him married,’ and with
another ‘Of course he is married, and didn't I tell you so?’

The present subjunctive is not shifted to a preterit in reports
of proposals made at meetings, etc.: He moved that the bill be
read a second time. Here the form be is felt as indicating futurity
and therefore as more adequate than were, which would rather
imply something unreal or hypothetical; in other verbs there
would be no difference in the preterit between the indicative and
the subjunctive, and so the form of the proposal is kept unchanged
in spite of the conjunction that. 1183294

In most cases of shifted tenses the main verb refers to some
time in the past; but we may have similar shiftings alter a main
verb in the future, though this will be rarer. When we imagine
a person, who is now absent, saying at some future date “I regret
I was not with them then,” we naturally say “He will regret
that he is not with us now.” But Henry V in Shakespeare (IV.
3. 64) uses the preterit that belongs to the direct speech of the
gentlemen concerned (though he says here, which implies his own
standpoint): And gentlemen in England, now a bed, Shall thinke
themselues accurst they were not here, And hold their manhoods
cheape, whiles any speakes, That fought with vs vpon Saint Crispines
day. This reminds one of the Latin “epistolary tenses,”
in which the writer of a letter transports himself to the time when
it will be read, and therefore uses the imperfect or perfect where
to us the present tense is the only natural one.

Shifting of Mood.

The shifting of mood from the indicative to some other mood
in indirect speech is not found in modern English and Danish,
but in other related languages. Latin makes an extensive use of
the accusative with infinitive in what in direct discourse would be
a principal clause, as well as in the more independent of subordinate
clauses, and of the subjunctive in other dependent clauses. Other
languages have other rules, and the use of the subjunctive, or
optative, mood in indirect speech shows such marked divergencies
in the various ancient languages of our family that it seems to have
developed independently at different places for different reasons.
T. Frank (in Journal of Engl. and Germ. Philol. 7. 64 ff.), while
rejecting earlier “metaphysical” explanations from the nature
of “subjectivity” and “potentiality,” gives good reasons for
supposing that the use of the subjunctive in the Gothonic languages
is a gradual extension by analogy from its use in clauses dependent
on such verbs as Goth, wenjan, OE. wenan, G. wähnen, which
at first meant “hope, desire” and therefore naturally required
the optative. It was retained when the verbs came to mean
‘imagine, think,’ and then transferred to other verbs meaning
‘think, say,’ etc.

Questions in Indirect Speech.

Here we meet with the chief difference between the two
kinds, dependent and represented speech. We shall speak first
of dependent questions.

When a question is reported the interrogatory intonation,
which is very often the chief indication that a question is meant,
is necessarily lost or weakened, but there is some compensation,
partly in the introductory (or inserted) formula, in which the verb
ask is used instead of say, partly in the use of an interrogative
conjunction where there is no interrogative pronoun. The conjunction
often originates in a pronoun meaning ‘which of two’:
E. whether, Icel. hvárt, Lat. utrum, but in other cases the origin
is different, and we frequently find the use of a conditional conjunction:
E. if, Fr. si, Dan. om, cf. G. ob. Very frequently the
difference between a direct and an indirect question is marked by
a different word-order: 1184 Who is she? — He asked who she was |
How can I bear to look any of them in the face? — … how he
could bear to look… | Hasn't he a right to spend his money? —
… whether he had not… In the same way in other languages,
e.g. Danish: Hvem er hun? — Han spurgte, hvem hun var | Hvor
kan jeg holde det ud? — … hvor jeg kunde holde det ud | Har
han ikke ret? — … om han ikke havde ret. French: Qui
est-elle? (Qui est-ce?) — Il a demandé qui elle était (qui c'était) |
Comment peut-on le souffrir? — … comment on pouvait le
souffrir | N'a-t-il pas raison? — … s'il n'avait pas raison. In
Danish there is the further difference that an interrogative pronoun
as the subject of the sentence requires the addition of der
in an indirect question: Hvem har ret? — Han spurgte (om) hvem
der havde ret? | Hvad er grunden? — … hvad der var grunden
(but if grunden is here treated as the subject, which is also possible
297the result is the inverse word-order: Han spurgte om hvad grunden
var).

Instead of the form peculiar to dependent indirect questions
it has become more and more frequent in English to use the form
also found in represented discourse, with no introductory if or
whether, and with inverted word-order. Thus: I know not yet,
was it a dream or no (Shelley) | he said was I coming back, and
I said yes; and he said did I know you, and I said yes; and he
said if that was the case, would I say to you what I have said,
and as soon as I ever saw you, would I ask you to step round
the corner (Dickens), In recent writers this is very frequent
indeed; it is mixed up with the dependent form in: they asked
where she was going, and would she come along with them?
(Carlyle). In German the same form is found, though rarely,
e.g. “man weiss nicht recht, ist er junggeselle, witwer oder gar
geschieden” (G. Hermann).

Besides being used in quotations of direct questions, indirect
questions are very often used (as “clause primaries”) after verbs
like know, doubt, see, etc., as in: I want to know if he has been
there | Go and see who it is, and try to find out where he comes
from | it is not easy to say why the book is so fascinating. — They
may also be subjects, as in “Whether this is true or not is still
an open question.” Sometimes the main sentence may be omitted,
and the (formally) indirect question thus becomes a (notionally)
direct question: If I may leave it at that? (I ask if… = May
I leave it at that?).

In represented discourse the only shiftings in questions are
those shiftings of person and tense that are common to all indirect
discourse; otherwise questions remain what they would be in
direct quotation. Thus the questions “How can I bear to look
any of them in the face now?” and “Hasn't he a right to spend
it?” in the passage from Pendehnis simply became “How could
he bear…” and “Hadn't he a right…” “What does she
see?” became “What did she see?” 1185 In French the imparfait
replaces the présent, in German the preterit indicative (not the
subjunctive) is used, etc.

Exclamations introduced by an interrogative word remain
unchanged except for the shifting of tense and person: “What
a nuisance it is to change!” becomes “What a nuisance it was
to change” both when it is dependent on such a verb as “He
said” and when it forms part of a represented speech.298

Indirect Requests.

Such requests (commands, etc.) as in direct speech are expressed
in the imperative have to be changed. In dependent speech
either the element of request is expressed in the main verb, e.g.
when “Come at once” is made into “He ordered (commanded,
told, asked, implored) me (her) to come at once” or the main
verb does not express the element of request, which must therefore
find expression otherwise in the dependent clause: “He said
(wrote) that I (she) was to come at once.” The latter is the form
generally employed in represented speech, though occasionally
the imperative may be retained, as in the following passage from
Dickens: “Mr. Spenlow argued the matter with me. He said,
Look at the world, there was good and evil in that; look at the
ecclesiastical law, there was good and evil in that. It was all
part of a system. Very good. There you were.” Imperatives
with let us are differently rendered in the two kinds of indirect
discourse: “He proposed that we (they) were to go” and “Let
us (them) go.”

Final remarks.

The distinction between direct and indirect speech is not
always strictly maintained. A direct quotation may be introduced
by the conjunction (‘that’) usually reserved for indirect quotation;
thus not unfrequently in Greek. The Greek “kai legōn
autōĭ, hoti ean thelēĭs, dunasai me katharisai” was imitated by
Wulfila: “jah qiþands du imma þatei jabai wileis, magt mik
gahrainjan” (Mark 1. 40, thus also ib. 1. 37). I take a modern
instance from Tennyson: “she thought that peradventure he
will fight for me.” 1186 In French we have “je crois que non,”
although non belongs to direct speech.

Human forgetfulness or incapacity to keep up for a long time
the changed attitude of mind implied in indirect discourse causes
the frequent phenomenon that a reported speech begins indirectly
and is then suddenly continued in the direct form. Examples
from Greek writers like Xenophon are given in handbooks of
Greek syntax. In Icelandic sagas they abound, e.g. Vols. 1:
segir at Breði hafi riðit frá honum á skóginn, ok var hann senn
ór augliti mér, ok veit ek ekki til hans ‘he says that B. rode from
him into the wood, and I soon lost sight of him, and I know nothing
about him’ | ib. 6 mælti at hann skyldi gera til brauð þeira, en
ek man sœkja eldivið ‘he said that he [the other] was to prepare
299their bread, but I will fetch fuel’ | ib. 9 hann spyrr, hverir þar
væri, eða hví eru-þér svá reiðuligir? ‘he asks who were there,
and why are you so angry.’ A different kind of mixture of the
two discourses is seen in Goldsmith Vic. 2. 166: But tell me how
hast thou been relieved, or who the ruffians were who carried
thee away?

German and Danish have a curious way of expressing what
is notionally an indirect discourse by means of the verb soll, skal:
Er soil sehr reich sein (gewesen sein) | han skal være (ha været)
meget rig ‘he is said (reputed, rumoured) to be (have been) very
rich.’ As soll, skal is in most of its uses a kind of weaker muss,
må, I think this usage may be classed as a kind of weaker counterpart
of the muss, ma, must of logical necessity or of compelling
conclusion, as in “he must be very rich (since he can give so much
to the poor).”300

Chapter XXII
Classification of Utterances

Que donnee nous fut parole
Por faire nos voloirs entendre,
Por enseignier et por aprendre.
Roman de la Rose.

How many Classes? Questions. Sentence.

How many Classes?

Brugmann (Verschiedenheiten der satzgestaltung nach massgabe
der seelischen grundfunktionen, Sächs, ges. d. wiss. 1918) has an
elaborate classification of sentences or utterances with the following
main divisions, most of them with up to 11 subclasses: (1)
exclamation, (2) desire, (3) invitation (aufforderung), (4) concession,
(5) threat, (6) warding off (abwehr und abweisung), (7) statement
about imagined reality, (8) question. 1187 In the treatment of these
classes historical considerations often cross purely logical divisions,
and it is difficult to see the rationale of the whole classification
as well as to see where such simple statements as “he is rich”
have to be placed. This criticism does not hinder one from
acknowledging the high value of many things in this book, one
of the last things the revered master of comparative philology
ever wrote. The older classification is much clearer: (1) statements,
(2) questions, (3) desires, (4) exclamations (see, e.g., Sonnenschein's
Grammar). But even this division is open to criticism;
the boundary between (3) and (4) is not clear: why are “God
save the King” and “Long may he reign” excluded from Exclamations,
and why are these latter confined to those that are
“introduced by exclamatory pronouns, adjectives or adverbs”
such as what and how?

A further objection to the classification given by Sonnenschein
is that it is expressly meant as a classification of “sentences”
only, i.e. such utterances as contain a finite verb. But obviously
utterances like “What fun!”, “How odd!”, “Glorious!”
or “Hurrah!” are “exclamations” just as much as those
301mentioned; “Waiter, another bottle!” cannot be separated from
“desires” containing an imperative; and among statements we
must reckon also the “nominal” sentences considered above
(p. 121). It might perhaps also be said that the term “desire”
is not the best term to include “commands, requests, entreaties,
and wishes,” and at the same time exclude “I want a cigar”
and “Will you give me a light, please?” etc. Notionally these
are really desires to be classed with the imperative “Give me,”
though formally they are “statements” and “questions.” The
classification is thus seen to be faulty because it is neither frankly
notional nor frankly syntactic, but alternates between the two points
of view: both are important, but they should be kept strictly
apart in this as in other domains of grammatical theory.

If, then, we attempt a purely notional classification of utterances,
without regard to their grammatical form, it seems natural
to divide them into two main classes, according as the speaker
does not or does want to exert an influence on the will of the hearer
directly through his utterance. In the former class we must
include not only ordinary statements and exclamations, but also
such wishes as “God save the King,” etc. With regard to this
class it is, of course, immaterial whether there is a hearer or not;
such an utterance as “What a nuisance!” is the same whether
it is spoken in soliloquy or to someone else.

In the second class the aim of the utterance is to influence the
will of the hearer; that is, to make him do something. Here we
have two subclasses, requests and questions. Requests comprise
many utterances of different forms, imperatives, verbless expressions
(“Another bottle!” | “Two third Brighton” | “A horse,
a horse!” | “One minute” | “Hats off”), formal questions
(“Will you pack at once!”) and formal “statements” (“You
will pack at once”) if the situation and the tone shows them to
be equivalent to commands, etc. Requests may range from
brutal commands through many intermediate steps (demands,
injunctions, implorations, invitations) to the most modest and
humble prayer (entreaty, supplication).

Questions.

A question also is a kind of request, viz. a request to tell the
original speaker something, to give him a piece of information
that he wants. Questions again may range from virtual commands
to polite prayers: the answer may be as it were exacted
or humbly solicited. The kinship between ordinary requests and
questions is seen in the frequency with which a question is tagged
on to an imperative: “Hand me that box, will you?” The
302question “Well?” means the same thing as the imperative “Go
on!” or “Speak!”

There are two kinds of questions; “Did he say that?” is
an example of the one kind, and “What did he say?” and “Who
said that?” are examples of the other. Many names have been
proposed for these two kinds: yes-or-no question or categorical
question v. pronominal question, sentence question v. word question,
totality question v. detail question or partial question, entscheidungsfrage
v. erganzungsfrage or tatsachenfrage, bestatigungsfrage
v. bestimmungsfrage. Noreen (VS 5.118 ff.) discusses and criticizes
these proposed terms and ends by proposing (in Swedish) “rogation”
v. “kvestion.” This distinction would be impossible in English
(and French), where the word “question” has to be used as the
common term; it has the further grave drawback that it is impossible
to remember which is which. An unambiguous terminology
may be easily found if we remember that in the former kind it
is always a nexus the truth of which is called in question: the
speaker wants to have his doubt resolved whether it is correct to
connect this particular subject with this particular predicate. We
may therefore call questions of this kind nexus-questions. In the
other kind of questions we have ah unknown “quantity” exactly
as in an algebraic equation; we may therefore use the well-known
symbol x for the unknown and the term x-question for a question
aiming at finding out what x stands for.

Sometimes there may be two unknown quantities in the same
equation, as in the colloquial: “Who shall sit where?” (But
“I don't know which is which” and “Who's who?” are different:
they really mean: ‘which (who) is one, and which (who) is the
other?’)

The answer to a nexus-question is either yes or no; to an
x-question it may according to circumstances be anything except
yes or no. With regard to tone it is the general rule that nexus-questions
have a rising and x-questions a falling tone towards
the end of the sentence. But there are certain questions which
in these two respects are like x-questions, and yet resemble nexus-questions
in their form. If we extend the question “Is it white?”
by adding “or black?” and alter “Do you drink sherry?” into
“Do you drink sherry or port?” we get disjunctive or alternative
questions, in which the rising tone is concentrated on the first
part as in the simple question, and the added “or white,” “or
port” has a falling tone. These questions are the equivalents
of pronominal questions (x-questions) of this type: “What colour
is it?” “Which do you drink, sherry or port?” But it is interesting
to notice that what are seemingly the same questions may
have a different meaning with a different intonation, if sherry or
303port is taken as one comprehensive term for strong wines, the
answer to this question (Do you drink [such strong wines as] sherry
or port?) is then naturally yes or no (cf. LPh 15. 54). Questions
with neither — nor (Have you neither seen nor heard it?) are nexus-questions
because neither — nor is a negative both — and, not a
negative either — or.

Mention may here be made of the phenomenon which I have
termed “questions raised to the second power” (LPh 15. 52).
One person asks “Is that true?” but instead of answering this,
the other returns “Is that true?” — meaning “How can you
ask?” Here most languages use the same form as in indirect
questions: “Om det er sandt? | Ob das wahr ist? | Si c'est
vrai?” 1188 though the sentences differ from ordinary indirect
questions by having a much more marked rising of the interrogatory
tone. I find the same form in Caxton (Reynard 21, imitation
from French?) “Loue ye wel myes [mice]? Yf I loue hem wel,
Baid the catt, I loue myes better than ony thing.” But otherwise
the English form of the question (inversion without conjunction)
is here the same as in direct questions; I have collected a great
many examples from the time of the earliest comedies to that
of the latest novels. As the retorted question generally implies
that it was superfluous to ask, it amounts to the same thing as
an affirmation: “Do I remember it?” = Certainly I remember
it, and the curious consequence is that it often does not matter
whether there is a negative or not in the question, as “Don't I
remember it?” is also equivalent to an affirmation.

Questions introduced by an interrogative word (x-questions)
may be similarly retorted, and here, too, most languages use the
form of indirect questions: Was hast du getan? — Was ich getan
habe? | Hvad har du gjort? — Hvad jeg har gjort? In French
we see a relative clause taking the place of the interrogative
clause: Ce que j'ai fait? Chaucer used an inserted that as in other
clauses: But wherefore that I speke al this? (Parl. 17). But
from the time of Shakespeare it has been usual in English simply
to repeat the question unchanged (except for the tone): “Where is
it? — Where is it? taken from vs, it is” (Shakespeare). — The change
in the character of the question by being “raised to the second
power” is shown also in the kind of answer required: “What have
you done?” — “What have I done?” — “Yes, that is what I wanted
to know.” Questions of this kind are thus always nexus-questions. 2189304

The formal means by which questions are expressed, are
(1) tone; (2) separate interrogative words, whether pronouns or
particles, e.g. Lat. num, enclitic -ne (originally the negative word),
Dan. mon (originally an auxiliary verb), Fr. ti (Lang. 358) — in
spoken French we may count [ɛskə] as an interrogative particle;
(3) word-order.

But it should be noted that what from a formal point of view
is a question very often is used for something which notionally
is not a question, i.e. a request to solve some doubt in the mind
of the speaker. Besides the so-called rhetorical questions, which
retain part of the notional value of questions, we must here mention
expressions of surprise, e.g. “What! are you here?” which
certainly is not said in order to be informed whether the other
person is here. Further “Isn't he stupid!” | G. “Ist das
unglaublich!” In exclamations of this kind the tone is modified,
and in so far they cannot be said to have the complete
form of questions. This is even more true of conditional
clauses having the same word-order as questions and developed
out of original questions, e.g. “Had he been here, I should
have given him a piece of my mind.”

Sentence.

The definitions of ‘sentence’ are too numerous and too diver
gent for it to be worth while here to reprint or criticize them all. 1190
In so far as they are not merely bogus definitions, in which technical
words are used to conceal the want of clear thought, these definitions
have taken as their starting point either formal or logical
or psychological considerations, while some of them have tried
to reconcile two or three of these points of view. But though
there is thus no consensus of theory, grammarians will generally
be more apt to agree in practice, and when some concrete group
of words is presented to them will be in little doubt whether or
not it should be recognized as a real sentence.

According to traditional logic every sentence forms a trinity
of Subject, Copula and Predicate. Logicians analyze all sentences
(propositions) with which they have to deal into these three components
and thus obtain one fixed scheme that facilitates their
operations. But even with regard to their purely intellectual
propositions the scheme is artificial and fictitious, and it does
not at all fit the great majority of those everyday sentences of a
305more or less emotional colouring which form the chief subject-matter
of the researches of the grammarian.

Instead of the old ‘threeness’ it is now more customary to
postulate a ‘twoness’: every sentence is said to be composed
of two parts, Subject and Predicate. In “the sun shines” the
sun is subject and shines predicate. Each of these two parts may
be composite: in “The youngest brother of the boy whom we
have just seen once told me a funny story about his sister in
Ireland” all the words up to seen constitute the subject, and the
rest the predicate. Opinions vary as to how this ‘twoness’ is
brought about psychologically, whether by the bringing together
of two ideas existing already separately in the mind of the speaker,
or by the breaking up of one idea (gesamtvorstellung) into two
special ideas for the purpose of communication. This question need
not, however, occupy us here. On the other hand, it is important
to keep in mind that tho two parts of the sentence, subject and
predicate, are the same as the two parts of a nexus, primary and
adnex, but that, as we have seen, it is not every nexus that constitutes
a sentence: only an independent nexus forms a sentence.

It is, however, being more and more recognized by linguists
that besides such two-member sentences as just mentioned we
have one-member sentences. These may consist of one single
word, e.g. “Come!” or “Splendid!” or “What?” — or of two
words, or more than two words, which then must not stand to one
another in the relation of subject and predicate, e.g. “Come
along!” | “A capital idea!” | “Poor little Ann!” | “What
fun!” Here we must first guard against a misconception found
in no less a grammarian than Sweet, who says (NEG § 452) that
“from a grammatical point of view these condensed sentences
are hardly sentences at all, but rather something intermediate
between word and sentence.” This presupposes that word and
sentence are steps in one ascending hierarchy instead of belonging
to two different spheres; a one-word sentence is at once a word
and a sentence, just as a one-room house is from one point of view
a room and from another a house, but not something between
the two.

An old-fashioned grammarian will feel a certain repugnance to
this theory of one-member sentences, and will be inclined to explain
them by his panacea, ellipsis. In “Come!” he will say that the
subject “you” is understood, and in “Splendid!” and “A
capital idea!” not only the subject (“this”), but also the verb
“is” is understood. In many exclamations we may thus look
upon what is said as the adnex, the subject (primary) being either
the whole situation or something implied by the situation (cp.
Ch. X). Most grammarians would probably analyze such Latin
306one-word sentences as “Canto” or “Pluit” as containing implicitly
a subject, however difficult it may be to say exactly what is the
subject of the latter verb. But grammarians should always be
wary in admitting ellipses except where they are absolutely necessary
and where there can be no doubt as to what is understood —
as, for instance, in “he is rich, but his brother is not [rich],” “it
generally costs six shillings, but I paid only five [shillings].” But
what is understood in “Watercresses!” or “Special edition!”?
“I offer you…” or “Will you buy…?” or “This is…”?

If the word “John!” forms a whole utterance, it may according
to circumstances and the tone hi which it is said be interpreted
in various ways: “How I love you, John,” “How could you do
that?”, “I am glad to see you,” “Was it John? I thought it
was Tom,” etc. How can these various “John!”'s be reduced to
the scheme subject-predicate, and how can ellipses assist us
in analyzing them? Yet it would not do to deny their being
sentences. Nor can we stop here. “Yes” and “No,” and interjections
like “Alas!” or “Oh!” or the tongue-clicks inadequately
spelt “Tut” and “Tck” are to all intents and purposes sentences
just as much as the most delicately balanced sentences ever uttered
by Demosthenes or penned by Samuel Johnson.

If we admit this — and I confess that I do not see at what point
of the chain between the Johnsonian construction and the click
we should draw the line, then the definition of a sentence is comparatively
an easy matter.

A sentence is a (relatively) complete and independent human
utterance — the completeness and independence being shown by
its standing alone or its capability of standing alone, i.e. of being
uttered by itself. 1191

In this definition the word “utterance” has been expressly
chosen as the most comprehensive term I could find. Generally
by an utterance is meant a piece of communication to someone
else, but this is not necessary (soliloquy!); however, in order to
be recognized as a sentence an utterance must be such as might
be a piece of communication were there someone to listen to it. 2192

Let us see what is implied hi the word “independent” in our
definition. “She is ill” is a sentence, but if the same words enter
into the combinations “He thinks (that) she is ill” and “He is
307sad when (if, because) she is ill,” they are no longer independent
utterances, but parts of sentences, either, as in the first example,
the object of thinks, or, as in the others, subjuncts (strictly speaking,
parts of subjuncts, as the conjunctions are also required).
These parts of sentences, which in English are generally termed
(dependent) clauses, are in German called “nebensätze” and in
Danish “bisætninger,” as if they were in themselves sentences
of a particular kind, which according to our definition they are
not. In the same way, while “What to do?” is a complete
sentence when standing alone, it ceases to be one and becomes
a mere clause in “He did not know what to do.” 1193

It is also a simple corollary of the definition that when “If
only something would happen!” stands alone and means “I
wish something would happen,” and when “If this isn't the limit!”
means “This is the limit,” these are (complete) sentences, no
matter how easy it is to see that they have developed from clauses
requiring some continuation to be complete.

It will be noticed that sentence as here defined is a purely
notional category: no particular grammatical form is required
for a word or a group of words to be called a sentence. I do not
even imitate those scholars who introduce the term “normal
sentence” (normalsatz) for sentences containing a subject and a
finite verb. Such sentences may be normal in quiet, easy-flowing
unemotional prose, but as soon as speech is affected by vivid
emotion an extensive use is made of sentences which fall outside
this normal scheme and yet have every right to be considered
natural and regular sentences.

Articulate sentences contain both components of a nexus, and
as the “nominal sentences” considered above, p. 122, are in the
minority, this means that the great majority of articulate sentences
contain a finite verb.308

In the practice of any speech-community there will always
be strong forces making for order and regularity, for uniformity,
for fixed patterns. Through wholesale imitation of the word-combinations
in most frequent use certain types will tend to
become practically universal. Hence some words which at first
may have been rare and have been thought more or less superfluous
become more and more frequent and at last may come to
be thought necessary because they make the whole sentence conform
to the most usual patterns. As most sentences have a
subject (Petrus venit), subjects come to be introduced where at
first there were none: je viens, il vient, il pleut as against venio,
venit, pluit, and in the same way E. I come, he comes, it rains. As
most sentences have something placed before the verb, the empty
there came to be used in there are many, etc. As most sentences
contain a verb, a verb was inserted in places where it was not at
first necessary to have one, hence the use of the ‘copula’ is and
of does in “So John does!” As some verbs generally take a
predicative, an empty so (G. es, Dan. del) is used, e.g. in “In
France the population is stationary, and in England it is rapidly
becoming so,” cp. also “To make men happy, and to keep them so”
(Pope). As most adjuncts are followed by a primary, one is used
to prop up the adjunct in “a grey horse instead of the white
one” | “birds love their young ones,” etc. In all these cases
we have practically the same tendency to round off sentences
so as to make them conform to a prevalent type.

Although this uniformizing tendency has not been carried
through with perfect consistency, it has nevertheless been made
the basis of the grammarian's assumption that every sentence,
or every normal sentence, must contain a subject and a finite
verb; but as soon as we see that it is merely a tendency, and not
a law of language, it becomes urgent to give a definition of
‘sentence’ which does not require the presence of those two
constituents.

In all speech activity there are three things to be distinguished,
expression, suppression, and impression. Expression is what the
speaker gives, suppression is what he does not give, though he
might have given it, and impression is what the hearer receives.
It is important to notice that an impression is often produced
not only by what is said expressly, but also by what is suppressed.
Suggestion is impression through suppression. Only bores want
to express everything, but even bores find it impossible to express
everything. Not only is the writer's art rightly said to consist
largely in knowing what to leave in the inkstand, but in the most
everyday remarks we suppress a great many things which it would
be pedantic to say expressly “Two third Brighton return”
309stands for something like: “Would you please sell me two third-class
tickets from London to Brighton and back again, and I will
pay you the usual fare for such tickets.” Compound nouns state
two terms, but say nothing of the way in which the relation between
them is to be understood: home life, life at home, home letters,
letters from home, home journey, journey (to) home; compare
further life boat, life insurance, life member; sunrise, sunworship,
sunflower, sunburnt, Sunday, sun-bright, etc.

As in the structure of compounds, so also in the structure of
sentences much is left to the sympathetic imagination of the hearer,
and what from the point of view of the trained thinker, or the
pedantic schoolmaster, is only part of an utterance, is frequently
the only thing said, and the only thing required to make the
meaning clear to the hearer. This is especially true of certain
types of sentences in which suppressions of the same kind have
occurred so often that at last no one thinks of what is left out, the
remainder becoming a regular idiomatic expression which the grammarian
must recognize as a complete sentence. There are two types
of suppression which require particular attention (cf. Lang. 273).

(1) The beginning of a sentence falls out by what we might
learnedly term prosiopesis: the speaker begins to articulate, or
thinks he begins to articulate, but produces no audible sound
(either for want of expiration, or because he does not put his vocal
chords in the right position) till one or two syllables after the
beginning of what he intended to say. Examples are such forms
of salutation as Morning instead of Good morning, G. (Guten)
tag, etc. Further: colloquial See? for Do you see? | (Do you)
remember that chap? | (Will) that do? | (I'm a)fraid not | (When
you) come to think of it | (I shall) see you again this afternoon |
(God) bless you! Similar examples occur in all languages.

(2) The end is left out: aposiopesis is the learned name for
what I have elsewhere (Language 251) more colloquially called
stop-short or pull-up sentences. After saying “If only something
would happen” the speaker stops without making clear to himself
how he would go on, were he to complete the sentence, whether
“I should be happy,” or “it would be better,” or “things would
be tolerable,” or whatever he might think of. But even without
any continuation the if-clause is taken at more than its face-value
and becomes, to speaker and hearer alike, a complete expression
of a wish. Other expressions of wishes are G. “Wer doch eine
zigarre hätte!” | Dan. “Hvem der havde en sigar!” | Span.
“Quién le diera!” Further examples of pull-up sentences:
Well, I never! | The things he would say! | The callousness of
it! | To think that he has become a minister! | Dire qu'il est
devenu ministre? | Tænke sig at han er blevet minister! | Figurarsi
310ch'egli è divenuto ministro! In all such cases the fact that something
is left out should not prevent us from recognizing the
utterance as sufficiently complete to be called a sentence.

In other cases, however, the suppression is so violent that
this condition is not fulfilled. I should not recognize as sentences
signboards (“J. C. Mason, Bookseller”), book-titles (“Men and
Women”), head-lines in newspapers (“New Conferences in Paris”
or “Killed his father-in-law”), indication of speaker in plays
(“Hamlet”), entries in diaries (“Tuesday. Rain and fog. Chess
with uncle Tom, walk with the girls”) and similar short expressions.
It is, however, important to observe that all these phenomena
occur in writing only and thus fall outside language proper:
spoken language may indulge in many suppressions, but the result
is always distinguished from that exemplified in this paragraph.

With regard to suppression a few final remarks may not be
out of place here. 1195 It has been said (C. Alphonso Smith, Studies
in Engl. Syntax, 1906, p. 3) that “verbs denote activity and
change: they are hustling and fussy,” and that therefore the
omission of verbs gives the impression of calm. This is exemplified
by Tennyson's In Memoriam, XI (Calm and deep peace on this
high wold, etc.). But as a matter of fact the impression there
is produced in the first place by the constant repetition of the
word calm and its synonyms, and secondly by the fact that the
verb omitted is one of rest, “is.” If verbs of motion are omitted,
their suppression may inversely strengthen the impression of
unrest, as in the following example: “Then rapidly to the door,
down the steps, out into the street, and without looking to right
or left into the automobile, and in three minutes to Wall Street
with utter disregard of police regulations and speed limits,” or in
Longfellow's description of Paul Revere's ride: “A hurry of
hoofs in a village street, A shape in the moonlight, a bulk in the
dark, And beneath, from the pebbles, in passing, a spark Struck
out by a steed flying fearless and fleet.” As in these cases a
feeling of terseness and of vigour is also produced by the omission
of verbs in a great many proverbial locutions, apophthegms,
party devices, and similar sayings. G. “Ende gut, alles gut”
is more pithy than E. “All is well that ends well,” Fr. “Tout
est bien qui finit bien,” Dan. “Nar enden er god, er alting godt.”
Cp. also: “Like master, like man” | “Every man to his taste”
“No cure, no pay” | “Once a clergyman, always a clergyman”
“Least said, soonest mended,” “One man, one vote,” etc. By
311leaving out what may seem superfluous one creates the impression
of hurry or stress of business which does not allow time enough
to round off one's sentences in the usual way: it is also of importance
that proverbs, etc., should be easy to remember and
therefore not too long. In these cases, however, it is not the
fact that a verb is omitted which produces the effect, for we have
other abbreviated proverbs, etc., in which a similar effect is produced
though they contain verbs: “Live and learn” | “Rule
a wife and have a wife” | “Spare the rod and spoil the child” |
“Love me, love my dog.” 1196 In both classes of sayings the usual
sentence-construction with subject and finite verb is abandoned
in favour of something which may be compared to a Japanese
drawing, in which the contours are not completely filled in; the
very boldness of such a drawing assists in bringing about an
artistic effect by leaving more to the imagination of the beholder.
And our grammatical phenomenon thus turns out to be one little
part of the ever-standing war between classicism and impressionism.312

Chapter XXIII
Moods

Classification. Imperative. Indicative and Subjunctive. Notional Moods.

Classification.

Many grammars enumerate the following moods in English, etc.:
indicative, subjunctive, imperative, infinitive, and participle.
It is, however, evident, that infinitives and participles cannot
be co-ordinated with the others; enough has also been said of
them in various other parts of this work, and we shall therefore in
this chapter deal with the first three moods only. These are sometimes
called fact-mood, thought-mood, and will-mood respectively.
But they do not “express different relations between subject and
predicate,” as Sweet says (NEG § 293). It is much more correct
to say 1197 that they express certain attitudes of the mind of the
speaker towards the contents of the sentence, though in some
cases the choice of a mood is determined not by the attitude of
the actual speaker, but by the character of the clause itself and its
relation to the main nexus on which it is dependent. 2198 Further it
is very important to remember that we speak of “mood” only
if this attitude of mind is shown in the form of the verb: mood
thus is a syntactic, not a notional category.

Imperative.

This is true even of the Imperative, though that mood comes
nearer than either the indicative or the subjunctive to being
notional. It is a will-mood in so far as its chief use is to express the
will of the speaker, though only — and this is very important — in so
far as it is meant to influence the behaviour of the hearer, for otherwise
the speaker expresses his will in other ways. Imperatives
thus are requests, and, as we have seen, these range from the
strictest command to the humblest prayer. But we saw also that
313requests are very often expressed by other means than the imperative
(“Another bottle!” | “Wollen wir gehen” | “You will pack
at once and leave this house,” 1199 etc.), and we may here remind the
reader of the use in requests of infinitives (“Einsteigen!” | “Nicht
hinauslehnen!” | “Non piangere!”) and of participles (“Vorgesehen!”
| “Still gestanden?” | “Wohl auf, kameraden, auf's pferd,
auf's pferd, In's feld, in die freiheit gezogen!”) — in other words,
imperative and request are not convertible or coextensive terms.

Nor can it be said that imperatives are exclusively used to
express requests. An imperative very often means permission,
which is not a request, because it does not say that the speaker
wants the hearer to behave in a certain way. But a permissive
“Take that (if you like)!” may also be expressed in other ways:
“I allow you to take that” | “You may take that” | “I have
no objection to your taking that” | “I don't mind if you take
that.” — On prohibition = negative command or permission see
Ch. XXIV.

A further use of the imperative is seen in Hamlet's “Vse euerie
man after his desart, and who should scape whipping” — the first
part is no more a real request to use every man after his desert
than the second is a real question; together the two sentences
mean: if we used…, no one would escape punishment. Other
examples: Spoil foc's'le hands, make devils (Stevenson) | Give
you women but rope enough, you'll do your own business (Richardson;
the use of you as an indirect object shows that no request
to the person addressed is meant).

As the imperative has no particular ending in English, one
might perhaps feel inclined to think that these sentences contained
infinitives (though how used?). Parallel uses in other languages
show us, however, clearly that they contain imperatives, e.g. G.
Sage das, und du wirst (so wirst du) verhöhnt | Dan. Tag hatten
op eller lad den ligge, i begge tilfælde far du prygl | Fr. Obligez
cent fois, refusez une, on ne se souviendra que du refus | Lat.
Scaevae vivacem crede nepoti Matrem: nil faciet sceleris pia
dextera (Hor.) | Gr. Dos moi pou stō, kai tēn gēn kinesō.

As imperatives in this function serve to express condition, we
can understand their occurrence in connexion with a preterit,
e.g. “Give him time, and he was generally equal to the demands of
suburban customers; hurry or interrupt him, and he showed
314himself anything but the man for a crisis” (Gissing), and the use
of a perfect imperative in “Soyez bon, pitoyable, intelligent, ayez
souffert mille morts: vous ne sentirez pas la douleur de votre ami
qui a mal aux dents” (Rolland). Note also the imperative in the
middle of a dependent clause, e.g. “Darwin tells us how little
curly worms, only give them time enough, will cover with earth
even the larger kind of stones” (Birrell) | an Alpine Avalanche;
which once stir it, will spread (Carlyle) | I thought that, take them
all round, I had never seen their equals (Butler). 1200

This use of what might be called the imaginary imperative 2201
helps us to explain the fact that some imperatives have become
prepositions or conjunctions, e.g. When you feel that, bar accidents,
the worst is over (Quiller-Couch) | I am not in the habit of beating
women at any time, let alone at a lunch-party (Hope) | Suppose
he were to come, what then? Dan. Sæt han kom, hvad så?

Indicative and Subjunctive.

If we pass on to the Indicative and the Subjunctive, the first
remark that obtrudes itself is that the treatment of this subject
has been needlessly complicated by those writers who speak of
combinations with auxiliary verbs, e.g. may he come | he may come
| if he should come | he would come, as if they were subjunctives of
the verb come, or subjunctive equivalents. Scholars would hardly
have used these expressions if they had had only the English language
to deal with, for it is merely the fact that such combinations
in some cases serve to translate simple subjunctives in German or
Latin that suggests the use of such terms, exactly as people will
call to the boy a dative case. Itus equally wrong to speak of bless
in God bless you as an optative, while the same form in if he bless you
is called a subjunctive; we should use the term ‘optative’ only
where the language concerned has a separate form, as is the case in
Greek — but there, of course, the optative is not exclusively an
“optative” in the sense just alluded to, i.e. a mood of wish, but
has other meanings as well. A precise terminology is a conditio sine
qua non if one wants to understand grammatical facts. 3202

The view here presented is in direct opposition to that taken by
Professor Sonnenschein. Though my objections to his treatment
of the theory of moods are essentially the same as those I had
315against his theory of cases, it may not be superfluous to review what
he says of moods, and to show the contradictions and difficulties
inherent in his conception of them. The term ‘mood’ must not,
he says, be taken to involve a difference of inflexion. Such a
definition would make havoc of the moods of any language; for
example, the Latin regam and rexerit and the German liebte may be
either indicative or subjunctive; and the Latin forms in -ere may
be either imperative or indicative or infinitive. — My reply is, of
course, that we recognize the Latin moods because the majority
of forms are distinctive: rego, regis, rexero, rexeras, and innumerable
other forms can only be one mood each, and if we substitute the
forms of another verb or another person of the same verb it is quite
easy to decide what is the mood of any ambiguous form in a given
context. If instead of G. liebte in one sentence we should say
hatte, it is the indicative; if we should say hätte, it is the subjunctive,
etc. 1203

Moods then, according to Professor Sonnenschein, denote
categories of meaning, not of form. The indicative mood speaks
of a matter of fact (S. § 211). But if I say “Twice four is seven”
I use the indicative to express the opposite of a fact. This objection
might be called captious, for the meaning evidently is that the
indicative is used to represent something as a fact; yet even in
that form the statement cannot be always maintained, cf. the
frequent use of the indicative in conditional clauses: “if he is ill,”
and after wish: “I wish he wasn't ill.”

Next, we are told that “the meaning of the subjunctive is
quite different from that of the indicative” (§ 214). Nevertheless
we read in § 315 that in “Take care that you are not caught” the
indicative is “used with the meaning of the subjunctive.” Similar
contradictions are found in other places: in § 219 the author
admits that it would be possible to use contest and falls instead of
the subjunctives in “stint not to ride, Until thou come to fair
Tweedside” and “Who stands, if freedom fall?”, but he says
that “these present indicatives would be used with a special
meaning; they would, in fact, be equivalent to subjunctives.”
Similarly in § 234: “the past indicative is sometimes used after
‘as if,’ but it always has the meaning of a past subjunctive.”
But as the distinction of moods is by definition one of meaning,
the simple inference is that this indicative is a subjunctive! Inversely,
in § 303 (note) S. speaks of a subjunctive without any clear
316difference of meaning from an indicative in when I ask her if she
love me. According to § 219 Obs. a present indicative is quite
impossible in noun-clauses which express that something is to be
done. We take his own sentence “Give the order that every
soldier is to kill his prisoners,” and we naturally ask, is this “is (to
kill)” an indicative or a subjunctive? How are thinking pupils to
find their way in this wilderness? 1204

If we start from the assumption that meaning is decisive in
these matters, it is also difficult to see the logic of Sonnenschein's
§ 215: “The reason why the subjunctive is not so common now
as it used to be is that we have got into the habit of expressing the
subjunctive meaning in other ways, especially by using the verbs
‘shall’ and ‘may’ with the infinitive instead of the subjunctive”
and § 219 “It is a mistake to say that the subjunctive mood has
practically disappeared from modern English… But it is true
to say that the equivalent expressions mentioned in § 215 are still
commoner,” for here “subjunctive” must necessarily be used of
the form if the paragraphs are to make sense.

If there are thus many divergences, there are also certain
general tendencies common to languages of our family. The
indicative is generally used in relative clauses and clauses introduced
by local and temporal conjunctions (where, when, while),
unless (in some languages) an intention is implied or the clauses
express the thought of some other person than the speaker or
writer. With regard to condition, the subjunctive is most often
required if impossibility is implied (in “clauses of rejected or,
better, of rejecting condition,” or “contrary-to-fact-condition”),
though even there English tends to get rid of the subjunctive;
greater hesitation is found when the possibility is admitted, but
the speaker “wants to guard himself from endorsing the truth or
realization of the statement” (NED); and finally the indicative
is required when the two ideas are not really meant as conditioning
and conditioned, but as equally true: “if he was rich, he was openhanded
too,” i.e. he was both, though these two things do not
always go together; the meaning of the conditional form may be
said to be: if you admit that he was rich, you must admit also that
he was open-handed; cp. “she is fifty if she is a day.” 1208 Similar
considerations hold good with regard to concession (though he were,
was, be, is).

Notional Moods.

Would it be possible to place all “moods” in a logically consistent
system? This was attempted by grammarians more than a
hundred years ago on the basis of first Wolff's and then Kant's
philosophy. The former in his Ontology had the three categories,
possibility, necessity and contingency, and the latter under the
head of “modality” the three of possibility, existence, and
necessity; Gottfried Hermann then gave the further subdivisions:
objective possibility (conjunctive), subjective possibility (optative),
objective necessity (Greek verbal adjectives in -teos) and subjective
necessity (imperative). It is hardly worth while following the
subsequent development of these theories (see the able paper
“A Century of Metaphysical Syntax,” by W. G. Hale, in the St.
Louis Congress of Arts and Sciences, 1904, Vol. III).319

Even Deutschbein's scheme is not exhaustive, and he does not distinguish
strictly enough between syntactic and notional categories.
As a tentative scheme of the purely notional ideas expressed more
or less vaguely by the verbal moods and auxiliaries of various
languages we might perhaps give the following list, to which I
cannot, however, attach any great importance. The categories
frequently overlap, and some of the terms are not quite unobjectionable.
The placing of the Conditional and Concessional also is
subject to doubt, and a “Subordinative” should perhaps be
added at the end of the list.

Each of these can be expressed linguistically by a variety of
means besides those mentioned.

There are many “moods” if once one leaves the safe ground
of verbal forms actually found in a language. 1209321

Chapter XXIV
Negation

Contradictory and Contrary. Some Tripartitions. The Meaning of Negation.
Special and Nexal Negation. Double or Cumulative Negation.
History of Negatives. Implied Negation.

Contradictory and Contrary.

Logicians distinguish between contradictory terms, such as white
and not-white, rich and not-rich, and contrary terms, such as white
and black, rich and poor. Two contradictory terms together comprise
everything in existence, as any middle term is excluded,
while two contrary terms admit one or more middle terms. For
contradictory terms language generally employs either derivatives
like unhappy, impossible, disorder or composite expressions containing
the adverb not: On the other hand, separate roots are
very often used to express the most necessary contrary terms.
Hence such pairs as young — old, good — bad, big — small, etc. Intermediate
stages may be expressed negatively, e.g. neither young nor
old, but in some cases we have special expressions for the intermediate
stage, e.g. indifferent in the comparatively recent sense of
‘what is between good and bad.’ Sometimes we have even a whole
long string of words with shades of meaning partially overlapping,
e.g. hot (sweltering), warm, tepid, lukewarm, mild, fresh, cool, chilly,
cold, frosty, icy; though each adjective at the head of this list is a
contrast to each of those at the tail, it is impossible to draw a sharp
line between two halves of the list.

A and C are absolute and imply certainty, B implies uncertainty,
and in that respect B is the negative counterpart of the two
positive sentences A “it is certain that he is rich” and C “it is
certain that he is not rich.”

It may shock the logician that the two sentences “John is
rich” and “John is not rich” are here treated as contrary and not
as contradictory, but I hope he will be relieved when I say that
evidently “rich” and “not rich” are contradictory and admit no
middle term: the tripartition given above refers only to the attitude
of the speaker to the inclusion of John in one of the two classes
“rich” and “not rich.” Our tripartition assists us in understanding
some linguistic facts with regard to questions, for a
question is an assertion of the class B + a request addressed to the
hearer to resolve the doubt. It is therefore immaterial whether
the question is couched positively or negatively: “Is John rich?”
or “Is John not rich?” are perfectly synonymous, because the
real question is double-sided: “Is John rich, or is he not?”
(Alternative question, p. 303, above.) In the same way, in offering
a glass of beer one may say either “Will you have a glass of beer?”
or “Won't you have a glass of beer?” Positive and negative
here mean the same thing, just as in “Perhaps he is rich” and
“Perhaps he is not rich.”

What is here said of questions is true of unemotional questions
only; a marked tone of surprise will make the two sentences
into distinct contrasts: for then “Will you (really) have a glass of
beer?” comes to mean ‘I am surprised at your wanting a glass
of beer?’, and “Won't you have a glass of beer?” the reverse.
While in English “Won't you pass me the salt?” would be
rude as implying unwillingness in the person addressed, in Danish
“Vil De række mig saltet?” is generally a command, and “Vil
De ikke række mig saltet?” a polite request (‘Would you mind
passing the salt?’). A Dutch lady once told me how surprised
she was at first in a Copenhagen boarding-house at these negative
questions, which she took as requests not to pass the salt. Very
often the particular interrogative form is chosen to suggest a
particular answer, thus especially in tag-questions (“He is rich,
isn't he?” | “He isn't rich, is he?”). Consequently questions
often come to mean assertions of the inverse: “Am I my
brother's keeper?” = ‘I am not’ | “Isn't that nice?” = ‘It
is very nice.’

As exclamations nave in many cases developed out of questions,
we now also understand how it is that very often it does not matter
whether not is added or not: “How often have I (not) watched
him!”323

Some Tripartitions.

B1 approaches A (all); B3 approaches C (none) and may even
in many cases be considered negative rather than positive; this
is especially true of the adverb little, e.g. in “They little think
what mischief is in hand” (Byron). The use of the indefinite
article to distinguish B2 and B3 is linguistically interesting; it
is not confined to English, cp. Fr. un peu, It. and Sp. un poco,
G. ein wenig. The difference is well brought out in Shakespeare's
sentence: “When he is best, he is a little worse than a man, and
when he is worst, he is little better than a beast.” B3 is felt as a
contrast to B1, but B2 rather to C; cp. “Few of the passengers
survived” and “A few of the passengers survived.”324

The imperative (“Take that!”) may mean either A or B, see
above under Requests.

The Meaning of Negation.

If we now want to inquire into the meaning of negation, the
first point of importance is to emphasize the difference between a
linguistic negative and a mathematical negative: — 4 means,
not everything different from + 4, but a point as much below
0 as 4 is above 0. A linguistic negative, on the contrary, changes a
term into the contradictory term, at any rate theoretically, for on
closer inspection we shall find that in practice this rule requires
some very important qualifications; to understand these the
division made above into A, B, and C-categories will prove useful
and should constantly be borne in mind. Let us first look at
quantities in the B-category (above, p. 324): neither all nor nothing.

Here the general rule in all (or most) languages is that not means
‘less than,’ or in other words ‘between the term qualified and
nothing.’ Thus not good means ‘inferior,’ but does not comprise
‘excellent’; not lukewarm indicates a lower temperature than
lukewarm, something between lukewarm and icy, not something
between lukewarm and hot. This is especially obvious if we
325consider the ordinary meaning of negatived numerals: He does
not read three books in a year | the hill is not two hundred feet high
| his income is not £200 a year | he does not see her once in a
week | the bottle is not half full — all these expressions mean less
than three, etc. Therefore not one comes to be the natural expression
in many languages for ‘none,’ e.g. OE. nan = ne-an, whence
modern none, no, further ON. eingi, G. k-ein, Fr. pas un bruit, etc.

But the same expressions may also exceptionally mean ‘more
than,’ only the word following not then has to be strongly stressed
(with the peculiar intonation indicative of contradiction), and then
the whole combination has generally to be followed by a more
exact indication: not lukewarm, but really hot | his income is not
two hundred a year, but at least three hundred | not once, but two
or three times, etc. Note that not once or twice always means
several times, as in Tennyson's “Not once or twice in our fair
island-story, The path of duty was the way to glory.”

Not above 30 means either 30 or less than 30. No more than
generally means ‘as little as,’ and no less than ‘as much as,’ e.g.
“the rank and file of doctors are no more scientific than their
tailors; or their tailors are no less scientific than they” (Shaw); note
the distinction between no and not in these combinations: no more
than three ‘three only’; not more than three ‘three at most’; he
paid no less than twenty pounds implies astonishment at the greatness
of the amount, which was exactly £20; he paid not less than
twenty pounds implies uncertainty with regard to the exact amount,
which at the very least was £20 (MEG II, 16. 84). In Latin both
non magis quam and non minus quam are favourite expressions for
equality, though, of course, used in different connexions: Cæsar
non minus operibus pacts florebat quam rebus in bello gestis | Pericles
non magis operibus pads florebat quam rebus in bello gestis (Cauer).

If we turn to the negatives of the terms given above as B1,
2 and 3, we see that negativing 1 turns it into three: not much = little;
not many = few. But a negative 2 becomes nearly synonymous
with 1 (or stands between 1 and 2): not a little = much, not a few
= many. B3 is not used idiomatically with not.

Next we turn to the A and C-categories, the two extremes.
Here we have the general rule that if the negative word is placed
first, it discards the absolute element, and the result is the intermediate
term: Not A = B; not C also = B. If, on the other
hand, the absolute term is mentioned first the absolute element
prevails, and the result is the contrary notion: A… not = C;
C… not = A.

Exceptionally the same effect (B) is obtained even though the
negative comes after the A-word in such sentences as “All that
glisters is not gold” (Shakespeare), and “Tout ce qui reluit n'est
pas or,” which correspond to the Danish and German forms of the
proverb: “Ikke alt hvad der glimrer er guld” and “Nicht alles
was glänzt, ist gold”; cp. also from the Bible: All things are
lawfull vnto mee, but all things are not expedient | all is not lost
(Milton, Shelley) | But all men are not born to reign (Byron) | For
each man kills the thing he loves, Yet each man does not die (Wilde);
similar examples abound also in the literatures of other countries;
they are easy to explain psychologically as the result of the two
tendencies, to place the subject first, and to attract the negation
to the verb. Tobler (VB 1. 197) tries to justify them logically
as saying “von dem subjekte ‘alles glänzende’ darf ‘gold sein’
nicht pradiziert werden.” This is true, but does not touch the
fact that the word-order makes us expect the meaning ‘nothing
of what glitters is gold’ (was glänzt, ist niemals gold; C) rather
than the intended meaning ‘only some part of what glitters is
gold’ (was glänzt ist nicht immer gold; B). 1211

Examples of A with a negative after it = C: Tous ces gens-là
ne sont pas humains (i.e. none of them is, Rolland) | the one [uncle]
I was always going to write to. And always didn't (Dickens).
This is rare except when the negative is in the form of a prefix or is
implied, e.g. they were all of them unkind; everybody was unkind
(= nobody was kind) | he was always unkind | they all failed
(= nobody succeeded).

Result C: he is not at all happy (he is not happy at all) | pas
du tout | slet ikke | gar nicht.327

Cp. from a recent newspaper: Germany's offer is entirely
unacceptable to the French and not wholly acceptable to the English
Government.

We next proceed to the three categories mentioned p. 325: A
necessity, B possibility, C impossibility. If we add a negative,
we see the following results: not necessary (A) = possible (B);
not impossible (C) = possible (B); it is impossible not to see =
necessary; no one can deny = everyone must admit | nobody need
be present = everybody may be absent | he cannot succeed = he
must fail | non potest non amare | il ne pouvait pas ne pas voir qu'on
se moquait de lui.

With regard to the further tripartition A command, B permission,
C prohibition, we have seen that the imperative may
mean either A or B. Therefore a negative imperative, e.g. Don't
take that! may mean either a negative command (= a prohibition),
or a polite request (or advice) not to take it; and on account of this
ambiguity there is in many languages a disinclination to use a
negative imperative. In Latin it is only found poetically, being
otherwise replaced by a paraphrase with noli (Noli me tangere)
or a subjunctive (Ne nos inducas in tentationem); in Spanish the
latter has become the rule (No vengas ‘don't come’). In Dan.
Tag det ikke is generally a piece of advice, and La vær å ta
det (Lad være at tage det) has become the usual form for
a prohibition. In other languages we find separate verb-forms
(‘jussive’) or else separate negatives (e.g. Gr. mē) used in
prohibitions.

Both may not and must not may be used in prohibitions. In
the former not logically belongs to may (the negation of a permission,
cf. G. du darjst nicht), but as the same combination is often
used in a different sense, e.g. in “He may not be rich, but he is
a gentleman” (where not goes with be: it is possible that he is not),
and as may is also felt to be too weak for a prohibition, the tendency
328is more and more to use the more brutal must not, except in questions
implying a positive answer (mayn't I = ‘I suppose I may’) and
in close connexion with a positive may, e.g. in answers (“May I
take that? No, you may not”). In you must not take that, the
negative logically belongs to the infinitive: it is a positive command
(must) not to take that; 1212 but the prevailing tendency to
attract the negative to the auxiliary verb leads to the usual form
you mustn't. In this way we get different auxiliaries in positive
and negative sentences, e.g. You may call me Dolly if you like;
but you mustn't call me child (Shaw) | You mustn't marry more
than one person at a time, may you? (Dickens). Now, however,
must is beginning to be used in tag questions, e.g. “I must
not go any farther, must I?” (G. Eliot), though it is not possible
otherwise to substitute Must I? for May I?

Special and Nexal Negation.

We have seen already that the meaning of a sentence sometimes
depends on the place of a negative element. In a more
general way we may say that the negative notion may belong
logically either to one single idea (special negation) or to the combination
of the two parts of a nexus (nexal negation). In the
former case we have either a negative prefix (as in never, tmhappy,
disorder), or the adverb not put before the word (not happy); in
some cases a single word without any negative prefix may be regarded
as containing a negative idea, e.g. lack (= have not), fail
(= not succeed; but we may also say that succeed is the negative
counterpart of fail).

When a nexus is negatived, the negative adverb is generally
attracted to the verb, in many languages in the form of a weak
ne or similar particle placed before the verb, and sometimes
amalgamated with it (cp. earlier E. nis, nill); in MnE we have
the do-combinations (does not come, doesn't come, etc.) except
with the well-known group of verbs (is not, isn't, cannot, etc.).

In the sentence “Many of us didn't want the war” the nexus
is negatived, but in “Not many of us wanted the war” not belongs
exclusively to many, which it turns into ‘few.’

In many cases it seems to be of no importance whether we
negative one notion only or the combination of that notion with
another; she is not happy may be analyzed either as a description
of what she is, viz. not-happy (= unhappy), or as a negativing
of her being happy (she is-not, isn't, happy). If we add very,
however, we see the difference between “she is very unhappy”
and “she is not very happy.”329

The general tendency is to use a nexal negative, even in some
cases where a special negative would be more apposite. By the
side of the logically impeccable “I came not to send peace, but a
sword” (Matt. 10. 34) we frequently find sentences like “I don't
complain of your words, but of the tone in which they were uttered”
(= I complain, though not…, but of…) | “We aren't here
to talk nonsense, but to act” (where “we aren't here” in itself is
a contradiction in terms). A particular case is found with because:
the sentence “I didn't go because I was afraid” is ambiguous and
may mean either ‘I went, but the reason was not fear,’ or ‘I
did not go, and the reason for not going was fear,’ though in the
spoken language the tone may show which is meant; cp. further
“I didn't call because I wanted to see her” (but for some other
reason), and “I didn't call because I wanted to avoid her.”

With infinitival and similar constructions it is often very important
to know which of two verbal notions is negatived; various
devices are used in different languages to make the meaning clear.
A few examples may suffice: She did not wish to reflect; she
strongly wished not to reflect (Bennett) | Tommy deserved not
to be hated | Tommy did not deserve to be loved | Dan. prav ikke
pa at se derhen | prav pa ikke at se derhen | il ne taohe pas do
regarder | il tfi.che de ne pas regarder | il ne peut pas entendre |
il peut ne pas entendre | (Will he come?) I am afraid not | I am
not afraid.

The tendency already mentioned to attract the negation to the
verb is not the only one found in actual language: we often find
the opposite tendency to attract the negative notion to any word
that can easily be made negative. In literary English “we met
nobody” is thought more elegant than the colloquial “we didn't
meet anybody”; cp. also “this will be no easy matter” and
“this won't be an easy matter.” In many cases we find words
like nothing used where a nexal negation would be more logical,
e.g. she loves you so well that she has the heart to thwart you in
nothing (Gilbert) | you need be under no uneasiness. Attraction
of this kind is seen also in the idiomatic use of “he was no
ordinary boy” in preference to “he was a not ordinary boy”
and in sentences like “you and I will go to the smoking-room,
and talk about nothing at all subtle” (= about something
that is not subtle, Benson), which most people would probably
censure as wrong.

Wherever it might seem possible to attract the negative element
to either of two words, it is nearly always put with the first. We
may say “no one ever saw him angry” or “never did any one see
him angry,” but not “any one never saw him angry” or “ever did
no one see him angry.” Cp. also Lat. “nec quisquam” (not “et
330nemo”), “neque ullus,” etc. Without any danger is preferred to
with no danger.

When the negative is attracted to the subject, the sentence is
often continued in such a way that the positive counterpart of the
first subject must be understood. In ordinary life this will cause
no misunderstanding, and it is only the critical, or hyper-critical,
grammarian that discovers anything wrong in it, e.g. Not one should
scape, but perish by my sword (= but all perish, Marlowe) | none
of them are hurtful, but loving and holy (Bunyan). Cp. also:
Don't let any of us go to bed to-night, but see the morning come
(Benson) | I quite forget the details, only that I had a good deal of
talk with him (Carlyle). 1213

Double or Cumulative Negation.

It seems to be an established view among theorists, logicians
as well as linguists, that two negatives ought to cancel one another,
because two negatives logically make an affirmative in the same way
as in mathematics -(-4) = +4. Languages, as well as individual
writers, are consequently censured if they use a double negative as
a strengthened negative. If this view were true, a consistent
logician would have to find fault with Chaucer's “He neuere yet no
vileynye ne seyde In al his lyf unto no maner wight,” because here
four negatives (thus an even number) are made to serve as a
strengthened negative expression, but not with the OE. example
“nan man nyste nan þing” (no man not-knew nothing), because
there are here three negatives, of which two should cancel each
other, leaving one over. But as a matter of fact no one seems to
calculate cumulative negation in this way, and this is perfectly
right from the point of view of linguistic logic.

Language is not mathematics, and, as already remarked, a
linguistic negative cannot be compared with the sign — (minus)
in mathematics; hence any reference to the mathematical rule
about two minus's is inconclusive. But neither are the attempts
made by some linguists to justify the use of double negation perfectly
satisfactory. Van Ginneken rightly criticizes the view of Romanic
scholars, who speak of a half-negation in the case of French ne —
an explanation which at any rate does not explain many of the
phenomena in other languages. His own explanation is that
negation in natural languages is not logical negation, but the
expression of a feeling of resistance; according to him the logical
or mathematical conception of negation, according to which two
331negatives are mutually destructive, has only gained ground in a
few centres of civilization and has never struck root in the popular
mind. I have my doubts as to the greater primitivity of the idea
of ‘resistance’ than that of negation understood exactly as we
understand it in such a simple sentence as “he does not sleep.”
Other writers speak of a difference between qualitative and quantitative
negation and imagine that this distinction finds a support in
Kant's table of categories, though as a matter of fact Kant ranges
all negation under the heading of “quality.” Anyhow the distinction
does not assist us at all to comprehend double negation. 1214

Language has a logic of its own, and in this case its logic has
something to recommend it. Whenever two negatives really
refer to the same idea or word (as special negatives) the result is
invariably positive; this is true of all languages, and applies to such
collocations as e.g. not uncommon, not infrequent, not without some
fear. The two negatives, however, do not exactly cancel one
another in such a way that the result is identical with the simple
common, frequent, with some doubt; the longer expression is always
weaker: “this is not unknown to me” or “I am not ignorant of
this” means ‘I am to some extent aware of it,’ etc. The psychological
reason for this is that the détour through the two mutually
destructive negatives weakens the mental energy of the listener
and implies on the part of the speaker a certain hesitation which is
absent from the blunt, outspoken common or known. In the same
way I don't deny that he was angry is weaker than I assert, etc.
Cp. also Fr. il n'était pas sans être frappé.

On the other hand, if two (or more than two) negatives are
attached to different words, they have not the same effect upon one
another, and the total result, therefore, may very well be negative.
We see this in a great variety of languages, where cumulative
negation in this way is of everyday occurrence. Examples from
Old and Middle English have already been given; they abound
in these periods, but are somewhat rarer in Elizabethan English;
in dialectal and vulgar English of our own day they are frequent,
and many examples may be culted from representations of popular
language in novels and plays, e.g. “Nobody never went and hinted
no such thing, said Peggotty” | “I can't do nothing without my
staff” (Hardy).

In other languages we find the same phenomenon more or less
regularly. Thus in Middle High German: nu en-kan ich niemanne
gesagen. In French: on ne le voit nulle part. In Spanish: aqui
no vienen nunca soldados ‘here not come never soldiers.’ In
332Slavic languages, Serbian: i nikto mu ne mogaše odgovoriti riječi
‘and nobody him not could answer word’ (Delbrück). Russian:
Filipok ničego ne skazal ‘F. nothing not said’ Greek: aneu
toutou oudeis eis ouden oudenos an humōn oudepote genoito axios
(Plato, in Madvig).

So also outside our family of languages, e.g. Magyar: sēmmit sēm
hallottam, or: nëm hallottam sēmmit ‘nothing not I have heard’
(Szinnyei). Congo (Bantu): kavangidi kwandi wawubiko, kamonanga
kwandi nganziko, kaba yelanga kwa-u ko ‘not did he evil not,
not feeling he no pain, not they sick not.’

How to account for this phenomenon, which is spread over so
many different languages? There is one very important observation
to be made, without which I do not think that we shall be able
to understand the matter, namely that repeated negation becomes
an habitual phenomenon in those languages only in which the
ordinary negative element is comparatively small in phonetic
bulk: ne or n- in Old English, in French, in Slavic, en or n- in Middle
High (and Middle Low) German, ou in Greek, s- or n- in Magyar.
These are easily attracted to various words (we have already seen
instances of such attraction in previous sections), and the insignificance
of these initial sounds or weakly stressed syllables makes it
desirable to multiply them in a sentence so as to prevent their
being overlooked. Under the influence of strong feeling the speaker
wants to make absolutely sure that the negative sense will be fully
apprehended; he therefore attaches it not only to the verb, but
also to any other part of the sentence that can be easily made
negative: he will, as it were, spread a layer of negative colouring
over the whole of the sentence instead of confining it to one single
place. If this repetition is rarer in modern English and German
than it was formerly, one of the reasons probably is that the fuller
negatives not and nicht have taken the place of the smaller ne and
en, 1215 though the logic of the schools and the influence of Latin
have also contributed towards the same result. It may also be
said that it requires greater mental energy to content oneself
with one negative, which has to be remembered during the whole
length of the utterance both by the speaker and the hearer, than
to repeat the negative idea whenever an occasion offers itself,
and thus impart a negative colouring to the whole of the sentence.

If we are now to pass judgment on this widespread cumulative
negation from a logical point of view, I should not call it illogical,
333seeing that the negative elements are not attached to the same
word. I should rather say that though logically one negative
suffices, two or three are simply a redundancy, which may be
superfluous from a stylistic point of view, just as any repetition in
a positive sentence (every and any, always and on all occasions),
but is otherwise unobjectionable. No one objects from a logical
point of view to combinations like these: “I shall never consent,
not under any circumstances, not on any condition, neither at
home nor abroad”; it is true that here pauses, which in writing
are marked by commas, separate the negatives, as if they belonged
to so many different sentences, while in “he never said nothing”
and all the other cases quoted from various languages the negatives
belong to one and the same sentence. But it is perfectly impossible
to draw a line between what constitutes one, and what constitutes
two sentences: does a sentence like “I cannot goe no further”
(Shakespeare) become more logical by the mere addition of a
comma: “I cannot goe, no further”?

As a separate variety of double negation must be treated what
might be called resumptive negation (Delbrück's erganzungsnegation).
This is especially frequent when not is followed by a disjunctive
combination with neither… nor or a restrictive addition with
not even: “he cannot sleep, neither at night nor in the daytime”
or “he cannot sleep, not even after taking an opiate”; cp. also the
addition in “loue no man in good earnest, nor no further in sport
neyther” (Sh.). Similarly in other languages, Lat. non… neque
… neque, non… ne… quidem, Gr. ou… oude… oude,
etc. In such cases, with ‘neither — nor’ and ‘not even,’ all languages
seem freely to admit double negatives, though, even here
precisians object to them. 1216

Closely connected with resumptive negation is paratactic
negation: a negative is placed in a clause dependent on a verb
of negative import, e.g. ‘deny, forbid, binder, doubt,’ as if the
clause had been an independent sentence, or as if the corresponding
positive verb had been used in the main sentence. Examples:
First he deni'de you had in him no right (Sh.) | What hinders in
your own instance that you do not return to those habits (Lamb).
It is well known how in some languages this develops to a fixed
rule, e.g. in Latin with ne, quin, quominus, in French with ne (which
now, like ne in other positions, tends to disappear). Here, too, we
have redundancy and over-emphasis rather than irrationality or
want of logic.334

History of Negatives.

The general history of negative expressions in some of the
best-known languages presents a curious fluctuation. The negative
adverb is often weakly stressed, because some other word in the
sentence has to receive a strong stress of contrast. But when the
negative has become a mere proclitic syllable or even a single sound,
it is felt to be too weak, and has to be strengthened by some additional
word, and this in its turn may come to be felt as the negative
proper, which then may be subject to the same development as
the original word. We have thus a constant interplay of weakening
and strengthening, which with the further tendency to place the
negative in the beginning of the sentence where it is likely to be
dropped (though prosiopesis) leads to curious results, which can
here be sketched only in the briefest outlines by examples taken
from a few languages.

In Scandinavian, too, the original ne was first strengthened by
additions and finally ousted by these, ON. eigi, ekki, Dan. ej, ikke,
which at first had no negative meaning.

In German we had first ni alone before the verb, then ni, ne
(or weakened n-, en-) before and nicht after the verb, and finally
nicht alone.

The strengthening of negatives is effected either by means of
some word meaning a small thing (not a bit, not a jot, not a scrap,
etc., Fr. ne… mie, goutte, point, pas), or by means of an adverb
meaning ‘ever’ (OE. na from ne + a = Gothic ni aiws, G. nit;
E. never also sometimes loses its temporal meaning and means
nothing but ‘not’). Finally the strengthening addition may be
a word meaning ‘nothing’ as Lat. non, E. not (a weaker form
of nought) or G. nicht; in ME. I ne seye not there is a double
negation.

The dropping or leaving out of a weak negative adverb changes
a positive into a negative word. The most characteristic examples
of this are found in French, where pas, personne, jamais and other
words are now negative — invariably so when there is no verb:
pas de doute | Qui le sait! Personne | Jamais de la vie, and in
vulgar and familiar speech also in sentences containing a verb,
where literary language requires ne: Viens-tu pas? | je le vois
jamais. With regard to plus, ambiguity has in some cases been
obviated by the popular pronunciation, [j ãn a ply] meaning ‘there
is no more of it’ and [j ãn a plys] ‘there is more of it.’ An isolated
Plus de bruit is a negative, but Plus de bruit que de mal a positive
expression, though the pronunciation is here the same. There is a
curious consequence of this negative use of plus, namely that moins
may occasionally appear as a kind of comparative of plus: Plus
d'écoles, plus d'asiles, plus de bienfaisance, encore moins de théologie
(Mérimée).

In other languages the transition from positive to negative is
found sporadically, as in Sp. nada ‘nothing’ from Lat. (res) nata,
nadie ‘nobody,’ and in the ON. words in -gi; in English we find
but from ne… but, cp. dialectal nobbut, and the curious more
for ‘no more’ in the South-Western part of England, e.g. “Not
much of a scholar. More am I” (Phillpotts).

Implied Negation.

As in other provinces of grammar, we have here cases of disagreement
between the notional meaning and the grammatical
expression. A notional negation is often implied though the
sentence contains no negative proper.336

A question is often equivalent to a negative assertion: Am I my
brother's keeper? (See p. 323.)

Combinations like Me tell a lie! = ‘I cannot toll a lie’ have
been mentioned, p. 130.

Conditional expressions may serve the same purpose, e.g. “I
am a rogue if I drunke to-day” (= I did not drink, Sh.) | I'm
dashed if I know; also with the conditional clause standing alone:
If there isn't Captain Donnithorne a-coming into the yard! (G.
Eliot; here, of course, the direct and the indirect negations cancel
each other, the result being positive: he is coming).

Further may be mentioned: (you) see if I don't | catch me going
there! | Mr. Copperfield was teaching me. — Much he knew of it
himself | When the devil was ill, the devil a monk would be; When
the devil got well, the devil a monk was he. Similar idiomatic
and ironical expressions seem to be frequent in all languages.

A notional negative is also implied in the use of the preterit
(subjunctive) in clauses of rejected condition (p. 265).

Note. — The whole subject of this chapter has been treated with much
fuller illustration from many languages and with discussion of some points
here omitted (negative conjunctions, prefixes, the contraction of not into -nt,
etc.) in “Negation in English and Other Languages,” Det kgl. Danske
Videnskabernes Selskabs Historisk-Filologiske Meddelelser I, 5 (Copenhagen,
1917).337

Chapter XXV
Conclusion

Conflicts. Terminology. The Soul of Grammar

Conflicts.

It is a natural consequence of the complexity on the one hand of
the phenomena of life which have to be expressed, and on the other
hand of the linguistic means available to express them, that conflicts
of various kind are bound to occur, in which the speaker has
to make a choice and then, possibly after some hesitation, uses
one form where someone else in the same situation might have
used another form. In some cases we witness a tug-of-war, as it
were, between two tendencies which may go on for a very long
period, during which grammarians indulge in disputes as to which
form or expression is “correct”; in other cases one of the conflicting
tendencies prevails, and the question is settled practically by
the speaking community, sometimes under protest from the Lindley
Murrays or Academies of the time, who very often prefer logical
consistency to ease and naturalness. Examples of grammatical
conflicts will be found here and there in this volume: the most
typical ones are perhaps those mentioned in Ch. XVII of rivalry
between the notional idea of sex and grammatical gender (leading,
for instance, to Greek neanias, G. ein fräulein … sie, Sp. el
justicia). In Ch. XIV we saw the competition between singular
and plural in the verb connected with a collective. Some other
conflicts of a similar character may be mentioned here.

In the Gothonic languages there is no distinction of gender in
the plural; but the want of an express indication of the “natural
neuter” in speaking of more than one thing leads to the employment
of what is properly a singular neuter ending in G. beides,
verschiedenes (cp. also alles); Curme GG 149 mentions alles dreies,
and Spitzer somewhere writes alles drei (“Sie sind weder germanen
noch gallier noch auch romanen, sondern alles drei der abstammung
nach”). Here, then, gender has been stronger than number.

Similarly the feeling for the neuter is often stronger than the
feeling for the proper case. In the dative there was originally no
difference between masculine and neuter; but in English from an
338early period we find for it, to this, after what, and finally these
nominative-accusatives were the only forms of the neuter pronouns
that were used. In German we see the same tendency, though it
has not prevailed as completely as in English: Goethe has zu
was; was wohnte er bei is common, and zu (mit, von) etwas is the
only form used; thus also mit nichts, etc. (a survival of the old
form is found in zu nichte machen, mit nichten); wegen was is used
colloquially instead of the ambiguous wegen wessen (Curme GG 198)
But the tendency has not been strong enough to allow mit das,
von welches, though mit dem, von welchem in a neuter sense is not
frequent (cp. damit, wovon), and the dative is required in an adjective
following the uninflected pronoun: “der gedanke von etwas
unverzeilichem.”

G. wem, like E. whom, is common to masculine and feminine,
but where a distinctive form for the female sex is desirable, a rare
and unrecognized form wer may be used: “Von Helios gezeugt?
Von wer geboren?” (Goethe) | “Da du so eine art bruder von ihr
bist. — Von ihr? Von wer?” (Wilbrandt, Curme GG 191). This,
however, is only possible after a preposition, as wer as the first
word of the sentence would be taken as the nominative; Raabe
therefore finds another way out: “Festgeregnet! Wem und
welcher steigt nicht bei diesem worte eine gespenstische einnerung
in der seele auf?” (= what man and woman).

On the other hand, case has proved stronger than gender in
the gradual extension of the genitival ending -s to feminines in
English and Danish, the chief reason being, of course, that the old
form did not mark off the genitive distinctly enough from the other
cases. In German the same tendency is sometimes found with
proper names; Frenssen thus writes: “Lisbeths heller kopf.”

A conflict between the ordinary rule which requires an oblique
case after a preposition, and the feeling of a subject-relation which
requires the nominative, sometimes leads to the latter idea gaining
the upper hand, e.g. E. “Me thinkes no body should be sad but I”
(Sh.) | “not a man depart, Saue I alone” (id.) | “Did any one
indeed exist, except I?” (Mrs. Shelley) | G. “Wo ist ein gott
ohne der herr” (Luther) | “niemand kommt mir entgegen ausser
ein unverschämter” (Lessing) | Dan. “ingen uden jeg kan vide
det,” etc. (cf. ChE, p. 57 ff.).

In a similar way we have in Sp. hasta yo lo sé ‘up to I, i.e. even
I know it’ (cp. Fr. jusqu'au roi le sait). It is really the same
principle that is at the bottom of the G. nominative in was für
ein mensch and the corresponding Russian što za čelovjek; finally
also in G. ein alter schelm von lohnbedienter.

The wish to indicate the second person singular is seen to have
been stronger than the desire to distinguish between the indicative
339and the subjunctive by the fact that combinations like if thou
dost and if thou didst became frequent at a much earlier period
than the corresponding uses of the indicative instead of the subjunctive
in the third person.

In Ch. XXI we have already seen the conflicts in indirect
speech between the tendency to keep the tense of direct speech
and the tendency to shift it into accordance with the main verb
(“He told us that an unmarried man was (or, is) only half a man”
| “he moved that the bill be read a second time”). In the sentence
“he proposed that the meeting adjourn” we may say that mood
has been stronger than tense, and the same is true in French, where
“il désirait qu'elle lui écrive” is now the only form used in ordinary
language instead of the earlier écrivisse. Inversely tense is stronger
than mood in colloquial French in a case like “croyez-vous qu'il
fera beau demain,” where old-fashioned grammarians would prefer
the present subjunctive fasse; Rousseau writes: “Je ne dis pas
que les bons seront récompensés; mais je dis qu'ils seront heureux”:
although after a negative main verb the ordinary rule is that the
verb is put into the subjunctive in the dependent clause.

In the matter of word-order there are a great many similar
conflicts, many of which fall under the head of style rather than
of grammar. Let me mention only one point of grammatical
interest: on the one hand prepositions are placed before their
objects, on the other hand interrogative and relative pronouns
have to be put in the beginning of the sentence. Hence conflicts,
which are often settled according to the more or less intimate
connexion between the preposition and its object or between the
preposition and some other word in the sentence: “What are
you talking of? | What town is he living in? or, In what town is
he living? | In what respect was he suspicious? | Some things
which I can't do without | Some things without which I can't
make pancakes.” I find an instructive example in Stevenson:
“What do they care for but money? For what would they risk
their rascal carcases but money?” By the side of “this movement
of which I have seen the beginning” (here it would be less natural
to say “which I have seen the beginning of”) we have the literary
“the beginning of which I have seen.” 1217 In French it is impossible
to relegate the preposition to the end of the sentence, hence it is
necessary to say “l'homme à qui j'ai donné le prix” and “l'homme
au fils duquel j'ai donné le prix.” As a genitive in English cannot
be separated from the word it belongs to, the object, which in
ordinary sentences comes after the verb, has to be placed before
the subject after wwwwhose in “the man whose son I met”; in French,
340on the other hand, there is no such inducement, and the object
comes at its usual place, though separated from dont, in “l'homme
dont j'ai rencontré le fils.”

Terminology.

Any branch of science that is not stationary, but progressive,
must from time to time renew or revise its terminology. New terms
must be found not only for newly discovered things like radium,
ion, but also for new ideas resulting from new ways of considering
old facts. Traditional terms often cramp the minds of investigators
and may form a hindrance to fertile developments. It is
true that a fixed terminology, in which the meaning of every single
term is plain to every reader, is a great boon, but if the terminology
is fixed only in so far as the same terms are used, while their meanings
vary according to circumstances or the usage of individual writers,
it becomes necessary to settle what would be the best meaning to
attach to these terms, or else to introduce new terms which are
not liable to misunderstanding.

In grammar terminological difficulties are aggravated by the
facts that many terms go back to pre-scientific ages and that many
again are used outside of grammar, often in meanings which have
little or no resemblance to the technical meanings attached to
them by grammarians, and finally by the fact that the same set
of terms is used for languages of different structure. It is, of
course, an advantage to the learner that he has not to acquire
a new set of terms for each new language he takes up, but this
is only of value if the grammatical facts covered by the same
terms are really analogous, and not so dissimilar that the use
of one and the same name may create confusion in the student's
mind.

The scorn of the oldest grammarians for a good terminology
is shown by their term verbum substantivum for the verb which
is the least substantial and farthest removed from any substantive,
further by the use of positive as the first degree of comparison, thus
not as usual opposed to negative, but to comparative, and by the
use of impersonal of some functions of the third “person.” It is
a great disadvantage that many grammatical terms have other
non-technical meanings, which sometimes make it difficult to avoid
such clashings as “this case [speaking of the nominative, for
instance] is found in other cases as well” or “en d'autres cas on
trouve aussi le nominatif,” “a singular use of the singular.” When
a grammarian sees the words “a verbal proposition” in a treatise
on logic, he is at first inclined to think that it has something to do
with a verb and may be opposed to a nominal sentence (nominal,
341by the way, is also ambiguous), until he discovers that it means a
mere definition of a word. Active, passive, voice, object, subject — I
have had occasion in various chapters to point out how the everyday
use of these words may mislead the unwary; the fact that subject
may mean ‘subject-matter’ has given rise to whole discussions
about logical, psychological, and grammatical subject which might
have been avoided if grammarians had chosen a less ambiguous
term. Neuter, besides its ordinary uses outside the province of
grammar, has two distinct meanings in grammar, of which one is
unavoidable (neuter gender), but the other can easily be dispensed
with: neuter verb — explained as “neither active nor passive;
intransitive” in spite of the fact that an intransitive verb is active
in the only sense in which the word ‘active’ should be used by a
consistent linguist. Besides this, the NED gives as an additional
meaning “neuter passive, having the character both of a neuter
and a passive verb” — confusion worse confounded!

A bad or mistaken name may lead to wrong rules which may
have a detrimental influence on the free use of language, especially
in writing. Thus the term preposition, or rather the unfortunate
knowledge of the Latin etymology of this word, is responsible for
that absurd aversion to putting a preposition at the end of a sentence
which many schoolmasters and newspaper editors profess
in utter ignorance of the principles and history of their own language.
These people do not consider the two possibilities which
the most superficial knowledge of general linguistics would have
brought to their notice, that the name may have been a misnomer
from the very first, or else that the value of the word may have
changed as has been the case with so many other words the etymology
of which is not, or is no longer, understood by the ordinary
users of the language. A ladybird is not a bird, nor a butterfly
a fly, and no one is the worse for it; blackberries are not black
till they are ripe; a barn may be used for other things than barley
(OE. bere-ærn ‘barley-house’) and a bishop has other occupations
than to ‘look at’ or ‘overlook’ (Or. epi-skopos). Why not, then,
admit postpositional prepositions, 1218 just as one admits adverbs
which do not stand by the side of a verb? (As a matter of fact,
very is always recognized as an adverb though it never qualifies
a verb.)

Terminological difficulties are sometimes aggravated by the
fact that languages change in course of time, and that therefore
terms which may be adequate for one period are no longer so for
a subsequent period. It is true that the case following the preposition
to in OE. to donne was a dative, but that does not justify us
in calling do in the modern to do a ‘dative infinitive,’ as the NED
342does (though under the word dative it does not mention this
use). It is even worse when the terms dative and genitive are
applied to modern prepositional groups like to God and of God;
see Ch. XIII.

It would evidently be utterly impracticable to throw the whole
traditional nomenclature overboard and create a totally new one,
for instance by an arbitrary system analogous to that of the old
Indian grammarians, who coined words like lat present tense, lit
perfect, lut first future, lrt second future, let subjunctive, lot imperative,
lan imperfect, lin potential, etc. (Benfey, Gesch. d. sprachw.
92: I omit the diacritics). We must take most of the old terms
as they are, and make the best use of them that we can, supplementing
them where it is necessary, and limiting the meanings
of all terms, old and new, as precisely and unambiguously as possible.
But this is no easy task, and I have the greatest sympathy
with Sweet, who wrote to me at the time when he brought out his
New English Grammar: “I have had most difficulty with the
terminology.”

In the preceding chapters (and earlier in my MEG) I have
ventured to introduce a certain number of new terms, but I make
bold to think that they are neither very numerous nor very difficult.
In both respects my procedure compares favourably both with the
wholesale coining of new grammatical terms and perversion of
old ones in Noreen's great work, and with the nomenclature of
certain recent psychologists. It should also be counted to my
credit that I am able to toss to the wind many of the terms used
in former grammatical works; thus I have no “use for” such terms
as synalepha, crasis, synæresis, synizesis, ekthlipsis, synekphonesis,
to mention only terms from one department of phonetic theory;
in the matter of “aspect” (Ch. XX) I am also more moderate
than most recent writers.

Among my innovations I should like to call special attention
to the terms connected with the theory of the “three ranks,”
where I think that the few new terms allow one to explain a great
many things more precisely and at the same time more tersely
than has been possible hitherto. Let me give one example that
has recently come under my notice. In Tract XV of the Society
for Pure English, Mr. H. W. Fowler speaks of the position of adverbs,
saying: “The word adverb is here to be taken as including adverbial
phrases (e.g. for a time) and adverbial clauses (e.g. if possible),
adjectives used predicatively (e.g. alone), and adverbial conjunctions
(e.g. then), as well as simple adverbs such as soon and
undoubtedly.” These five lines might have been spared if the writer
had made use of my simple word subjunct.343

The Soul of Grammar.

My task is at an end. A good deal of this volume has necessarily
been taken up with controversial matter, but it is my hope
that the criticism contained in it will be found to be constructive
rather than destructive. And let me add for the benefit of those
reviewers who are fond of pointing out this or that little article
in some recent periodical or this or that doctor's thesis which has
been overlooked, that I have very often silently criticized views
which appear to me to be wrong, without giving in each particular
case chapter and verse for what I take exception to. My theme
is so comprehensive that the book would have swelled to unwarrantable
dimensions had I treated at full length all the varying
opinions of other scholars on the questions I deal with. Those
who are interested in the great problems at issue rather than in
grammatical detail will perhaps think that I have quoted too much,
not too little, from the ever-increasing flood of books and articles
on these questions.

My endeavour has been, without neglecting investigation into
the details of the languages known to me, to give due prominence
to the great principles underlying the grammars of all languages,
and thus to make my contribution to a grammatical science based
at the same time on sound psychology, on sane logic, and on solid
facts of linguistic history.

Psychology should assist us in understanding what is going
on in the mind of speakers, and more particularly how they are
led to deviate from previously existing rules in consequence of
conflicting tendencies, each of them dependent on some facts in the
structure of the language concerned.

Logic as hitherto often applied to grammar has been a narrow
strictly formal kind of logic, generally called in to condemn certain
developments in living speech. Instead of that, we should cultivate
a broader-minded logic which would recognize, for instance,
that from the logical point of view the indirect object may be made
the subject of a passive sentence just as much as the direct object,
the question as to the permissibility of such sentences as “he
was offered a crown” being thus shifted from the jurisdiction of
logic to that of actual usage. Fr. “je m'en souviens” was only
illogical so long as the original meaning of souvenir was still felt —
but at that time people still said “il m'en souvient,” and the
new construction is the outward symptom of the fact that the
meaning of the verb has changed (cp. the change from me dreams
to I dream): when souvenir has come to mean ‘have in one's
memory’ instead of ‘come to one's memory,’ the new construction
is the only one logically possible. The paragraphs devoted in
344Ch. XXIV to double negation also show us the applications of
mistaken logical notions to grammar, and our conclusion if not
that logic cannot be applied to grammatical questions, but that
we should beware of calling in a superficial logic to condemn what
on a more penetrating consideration may appear perfectly justifiable.
On the other hand, of course, logic is of the greatest value for the
building up of our grammatical system and for the formulation of
our grammatical rules or laws.

The study of linguistic history is of the utmost importance
to the grammarian: it broadens his mind and tends to eliminate
that tendency to reprobation which is the besetting sin of the non-historic
grammarian, for the history of languages shows that changes
have constantly taken place in the past, and that what was bad
grammar in one period may become good grammar in the next.
But linguistic history has hitherto perhaps been too much occupied
with trying to find out the ultimate origin of each phenomenon,
while disregarding many things nearer our own days which are still
waiting for careful investigation.

Grammatical phenomena can and should be considered from
various (often supplementary) points of view. Take the concord
between a substantive and its adjective (in gender, number and
case) and between a subject and its verb (in number and person).
The traditional grammarian of the old type states the rules and looks
upon deviations as blunders, which he thinks himself justified in
branding as illogical. The linguistic psychologist finds out the
reasons why the rules are broken in this or that case: it may be
that if the verb comes long after its subject, there is no more mental
energy left to remember what was the number of the subject, or that
if the verb precedes the subject, the speaker has not yet made up
his mind as to what the subject is to be, etc. The historian examines
his texts over various centuries and finds a growing tendency
to neglect the forms distinctive of number, etc. And then the
linguistic philosopher may step in and say that the demand for
grammatical concord in these cases is simply a consequence of
the imperfection of language, for the ideas of number, gender
(sex), case and person belong logically only to primary words and
not to secondary ones like adjective (adjunct) and verb. So far,
then, from a language suffering any loss when it gradually discards
those endings in adjectives and verbs which indicated this agreement
with the primary, the tendency must, on the contrary, be
considered a progressive one, and full stability can be found in
that language alone which has abandoned all these clumsy remnants
of a bygone past. (But don't let me be tempted to say
more of this than I have already said in the fourth book
of Language.)345

My concern in this volume has been with what might be called
the higher theory of grammar. But it is clear that if my views are
accepted, even if they are accepted only partially, they must have
practical consequences. First they must influence those grammars
that are written for advanced students (the second volume of my
own Modern English Grammar already bears witness to this influence,
as does August Western's Norsk Riksmaalgrammatik); and
through such grammars the new views may also in course of time
penetrate to elementary grammars and influence the whole teaching
of grammar from the very earliest stage. But how that should be
brought about, and how many of the new views and terms may
advantageously be adopted in primary schools — those are questions
on which I should not like to pronounce before I have seen how
this book is received by those scholars to whom it is addressed. Let
me only express the hope that elementary teaching of grammar in
future may be a more living thing than it has been up to now, with
less half-understood or unintelligible precept, fewer “don't's,”
fewer definitions, and infinitely more observation of actual living
facts. This is the only way in which grammar can be made a
useful and interesting part of the school curriculum.

In elementary schools the only grammar that can be taught is
that of the pupils' own mother-tongue. But in higher schools and
in the universities foreign languages are taken up, and they may
be made to throw light on each other and on the mother-tongue.
This involves comparative grammar, one part of which is the
historical grammar of one's own language. The great vivifying
influence of comparative and historical grammar is universally
recognized, but I may be allowed to point out here before I close that
the way in which the facts of grammar are viewed in this volume
may open out a new method in comparative grammar, or a new
kind of comparative grammar. As this subject is always taught
now, it starts from the sounds and forms, compares them in various
related languages or in various periods of the same language in
order to establish those correspondencies which are known under
the name of phonetic laws, and to supplement them by developments
through analogy, etc. In the scheme given above in
Ch. III, this means starting from A (form), and proceeding to
B (function) and C (notion or inner meaning). Even Comparative
Syntax goes in the same direction, and is tied down by forms, as
it is chiefly occupied in examining what has been the use made by
different languages of the forms and form categories which Comparative.
Morphology has ascertained. But we can obtain new and
fruitful points of view, and in fact arrive at a new kind of Comparative
Syntax by following the method of this volume, i.e. starting
from C (notion or inner meaning) and examining how each of the
346fundamental ideas common to all mankind is expressed in various
languages, thus proceeding through B (function) to A (form).
This comparison need not be restricted to languages belonging to
the same family and representing various developments of one
original common tongue, but may take into consideration languages
of the most diverse type and ancestry. The specimens of this
treatment which I have given here may serve as a preliminary
sketch of a notional comparative grammar, which it is my hope
that others with a wider outlook than mine and a greater knowledge
of languages may take up and develop further, so as to assist us
in gaining a deeper insight into the innermost nature of human
language and of human thought than has been possible in this
volume.347