CTLF Corpus de textes linguistiques fondamentaux • IMPRIMER • RETOUR ÉCRAN
CTLF - Menu général - Textes

McCawley, James. Philosophy of Grammar – T02


by James D. McCawley

I began my introduction to the University of Chicago Press's reprint of
Jespersen's Analytic Syntax by saying that with its reappearance in print I
could cross it off the top of my list of important out-of-print linguistics
books. The book that now holds that position of honor is the same author's
The Philosophy of Grammar, and I am delighted to play a hand in
speeding its removal from that particular pedestal: The Philosophy of
(henceforth, PG) belongs in an easily accessible edition that
will enable new generations of readers to experience the pleasure and awe
that it has aroused in earlier readers. 1

Otto Jespersen (1860-1943) spent his entire academic career at the
University of Copenhagen, first as a student and then as a professor (with
brief stints as dean of the faculty of arts and as rector), though he traveled
extensively (including a year as a visiting professor at Columbia University
in New York, in 1909-10) and had contacts the world over. His list
of publications is unusually long (he published prolifically for sixty years)
and covers virtually every area of linguistic research. The amazing thing
about Jespersen's output, however, is not its prodigious size but the diverse
range of fields in which he published works whose striking originality
makes them still major sources of insight and information as the
twentieth century nears its end. Jespersen was decades ahead of his time
in the fields of syntax, sociolinguistics, language acquisition studies, language
pedagogy, phonetics, and the theory of language change, and his
contributions to historical linguistics, especially to the history of English,
were monumental.

In PG, Jespersen provides a critical examination of a large number of
grammatical concepts, distinguishing clearly among notions that linguists
often confuse even today (such as the notions of “subject,” “topic,”
and “agent,” which even Leonard Bloomfield couldn't keep straight) and
paying attention to a far broader range of examples than scholars of his
time or even of ours were accustomed to dealing with (as when he discusses
examples such as Catch me going there in his chapter on “Negation”).
He notices the implications of details in standard definitions and
shows how they often misrepresent the facts, e.g., traditional definition
of third person as “person or thing spoken of” is inaccurate, since one
1still uses first and second person forms when speaking about oneself or
one's addressee; here an accurate definition must be negative: “does not
include speaker or addressee,” and even that definition must be qualified
to allow for forms such as your humble servant and madame that are “notionally”
first and second person but “formally” third person. The thoroughness
of Jespersen's coverage of many classes of syntactic constructions
in English is astonishing, as in his treatment of full and reduced
sentential complements in chapter 9. There are also gratifyingly thorough
treatments of syntactic constructions that linguists have largely ignored,
as in his brief but highly comprehensive treatment of “comparative
conditional” sentences such as The more he gets, the more he wants,
which includes the only successful attempt that I am aware of to assimilate
the semantics of sentences such as It grew darker and darker to that of
comparative conditionals.

The unusual amount of effort that Jespersen put into PG is reflected
in the fifteen years that he spent working on it and the three changes of
title (for which, see Jespersen's preface) that it underwent during its long
gestation. As in his Language, the other large semipopular work that he
completed as he was approaching his retirement in 1925 from the faculty
of the University of Copenhagen, he cannibalized many of his earlier
works in writing it. Each of the chapters deals with a topic on which he
could easily have written an entire book, and in one case he already had
written one: his 150-page 1917 monograph, Negation in English and Other
, of which chapter 24, “Negation,” is a distillation.

In his reconsideration of grammatical concepts, Jespersen was particularly
concerned with identifying the “notional” categories that they corresponded
to, and with the details of the mismatches between the various
“formal” categories (such as “plural” and “masculine gender”) and their
“notional” counterparts (such as “referring to more than one entity” and
“referring to male beings”). His view of notional categories is quite reminiscent
of views that Benjamin Lee Whorf (1897-1941) was shortly to
develop, in that he regarded notional categories as existing only in virtue
of the role that they play in languages and as being open to interlinguistic
variation: “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” (p. 57).

Jespersen's discussion of mismatches between formal and notional categories
2goes far beyond merely noting obvious points such as that all
three genders in German include words denoting men, words denoting
women, and words denoting things other than human beings. He takes
up cases in which a word simultaneously shows the behavior of two different
genders — as in the Swedish Statsrådet är sjuk/*sjukt, ‘The councillor
of state is sick,’ where the noun Statsråd still takes the neuter article
-et that råd, ‘advice,’ calls for, but where the predicate adjective must be
in the masculine, not the neuter, form (p. 231) — and inquires into factors
that affect the likelihood of rectification of a mismatch, noting, e.g., that
mass nouns have an even stronger tendency than other nouns with inanimate
referents to take on neuter gender. The Swedish example shows
that where the factors relevant to determining an item's gender have conflicting
implications, different syntactic contexts may weight the factors
differently. Jespersen's conception of notational categories also takes in
notional counterparts of particular syntactic constructions, as when he
speaks of That book sells well, The door opened, and respectable as “notional
passives” and of an hour before sunrise, outlast, prefer A to B, and my senior
by two years
as “notional comparatives.”

PG provides ample evidence that Jespersen's thinking about syntax was
far ahead of its time. Indeed, syntacticians who are familiar with the
scholarly literature since the 1960s will see many of the discoveries of
their and their teachers' generations presented with admirable clarity in
PG. The evidence that he offers in support of his various conclusions and
analyses is often of types that syntacticians did not regularly employ until
well into the 1960s, as when he cites the form of the reflexive in such
expressions as to deceive oneself as evidence that they have an understood
“generic” subject (p. 143) and uses the form of the tag in And bring out
my hat, somebody, will you?
as grounds for saying that imperatives and
vocatives are in the second person (p. 214). Historians of linguistic ideas
will observe with interest that the conception of the lexicon as the repository
of a language's irregularities, which is commonly associated with
Bloomfield, appears here (p. 32), attributed to Henry Sweet (1913:31), as
does the Prague-school notion of “new information” (pp. 116, 145); and
Jakobson, in his celebrated paper on “shifters” (1957) credited Jespersen
with introducing both the notion and the term in PG.

Jespersen's chapters on parts of speech are one of the few places in the
entire vast corpus of his writings where his usual insight and perceptiveness,
though still abundantly in evidence, are diluted by some surprising
sophistry. He is able to argue for a system of five parts of speech only by
stacking the cards in favor of a system having few distinct parts of speech:
he accepts some distressingly weak arguments for assigning words to the
same part of speech and never makes clear how one could argue for assigning
two words to different parts of speech. His argument that numerals
3should be treated as a subclass of pronouns is that the indefinite article
is a weak form of one and “if its counterpart the ‘definite article’ is justly
reckoned among the pronouns, the same should be the case with a, an,
Fr. un, etc.” (p. 85). While a strong case can be made for the proposition
that definite articles are pronouns, Jespersen evidently had in mind not
the sort of synchronic syntactic case that Postal (1966) makes for it but
merely the diachronic morphological identity of definite articles with
some sort of pronouns; moreover, the argument depends on Jespersen's
gratuitous assumptions that the different kinds of pronouns all belong to
the same part of speech and that a/an necessarily belongs to the same part
of speech as the word of which it is in some sense a “weak form.” He goes
on to remark that “to establish a separate ‘part of speech’ for the two
‘articles’, as is done in some grammars, is irrational,” but he never says
why that is irrational; I suspect that here Jespersen was simply expressing
the remarkably common prejudice that the number of parts of speech
must be small (a prejudice that is as unfounded as the once equally common
prejudice that the number of chemical elements must be small) and
its corollary, that the number of items that belong to any given part of
speech should be large.

Jespersen sometimes supports his decisions to put things into the same
part of speech with arguments based on positive characteristics shared by
the items in question, and those arguments are generally solid and perceptive,
as where he argues (pp. 88-89) that words such as before belong
to the same part of speech irrespective of whether they are used as prepositions
(before breakfast), as “adverbs” (He had been there before), or as
“subordinating conjunctions” (before he had breakfasted). 2 However, in
setting up his category of “particles (comprising what are generally called
adverbs, prepositions, conjunctions — coordinating and subordinating —
and interjections)” (p. 91), Jespersen has adulterated such well-supported
conclusions by combining them with uncritical acceptance of traditional
categorizations (in which coordinators such as and and or are conflated
with subordinators such as while and because in a category of “conjunctions”)
and vague considerations of “convenience” (e.g., “Those interjections
which cannot be used except as interjections may most conveniently
be classed with other ‘particles’” [p. 90]).

While Jespersen raises solid objections to most of the most popular
definitions of the various parts of speech, he never takes up the fault that
disqualifies them from yielding a system of parts of speech, namely that
(as in the Chinese zoological taxonomy of Borges's “The Analytical Language
of John Wilkins”, 1964) the categories defined don't belong in the
same classificatory scheme: “noun” and “verb” are commonly defined
in terms of the denotations of words of those categories, “adjective”
and “adverb” in terms of what kinds of things they modify, “pronoun”
4in terms of what the words of that category “stand for,” “interjection” in
terms of the roles of those words in discourse structure, etc. None of the
authors whose definitions of part-of-speech terms Jespersen criticizes applied
any classificatory principle systematically; e.g., by recognizing a different
class of modifiers for each category of expressions that could be
modified (rather than lumping all modifiers other than adjectives into the
category to which the term “adverb” is misleadingly applied), or by recognizing
a different class of “pro-forms” for each category of expressions
that allowed replacement by a pro-form. Jespersen indeed toys with the
latter possibility and, after going through a number of examples, says “in
this way we should get a class of substitute words which might be subdivided
into pro-nouns, pro-adjectives, pro-adverbs, pro-infinitives, proverbs
(and pro-sentences as so in the second instance above)” but retreats,
saying that the class of pro-forms “could hardly be called a real grammatical
class” (p. 83). It is not clear, though, why he did not take the alternative
step of recognizing a class of pro-forms as a subclass of each of the
categories that he did recognize.

Jespersen accepted uncritically the popular practice of applying the
term “pronoun” not only to personal pronouns and demonstratives but
also to interrogative, “indefinite,” and relative pronouns, and criticized
the definitions of earlier authors for failing abysmally to characterize the
class of words to which the term was generally applied: “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” (p. 82). In view of
Jespersen's unwillingness to distinguish sharply between morphology and
syntax (chapter 2), one might have expected him to take up the possibility
that the -body of nobody is literally a pro-noun (standing for, e.g., “person”),
combined with the same no as in No man is an island, and that nobody
has been misclassified by those who put it in the same category as he and
this, but somehow he did not.

Fortunately, though, the ideas that do the most work for Jespersen are
not his dubious conclusions about parts of speech but his notions of
nexus and rank, and these ideas he exploits with great flair. A nexus is a
combination of something with items that stand in grammatical relations
(such as “subject” or “object”) to it. 3 The “head” of a nexus need not be
the predicate element of a full clause (e.g., it can be a nominalization, as
in the doctors cleverness, or the predicate element of a “small clause,” as in
He makes her happy), and by allowing the notion of nexus to be largely
independent of notions of part of speech or of clause structure, Jespersen
became able to make use of it in a range of cases that go far beyond
common invocations of “subject” and “object,” as in his section “Nexus
without a Verb” (pp. 120-22), where he takes up such examples as Small
wonder that we all loved him exceedingly
, and in his treatment of No news is
5good news
(which he contrasts with No news arrived on that day) in terms of
a reduced subordinate nexus in which no is the head and news its subject
(p. 126).

The notion of dependency that has traditionally figured in syntax is a
notion of dependency between words (e.g., in Johns elder brother lives next
, brother was said to be the subject of lives, and not John's elder brother
the subject of lives next door; cf. Percival 1976). At the time when Jespersen
completed PG, linguists such as Leonard Bloomfield, partly under
the influence of the psychologist Wilhelm Wundt (1832-1920), had
started doing syntax in terms of immediate constituent structure, i.e.,
describing sentences and other multi-word syntactic units by specifying
what smaller syntactic units each was made up of, without regard to dependency
relations. Jespersen's conception of syntactic structure (which
he made more explicit in such later works as his Analytic Syntax) gave to
notions of dependency (i.e., to ways in which one linguistic unit could
“complete” or “modify” another unit) the centrality that they had had in
traditional grammar, in that “dependency” is presupposed by his notion
of “rank,” but rejected the traditional notion of single words as being the
units between which dependency relations held. The fact that the notion
of “word” did not have a central place in Jespersen's conception of syntax
is illustrated by his response to Sweet's descriptions of utterances such as
Splendid! as “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”
(p. 306). 4 One possible vestige of the traditional idea of single words as
the only (or at least, the main) units relevant to syntax is something that
I regard as the sole major defect in Jespersen's writings on syntax,
namely his failure to draw the distinction between noun and noun phrase.
In his discussion of “ranks” both here and in Analytic Syntax, he uses the
term “primary” (or the symbol “1”) indiscriminately for nouns and for
NPs and thus fails to distinguish, e.g., between those pronouns that
“stand for” an NP (personal and demonstrative pronouns) and those that
“stand for” a noun (one, as in his own example a grey horse and two black
). 5

At the time Jespersen wrote PG, the most prestigious kind of linguistics
was historical linguistics, and Jespersen felt obliged to provide a defense
of descriptive linguistics against those who accorded it little prestige
(pp. 30-31). He described the relationship between descriptive and
historical linguistics as like that in physics between statics and dynamics,
and maintained that neither enterprise can get very far without the other.
While Jespersen deals primarily with synchronic questions in PG, there
6are numerous places where he takes up diachronic issues, and here his
analogy between historical linguistics and dynamics becomes particularly
apt, since his concern is with identifying the factors that cause languages
to change in particular ways, as in his explanations of cross-linguistic differences
such as that in the use of singular and plural forms in the eighteenth
and nineteenth centuries
, les siècles dix-huitième et dix-neuvième, and
das achzehnte und neunzehnte Jahrhundert (p. 190) in terms of ways in
which the different word order systems and agreement systems of the
three languages would make the choice of a singular or a plural form be
felt as incongruous.

A “dynamic” perspective on Jespersen's “statics” is provided by his
perceptive discussion of the factors that motivate the use of different constructions,
as in his discussion of the passive, where he notes a highly
heterogeneous set of factors that motivate its use: in a passive, the by-phrase
can be omitted, and thus the construction is favored by any of the
diverse factors that provide grounds for omitting the underlying subject
(it isn't known who fills that role; or it's obvious who filled it and thus
superfluous to mention it; or it would be immodest for one to mention
one's own role); by changing what the subject is, the passive construction
opens up different possibilities for combining elements in coordinate
structures; and since a subject can most readily be interpreted as what
Kuno (1987) subsequently called an “empathy focus,” passive is favored
when the derived subject is something denoting a person or thing with
which one wishes to express “empathy.”

To help modern readers of PG to utilize it to the fullest, I have prepared
a new index to replace the quite unsatisfactory index of the original
edition, and have supplemented Jespersen's own terminology (some of
which has never gained any currency) by entries reflecting the terms by
which modern readers know many of the topics that Jespersen discusses.
Thus, there are entries for “complementizer,” “dislocation,” and “Equi-NP-deletion”:
these are notions that Jespersen took up, albeit under different
names or under no particular name at all, and an index to a modern
reprint of PG ought to enable readers to look up Jespersen's thoughts on
those topics even if they are not very conversant with the terminology
that Jespersen often used. I should add, though, that Jespersen was the
greatest coiner of terminology in the history of linguistics, and dozens of
modern linguistic terms, ranging from “existential sentence,” “extrapositon,”
“tag question,” and “resumptive pronoun” to “Great Vowel Shift,”
were introduced by Jespersen. Readers may derive some amusement
from looking through the index and trying to guess which terms were the
ones that Jespersen actually used. The index of names reveals a number
of things that were not apparent from the original edition's skimpy index,
notably the sheer length of the list of names, which reflects Jespersen's
7formidable erudition, and the frequency with which he cites certain
scholars, especially Henry Sweet and Adolf Noreen, whose influence on
Jespersen is worth exploring. 6

But I don't mean to end this introduction on a note of scholarly piety.
Besides being one of the most perceptive observers and original thinkers
that the field of linguistics has ever known, Jespersen was also one of its
most entertaining writers, and reading PG is fun. Read it, enjoy it, and if
you have a serious interest in language, keep it where you will easily be
able to find it on the many occasions when you'll want to refer back to it.8


Borges, Jorge Luis. 1964. “The Analytical Language of John Wilkins.” In Borges,
Other Inquisitions. Austin: University of Texas Press, 101-05.

Jakobson, Roman. 1957. Shifters, Verbal Categories, and the Russian Verb (Cambridge,
Mass.: Harvard University). Reprinted in Roman Jakobson, Selected
, vol. 2. The Hague: Mouton, 1971, 41-58.

Jespersen, Otto. 1917. Negation in English and Other Languages. Hist.-Fil. Meddelelser
1, 5. Copenhagen: Det kgl. danske Videnskabernes Selskab.

Jespersen, Otto. 1922. Language, Its Nature, Development, and Origin. London: Allen &

Jespersen, Otto. 1937. Analytic Syntax. Allen & Unwin. Reprinted 1984, Chicago: University
of Chicago Press.

Kuno, Susumu. 1987. Functional Syntax. Chicago: University of Chicago Press.

Percival, W. Keith. 1976. “On the Historical Sources of Immediate Constituent
Structure.” In McCawley (ed.), Notes from the Linguistic Underground. Syntax
and Semantics 7, New York: Academic Press, 229-42.

Postal, Paul M. 1966. “On So-Called ‘Pronouns’ in English.” Georgetown University
Roundtable on Languages and Linguistics
, 177-206. Reprinted in D. Reibel
and S. Schane (eds.), Modern Studies in English. Englewood Cliffs: Prentice-Hall,
1969, 201-24.

Sweet, Henry. 1913. Collected Papers. Oxford: Oxford University Press.9

1.voir However, the reappearance of PG does not remove Jespersen's name from
the top of my list of out-of-print classics of linguistics: with the appearance
of this reprint of PG, his Language moves up from second to first place on
that list.

2.voir One particularly good argument that Jespersen gives for this identification is
the possibility of using the word both ways at the same time in coordinate
examples such as Thackeray's “After the Baden business, and he had dragged
off his wife to Champagne, the Duke became greatly broken” (p. 89).

3.voir The centrality of nexus and grammatical relation to Jespersen's thinking
about syntax makes his approach in many ways a forerunner of Relational

4.voir Jespersen classes a broad range of “sentence fragments” as sentences: “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” (p.
311). He draws the line only at such things as “signboards (‘J. C. Mason,
Bookseller’), book-titles (‘Men and Women’), head-lines in newspapers
(‘New Conference 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.” However,
he notes that “all of 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.”
I do not share Jespersen's view of writing as “outside language
proper,” nor his confidence that nonsentential utterances are confined to
writing, e.g., I see no grounds for denying the status of sentence to “Taxi”
painted on the side of a car that would not also apply to “Taxi!” shouted in
an attempt to hail a cab.

5.voir Actually, what one stands for is not a noun but an ‘N’: if a noun has any complements,
it is the combination of noun and complements that is available to
serve as antecedent of one, as in His old theory of syntax made more sense than his
new one

6.voir Readers who wish to see where Jespersen went from here in his thinking on
the many topics that he takes up in PG are urged to consult pages 156-57 of
Analytic Syntax, where he gives the pages on which he reexamines various
matters, major and minor, that he dealt with in PG.