Stetson, Raymond. Motor Phonetics – T01

I
The basic conceptions of phonology

1. The syllable as the basic unit

For a motor phonetics the primary concern is the
train of movements of speech: the articulations and
the larger movement units which incorporate the
articulations. Speech can be indicated for a native
by a row of signs in a line; but the signs give no
hint of the pattern and connections of the articulatory
movements. (Note 1)1

One fundamental contrast has survived all forms
of notation and all sorts of theories, the contrast
between the vowel and the consonant. There have
been occasional efforts to whittle down the contrast
or to be rid of it; but the very people who try to
ignore the contrast use the distinction at every
turn. One common pair of terms, “syllabic and
non-syllabic” recognizes the function of the vowel
(and vowel substitutes) in the syllable.

It has seemed to some-phonologists that a row
of “phonemes” is easier to handle than a train of
syllables. But the syllable is always to be considered;
it embodies the syllabic and the non-syllabic;
it is the unit for the word- and the sentence-stress;
it is the unit for the ‘tones’ of the tone languages;
it is the stressed and unstressed unit of the feet of
patterned verse. The syllable is the unit which
incorporates the syllable factors, and in turn is
incorporated in the units of the foot and breath
group.

The mechanism of speech is often compared to
the action of an organ: the chest is the bellows,
furnishing air under pressure to the vocal folds
which functions as the vibrating reed; the throat-
and mouth-cavities constitute the resonating pipe,
and the glottis or the consonant constriction acts as
the palate (valve) to start and stop the flow of air
through the vocal folds.

Careful observation shows that the action of an
organ, bellows, reed, pipe, must be radically modified
to fit the mechanism of speech. When the chest
is slightly inflated for speaking, the air is not under
pressure; like a hand bellows for blowing the fire,
the volume is increased, but the nozzle is open and
there is no flow of air; so the mouth and the glottis
may be open as in whispering, but there is no flow
of air from the inflated chest.

If one makes quick strokes of the hands while
holding the inflated hand bellows, the nozzle emits
little pulses of air; as the quick strokes are repeated
the air pulses reduce the volume of air, and the
arm muscles must bring the boards of the bellows
closer and closer to accommodate for the loss of
air. There has been no valve action; no palate has
started or stopped the flow of air.

If the hand bellows is connected with the reed
of an artificial larynx, the reed will sound for each
air pulse from the nozzle; it is the quick movements
of the hands which release and arrest the air pulse
voiced by the reed.

In much the same way the vocal apparatus makes
a series of vowel syllables, “Oh, Oh, Oh…”. The
chest is inflated by the larger muscles, the quick
strokes for each air pulse are made by the short
muscles between the ribs, and as the chest volume
gets less and less, the larger outside muscles and
the abdominal muscles accommodate the walls and
floor of the chest to the changing volume.

The muscles between the ribs (intercostals) produce
1the syllables “Oh, Oh, Oh…” as units
included in the slower movement of the breath
group which is made by the larger muscles of chest
and abdomen. The rapid muscle contractions are
like ripples on the wave of the expiratory movement
of the breath group.

No air pulse from the nozzle of the hand bellows
can be made without inflating the bellows and
maintaining the position of the boards of the
bellows while the hands make the little strokes for
the air pulse. No syllable can be uttered without
inflating the chest and maintaining the chest-abdomen
position while the rib muscles (intercostals)
make the quick strokes of the syllable. In
other words a single syllable, “Oh” must be part
of the movement of a breath group. The air pulse
for the syllable “Oh” is released and arrested by
the rib muscles, but the supporting coordination
involves the large ribcage muscles, the diaphragm,
and the abdominal muscles for this one-syllable
breath group.

The hand bellows connected to an artificial
larynx is not fitted with anything to correspond
to the constriction of a consonant in the syllable.
What happens when “beau, ope, pope” are uttered?
Experiment shows that in uttering “beau, pope”
the lips close with the contraction of the rib
muscles for the syllable pulse, and open to release
the syllable pulse and the air flow sets the glottal
folds in action for the vowel. In uttering “ope,
pope” the lips close at the end of the syllable,
arresting the pulse of the syllable and stopping
the air flow through the glottis.

Careful experimentation with all the types of
syllables and consonants makes it certain that every
syllable has its chest pulse delimited by the chest
muscles (intercostals) or by the constriction (complete
or partial) of the consonant, or by both.

The syllabic pulse, like any rapid movement,
consists of two strokes, a beat stroke and a back
stroke; the beat stroke releases the pulse with or
without the auxiliary consonant movement; the
back stroke arrests the syllabic pulse with or without
the auxiliary consonant movement. (Note 2)2

The consonant movement is always an auxiliary
movement. The consonant functions only in a
syllable. Recordings show that demonstrators and
teachers who believe that they are producing
isolated “sounds” are actually producing syllables.
A chest pulse is necessary; there is cither a silent
vowel which the supposedly isolated consonant
actually releases or arrests, or a continuant that
provides the vocal-canal shape through which the
pulse is emitted.

It is impossible of course to make a given vowel
without releasing and arresting a pulse from the
chest; shaping the vowel canal prepares for the
syllable pulse of the vowel but the pulse must occur
if there is to be a vowel. The lips or tongue may
be in position to make the releasing or arresting
stroke of a given consonant but that does not
constitute a consonant; the consonant is produced
when the stroke releases or arrests the air pressure
of the syllable pulse. The vowel is an articulation
which has the function of shaping the vocal canal
for the chest pulse; the consonant is an articulation
which has the function of delimiting the chest
pulse of the syllable. It is only in the coordination
of the syllable pulse that they act as vowel and
consonant. They may be named as if separate, but
that is merely a convenient abstraction. A vowel
is a specific quality of a syllable when uttered
through a specific vocal-canal shape; a consonant
is a specific way of releasing or of arresting a
syllable. Details of the processes involved come
later.

Any one of these syllable factors may be characterized
by one or several significant aspects; these
are the phonemes. A syllable may have no significant
aspect, no phoneme, or it may have as many
as seven significant aspects of the syllable factors,
phonemes. There are compound consonants of two
or three aspects, and there are compound vowels,
diphthongs (and a few triphthongs). Details of the
characteristics and processes involved in compound
syllable factors come later. Cf. Appendix II: Artificial
Larynx; Appendix III: Historical Aspects of
the Syllables.

2. The foot and breath group as related to the
syllable

Often the phoneme has been considered without
reference to the syllable; and often the syllable has
been considered without reference to the foot,
breath group and phrase in which the syllable is
always incorporated. The syllable, the foot, the
breath group, and the phrase, are elaborations of
the movements of expiration. They must be audible;
if the expiration is to be widely audible it must be
through the mouth, and nasal outflow will be
secondary along with the minor noises of articulation.

It is rather common but false to assume that the
consonants and vowels float on the continuous
pressure of the breath group. The chest is mistakenly
thought of as the wind chest of an organ
maintaining a steady flow under pressure throughout
the breath group, interrupted by the constrictions
(partial or complete) of the consonants, and
colored by the shaping of the vowels; the working
of the artificial larynx controverts such a notion.

Experimental study of the speech mechanism
shows on the contrary than the chest does not
maintain a steady pressure throughout the breath
group. Instead, the chest muscles (intercostals)
produce a separate pulse of pressure for each
syllable; the pressure falls between the syllable
pulses. The chest retains an overall posture which
maintains its volume of air, but the minute contractions
of the short muscles between the ribs
(intercostals) force out little pulses of air which
constitute the separate syllables. Even in a language
like Hottentot in which suction “clicks” are prominent,
there occur the usual chest pulses of the
syllables grouped in the usual breath group. (Note 3)3

A slight pressure is generally maintained during
the breath group, at least in English utterance, but
the chest pulses of the syllables rise from this level.
Between the breath groups the pressure goes to zero
although the over-all posture and the chest volume
are maintained; if there is an intake between breath
groups, the pressure becomes negative.

Careful experimentation proves the breath group
to be due to an abdominal movement with its culminations
which mark the stresses of the constituent
feet. One of these stresses constitutes the main stress
of the breath group, while the syllable pulses are
produced by the intercostal muscles of the rib
cage.

The “foot” is the smallest unit group incorporating
the syllables; it is due to an abdominal pulse
which integrates a single stressed syllable or a few
syllables grouped about a single stressed syllable.

After each movement of expiration for the breath
group an intake may or may not occur; but there
is always a readjustment to the slight change in
volume of the chest. As the capacity of the chest
decreases, the rib cage descends slightly arid the
abdominal muscles raise the diaphragm slightly to
compensate for the outgo of air. Cf. Fig. s 50, 57.

Any utterance however simple has all the essential
movement units: the syllable is an integration of
3the three syllable factors; the syllable or syllables
are components of a foot; the foot or feet are
components of a breath group; the breath group or
breath groups compose a phrase. Thus the simplest
utterance will be a phrase of one breath group consisting
of a monosyllabic foot. The experimental
study of speech movements shows that all these
units are actually present in the utterance of a single
syllable, and of any group of syllables.

A. Movement units in speech

There has been some general recognition of the
articulation as the fundamental event in phonology,
rather than the acoustic pattern. These articulatory
events which figure as signals and symbols in speech
must be classified and handled with reference to
their actual function, in the processes of utterance.
The various lists and classifications which prove
useful in phonology are derived from the functions
of consonants and vowels in the syllables of speech.

Articulate language involves the two phases: 1)
A universal apparatus for speech common to all
speakers; and 2) A specific, acquired language
mechanism peculiar to the given language. The
experience in learning foreign languages shows that
the apparatus for speech is universal; any child
learns any language or pair of languages with the
same ease. The program of a general phonemics
based on comparative phonetics proposed by Trubetzkoy
(Anleitung, '35) (98) and the project of a
universal auxiliary language assume that all speakers
have a common apparatus for speech. But there are
also the peculiar characteristics of the specific native
language which are apparent in the “accent” of the
adult foreigner.

The “acquired language mechanism” of a specific
language is evidently something more than a row
of special “sounds”. The practical teaching of
language as well as the needs of linguistic analysis
have led to the consideration of larger and larger
units.

It is impossible to teach the “separate sounds” of
a language (the sounds cannot be separated), and
then assembled into syllables (syllables are not
assemblies), and the proper stresses, pauses, and
“intonation” added (they are not additions but basic
over-all traits). Everyone realizes that “word stress”,
“phrase and sentence stress”, “intonation” all
involve units of articulation larger than the supposed
“sound”, or the syllable. And everyone realizes
that the smallest units of articulation are parts of
the larger system of utterance; everything hangs
together. Concepts like “base d'articulation”,
“Mundstellung” grow out of the fact that the
“accent” of a language involves the whole language
mechanism.

Every utterance is a movement consisting of the
phrase which is the larger, inclusive unit of which
the breath group, foot, and syllable are organic
parts. For this movement the posture is the adjustment
of the abdomen-diaphragm and external chest
muscles for regulated expiration; the movements
of the breath group, and of the feet within the
breath group are due to the abdomen-diaphragm
movements of expiration. The series of small
expiratory movements, the breath groups, constitute
the phrase. The phrase is followed in the breathing
cycle by a rapid inhalation. Although the utterance
may consist of only one or two syllables, all the
fundamental movement units are present.

The feet and the breath groups are long enough
to have varying culminations in the stresses and
grouping for the “word accent”, and for the breath-group
stress. The syllable however is a single ballistic
movement; it is impossible to have two stresses
within one syllable. (Note 4)4

The various boundary markers (“Grenzsignale”),
stress and intonation patterns which have been
noted, are not independent traits, appearing isolated
as members of a series of symbols; they are rather
cues to these basic, coordinated movement units
which make up connected articulate speech. (Note 5)5

B. Summary of the movement units of a motor
phonetics

Even though the utterance consist of only one
syllable, it constitutes a foot of a single syllable, a
breath group of a single foot, and a phrase of a
single breath group; the posture and the abdomen-diaphragm
movements, as well as the articulations
of the single syllable all actually occur, as observations
of boundary markers, stresses, etc., and
laboratory tracings from the speech apparatus,
demonstrate. (Note 6)6 Cf. Appendix V: Conditioning
and the units of speech; Appendix IV: Earlier
stages of modern phonetics; Appendix VI: Reduction
of a language to phonemes.

3. Phonetics and phonemics

Since the concepts of the “phoneme” and of aspects
of speech which are “phonemic” are current, it is
important to contrast “phonetics” and “phonemics”
(“phonologie” on the Continent).

The phonemic method of analysis is an important
method and often applicable, but it is an analysis
of speech to determine the signals (symbols) and
not the process by which signals are produced.
Language has to do primarily with meanings; the
syllables are primarily meaningful syllables. They
are products of the system of habitual movements
of speech in a given language. The articulations and
their organization for expressing meanings in the
given language constitute the “language mechanism”.

Therefore the signals with meaning are primary;
the syllables are the minimal units. “Phonemic”
means that a given articulatory change of the
syllable affects the meaning. The type of change
and the amount of change necessary to affect the
meaning of a syllable depend on the language. The
users of a language may sense a very small change
as significant, or they may be indifferent to a very
extensive change; and this differs from language to
language, because the changes and their significance
are learned with the language.

It is sometimes said that a ‘phoneme’ is not an
entity, but that it stands for a class; which is true
of any concept in science; the phoneme stands for
a class of syllables which are released in a given
way, or are arrested in a given way, or have the
vowel canal shaped in a given way. And again the
term phoneme is very convenient as a name for the
group of variants which the characterized syllable
factor (phonemic signal) undergoes with the changes
of syllable rate and stress. The given way of releasing,
arresting, or shaping a syllable changes with
rate and stress, and when the original rate and stress
are restored, .these variations revert to the slow
careful form by which the phoneme is identified.
The slow, careful forms of the units in “apart” and
“opinion” make the first syllables unlike; at a rapid
rate both the “o-” and the “a-” approach shwa.
“Telegraph” and “telegraphy”, “monotone” and
“mono'tonous” show changes due to shifting stress.
To distinguish the forms “an icehouse” from “a
nice house” requires a change in grouping of the
foot image for “a nice house”, and image, for
“an icehouse”.

The phonemic method of reducing a language
to its units is an obvious method; but it is not to be
5assumed that reference to the meanings embodied
is the only way of reducing language to its units.
The cryptographer deciphering an intercepted and
recorded oral communication in an unknown
vernacular obviously cannot depend on meanings.
The units must be isolated and assembled to determine
the language and find the meaning. The
paleographer dealing with an undeciphered inscription
is in the same case; the syllables and/or their
factors may be determined while the meaning
remains unknown; so Etruscan and Minoan.

The terms “phonetic”, “sound”, “articulation”
refer to articulate processes of speech without regard
to meaning. The articulations and their variations
and combinations when identified in a given
vernacular constitute the basis for phonetics. The
habits of utterance of the language constitute an
organized system of skilled movements which may
produce meaningful syllables, or may produce
nonsense syllables of the same type.

The purposes of phonemics and phonetics are
often quite unlike. It is convenient to distinguish
the phoneme by a trait or traits which will be
differential. For taxonomic purposes diagnostic
items are important. The context in which the
differentiating traits are perceived and which is
essential to their production and perception may
be ignored. In phonemics identification and classification
are usually primary; and the movement
complex producing the significant trait is hardly
considered or is ostentatiously ignored, although
it is obvious that uttering the phonemes is quite
as important as perceiving the phonemes in this
the most important form of social communication.

A breath group can often be recognized by only
a few distinguishing items in the pattern context
of the breath group with its stress and rhythm. If
these distinguishing items, however, were somehow
presented alone (sometimes possible with proper
apparatus), even though properly timed, they could
not be perceived. And as for producing such
significant items, that is impossible except as
indivisible aspects of syllables in the overall rhythmic
pattern of the breath group.

It may be enough for the classification of significant
traits or for an orthography intended for
native use, to isolate and indicate the significant
items or groups of items simultaneous and successive.
But for understanding the synchronous
changes and the process of producing the significant
traits and their combinations, and for controlling
the learning of a language, the working of the
language mechanism is essential.

Motor phonetics is the study of the skilled
movements involved in the process of handling
articulatory signals. Motor phonetics deals with
the organized series of actual syllables or nonsense
syllables shaped by the language mechanism.
“Phonetic” refers to such physiological processes
in the study of the signals of a language independent
of the meanings of the signals. The signals are
physiological; they are uttered and perceived by
kinesthetic cues from the vocal organs, and by
auditory cues; but they may be uttered and perceived
by kinesthetic and visual cues in the case of
the deaf; the “sounds” are not essentially auditory.
A phonetic change is a mechanical change which
occurs in the syllable due to context and to change
of rate and stress in the utterance. Since these coordinated
movements of articulation are the medium
for the signals of an articulate language, the mechanical,
physiological changes are important and affect
the phonemes. Phonologists have often resorted to
nonsense syllables for convenience in observation
and experiment: F. de Saussure “appa, apta”; Rousselot
“afa, af-fa”; Sievers “alla, ar-a”; recent writers
“pup, aftpa, faf, vav, gug”, etc.

When the language signals of different languages
are compared, the discussion must strictly be
phonetic. There are no consistent meanings and
the syllables compared are in effect nonsense
syllables treated as physiological processes. The
syllables must be handled independent of the meanings,
although meanings gave rise to the syllables in
each language in question. Comparative phonemics
becomes comparative phonetics.

Since phonetic change occurs as a phase of the
acquired language mechanism, it is observable by
6the users of a language though the change may have
no meaning. Phonemic change on the other hand
is a matter of conditioning and may be so slight as
to tax phonetic discrimination, or so extensive as
to be very obvious — if it does not affect the
meaning, however, it is negligible as a phonemic
change.

Phonetic change is physiological, and occurs
within the acquired language mechanism. Phonemic
change is significant; it always involves the conditioned
meaning of the signal. Cf. Appendix VII:
Segmentation. Appendix VIII: Reduction of Trubetzkoy's
Phonologie to Motor Terms. Appendix
IX: American Version of Phonemics (Phonologie).

Notes

II
Methods and apparatus in experimental phonetics

All the theories of phonemics and phonetics assume
articulate language as the medium to be considered.
Even the basic phonemic distinction of “la langue
et la parole” was made within the domain of articulate
language. “Ce qu'on s'imagine prononcer”,
“ce qu'on prononce en réalité” both assume the
process of pronouncing, either in physiological or
mental terms. The practical phonemicists handle
“transcriptions”. One of the last papers of Trubetzkoy
deals with a list of universal “phonemes”
for the utterance of an international, auxiliary
language. Since mentalism has been abandoned in
the main in phonology, articulate language in action
is the concrete medium for observation and for
experimentation which can be repeated.

There are two general approaches to the problem
of phonetic analysis: a) the physiological and b)
the acoustic. Both methods have been employed
in obtaining the materials presented in this book.
The primary emphasis, however, is upon the
physiological analysis of the speech mechanism in
action. The physiological aspects of speech may
be inferred from the data obtained in an acoustical
analysis of the speech sounds, but such inferences
have often led to serious error with regard to the
functioning of the several parts of the complex
speech mechanism. Acoustic recordings are valuable
for some purposes however, and have been used
with success by careful investigators. Both of the
methods will be described in this chapter in so far
as they have been employed in obtaining the data
here presented.

Historically experimental phonetics moved toward
an exact science during the last quarter of the 19th
century. The physiological methods used were
adapted from those already in use in physiological
laboratories. Rosapelly, working in Marey's laboratory,
is credited with having first made recordings
of speech processes by the kymographic method.
He obtained and studied simultaneous records of,
a) speech breathing movements, b) vibrations of the
larynx, c) movements of the lips in articulation,
and d) air pressures from the mouth and nose
during the processes of speaking (66, 71). Oakley-Coles
developed the artificial palate which often
gives important data as to bearings in mouth (49,
50, 58).

Rousselot gives credit to Rosapelly for his
pioneering. Rousselot developed the methods and
became the leader in the field of experimental
phonetics and may be considered the founder of
the science of experimental phonetics. Rousselot's
methods have been adopted with modifications by
everyone working in the field since his time.

Early efforts to handle the problems of phonetics
by acoustic methods were confined largely to the
study of the vowel structure. Helmholtz, Hermann,
Stumpf, and others employing methods then
available, set the pattern for much of the work that
followed. The development of the phonograph
enabled investigators to make relatively accurate
and permanent records of speech sounds which
could be studied at leisure. Thus Marichelle studied
phonographic grooves with a microscope and worked
out a thorough-going analysis of the relationship
of vowels and consonants within the train of
syllables of speech. (50) The development of the
telephone and the practical needs for further knowledge
10concerning the acoustic aspects of speech,
especially as related to telephonic communication,
has given further impetus to research in experimental
phonetics. Advances in technology, especially
in the application of electronics to the problems,
have motivated research in this field.

1. The recording methods applied to the
processes of speech

The apparatus and methods which have been
used to obtain graphic records of the action of
these several parts will be considered separately.11

A. The breathing mechanism in speech

Two aspects of the breathing mechanism have been
recorded in parallel: movements of the chest muscles
for the syllable pulse and movements of the
abdominal muscles for the foot and breath group.

1. Movements of the chest muscles for the syllable
pulse

The rapid pulses of the intercostal muscles for
individual syllables have been generally overlooked
by experimenters. The result is that the existence
of the syllable as a physiological unit has often been
denied or ignored. It is possible to obtain recordings
showing the syllable pulse both in terms of the
direct movements of the muscles themselves and
indirectly in terms of the rapid fluctuations of air
pressure within the chest. Four methods have been
employed in studying this phenomenon. Three of
these rely on pneumatic recording and involve (1)
direct recordings of pressure changes within the
chest from the tracheal tube of tracheotomized
subjects, (2) pressure changes within the chest
recorded by means of a rubber balloon in the
stomach, (3) recordings of the pressure changes
transmitted through the chest walls by means of a
negative pressure applicator placed on the body
wall, and (4) recording of action potentials from the
intercostal muscles during speech. Each of these
methods will be described in some detail below.

(1) Chest pulses recorded from the trachea.
Disease of the larynx may make it necessary to
provide an orifice for breathing below the larynx.
Just above the sternum a tracheotomy tube is
inserted into the trachea through which the subject
breathes.

With such tracheotomized subjects who are able
to speak normally, it was possible to get tracings
directly of the sub-glottal variations of air pressure.
A tight joint, easily separated for taking breath, led
from the tracheal breathing tube to a taut tambour
of ca. 5 cm. diameter. The membrane must resist
pressures of 250-300 mm. water; the most satisfactory
membranes were built of thin rubber layers
cemented together. It is difficult to have such
membranes as responsive at high as at low pressures;
and yet it is important because the oscillations for
the chest pulse may on occasion take place at a
fairly high over-all pressure of the abdominal breath
pulse.

Tracings in great detail of sub-glottal pressures
of this type were taken also from four laryngectomized
subjects speaking with an artificial
larynx.

An artificial larynx can be set up for the normal
speaker, with a source of air pressure controlled
by the hand, with the tube of the artificial larynx
leading the tone and air pressure into the corner of
the mouth. The subject “holds his breath”, i.e. closes
the glottis, while he articulates the syllables, the
pulses of which he makes with the hand. As a
source of air pressure a fire bellows is simple and
convenient. If the bellows does not have a volume
of 3-4 liters, a flask serving as an air chamber
must be let into the line. The process requires an
hour or so of practice but is not difficult. Tracings
from such utterances with manual syllables correspond
to tracings of the speech of laryngectomized
subjects. Such a device is not expensive, and will
settle once and for all the relations between the
“chest pulse” and the “articulations”.

Immediately after recovery from removal of the
larynx a patient cannot utter the slightest sound
by manipulating the articulatory apparatus; yet in
a few hours he can learn to speak again with an
artificial larynx. The air pulses from the trachea
into the artificial larynx can easily be demonstrated
by kymograms. Cf. Fig. 15, CT. When the tube
from the artificial larynx leads the air flow and
tone into the front of the mouth the movements
of the articulations and the shaping of the mouth
produce the consonants and vowels just as they did
before the operation when the air flow and tone
came into the mouth through the intact larynx.
Although the reed of the artificial larynx is not
subject to manipulation, the subject easily learns
to control its vibration by adjusting the pressures
above and below so as to start and stop the air flow
through the reed.

(2) Chest pulses recorded from a gastric balloon.
The subjects swallowed a small rubber balloon
which was then inflated and rose to contact with
the diaphragm; a small rubber tube led to the
recording tambour; the pressure of the inflated
balloon was compensated by connecting a rubber
balloon in a flask outside which was inflated to ca.
200 mm. water pressure; a septum was introduced
into the tube leading to the recording tambour.
The pneumatic tracings of the transmitted pulses
parallel precisely the tracings taken directly of the
air pressure pulses in the trachea. Cf. Fig. 16 GB.

(3) Chest pulses recorded through the body
wall. Indirect records of the changes of pressure
within the chest can be obtained by means of an
applicator placed on the chest-or abdominal wall
and held in position by negative pressure. Tracings
thus obtained parallel very closely those obtained
by means of either the tracheal tube or the gastric
balloon. The method has the advantage of not
requiring specially selected subjects. Tracheotomized
patients are relatively rare, and subjects have to be
trained to swallow the gastric balloon.

The most responsive area for placement of the
negative pressure applicator proved to be the epigastric
region. Here the rectus abdominis musculature
fans out leaving a relatively thin-walled area.
Tracings have been obtained however from various
areas over the entire trunk. A shallow aluminum
cup with a narrow flange at the rim shaped to fit
the contour of the body wall serves as an applicator.
Fig. 1 CE, Fig. 11 E. Aluminum funnels are
convenient material for the construction of such
applicators. The applicator is held in place by a
light negative pressure of 300 to 600 mm. of water
provided by a siphon respirator. Fig. 11, A, C.

Tubing with heavy walls connected the applicator
with the aspirator, and the recording tambour.
Between the applicator and the aspirator was a
tiny check-valve made by grinding smooth the
end of a bit of heavy-walled glass tubing and
attaching a disk of very thin rubber over the lumen
of the tube with a bit of rubber cement applied
at one point. The valve was set to close when the
pressure in the applicator fell below that of the
aspirator. The check valve may be replaced by a
capillary tube. Fig. 11, D.

In the transmission line between the applicator
and the recording tambour a septum of thin rubber
17under tension was introduced into the tube to
withstand the negative pressure; the distended,
elastic septum balanced the negative pressure, so
that the pressure was neutral in the section of the
tube leading directly from the septum to the
recording tambour. This makes a more sensitive
recording system than is possible if the negative
pressure is transmitted to the recording tambour.
A convenient form of septum consisted of a rubber
diaphragm stretched over the mouth of a thistle
tube 3-6 cm. diameter; the thistle tube was inserted
in a funnel and cemented in place with plastocene.
Fig. 11, F. Hudgins have obtained good negative
pressure tracings by connecting the applicator
directly with a pneumodeik (34); the negative pressure
in the line is balanced by a light coiled spring
attached to the metallic diaphragm.

If a fairly large, triangular applicator is attached
to the epigastrium, it is not affected by the gross
movements of the muscles and it is possible to get
rather simple and clear tracings of the chest pulses;
the fixation of the abdominal muscles is likely to
interfere at rapid rates. At the navel level a large,
oval applicator with the longer axis across the
rectus abdominis will often give simple chest-pulse
tracings, unaffected by the large abdominal contractions
which move the applicator bodily without
producing a record.

The detailed correspondence of tracings made
from these negative pressure applicators to the
tracings of the variations of air pressure within the
chest is surprising; not only do the individual pulses
appear but also the groupings of the pulses which
are a marked feature of the air pressure tracings.
Cf. tracings of the air pressure in chest with
negative pressure tracings (87, p. 92-3, 183; 91,
Figs. 17 and 21).

(4) Oscillograph recording of action potentials.
Action potentials from the muscles can be made
to give direct records of the actual contractions of
the intercostals which originate the syllable pulse.
The placement of the electrodes for simultaneous
recording of the muscles involved in speech is shown
in Fig. 14.

With proper amplification it is possible to get
simultaneous oscillographic tracings of the contractions
of the various groups of muscles. For details
of electrodes, amplification, and procedure, cf. 9,
79, 90.

The oscillograms show clearly the separate pulses
of the intercostal muscles, and the grouped contractions,
especially at rapid rates, of the abdominal
muscles (breath groups). (85, p. 9, Fig. 3, p. 11, Fig.4)

There has been an exaggerated notion of the
authority of an oscillogram. Any precision of the
oscillogram must derive from a like precision of
the contraction of the controlled muscles of the
speech apparatus.

2. Movements of the abdominal muscles for the
joot and breath group

(1) Tracings of the positive movement of the
body wall in a given spot.

Such tambours must be supported independent
of the body wall on a rigid frame not attached to
the subject. Various types have been used: chair
(Fig. 12) with heavy metal strap to which supports
could be bolted; and several forms of “stanchion”,
are figured in cuts. The stanchion (Fig. 1) provided
two adjustable points of support for the spine, and a
vertical rod to which the applicators and the tubes
for chest and abdomen could be attached. Such a
frame accommodates the subject standing or sitting.

The contraction of the musculature of the body
wall will cause an outward movement, although
the resulting compression produces an expiratory
pulse. During the larger breathing movement of
the phrase, however, the abdominal walls and the
rib cage show a gradual recession, followed by the
inflation due to the inspiration for the next phrase.

(2) Negative pressure tracings of the pulses of
the feet and breath groups. (The method is described,
p. 17, 18.)

When properly placed on the abdomen a negative
pressure applicator records the pulses of the feet
and the breath groups which originate in the
abdominal musculature. It is possible with care
to place one applicator on the abdominal wall so
that it records primarily the foot and breath-group
pulses. A simultaneous tracing of the chest pulses
of the syllables is neccessary to show the timing
of the syllables in the foot and breath group
movement, so that the coordination of the constituent
syllables in the foot and breath group can
be studied.

(3) Oscillograms of the action potentials from
the abdominal musculature. (The method is described,
p. 19.)

Electrodes were placed on the abdominal muscles,
the intercostals, and in the proper area for the
diaphragm, so that contractions of the intercostals
for the chest pulses and the contractions of abdominal
muscles for the foot and breath-group pulses
could be recorded simultaneously. As the action-potential
method is the one way to get the detail
of the intercostal contractions for the chest pulses
of the syllable, the simultaneous recording of the
abdominal contractions is of primary importance
for the study of the detail of the coordination of
the larger units with the constituent syllables.

B. Action of the larynx

The methods of recording laryngeal vibrations
directly from the walls of the larynx have been
described by Rousselot p. 97. It is important for
some purposes to have a direct recording showing
the incidence and duration of the laryngeal tone.
Electric applicators can be used to pick up the
tone. The same purpose can be served, however,
by tracings of the air pressures both inside and outside
of the mouth. (33)

Vertical movements of the larynx have proved
to be significant factors in the articulation of surd
and sonant stops. To record these movements the
recording lever must be rigidly supported independent
of other moving parts. Hudgins and Stetson
(35) using a modified version of the apparatus
described by Rosapelly (70, p. 98) and Zwaardemaker
(107, p. 357) succeeded in anchoring the
supporting device firmly to the clavicles below and
the upper jaw above. Sen, using the same device in
Hudgins' Laboratory (77) at the Clarke School,
studied the vertical movements of the larynx as a
factor in differentiating between aspirated and non-aspirated
stops in Bengali dialects.

C. Articulatory processes. contacts of
articulatory members to opposing surfaces

Movements of the articulatory organs may be
recorded in terms of their contacts with the opposing
surfaces. Such records give important clues to
the action and timing of the articulatory member,
both with relation to the syllable pulse from the
chest and with relation to other articulatory
members. Contacts of the lips, lips and teeth, and
tongue with the alveolar ridge and with the hard
20palate were recorded by means of lip- and tongue-markers.

Lip- and tongue-markers were constructed to
record the strokes of lip and tongue in the production
of the consonants. The markers were made with
a thin wedge-shaped cross-section. It is essential that
they be very sensitive but that they do not collapse
whatever the pressure. Lip markers were constructed
of thin mica bound at the thin edge of the
wedge with thin “fish-skin” (animal membrane)
fastened with rubber cement, and with sides of
“fish-skin”. Another satisfactory form which became
routine was made by stretching very thin
rubber tightly over a flat steel wire loop of the
proper shape, the ends of which were thrust into a
flattened cork which constituted the thick end of
the wedge. The cork received the small tube leading
to the kymograph. The thin wedge of taut
rubber was coated repeatedly with ricinated collodion.
The collodion dried to a wall more rigid
than the rubber which was merely a temporary
support for the collodion wall.

The tongue-markers also were constructed with
a collodion wall laid on over taut rubber. The
frame of flattened steel wire was shaped to a plaster
cast of the subject's palate, and thin rubber was
stretched over the wire frame mounted so as to make
a thin wedge. The wall next to the palate was
repeatedly coated with collodion until quite rigid;
the surface receiving the stroke of the tongue was
given one or two coatings to make a firm but
resilient wall. The tube from the tongue-marker
was led out through a convenient gap in the teeth.
The tongue- and lip-markers stand use better if the
edges at the wire are bound with narrow strips of
“fish-skin” laid into the first moist coat of collodion.
Cf. Fig. 2.

It did not prove possible to get satisfactory
tracings with lenticular rubber bulbs, or with
capsules of thin, taut rubber such as Rousselot
used; they do not respond to the rapid variations
of pressure, and the tracing of the double or of the
abutting consonants does not show the two maxima.
Lip- and tongue-markers must respond to fluctuations
of 25 per sec. The tongue-tip trill is a
convenient signal for testing recording apparatus
up to 30-40 per sec.

The wedge-shaped markers with resilient walls
were thin enough so that they did not introduce
a thickness of more than a millimeter between the
lips or between the tongue and the palate. These
markers and their fittings had of course to be kept
air-tight; a leak betrays itself very quickly in the
tracing. The tongue-marker must fit the palate;
a layer of dental wax may be necessary so that air
does not pass to falsify a tracing of outside air made
concurrent with a tracing of the tongue strokes.
The lip- and tongue-markers, the mouth pressure
tubes, and the mask for outside air were held in
position by the subject. (Fig. 2.)

The tracing of the consonant can be called a
record of the consonant movement only in so far as
the movement of the consonant results in contact of
lip or tongue with the opposing surface. In general,
the tracing is a record of consonant contacts, with
the variations of pressure. The tracings of the consonant
contacts for abutting consonants (including
doubles) delineate the back stroke of the arresting
doublet and the beat stroke of the releasing doublet
which occur in contact and at rapid rate. It was in
this case that Rousselot's apparatus failed to respond.

Contacts of the tongue for the lingual consonants
have also been recorded by a method that employs
the familiar artificial palate. The original purpose
of this device was the combination of the palatographic
and the kymographic methods of recording
(92). The modified artificial palate, however,
may be readily employed to replace the tongue
markers described above for recording lingual contacts
both in the anterior and posterior areas of the
mouth. Artificial palates molded to fit the palates
of the individual subjects were made of vulcanite
following the dentist's technique. Two “windows”
were cut in the finished mold, one in the region of
the alveolar ridge, the other at the posterior edge of
the hard palate. Thin walled phosphor bronze
tubes, 1.5 mm. diam. were shaped to fit the contour
21of the palate and embedded in the vulcanite during
the process or vulcanizing. The tubes provided outlets
for the “windows” within the palate with their
outer ends leading to rubber tubes connected to
recording tambours. (Figs. 2 A, 6, 7.)

The windows were covered on the upper surface
in contact with the hard palate with a layer of
animal membrane reenforced with a heavy layer of
collodion to give them a rigid backing. The “windows”
were covered on the lower side with which
the tongue comes in contact by a membrane of
condom rubber. A thin coating of collodion adds
resiliency to the membrane of these windows which
become capsules formed in the wall of the artificial
palate. The rapid strokes of the tongue for the
front linguals “t, d, n, s”, and “l” and the posterior
linguals “k, g, and ng”, strike the flexible membranes
of the capsules and the contacts are recorded.
(Fig. 1, Stetson, Hudgins & Moses, Pal. changes,
1940.) (92, Fig. 1.) Figs. 6 and 7 show the artificial
palate lying in a cast of the subject's upper jaw. The
dotted lines of Fig. 2 A indicate the position of a
metal tube which may be used to record the air pressure
behind the consonant constriction along with
the contact. Figures 8 and 9 (Stetson, Hudgins &
Moses) show a palatogram-kymogram combination
obtained simultaneously by this method.

D. Recording air-pressures

The pressure changes in the chest are direct functions
of the action of the breathing muscles. The
chest pressure is only slightly affected by the articulatory
processes which occur above the larynx.
Records of chest pressure changes during speech
obtained by methods described above show only
minor differences in amplitude for syllables having
consonants and vowels, as compared with syllables
having vowels alone. (Fig. 18, 19, 20, 21.)

Pressures above the larynx, on the other hand,
although due to the chest pulse for the most part,
are radically modified by the articulatory processes.
Hence tracings that show the pressure changes
above the larynx taken simultaneously with pressure
changes below the larynx provide important clues
to articulatory processes.

Kymograph recording of the pressure just outside
the mouth has been a standard procedure since
Rosapelly and Rousselot. The technique of Rousselot
was modified in that the volume of the mouthpiece
was increased and an escape for air provided.
A firm rubber mouthpiece was fitted about the
mouth, with an elastic border which did not interfere
with the movements of jaw and lips. The
border was made elastic by slicing the edge into a
comb; the border was made air-tight by cementing
the wrist section of a rubber glove inside the mouthpiece
and turning the cuff out over the sliced edge.
The mouthpiece had a volume of ca. 200 cc., and
ventilation was always provided in the mouthpiece
and sometimes near the recording tambour to be
rid of reverberation and to prevent the air pressure
from backing up while speaking. The tube leading
to the recording tambour was large (1 to 1.5 cm.).
The recording tambour was of very thin, taut rubber,
diam. of 5 cm. The mouthpiece was pierced to
provide for lip- and tongue-markers and for the
mouthpressure tube at the corner of the mouth.
The range of oscillation provided was larger than
that of Rousselot's tracings of outside pressure.
Rousselot's rather rigid membrane was intended to
show sound waves. These tracings were intended
to show variations of pressure rather than sound
vibration. However, the sound vibrations were
usually present in one of the three tracings: 1) of
the consonant, 2) the mouth pressure, or 3) the outside
pressure. The tracings of pressure just outside
the mouth are not comparable to tracings of pressure
in the chest; the constriction of the larynx and
the variations of mouth volume make the series unlike.

Tracings of air pressure from the nose, which indicate
the action of the velum, were recorded by the
familiar nasal olives. This is a standard practice in
experimental phonetics. Usually a single olive, a
small blown glass bulb open at both ends which fits
snugly into the nostril, gives an adequate record of
the pressure changes in the nose. The single olive
22does not completely obstruct the air flow from the
nasal cavity, and thus interferes less with the normal
speech process, than do olives in both nostrils.
Tracings of air pressure changes inside the mouth
provide records of another important aspect of the
process of articulation. Pressures build up behind
the constrictions of both the releasing and the arresting
consonants. The rate of rise of this pressure, its
maximum amplitude, and the terminal shape of the
pressure curves provide significant information. A
record of the inside air pressure (A in the records)
is readily obtained by inserting a small metal tube
into the mouth so that it taps the oral cavity behind
the consonant constriction. The tube is placed so
that it does not interfere with the consonant constriction.
It may enter the mouth at the corner for
a recording of the labial and front lingual consonants.
In order to tap the pressures behind the posterior
sounds “k and g”, however, the tube must
reach well back into the mouth. For this purpose
a tube is laid along outside the lower molars and
curved sharply at the end to enter the oral cavity
posterior to the last molar. The dotted line, Fig. 2 A,
illustrates the position of the tube. Indeed, the pressures
for all of the consonants can be recorded with
the tube in this position and there is practically no
interference with the articulatory processes.

E. Recording the movements of the jaw

The movement of the lower jaw plays an important
role in both the shaping of the vowel cavities and
in the articulation of consonants. In ordinary speech
a rapid opening of the jaw occurs with each vowel
(syllable). For the syllables with vowels having a
narrow front orifice the jaw movement is of small
amplitude but none the less present, while for the
syllables with open vowels the movement of the jaw
is relatively large. There is evidence also that the
jaw is a factor in the articulation of labial consonants.
Studies of the maximum rate of movement
of articulatory organs have shown that the jaw is
capable of a higher rate than the rate of the lips
with the jaw fixated. The implication is that at
rapid rates of syllable utterance the jaws carry the
lips in the articulation of labial consonants. (36)

Records of jaw movements are readily obtained
by several methods. The simplest device supports
the active recording element on a rigid pair of
plastic spectacle frames. The recording element is
similar in structure to the rubber-covered coiled-spring
pneumograph commonly used in recording
respiration in physiological laboratories; but it is
much smaller and more delicate. A coiled spring
10 cm. in length and 1.5 cm. in diameter was covered
with thin-walled rubber tubing made of either
condom rubber or of surgical drainage tubes. Cork
stoppers, into which the ends of the spring were
inserted, serve both for the attachment of the tubing
and for the introduction of a glass tube outlet to the
connecting rubber tube of the recording tambour.
The ends of the recorder are attached to the eyeglass
frame and to a tape looped under the chin
with a slight tension when the mouth is closed.
Movements of the lower jaw distend the “pneumograph”
longitudinally, inducing pressure changes
which are communicated to the recording tambour.

F. Methods of recording vowels

There has been some instrumental observation of
jaw movements, and of the position of the tongue
with simple apparatus; but the bulk of the recording
of vowels and of vocalization has been done
with acoustic apparatus.

1. Physiological

The artificial palate has been used to delineate
the position of the tongue where it is in contact
with the hard palate; this gives a slight indication
of the differences in position for the vowel.

Direct measurements with disks have been used
to show the fundamental tongue positions. (20)

Extensive X-ray studies have been made of the
longitudinal section of the vocal canal during the
utterance of a vowel. The median line of the tongue
is indicated by a metal cord or chain, and a barium
paste outlines the surfaces. (73, 74)

The reciprocal relations of the front and back
orifices in the utterances of the vowels were demonstrated
23by Marichelle, in an unpublished study, in
which the front orifice was defined by wedges between
the teeth, and the back orifice was gauged by
use of the artificial palate.

2. Acoustic

The acoustic analysis in the main has been concerned
with the modulation of partials by the varying
cavities. The vowel has been treated as an acoustic
quality, though it is usually recognized that the
basic explanation must be in terms of cavities and
movements.

Much of the work by Helmholtz, Stumpf, D. C.
Miller, Crandall, Gemelli has been done with prolonged
vowels; the vowel has reached an approximately
steady state, and the analysis by Fourier's or
Vercelli's method has seemed adequate. A system
of one and two “formants” has been worked out
and has general vogue.

But is is apparent that the usual vowel of actual
speech does not come to a steady state; and the fact
that the vocal folds constitute a “relaxation oscillator”,
in which there is a damping or accelerating
factor in each cycle, makes the application of a
mathematical method out of the question.

There is nevertheless an acoustic modification of
the sound complex which issues from the larynx. It
may be that interference and absorption play a part
along with resonance.

It is possible to segment the vowel, and show that
the vowel quality inheres in the beginning, the
middle and the end of the acoustic pattern. Playing
back the same vowel of two identical variable-density
films on two photoelectric playbacks, in varying
phases, reversing one of the films so that the end of
the one film of the vowel is heard along with the
beginning of the other film of the vowel, or mixing
the segments of the vowel, do not affect the vowel
quality.

Potter and his colleagues at the Bell Telephone
Laboratories have shown that the sound spectrum
which consists largely of “white noise” is radically
modulated by cavities and movements. They have
noted empirically the regions in the spectrum most
subject to this modulation. It is especially, noteworthy
that their recordings show the changes of
the beginning and end of the vowel due to the
opening and closing of the orifices made by consonant
constrictions. The “kinks” in the ‘bars’
representing the zones of modulation are due to the
position of the opening or closing consonant constriction,
and define the consonant as front (lip),
median (front tongue), or posterior (back tongue).
(42, 43, 64, 82)

2. Analysis and interpretation of the records
from the process of the speech mechanism

A. Pneumatic recording systems

The various systems which employ pneumatic
kymograph recording must include a recording
member. For much of the work the recording tambours
were of rubber. For the lip- and tongue-markers,
and for the negative pressure tracings, they
were small, 1-1.5 cm., flexible but not elastic.
Perished rubber, or rubber in which a permanent
dimple had been made by deforming the sheet rubber
over a boss of the size of the diaphragm, was
used. The rubber was derived from condoms, and
the “animal membrane” used was from “fish-skin”
condoms. The recording diaphragms for the air
pressure tubes and for the other applicators were
taut rubber of 5-8 cm. diameter. The recording
parts of the commercial Marey tambours were rebuilt.
The recording stylus and the post vertical
to the membrane were made of light bamboo. The
writing stylus was short, ca. 6 cm., pivoted on an
axle made of the tip of a small cambric needle and
tipped with a glass or thin celluloid film; mass was
reduced to minimum and vibration and overthrow
were eliminated by damping.

Reference marks were made at the beginning of
each series to show the exact vertical relation of
the stylus tips: these marks were used in determining
simultaneous points. All tracings were taken with
a time line from either a small 50-per-sec. fork, or
a Jacquet time marker giving fifths of a second. In
measuring the tracings these fifths were divided into
24tenths and the hundredths estimated. The measurements
were made with a protractor in reference
to the nearest fifth indicated by the time line. Thus
the unit of measurement was 0.02 sec., with fractions
estimated. This is sufficiently minute for the
purpose and as close as the limits of error of the
method warrant.

It is important to note that in making such
records the pneumatic transmission at such rates
consists of a low-pitched sound wave, a travelling
condensation-rarefaction, not an air flow. There is
no measurable lag in such apparatus, the volume of
air is not important, but tubing with rigid wall and
smooth bore and with few or no sharp angles is
important; rarely reverberations may give difficulty.
These can be filtered out by inserting lengths of
capillary tube in the transmission lines. (8)

Many of the later recordings were made with a
substitute for the Marey tambour, the pneumodeik,
which was developed in the course of the work in
phonetic recording. The pneumodeik, with its metal
diaphragm, is durable, needs little attention, and is
sensitive enough for practically all recording. It is
not quite as sensitive as a small, loose rubber
diaphragm in skilled hands. But since it is quite
stable, and can be calibrated, it has been used
extensively in the laboratories of both Stetson and
Hudgins. (34)

B. Processing of kymographic tracings.

The figures in the text are photographic reproductions
of the kymograph tracings. It is now
convenient, with reversible film negatives, to make
photographic prints from smoked paper with black
lines on white ground. These prints were produced
by using the smoked paper itself as the negative.
The ground was cleared in some cases. Often both
negatives and positives were made of a very “hard”
thin printing paper, clearing by managing exposure
and development. In a few cases the time line has
been retouched, but all other tracings have not
been retouched.

A number of subjects were always used, and a
large number of measurements were made of each
type of record. The kymograph sheets used for
cuts were not mutilated, and the entire series of
records taken is filed for convenient reference, and
has proved valuable. The detail of procedure, and
of the apparatus used, was written on the smoked
sheets, usually before they were fixed; and if
necessary, notes were added during the measuring
process.

If records are made and preserved in large
quantities, the method of handling becomes important.
A type of thin “label” paper has been used,
selected by testing so that it would not curl when
dipped in alcohol. The records are attached about
the drum with as little starch paste as possible;
the later bath in alcohol granulates the paste and
it gives no trouble; mucilages are to be avoided.

The kymograms were fixed by passing them face
down across the surface of the fixative, without
wetting the back of the sheet. The fixative consisted
of a solution of bleached shellac in alcohol, with a
cc. or so of glycerin or sorbitol, and a little castor-oil,
per litre. The solution is kept as dilute as
possible. It is tested occasionally, as it evaporates,
with fresh smoked paper. There should be barely
enough shellac to fix the smoke so that it will not
rub off. Such a record is thin and flexible and
convenient for filing. It will dry in a minute or so
and is ready for immediate study. The reference
lines for measurement, and any further comments,
are scratched in the fixed smoke with small chisels
ground out of No. 5 needles. Care is taken that
these etched lines do not cross any tracing of the
record. The records are labeled, each one in full,
and adequate titles written in large at the top of the
record. They are filed vertically in modified filing
cases, with labeled manila dividers, so that they
can be easily consulted. Many questions can be
answered by an inspection of a series of records
side by side.

For some of the later recording, “Teledeltos”
paper has been used in a polygraph. It is excellent
for photographing; annotations can be made in
pencil or in ink. The paper is furnished by the
Western Union Telegraph Company and can be
25used with any form of kymograph or polygraph.

Syllables, both open and closed, when in series,
have been measured from the détente of the releasing
consonant to the détente of the initial consonant
of the next syllable. Isolated syllables if closed have
been measured from the détente of the releasing
consonant to the détente of the arresting consonant;
if open from the détente of the releasing consonant
to the end of the vowel. This method of measuring
syllables accords best with other studies, and has
the advantage of giving an accurate rate per sec.,
and of eliminating the erratic length of contact of
the initial consonant of a group or series.

In the tracings the letters and combinations have
their usual English values. The vowels are the
English short vowels unless otherwise indicated.
Occasionally digraphs are used for a single vowel:
“ee” or “ea” = long “e” (tea); “ay” = long “a”
(pay).

C. Reading the tracings of motor phonetics

It is essential to read the evidence for an experimental
phonetics in the actual data presented; only
a little can be put in the form of phonetic characters,
and statistics. The actual events and their
concurrence can only be presented in the form of
annotated simultaneous tracings. Great pains have
been taken to label the tracings in full and to supply
them with legends.

The kymograph tracings represent both 1)
movements and contracts, and 2) changing air
pressures, by means of the rise and fall of line
tracings.

1) A positive stroke to contact, the increase in
pressure of the lips or tongue, the opening of the
jaw, the bulge of a muscle, the rise and expansion
of the chest are all marked by the rise of the curve;
the reverse, of the stroke leaving contact, the
diminishing pressure, the closing of the jaw, the
recession of chest or abdominal wall, are all marked
by the fall of the curve.

2) In the chest (C), the mouth (A), and outside
the mouth (AO), rising air pressure is indicated by
a rising curve and falling air pressure by a falling
curve. If the pressure becomes negative, as in the
intake of air, the air pressure curve falls below the
base line.

The oscillograph and microphone tracings indicate
intensity by variations in amplitudes; in some
cases a volume meter has been used. It is important
to note that the level of ground noise due to minor
disturbances is well below threshold.

In action-current recordings there is usually some
leakage of the stronger muscle pulses into the
tracings of neighboring muscles. This is noted in
the legends of the figures.

Constant reference to the time line is important.
Any change which occurs within an interval of
0.05 sec. must be discontinuous and figure as a
stroke, not a changing shape. Simple movements
at a maximum of 10 sec. consist of beat- and back-strokes
of 0.05 secs. each. 2.5-3 per sec. is the upper
limit for a series of syllables with abutting or
doubling consonants (“put-put-put-, pup-pup-pup-”).
Cf. Appendix X: Demarcation of Syllables.26

III
Coordination of the movements of speech

1. The movement of the syllable as the
fundamental unit of experimental phonetic
analysis.

Since the syllable is the simplest possible utterance
in a monosyllabic breath group, it is convenient
to consider first the syllable pulse. The syllable
pulse is the simplest basic movement and is essential
to the consonants and vowels which constitute the
phonemic units so often assumed. If movements
are the essentials in speech, why not consider the
movement of the “sound” as the fundamental unit,
as Rousselot actually did in his tension-tenue-détente
analysis? The answer is that the “sounds”
and their movements cannot be discussed without
involving their function in the syllable. The nature
of the “sound” depends on what it does in the
syllable. The distinction between the vowel and
the consonant, or between a “syllable-forming- sonant”
and a true consonant must depend on the
part the “sound” plays in the syllable. In this
Saussure was nearer right than Rousselot. The
distinction between consonants, implosive and
explosive (“appuyante et appuyée”), and their
different fate in the course of phonetic changes, can
be explained only by reference to their function in
the syllable. The movements of the “sounds” often
overlap and often fuse; they are not always separate.
But by common consent a syllable is always a
separate event in the speech series. And while
investigators may not find it easy to determine the
principle of syllable division, no one has assumed
that two syllables could in any sense overlap or
coincide.

When teachers and demonstrators give what they
think are “separate sounds”, they are actually uttering
syllables; the vowels and on occasion the liquids
and nasals constitute separate syllables, as in “oh, a,
I, rr…, ll…”; . long drawn out fricatives,
“sss…” etc., become vowel substitutes, the other
consonants are given either with a brief vowel which
the consonant releases as in “buh, puh”, or which
the consonant arrests as in “eff, ess”. The syllable is
the smallest unit which can be uttered.

The fundamental unit for phonetic analysis is
the movement of the syllable, a single breath pulse
which is sometimes released by a consonant
movement and sometimes arrested by a consonant
movement. The syllable-pulse may be released or
arrested by the chest muscles alone (“Oh, a, I, tea,
eat”). The division of the syllables is the division
of these chest pulses one from another. The delimiting
consonants are integral parts of this chest-pulse
movement; they either release or arrest the chest
pulse; the vowel on the other hand is due to the
shaping of the vocal canal by an articulatory
movement to emit the chest pulse. These articulatory
movements are auxiliary movements to the
chest pulse. When abutting consonants appear
between syllables, the constriction is maintained
throughout the pair; the first consonant of the pair
arrests the chest pulse of the first syllable and the
second consonant releases the chest pulse of the
second syllable. A “double consonant” is a case in
which the arresting consonant of the first chest
pulse and the releasing consonant of the next chest
pulse are the same consonant repeated. It is possible
to demonstrate these movements experimentally,
and to determine their coordinations.27

2. The three types of skilled movement and
application to the movement of the syllable

A study of the action of the muscles in such
ballistic movements shows that the movement is
started by a sudden contraction of the positive
muscle-group which immediately relaxes. During
at least half of the course of the movement neither
of the antagonistic muscle-groups is contracted, so
that the moving member flies free. At the end of
its course the movement is usually arrested by the
contraction of the negative muscle-group. The
movement is a movement by momentum.

When a tennis stroke is made, a sudden contraction
of the extensors of the arm starts the stroke; the
extensors immediately relax, during the first quarter
of the excursion, and the arm and racket are carried
through the stroke by their momentum; at the
close of the stroke the flexors of the arm may come
into play and arrest the arm. All rapid movements
are of this type. In speech, the rapid movements of
articulation and the syllable pulse are ballistic.
If one is unfamiliar with a language, one is always
impressed by the high speed of its movements (86,
p. 18 f; 90).

3. Analysis of the ballistic and tense movements
with special reference to the syllable

In the simplest possible case, the ballistic movement
is stopped by the muscles themselves, as is the
movement of the conductor's baton, of the violinist's
bow, or of the writer's pen. The antagonist muscles
involved divide into two groups: the positive group
which starts the movement, and the negative group
which arrests the movement; in case the movement
is repeated, this negative group returns the member
to the starting point.

When the arrest is almost instantaneous, the
member is returned at once to the starting point,
as in the case of a rapid, repeated movement, the
action of the positive muscle group results in a beat
stroke, and the action of the negative muscle-group
results in a back stroke which returns the member
to the original position.

The movement of the conductor's baton or of
the violinist's bow is arrested by the antagonist
muscles. So the syllable movement of “oh” or “a”
is arrested by the chest muscles. The movement of
the hand in clapping is arrested by the movement
of the opposite hand; the stroke of the cymbal in
the right hand is arrested by the movement of the
cymbal in the left hand. So the movement of the
28syllable pulse in “it” or “up” is arrested by the
articulatory movement of the tongue or of the lip,
which closes the vocal canal and compresses the air
column and so acts on the moving chest walls. If
the member when arrested is held in position at the
lower limit of the movement, the interval required
to take up the momentum and relax the arresting
negative muscle group is as long as the interval
required for a back stroke to return the member
to its original position for the next beat stroke.
Such a movement is to be seen in beating three-four
time so that the baton follows a triangular path.
This interval during which the negative muscle-group
contracts, arrests the movement and then
relaxes, may be called “the relaxation phase”. It
is an interval equivalent to the duration of the back
stroke, and is sometimes called a “condensed back
stroke” or a “back stroke in contact with an
obstacle”. This back stroke or relaxation phase
includes the preparation for the next beat stroke.

The beat stroke is always ballistic; it occurs very
rapidly and can hardly be longer than 40-100
milli. sec. On the other hand, the interval involved
in the back stroke varies greatly. In the slow
beating of time, beats may occur at the rate of two
in three seconds, i.e. with a period of 1.5 sec. per
movement. Assuming that the interval of the beat
stroke is 0.050 sec, the back stroke, or relaxation
phase, would be at this maximum, 1.45 sec in duration.
At maximum speed, a movement can be
repeated ca. 10 per sec. At such high speed the
back stroke as well as the beat stroke is ballistic,
and the back stroke has a duration of 0.050 sec.
The back stroke, or relaxation phase (including
preparation for the coming beat stroke) may vary,
then, from 1.45 sec. to 0.050 sec.

4. Speech as the activity of a moving member

The vocal apparatus constitutes a tubular member
inside the trunk which acts like the arm and hand
in the quick raps of the coppersmith or the rapid
strokes of the penman.

The apparatus for a skilled movement, like the
taps of the hammer or the strokes of the pen, is
not of course the musculature of the moving hand
and fore-arm merely; there must be supporting
movements of the upper arm, and a shoulder girdle
posture. The strokes are delivered with shifts of
these posture muscles of the shoulder and upper
arm which support and direct and furnish pivots at
the shoulder and elbow for the obvious ballistic
movements of the hands.

The connections of the moving member of arm
and hand for posture and support consist of bony
levers, pivot joints, and tendons. Some of the
connections of the tubular member of the speech
apparatus are more unusual. The pivots of the ribs
at the spine, and of the jaw are obvious, but the
rapid and nicely adjusted movements of the tongue
are peculiar. The abdominal muscles and the
diaphragm are connected by the liquid mass of the
viscera; and the chest is connected with the articulatory
organs by a column of air under varying
pressure (Cf. Fig. 13).

The apparatus of the skilled movement of the
syllable is not of course the musculature of articulation
merely; there must be a posture fixation of
the rib cage and abdomen for the beginning of
utterance, and a slow movement for the descent of
the rib cage and the recession of the abdomen as
the phrase progresses. The syllable pulse is supported
and integrated by the movements of the abdominal
muscles for the foot and the breath group.

The functions of the speech apparatus, however,
the time constants, the relation of posture (fixations),
the slow tense supporting movements, and
the ballistic pulses are all characteristic of a single
moving member. The maximum rate is ca. 8-12
29per sec. (not 16-18 per sec. as would be the case
if two or more ballistic members were alternating
as in piano- or in flute-playing).

The unit movement of speech is the pulse which
produces the syllable, a pulse of air through the
glottis made audible by the vocal folds in speaking
aloud and stopped and started by the chest muscles
or by the auxiliary movements of the consonants.
The larynx itself does not initiate the syllable nor
control the process of articulation; this is quite
apparent in the case of speech with an artificial
larynx. The larynx makes possible the vocalization
of speech. In just one thing is the larynx a prime
mover; it determines the pitch of the tone. The
initiation of the tone is due to the pressure differential
above and below the larynx. The principal
problems of the process of speech lie above and below
the larynx

A familiar way of treating speech divides the
physiological apparatus into resonators, reed, and
bellows. But the bellows is not like that of an
organ, furnishing air under pressure to the glottis.
Negus' statement that the larynx is a valve is easily
misunderstood; only in the case of the glottal stop
(which does not occur in English) does the larynx
figure as a valve in speech. The chest-abdomen
musculature is like a bellows for blowing the fire;
when the bellows is inflated there is an increase of
volume but there is no increase or pressure within.

Thus the intercostals make a rapid series of
syllabic pulses at rates as rapid as 8-10 per sec,
and the larger movements of the abdominal-thoracic
musculature group and stress the syllables.
These breath groups (of a single foot) may follow
each other at the rate of 3-4 per sec. ; There is
kymographic and action-potential evidence of these
two musculatures (Figs. 3 and 4, DIA, RA, II, and
EI). The pulses produced by the intercostals and the
abdominal muscles have been recorded in various
ways (cf. p. 11). Figs. 15, 16, 17, 20, 28 etc.

The breath group movement gives the overall
form of the foot and of the breath-group series
of syllables, which shows in the stresses of the feet,
and in the elimination and restoration of syllables.
It is evident that the syllables are parts of the breath
group, and variations in the rate or the stress of the
breath group may modify the component syllables
radically. Since the syllables are parts of the breath-group
pattern, the pattern conserves the individual
syllables; they may be reduced, or even eliminated,
but the proper modification of the rate and stress
of the breath group will restore them.

The breath group may consist of one or more
feet; the feet may consist of one or more syllables.
Thus it is possible to have a breath group composed
of a single syllable, or of 10-15 syllables.

In speech, the rapid movements of the consonant
and the movement of the syllabic pulse are ballistic.
The moving member in the case of the syllabic
consists of the driving and arresting muscles of the
chest wall and the air column on which they act;
the air column connects the chest muscles to the
articulatory muscles of the consonant.

If the syllabic consists of a vowel only, OVO, the
positive chest muscles (Inter. intercostals) start the
wave of compression in the air column and the
negative chest muscles (Exter. intercostals) stop the
compression and take up the momentum of the
chest wall. (The compression backs up slightly
against the resistance of the glottis, if it is phonating,
and of the tortuous vocal canal.)

If the syllable consists of a consonant and a vowel,
CVO, the positive chest muscles start the wave of
compression in the air column, which is momentarily
blocked and then released by the consonant
stroke, and the negative chest muscles stop the
compression and take up the momentum of the
chest wall.

If the syllable consists of consonant, vowel, consonant
CVC, the positive chest muscles start the
wave of compression in the air column, which is
momentarily blocked and then released by the consonant
stroke, and the arresting consonant stroke
stops the compression wave and the compressed air
column helps to take up the momentum of the
chest wall.

5. Characteristics of the syllable as a movement

The syllable then is constituted by a ballistic
movement of the intercostal muscles. Its delimitation
is not due to a “point of minimum sonority” but
to the conditions which define a movement as one
movement. In the individuality of the syllable the
sound is secondary; syllables are possible without
sound. Speech is rather a set of movements made
audible than a set of sounds produced by movements.

The consonants are not mere noises floating in
the stream of sound. They are auxiliary movements,
connected by the compressed air column with the
chest muscles, and they have a function in releasing
and arresting the chest pulse which constitutes the
main movement of the syllable.

Fig. 23 gives tracings of a syllable composed of a
repeated vowel in comparison with a similar syllable
with a releasing consonant. The variations in the
chest pressure show the separate syllables when
with and when without the consonant.

The articulatory conformation may not be that
of a vowel; any continuant will do; “f, s, v, z, r, l,
n, m” will all form syllables provided there is a
chest pulse. If the articulatory conformation is
maintained, as many syllables are formed as there
are new ballistic chest pulses. Thus the series of
“n” 's discussed by Sievers in “berittenen” (75,
p. 94).

The “n” can be developed into an indefinite
series; one has only to fixate the vocal apparatus
in position for “n”, and give a series of chest pulses,
and a series of syllables is the result. In Fig. in
“runnin' an' neighin'” and Figs. 88 and 89, “Lil' 'll
lie low”, the tracing of the chest pressure shows that
the pressure docs not rise because of the consonant
closure; on the contrary, the minimum pressure
occurs during the releasing consonant and the pulse
is only partly under way when the releasing
consonant opens. The maximum chest pressure
occurs between the consonants of the syllable. In
forms like “runnin' 'n' neighin' ” it seldom happens,
as Sievers assumes, that the “n” position is retained;
this does occasionally take place, but as a rule the
tongue makes a definite stroke for each syllable.
The same observations hold for the series of “l” 's
in “lil' 'll lie low”.

Klestadt gives pneumographic tracings of speech
in which each syllable is clearly indicated by a chest
pulse. (41)

Bloomer and Shohara published a preliminary
study of respiratory movements in which the syllables
are evident. (7)

These syllables, OVO, are ballistic movements
with chest release and chest arrest. If the movement
were not chest-arrested it would trail off in a sigh
or a moan. If a consonant is incorporated, we may
have the chest pulse released by the consonant, or
the chest pulse arrested by the consonant. All
phoneticians are agreed that the releasing and arresting
consonants are unlike. Compare Fig. 27 where
“eat…” becomes “tea…”.34

The rate of the syllable movement is rapid; it is
possible to repeat a single vowel like “a, a…” at
the rate of 5-7 per sec. The highest speed is obtained
when the syllable has an occlusive releasing
consonant; “ta, ta…” can be repeated at the rate
of 7-12 per sec; the rate depends slightly on the
rhythmic grouping and is equivalent to the maximum
repetition time (“tapping time”) of the subject.

It is interesting that the chest muscles are capable
of producing 8-12 pulses per sec. This is the
highest speed of small and well-coordinated muscles
like those of the hand. Tracings such as those
shown in Fig. 24 prove that the muscles involved
in the syllable movement cannot be the larger
muscles of diaphragm and abdomen; the intercostal
muscles of the chest must function.

“Long” vowels are distinctly slower, if the characteristic
quality is preserved. “Tea, too, tay” can be
repeated at the rate of 4.8-5.2 syllables per sec.

Rousselot's reading of a selection from the Chanson
de Roland shows an average rate of 4.9 per sec.
for 110 syllables; a reading of the same selection
some five years later gives an average of 4.5 syllables
per sec. (70, p. 1055; 72, p. 96)

The duration of the syllable depends somewhat
on the force with which it is uttered. Increased
35force of utterance involves a greater contraction of
the positive muscles of expiration, and this requires
a longer time for the process of arresting the movement.
Sometimes this increase in length shows in
the vowel, when the negative chest muscles which
arrest the movement come into play during the
latter part of the vowel. This is unusual in the
Romance languages, but quite common in English
in which most of the “long” vowels have a “vanish”
during which the arrest of the syllable movement
begins.

It is possible to prolong a syllable at will, if the
syllable is “open”, or if the syllable ends with a
voiced continuant like “m, n, l, r, etc.”. In beating
time, so that the baton describes a “figure-eight”, the
movement may be prolonged after the pulse of the
beat; in the same way the beat stroke of the syllable
may be arrested by a back-stroke process which
continues the course of the movement in a slow,
“controlled” form. Movements of this sort, which
begin with a ballistic pulse and are continued as a
“controlled” movement, are very common in all
fields. In unusual cases a vocal movement of the
“controlled” type begins, perhaps in the utterance
of a fricative, and is finally merged into a ballistic
pulse; in such cases the preceding movement is
counted part of the syllable generated by the ballistic
pulse. It is after this fashion that the preliminary
slow movement of the orchestral conductor merges
into the entering beat stroke of the selection.

6. Function of the vowel in the movement of the
syllable

The characterized factors of the syllable, the vowel,
and the consonant, constitute the familiar phonemes
which the phonemic systems are concerned with.

The compound consonants which function as a
single syllabic factor, and the abutting consonants
at the syllabic frontier will be handled in the discussion
of the movements and functions of the consonants.

It will be convenient to postpone the detailed
classification of the vowels and consonants until the
processes involved have been studied in detail.

In popular phonetics the vowel is counted the
core of the syllable and is often identified with the
syllable. Such a “sound” often constitutes the bulk
of the syllable in point of duration and is quite
prominent as a matter of quality. So much so that
Marage (48) and Devaux-Charbonnel (16), reasoning
from the prominence of the vowel in acoustic
records, have counted the vowel the important item.
The importance of the vowel in the syllable is reflected
in the common term “syllabic”. The very term
“consonant” stresses the importance of the vowel.
A. G. Bell (2) was surprised to find, however, in dealing
with the speech of the congenital deaf, that the
specific vowel is less important than the consonants.
But he showed that the function of the vowel is
important even if the quality is negligible. He
demonstrated that a reading is quite comprehensible
if a single indeterminate vowel is substituted for all
the vowels of the passage. The Semitic writing systems
seem to treat the vowel as secondary by putting
the consonants on the line and marking the
vowels, ad libitum, as mere “points”.

For audibility, however, the vowels are essential.
In singing the vowel becomes primarily important,
and the consonants often lose their function and are
neglected.

Sievers and others have stressed the fact that a
syllable may be composed of a consonant like “m,
n, l, r” counted in some sense as a vowel. One can
go further. The action of the vocal folds is not
essential to the syllable. A syllable may be composed
of “sh…” or of “pst”. In whistling through
the teeth we have a series of syllables composed of
nothing but a long drawn “s”.

The syllable, however, involves a chest pulse and
the vowel is a more or less open conformation of
the vocal canal for emitting this pulse. The chest
36pulse and the delimitation of the pulse by consonants
are made audible by the action of the vocal
folds.

The function of the vowel in the ballistic pulse of
the breathing apparatus which produces the syllable
is entirely different from the function of the consonants.
The consonant has a function in the movement
of the syllable, it constricts the vocal canal to
delimit the chest pulse; the vowel on the other hand
is a shaping movement to emit the chest pulse. The
articulatory movement which produces the vowel
conformation usually involves a movement of the
jaw as well as of the lips and tongue. The coordination
is simple; just as in violin playing the fingering
changes the pitch of the tone produced by the bow,
so the vowel shaping determines the timbre of the
tone produced by the chest pulse. When the syllable
disappears the vowel perforce goes with it; but the
consonant as a distinct auxiliary movement may
transfer to a neighboring syllable.

Rousselot has shown that the vowel is heard not
only when the position has been reached, but also
while the organs are quitting position and moving
to a new vocalic formation. Cf. Potter, Kopp and
Green. Visible Speech (64). In speech at the
ordinary rate of 4-7 syllables per sec. it is impossible
to hold the vowel position; there must be
constant change of the shaping movement and the
vowel cannot be acoustically a steady-state tone. It
is apparent that the diphthong is a compound vowel
in which the changing movement is vocalized during
the syllable pulse. Many of the English vowels are
really diphthongs, not only such recognized forms
as “the long i” and “oi” but also the “long” vowel
with “vanish”, which is characteristic of the English
pronunciation. The apparent simplicity of the
vowel sounds of a language when produced and
when heard is due to the simple, definite articulatory
movement which produces the vowel. At the rate
of 12-18 per sec. successive events fuse to qualitative
aspects of the mass of sound. The rate of
succession of the syllable factors is well within that
range. Every emitting vowel shape, however complex
it may seem on analysis, is physiologically a
simple movement, capable of a rapid rate of utterance,
of 7-10 syllables per sec. Only a movement of
that type can be handled in speech. In all languages
the stress and rate, however light and approximately
uniform, modify the duration of the syllable and
vary the vowel quality. But the syllable is recognized
as the same syllable with the same emitting vowel,
in spite of variations. The inevitable changes which
occur constantly in producing syllables, and the
conditioning when the language is learned, lead the
user to recognize the vowel as the same in spite of
extensive change.

A. . The functions or the vocal folds

For various reasons, the action of the vocal folds
has never been counted to be like that of the other
organs of articulation. It is true that the glottis
closes or opens the vocal canal, just as do the other
organs of articulation. But in the main this closure
is not absolute and there is a tendency to treat the
action of the vocal folds as if they merely added a
quality to the outgoing breath pulse. The breath
pulse is said to be “voiced” or “unvoiced”.

There are very real reasons for this usage. It is
only rarely, as in the “glottal stop” (coup de glotte),
that the larynx acts as a true consonantal organ.
In the production of the vowels and of the voiced
consonants, the action of the folds is not consonantal.
The vocal folds have no part in the delimitation
of the syllabic movement, as has the movement
of a consonant. In fact, the larynx does not
initiate any syllabic or consonantal or vocalic movement.
It is an open question whether the larynx
can be said to initiate a tone. While the pitch is due
to the action of the larynx, the inception of the
tone is due to the pulse of the breathing apparatus
for which the vocal folds may be adjusted in the
various types of vocal attack. At first sight it
would seem that the difference between a voiced
occlusive (“b, d, g”) and an unvoiced occlusive
(“p, t, k”) must be due to the action of the larynx
alone. And yet both A. Graham Bell and Rousselot
report cases in which speech with an artificial
larynx presented satisfactory voiced and unvoiced
37occlusives, when in these cases the difference between
the voiced and unvoiced occlusives must have
been due to the management of the pressures above
and below the artificial larynx, and not to the
larynx itself, the reed of which is not under control
but is ready to sound at any and all times. There
can be no question as to the accuracy of these two
observers — though oddly enough the importance
of the case for the theory of the voiced occlusives
occurred to neither of them. Bell (2) might have
found important evidence against his theory of continuous
breath-pressure during speech. Rousselot
(69) was concerned solely with the question of the
vowels.

The notion of the vowel as a “position” has more
in its favor than the notion of the consonant as a
position; the vowel prolonged does become a position.
The classification of the vowels based on
systematic observation by A. Melville Bell and
Henry Sweet was in terms of tongue position supplemented
by the position of the lips and larynx mass.

The development of methods of acoustic recording
led to the theories of the vowel quality; in
acoustic terms by Helmholtz (29), C. Stumpf (93),
D. C. Miller (53), Crandall (12), Gemelli (19), et al.

Working with prolonged vowels and applying
harmonic analysis they came to fair agreement as to
the “formants” of the vowel. The prolonged vowel
is characterized by the prominence of one, or two
zones in the frequency spectrum which are definite
for each vowel, and which are independent of the
pitch of the “larynx tone” produced by the glottis.
Gemelli's analysis was based on the disposition of
the characteristic regions of the frequency spectrum
in relation to each other, and does not assume the
formant bands (using the analysis of Vercelli, which
includes anharmonics).

The assumption in all these cases was that the
band of frequencies, reinforced for the formant, was
composed of (changing) partials of the periodic
larynx tone. There was some obvious evidence
which controverted this view. Whispering is a
common form of speech with intelligible vowels,
obviously without a larynx tone. In the artificial
larynx a free-swinging reed with well-defined period
and partials does not give satisfactory vowels. A
weighted reed resting against a seat which gives
many close-spaced frequencies and indeterminate
noise within a limited range, and which speaks and
stops abruptly with variations of pressure, does give
fair vowels. Even in the slow, careful speech studied
by Gemelli (19), Curry (13), et al., it was difficult
at times to locate in the oscillogram the few “typical
cycles” (cicli tipici) which the theory demanded for
the specific vowel quality; the effect of the releasing
consonant often extends well into the vowel, and
the effect of the arresting consonant appears so
early in the vowel that little remains of a “pure
vowel” unaffected by either consonant.

The notion that the vowel quality depends on the
partials of definite cycles is barred by testing the
phase relations. If two films printed from an original
vowel, and therefore identical, are played back
simultaneously in various phase relations, the vowel
quality is not affected by any phase relation between
the two series of vowel oscillations. One of
the identical films may be played back in reverse
concurrent with the other film, without disturbing
the vowel quality.

The shape or movement heard in the vowel is not
due to an overall pattern involving the entire vowel,
but to cues which appear in any part of the vowel
duration. Perhaps this might have been expected
in view of the fact that the vowel starts from a
variety of preceding cavity shapes and consonant
constrictions, and the rate of movement of the
different members involved in the incipient vowel is
different; the jaw, the lips, the back and the front
of the tongue do not move at the same rate (36).
After having approximated the normal shape, the
vowel pattern shifts to prepare for a variety of consonant
constrictions and cavity shapes which may
follow a spoken vowel. Therefore a single overall
pattern is not possible and the vowel must be recognized,
though modified, by various interpenetrating
movements.

The presence of two formants in many prolonged
vowels, and the earlier formulations of vowel positions,
38provoked the study of the cavities of pharynx
and mouth. Grandgent (20) in 1890 had undertaken
actual measurements showing a two-cavity system
for the vowels.

Paget developed plastocene models of coupled
resonators to represent the cavities of (steady-state)
vowels. X-ray photographs were used by G. O.
Russell and others to determine the positions of the
organs in the vowel series. The cavities are formed
in the front and back regions of the mouth and
pharynx, modified as to volume by the tongue and
jaw movements, and modified as to orifices by the
lips and by the tongue approaching the palate in
the median range.

Meantime the evidence from whispering and from
the artificial larynx was reinforced by observations
which showed that the vocal folds constitute a
relaxation oscillator of irregular cycle to which an
analysis of the Fourier or Vercelli type does not
apply; and specific experiments show that vowels
can be produced from noises unrelated to the human
voice. The essential is a range of frequencies 250-3500
within which most of the frequencies are
represented; the acoustic material for the vowel may
be a white noise rather than a complex tone. See
Fig. 10 showing five cycles of a vowel, mixed and
integrated.

The brief duration of the vowels in ordinary
rapid speech makes it impossible for resonances to
develop from varying partials of a specific larynx
tone of changing frequency. Moreover, the glottal
output is not a series of regular cycles but a series
of irregular cycles of damped oscillations and often
without pitch.

The term “modulation” used by the Bell engineers
is better than the term “resonance” as applied to the
process whereby the energy distributed through this
range of the sound spectrum is concentrated in
certain zones. The changing shape of the cavities
plays the principal part in intensifying the bands of
frequency which the Bell engineers call “bars”. The
work has been empirical, searching out the ranges
of frequency which seem characteristic and which
show changes most obviously. There is as yet no
precise indication of the changing shapes of mouth
and pharynx, or of the interaction of the cavities
responsible for the modulation. And there is no
question that the shapes change.

B. Relations of vowel and consonant

Most striking are the changes of the modulatory
bars by the influence of the releasing and arresting
consonants. (3,64) Where a consonant opens from
its constriction it changes the modulating cavity by
producing a rapidly enlarging orifice. This change
of the modulating cavity changes the vowel “bars”;
the most indicative bars are said to be “kinked” at
the beginning by the consonant. When a consonant
arrests the syllable, the consonant produces a rapidly
diminishing orifice, changing the modulating cavity
and “kinking” the vowel bars in the opposite direction.
In forms like “gag” or “bob” the reversal of
the warp is very apparent, and at no point in the
vowel are the bars free from the warping of one
consonant or the other

The function of the vowel shape is not only to
emit the syllable with a specific quality — often
variable in many languages — but also to make
audible the releasing and arresting factors of the
syllable. The earlier analyses of acoustic records
(Marichelle (50), Gemelli (19), e.g.) show the influence
of the consonant on the vowel, but the
visible-speech patterns make apparent the place of
consonant constriction from which the vowel shape
opens or the place of consonant constriction to which
the vowel shape closes. And for the first time in
acoustic recording the syllable unit is usually apparent.
The treatment by Potter, Kopp, arid Green
confuses consonants, which are vowel-like in appearance,
with the vowels, and does not take note of the
rapid rate-of-change which must be heard as a discontinuity,
and which requires a ballistic movement
to produce. But when allowance is made for those
two ambiguities, the patterns of visible speech are
significant confirmation of kymograph and action-current
recordings of the syllable unit with the syllable
factors.

The vowel makes the releasing factor and the
39arresting factor of the syllabic audible. The consonant
constriction may involve airflow and sudden
contacts which produce noises to be heard at close
range; but such noises are mainly inaudible across
the ordinary room, and the place of the consonant
constrictions is indicated in the changes of the vowel
bars of the visible-speech pattern. (Fig. 36.) .Although
the kymograph, oscillograph, and visible-speech
recordings all show the chest-release and the
chest-arrest of syllabic forms OVO, CVO, OVC,
the usual phoneme writing as V, CV, VC obscures
the syllable factors. The syllable pulse, in the chest-released
forms, is released by the chest muscles (the
internal intercostals) into the vocal canal which is
shaped for the vowel. Meanwhile the air pressures
in the mouth and just outside the mouth are equal,
and the vibration of the glottis starts as the oscillograph
oscillations, and the ‘bars’ of visible speech
assume the vowel pattern at once. The syllable
pulse, in the chest-arrested forms, terminates with
the opposing contraction of the antagonist chest
muscles (the external intercostals) with an abrupt
fall of the air pressures, and with a vowel pattern
which diminishes in intensity, but shows little change
in form. In a language with heavy stress, like
English, the stressed vowels in slow, careful type of
utterance tend to diphthongize in the familiar
“vanish”; the quality of the end of the vowel is
modified toward shwa. The chest release and the
chest arrest may or may not be phonemic.

Sequences like OVC OVC OVC (VC VC VC)
“ape, ape…, eat, eat,…” show the arrest by
the consonant and introduce an intersyllabic pause,
if the syllables ape, and eat are maintained; if the
rate per sec. is increased beyond 3.5 per sec, the
syllables become “pay, pay… and tea, tea….”
(Fig. 27) which are now chest arrested, while the
arresting consonant has shifted to the releasing function
in the next syllable. The consonant is fused
with the vowel in the form OVC, “ape, eat”; but
the consonant does not affect the following vowel
unless the syllable is re-organized as “pay, tea”, with
the consonant functioning in the releasing position.
Thus the vowel in its interaction with the consonant
helps in indicating the structure of the syllable.

C. Confusion of vowel and consonant

The semi-vowels are related to the vowels in this
fashion: If the vowel shaping of /i/ is made a stroke
to the limit of the constriction, and suddenly released,
the consonant y (j) results. “E, E”, — OVO,
and “ye, ye” — CVO gives the contrast. If the
vowel shaping of /u/ is made a stroke to the limit
and suddenly released, the consonant “w” results.
“Ou, ou” — OVO, and “woo, woo” — CVO gives
the contrast.

In a series of syllables in which the conditions of
rate and stress force an elimination of syllables, the
/i/ and /u/ are often reduced to the related consonant
strokes, e.g., “petunia, Rouen”.

The liquids and nasals, “l, r, m, n, -ng”, are
continuants with slight constriction, and therefore
may function as vowels, i.e. as “syllabic consonants”,
e.g., Hrdlička, symptom. Though any continuant
may be given the function: “whew” (aspirated
whistle), “shhh”, “pst”.

If the functions of the factors in the syllable are
kept in mind, the difference between the vowels and
consonants is always clear, and the reasons for a
change of function can always be traced.

The apparent interchange of function of consonant
and vowel has helped to confuse the distinction
between consonant and vowel for those who look
on the phoneme as the independent, basic unit, and
do not realize the basic function in the syllable
which is the distinguishing trait of consonant and
vowel. If an articulation emits the syllable pulse,
whether it is characterized by consonant timbre, or
even by an extraneous white noise, it is a vowel.
If an articulation, whatever its timbre, releases or
arrests the syllable pulse, it is marked by an auxiliary
stroke and is a consonant.

D. Tense and lax vowels

The contrast noted by Passy (61, p. 96) between the
“tense” vowel of the French “site” and the “lax”
vowel of the English “sit” is due to the fact that
the French “site” is arrested partly by the chest
muscles, so that the “-te” may tend to become
a separate syllable in emphatic speech. The English
“sit”, on the contrary, is arrested by the consonant
movement alone. The complete “relaxation” of the
English vowel is of the arresting chest muscles
(external intercostals). In the French “site”, on the
contrary, there is a contraction of the arresting chest
muscles (external intercostals) along with the consonant
movement at the close of the chest pulse.

E. Duration of vowels

It has long been customary to speak of the “length”
of the vowel. The study of oscillograms shows that
the vowel cannot be separated from the consonant
of the syllable. Gemelli's records give the gradual
transition of the acoustic pattern of the consonant
to the acoustic pattern of the vowel; the change is
continuous; and the vowel flows gradually into a
following consonant. In reality, the “long” and
“short” apply to the syllable not to the vowel.

The ancient doctrine of vowels (syllables), which
are long by position, implies that an arresting consonant
which belongs to the syllable in question
42lengthens the vowel (syllable); and the studies of
syllables in the laboratory confirms this lengthening.
If the consonant (or compound consonant, “cluster”)
belongs to the following syllable it does not lengthen
that syllabic. Experimental studies show that such
releasing consonants overlap the vowel process and
do not add to the duration of the syllable.

Languages with a heavy word stress have syllables
of varying duration, and the quality of the vowel
is affected. In the stressed syllables the vowels tend
to lengthen, and even to diphthongize (Brechung);
in the unstressed syllables, which are uttered at rapid
rate, the vowels are reduced and may even disappear
with the syllable.

It is a familiar fact that the English “long” vowel
may be shortened and reduced, but that the English
“short” vowel cannot be prolonged; the quality of
the “short” vowel depends on the rapid utterance
of the vowel. If the short syllable is heavily stressed,
an arresting continuant will be prolonged; or the
intersyllabic interval will be prolonged. The time is
necessary for neutralizing the momentum of the
heavy stress. The “e muet” of the French and the
“dark e” of the English are cases of a vowel reduced
to its lowest terms; there is as little as possible
vocalic movement from the neutral or “resting”
position of the vocal apparatus, and the vowel is as
brief as possible. Since the “short” English vowels,
cf. “pat, pet, pit, put”, are also very brief, they
seem to approach the “dark e” in quality; and they
also cannot be prolonged. The difficulty of singing
words like “mother” and “heaven” is a familiar
illustration of the change in vowel which must occur
if the vowel is prolonged.

The French vowels are all rather brief and there
is little apparent change during the vowel. In the
utterance of the “long” English vowel, on the other
hand, the vocalic apparatus is shifted to the next
sound, while the vowel is still sounding, and a
secondary quality is always heard at the close. This
is very marked in vowels having the heavy word
stress common to the Teutonic languages.

The movement of the jaw is often part of the
characteristic of certain vowels, and this involves
a variation in the duration of the syllable. (21, 27,
28, 52, 59)

In a language like French, where the word stress
is very light and variable, it is possible for an observer
like Verrier to assume “isochrony”, i.e. that
the syllables tend to be of equal duration.

In some languages the vowel duration is phonemic;
e.g. Finnish. The contrast between long and short
is observed. Often the conditioning from generation
to generation preserves such a contrast, although
the qualities have changed radically. Thus in English
the traditional “long i” (as in “pine”) contrasts with
the “short i” (retained as in “pin”) and the traditional
“long e”. (as in “he”) contrasts with the
“short e” (as in “hem”). This illustrates the effect
of conditioning in determining the habitual associations
of the articulations.

F. The vowel system

It is customary to speak of the “vowel system” of
a language, of which such traditional contrasts are
a part. The extreme vowel positions may be thought
of as /i/, /a/, and /u/; and the specific series of vowel
positions can be indicated between them, and there
may be combinations in which a position of one
series is fused with a position of the other series.
Each language has its “cardinal vowels” which
usually include the extreme positions. Although
the classification of the vowels is physiological, not
acoustic, the vowel positions chosen in a given
language are less obvious than the bearings of the
consonants. Within this range of recognized vowels,
various changes due to rate and stress appear, which
are notated in terms of the system, though the
gradations may actually be continuous. Between
the stressed form “the” and a completely unstressed
“the” there is an indefinite series, unrecognized in
any phonetic notation.

G. Physiological traits of the vowels

There is still very little exact knowledge of the
movement complexes of the vowel. Although
palatograms show some of the bearings, kymograph
tracings show the movement of jaw and velum,
43moving pictures show lip positions (early, Marey-Marichelle),
kymograph tracings show details of
air flow and air pressure in the chest and vocal
canal, X-ray profiles, both ‘stills’ and movie films
(too few frames per sec.) show something of the
changing shapes of the vowels, and the visible-speech
patterns show changes, produced by the
opening and closing of the consonants, the data
are not adequate; the changes are rapid and often
millimeters are significant.

For the systematic indication of the vowels the
older physiological descriptions are revised, or
standard words are cited. Some have suggested
that the visible-speech patterns developed in the
Bell Telephone Laboratories might well become
standard indication of the stable vowels. They do
not at present indicate the significant detail of the
vowel range of the various languages; but it is
possible that the method can be made more delicate
especially for particular purposes.

However objective a frequency spectrum indication
of the vowels may be, or however clearly
differentiated, such a record might define standard
vowels in given conditions but it would do nothing
toward explaining the movements which underlie
the vowels and which determine the relations and
the changes of the vowels in every language. It is
not a matter of “diachronous change” but of the
synchronous changes which figure from sentence
to sentence in everyday speech. The same vowel
may appear in a single English sentence as i) prolonged
with ‘vanish’, 2) as approaching shwa, and
3) as a nul form; changes of rate and stress in that
same sentence will give a new series for that same
vowel. It is certain that such changes and reversions
are not matters of the relations of acoustic patterns,
but are matters of the relations of movements.

7. The function of the movements of the jaw

The jaw figures in the movement of the lips for
the labials, and may be primary at maximum rates,
because the jaw movement is more rapid than that
of the lips. And the jaw has a secondary part to
play in some linguals since it shifts the general
position of the tongue.

But the primary function of the jaw is in opening
the vocal canal for the emission of the syllable.

The releasing consonant stroke closes the vocal
canal so that the action of the chest muscles compresses
the air; then the rapid back stroke of the
lip or tongue, with the opening of the jaw, releases
the air. The beat stroke, and often the back stroke,
occur during the beat stroke of the chest. At rapid
rates the back stroke of the consonant, the opening
of the jaw, and the back stroke of the chest, tend
to coincide, as well as the beat strokes. The opening
and closing of the jaw marks the series of syllables.

In a series of syllables with arresting consonants
like “up, up…” the syllabic pulse is started by the
chest muscles; it is “chest-released”. The jaw parts
as the mouth is opened for the syllable emission,
although the vowel movement is at minimum.
But the syllable pulse is stopped, arrested, by the
consonant stroke; the jaw movement is essential to
the arresting “p”, and the opening and closing of
the jaw indicates the syllable movement.

The shaping movement of the vocal canal, which
gives rise to the vowel quality, is often accompanied
by the opening of the jaw.

As the syllable closes, the jaw often closes because
of the adjustment for the new vowel and for the
arresting and releasing consonant movements at the
syllable frontier.

The neutral vowel, shwa, and the reduced
vowels which approach shwa, have the least vowel
movement and such syllables may be produced
without any jaw action whatever.

“Long” vowels and diphthongs involve a more
pronounced opening of the jaw. The common
tendency in English to diphthongize all stressed
“long” vowels makes the jaw movement prominent.

8. The movement of the consonant

Although the same “sound” has sometimes been
counted both a vowel and a consonant, in one form
or another, a distinction has always been drawn
44between the vowels and the consonants. It is the
custom to recognize a vowel “i”,. and a consonant
“i” (the consonant “i” in English is often spelled
“y” as in “yes”; in French it is often unchanged, as
in “rien, dieu”); a vowel “ou” and a consonant
“ou” (written “w” in English but unchanged in the
French as in “oui”). The difficulty of making a
distinction has given rise, as we have seen, to the
term “semi-vowel” which is sometimes applied to the
liquids “l”; and “r” and to the nasal “m” and “n”.
Sievers has been at pains to point out that the same
“sound” may function in rapid alternation as a
consonant and as a vowel; “berittenen” has already
been cited. But aside from the practice of dividing
“sounds” into the “syllabic”, which is supposed to
be the core of the syllable, and the “con-sonant”,
which is supposed to be added to the vowel, there
has been little attempt to define the function of
the consonants. Often the consonant has been
considered quite secondary to the vowel. Marage
defined the consonant as a supralaryngeal noise
preceding or following the aero-laryngeal vibration
of the vowel; he felt that the consonant occupies
but a small place in the syllable, and therefore
should be subordinated to the vowel. (16, 48)

Rousselot applies his general analysis of the tension-tenue-détente
to the consonant; this describes
the position of the organs and can be made to
explain the important difference between the
implosive (arresting) and the explosive (releasing)
functions of the consonant. Fig. 25. The study of
historic phonetics has shown that there is an
important distinction between the consonant which
closes a syllable and the consonant which opens a
syllable. The implosive consonant which closes the
syllable is called appuyante when it abuts the
explosive consonant of the next syllable which is
called appuyée. The different fate of these consonants
in the evolution of the Romance languages
led to this distinction. The terms imply that the
explosive consonant of the abutting pair is somehow
supported and so buttressed against change, while
the implosive is supporting and may disappear.
Rousselot defines the explosive consonant as consisting
primarily of taking-position and holding-position,
while the implosive consists primarily of holding-position
and quitting-position. F. de Saussure
counts the implosive-explosive junction important
for the division of the syllables. But these analyses
do not touch the fundamental question of the
relation of the movement of the consonant to the
movement of the syllable.

The consonants are to be divided into releasing
consonants which start the syllable movement and
arresting consonants which stop the syllable
movement. “Explosive” and “appuyée” are other
names for the releasing consonant; “implosive” and
“appuyante” are other names for the arresting
consonant.

A. The arresting consonant

In a series of syllables like “up, up…” the syllable
pulse is started by the chest muscles; it is “chest
released”. But the syllable pulse is stopped, arrested,
by the consonant stroke. The consonant closes the
vocal canal, and the rise of air pressure in the
pharynx and chest takes up the momentum of the
chest wall and brings the movement to a stop.
Figs. 26, 27 and 28 show that the highest chest
pressure of the pulse occurs when the consonant
stroke closes the vocal canal. The lowest chest
pressure between the syllable pulses occurs well after
the consonant. The next syllable pulse does not
begin until after the back stroke of the arresting
consonant. There is a hiatus in the sound, for
vocalization of the vowel of the next syllable does
not occur until the delayed rise in chest pressure.

The pressure in the mouth has the typical arresting
form; the closure of the lips brings the rise in
mouth pressure which accompanies an arresting
consonant. The curve has the rounded form due
to the gradual subsidence of pressure; the chest
pressure sinks as the chest muscles prepare the next
pulse.

When the pressure just outside the mouth is
recorded with a modified form of Rousselot's
apparatus, the tracing shows something of the
variations in the chest pressure, interrupted of
45course during the closure of the consonants. The
pressure just outside the mouth varies with the
mouth pressure which derives from the chest
pressure, but is radically modified by the valve of
the glottis and the changing volume of the pharynx-mouth
cavities. The mid-sag in the pressure outside
is not due however to any change in the chest
pressure, but to the enlargement of the mouth
cavity as the mouth opens which affects the pressure
in the mask before the mouth. There is a corresponding
rise in pressure in the mask when the
mouth cavity is rapidly diminished as the mouth
closes. The point of minimum chest pressure is
well marked in the tracing of pressure outside, and
the rapid rise for the next vowel is clearly indicated;
the vocalization usually appears in this tracing
if the membrane is sensitive. Tracings of whispered
syllables do not often show the mid-sag; the glottis
is open enough to sustain the mouth pressure in
spite of changes in volume and the curve of pressure
outside merely shows a steady decrease in pressure
during the syllable. Fig. 26 may be compared with
Fig. 29.

Passy notes that the English can differentiate
“an aim”, and “a name”, i.e. an arresting consonant
may persist in English where it would shift to the
releasing position and transfer to the next syllable
in French. (61) This is true, but only at a moderate
rate. If “an aim” is said faster and faster it soon
becomes “a name”. The “n” has shifted from the
arresting function in “an” to the releasing function
in “name”. The arresting consonants adds to the
duration of the syllable and, as the rate increases,
tends to shift to the releasing position in the next
syllable where it does not require extra time. Binet
and Henri (4) found that at maximum speed the
syllables of the digits with consonant arrest required
more time than those with open syllables. This
change can easily be studied in experimental series.
Fig. 26 shows “op, op…” becoming “po, po…”
and as the rate slows reverting to “op, op…”.
Fig. 27 shows “eat, eat…” becoming “tea,
tea…”. Fig. 28 shows “up, up…” becoming
“pu, pu…” and reverting to “up, up…”. The
reversion to the original form, and the increase in
chest pressure and decrease in the consonant stroke
at rapid rate will be considered later.

B. The releasing consonant

This consonant releases the syllable pulse. The
stroke of the expiratory chest muscles and the beat
stroke of the consonant occur at the same time.
The consonant stroke closes the vocal canal so that
the action of the chest muscles compresses the air,
then the rapid back stroke of the consonant releases
the air. The beat stroke and often the back stroke
of the consonant occur during the beat stroke of
the chest. At a rapid rate the back strokes of the
consonant and chest muscles, as well as the beat
strokes, tend to coincide. The releasing consonant
never adds to the length of the syllable and it
actually accelerates the syllable movement. The
coordination is much faster than that of the arrested
syllable. The rate may be as high as 9-12 per
sec., which is the maximum rate of a skilled
movement.

Tracings show that the pressure in the chest
continues to fall during the arrested syllable when
the consonant closure occurs, while the pressure in
the mouth rises. Although the consonant muscles
and the chest muscles contract at the same time
at the releasing consonant, the chest pressure does
not begin to rise until the middle of the consonant
contact. For the actual relations cf. Fig. 48, syl.
10-12, Fig. 50, syl. 4-5, pp. 66, 68. This lag of
.03-.05 sec. is due to the size of the cavities
involved. In speech the consonant muscles quickly
close a cavity of 200 c.c. at most, while the chest
muscles act on a limited area of the walls of a
cavity of 2500 c.c. at least. The pressure conditions
in a rapid series of open syllables permit vocalization
to begin immediately after the releasing consonant
and to continue up to the releasing consonant of
the next syllable. This has led some to assume that
the median releasing consonant belongs to both
syllables. (103, p. 357) But the consonant belongs
to the syllabic in which it functions.46

C. The releasing and arresting consonants
as two functions of one movement

The consonant stroke seems now to open and now
to close the vocal canal; how can it be counted the
same thing in these different cases? ; When the
tongue moves from the neutral position and makes
a quick stroke to the palate as in “a-la” or “hut”
there seems little question that the stroke is made to
the palate. In the word “hut” the stroke must occur
in that fashion; but in the case of “la” or “to” the
tongue may be in contact with the palate before the
essential stroke occurs. While the same spot is involved,
there seems at first sight a radical difference
between the essential stroke of the releasing consonant
in “ta” and “la” which seems to be from the
hard palate, and the essential stroke of the arresting
consonant in “at” and “al” which is to the palate.
Apparently the arresting and releasing strokes are
in opposite directions, although they move over
precisely the same path. But the movement shifts
from arresting to releasing and back again, and at
fairly rapid rate; therefore it is improbable that the
entire coordination of the movement changes with
this change of function.

The consonant movement involves an obstacle,
an opposing surface to which and from which the
stroke of the consonant plays. The detail of such
movements to an opposing surface will help in
understanding the arresting and releasing functions
of the consonants. It is sometimes said that a rapid
movement is likely to be followed by a “rebound”;
in reality, the return movement, the back stroke, is
not due to elasticity, but to the accurately timed
contraction of the negative muscle-group which
arrests the member and returns it to the starting
point. It is this rapid back stroke which is sometimes
called the “rebound”. In case the movement
is arrested by an obstacle, the opposing surface,
precisely the same sequence may occur. The momentum
of the movement is taken up by the
obstacle, and the negative muscle-group acts at the
proper time and returns the moving member. The
time consumed proves to be precisely the same as
if the opposing surface had not intervened. Plotted
against time, the movement would have presented
a u-shaped or v-shaped path at the end of the stroke.
The opposing surface truncates this u-shaped or
v-shaped curve and gives a flattened form, Fig. 30,
but the time relations are the same as if the
movement had been completed with self-arrest and
normal return, Fig. 30. When the movement barely
reaches the obstacle, barely flicks the opposing
surface, the curve of the complete movement is
practically normal, and the truncation is eliminated;
the member does not remain in contact with the
surface, Fig. 30. This means that although the
movement is to a surface, the holding in position is
eliminated. Many of Rousselot's records show such
forms. (70, p. 597, Fig. 387, p. 390, 949. Fig. 631,
p. 952, Fig. 635, 636; 86, p. 29),

This normal movement may be modified by
eliminating either the excursion to the opposing
surface, or the excursion from the opposing surface.
A blow can be delivered to an opposing surface
when the member is in actual contact with that
surface. The essence of the blow, of the beat stroke,
is the sudden contraction of the positive muscle-group
which exerts pressure on the opposing surface.
Such blows delivered to a surface with which the
member is in contact are very common. In piano-playing
and in typing the finger is often in contact
with the key struck; precisely the same movement
can be executed with the finger resting on the
surface of a table. In the case of the stroke to the
piano- or typewriter-key, the back stroke may be
of large amplitude, and may be the most obvious
thing in the movement; this is true of “staccato”
playing.

Translated into terms of phonetics this would
mean in the case of the explosive “t” that the tip
of the tongue may be in actual contact with the
palate when the blow is delivered to the palate.
The beat stroke actually occurs, although the tongue
tip is in contact with the opposing surface at the
time of the sudden contraction of the positive
muscle-group. The methods of recording employed
47to date have not been very satisfactory for showing
this sudden pressure against the opposing surface,
nevertheless it does appear in some records published
(72, p. 77, Fig. 68). In the word “pap-pa” the
lips are still in contact when the “p” of the second
syllable occurs, but the sudden pressure which is the
beat stroke of the releasing “p” shows clearly. The
beat stroke is not eliminated although the excursion
of the lips does not return the member to the original
position; contraction of the positive muscle-group,
and a pressure is a stroke.'

The “condensation of the back stroke”, the
elimination of the excursion from the opposing
surface, is as often observed. When the blow is
struck to the opposing surface there may be no
“rebound”; the momentum may be taken up by
the obstacle and the negative muscle-group does
not return the member to the original position.
The relaxation of the positive muscles and the
preparation of the next movement may occur
while the member remains in contact with the
surface to which the blow is struck. In the
pronunciation of the word “hut” the tongue may
deliver the stroke of the “t” to the palate and
remain in contact during the relaxation process.
This is the “condensation of the back stroke”. The
tongue may or may not leave the palate at once; in
any case the excursion from the palate has nothing
to do with the consonant.

This analysis of the consonant stroke makes the
arresting consonant and the releasing consonant
modifications of the one movement. The essential
thing is a stroke to the opposing surface, which
may be given in contact, and which may or may
not be followed by a back stroke from the surface.
When the beat stroke occurs in contact the only
apparent movement may be from the palate in the
releasing consonant. When the back stroke occurs
in contact the only apparent movement may be to
the palate in the arresting consonant.

The study of the tracings proves that all
consonants, whether continuants or occlusives, have
essentially the same function of releasing or arresting
the syllable movement. Cf. later material on
the double consonant and on other abutting pairs.

E. W. Scripture quotes similar values from
Grégoire (21, p. 161, 263, 418; 76).

These values compare very well with the intervals
of rapid movements in other fields. The interval
.180 sec. is equivalent to a rate of 6.6 per sec.;
.120 sec. is equivalent to a rate of 8.3 per sec. 6.6-8
per sec. is well within the limits of the “tapping
time”, or “repetition time” as measured for rapid
movements of the hand. Such repeated movements
involve, of course, the beat stroke and the back
stroke. The ordinary method of taking “tapping
time” involves a stroke to an opposing surface. The
records, however, are like those taken of a free movement
with self-arrest. The ballistic stroke (contraction
phase) of most of the consonants can be executed
with great rapidity, in .020-.040 sec, a fact
which has important bearings on the functions of
the consonant in the syllable movement.

D. Detail of consonant movement

The lingual occlusives “t/d” and “k/g” show the
stroke and the opposing surface very clearly but
these are not so apparent in the case of the other
linguals.

The lingual continuants “s/z, sh/j (French ch/j),
ch/ch (French tch), German ch, etc.” all constrict
the vocal canal but may not close it. The organs
however spring into position with a ballistic stroke
and quit the position with precisely the same type of
back stroke as occurs in the case of the occlusives.
The fundamental distinction lies in the fact that
they may not completely close the canal; in other
48respects their action and their time relations are
much the same. The fact that the breath is escaping
gives them a continuous sound during the movement
while the corresponding closure of the occlusive is
silent; but this does not affect the function of the
continuants in the least. Cf. Fig. 45. Figs. 31 and
32 show forms in which the pressure in mouth rises
for the continuants just as for the occlusives, and
there is often little difference in the consonant contact.
It is the difference between a tight and a leaky
closure.

The labials include the occlusives “p/b” and the
continuants “f/v”. The movements of the lips are
fairly obvious; in reality, the lower lip is primarily
the moving member.

The tongue may be substituted for the lower lip.
With a little practice, all the labials can be made
successfully with the tongue striking the upper lip
(or the lip and teeth).

The base of the tongue plays against the palate
and the velum in producing the gutturals “k/g”,
German “ch”, (hard), Greek gamma, and French
“r grasseyée”. The tongue is primarily the moving
member.

The glottis figures in two sounds. The aspirate
“h”, common in English and occurring on occasion
in French (60), is, strictly speaking, not a consonant.
Statements like Clédat's (10, p. 73) are merely
traditional. It is really a modification of the vowel
and cannot occur in the arresting position. The
rarer glottal stop (coup de glotte) figures as a consonant.

E. Voicing of consonants

As a whole, the consonants are divided into two
parallel series, the “voiced” and “unvoiced” (sonant
and surd). It is evident that the quality is due to
the action of the vocal folds which are often supposed
to act independently, like the other organs
of articulation. But the vocal folds do not come
into play as a separate factor. Instead, they are
activated by the pulse of air from the chest. And
there are reasons for thinking that the difference
between the voiced and unvoiced consonants is not
due to the direct action of the vocal folds, but rather
to the coordination of the articulatory apparatus
and the chest. When the pressure in the buccal
cavity is practically that of the chest, there is no
flow of air to activate the vocal folds. When the
pressure in the buccal cavity is reduced by the consonantal
release of air, or when the expiratory
muscles increase the chest pressure, the vocal folds
are activated. The experience of the subjects with
an artificial larynx has already been cited.

A study of the mouth pressure for surds and
sonants in rapid speech shows that the difference in
pressure, expressed by the terms “fortis and lenis”,
is more fundamental than the voicing of the consonants,
and persists after the voicing distinction is
lost, i.e., after both surd (fortis) and sonant (lenis)
are voiced. Musehold's statement that the velum is
momentarily lowered for “b, d, g” is quite mistaken.
The airflow necessary for the occlusives “b, d, g”
is provided by the descent of the larynx mass. The
velum is closed for all consonants except the “nasals”.
“n, m, -ng”. The relation of these accessory movements
which change the quality of the consonants is
somewhat the relation of the movements of the
vowel to the syllable movement. The movements
occur together but they do not condition each other,
though they cooperate in producing the given consonant
(35; 56, p. 51).

9. Types of syllables as determined by the
consonants

Elaborate combinations of movements are frequent
in complex habits. In the playing of wind instruments
auxiliary movements sometimes release the
chest pulse as in the case of “triple-tonguing”; the
movement is very seldom arrested by an auxiliary
movement. A whistled note is sometimes brought to
a close by a consonant stroke resembling the “t”
stroke.

In flicking a coin or shooting a marble, the finger
holds the thumb (or the thumb holds the finger)
during a growing tension which is finally released
by the auxiliary movement. In coughing, the pressure
of the chest compression is finally released in
the throat.

The movement of shaking down the mercury in
a clinical thermometer involves an arrest produced
by the movement of the other hand which intercepts
the beat stroke of the hand holding the thermometer;
the movement of the left hand is actually
substituted for a self-arrest by the muscles of the
right hand. It is possible to produce somewhat the
same shaking movement with the right hand alone,
with self-arrest. A refractory fountain pen is often
subjected to the same treatment. In movements of
plucking and picking, the thumb and fingers arrest
each other's movements in closing on the object.

In the ordinary forms of syllable movement with
chest-arrest, the muscles of the breathing apparatus
are connected with each other by an intervening
system of bones, cartilages and joints. This is the
rule with antagonistic muscle-groups. At first sight
it seems difficult to think of the muscles of the
tongue or of the lips functioning as the negative
muscle-group for the breathing apparatus; the
muscles of the tongue and lips are not attached to
the chest. How can the lips and tongue act as
antagonists to the chest muscles? An examination of
the entire speech mechanism shows that the column
of compressed air in the chest and vocal canal becomes
the intermediary between the expiratory
muscles of the chest acting as the positive group and
the muscles of the consonant apparatus acting as the
negative group. At the end of the syllable, the
sudden closure (or constriction) of the vocal canal
by the consonant movement stops the escape of air,
raises the air-pressure, and thereby arrests the beat
stroke and takes up the momentum of the movement.
The movement of the consonant apparatus is
harnessed to the movement of the breathing apparatus
as effectively as the organist's finger is connected
with the valve at the pipe in an organ with
pneumatic action. When the syllable pulse is chest-arrested,
the muscles of the rib-cage act directly on
the moving mass of the chest walls to take up the
momentum and arrest the syllable movement. But
when the syllable pulse is consonant-arrested the
muscles of lips, tongue, velum, do not act directly
on the mass of the chest walls. Instead, they close
the vocal canal, and cause compression of the air
column, which reacts on the moving mass of the
chest walls. This consonant-arrest therefore takes
more time since the air pressure must rise before
force is exerted. The compressed air must be released,
usually by the détente, and the process adds to
the duration of the syllable. Often the syllable
arrest is a combined consonant- and chest-action.
But as the rate of the syllable utterance increases,
there is no time for the delaying process of consonant-arrest
and the arrest must become a chest-arrest.
This is the determining factor in the shift of
the arresting consonant to the releasing position in
the next syllable, or in the dropping of the arresting
consonant if the shift is not possible.

Fig. 33, A, B, represents schematically the possibilities
of coordination of the articulatory apparatus
with the chest movement.

The slow, continuous movement of expiration,
punctuated by the consonantal movement, does not
occur in normal speech. There is no experimental
evidence of such a correlation, and the series of
syllables in “lil' 'll lie low”, and “runnin' 'n'
neighin'” (Figs. 88, 89) do not show any tendency
51to maintain continuous chest pressure while the
syllable is initiated by the “n” or the “l”. The early
efforts of the deaf to speak may approximate this
form; but it cannot give intelligible speech.

In some cases the articulatory apparatus is fixated
as in A II, and the syllable is due entirely to the
breathing apparatus. This is the case with a series
of vowels; cf. Fig. 23.

But in most cases, the chest pulse is released or
arrested, or both, by the movement of the consonant.
The scheme B I, II represents a series like “tat, tat,
tat…”. For details of such movements, cf. Fig. 36.

The events of a series of syllables as represented
by various investigators are presented in Fig. 36.
The observations of F. de Saussure of the implosion-explosion
division of syllables and of the explosion-syllable-point-implosion
structure of the syllable itself,
are not based on experimental work; but the
concepts are essentially physiological, as he notes in
the text (75, pp. 79-80). The diagrams of Fig. 36
are drawn from experimental studies.

Marichelle bases his diagram (not shown) on his
analysis of wax phonograph grooves. He makes the
consonants integral parts of the syllable, and counts
the transition from the constriction of the consonant
to the more open vowel and the reverse process the
significant articulations. He does hot attempt to
indicate this movement of articulation on his diagram;
but he does stress the effect of the consonant
constriction oh the cycles of the vowel, both in the
explosive and implosive articulations (50, pp. 80-83).
Marichelle is aware that the consonant is perceived
in the totality of the syllable, and not as a
separate unit: he gives detailed experimental confirmation
of this view (50, p. 83, Pl. 10).

Rousselot's scheme is derived from Saussure's explosion-implosion
conception of the syllable, but
applied by Rousselot to the basic articulation which
he conceives to be the “sound” (70, p. 334-5,
Figs. 116, 117; 72, p. 25, Fig. 12).

In the first curve of Rousselot, Fig. 36, it
will be noted that the “détente” phase of one
articulation is identical with the “tension” phase of
the next (c₁ and a₂; c₂ and a₃ etc.). If the syllables
“ab-ba” had been represented as separated there
would have been a complete cycle for the articulation
of each “b”, and the series would have been
a₁ b₁ c₁ and a₂ b₂ c₂. If, however, the syllables had
been “at-la” the overlapping of phases would have
been more extensive than in the preceding figure,
for the “tenue” of the “t” would be identical with
that of the “l”, the “tension” of the “t” would
serve for the “l”, and the “détente” of the “l” would
serve for the “t”.

Fig. 36, “Motor Phonetics”, gives a group of
curves representing the events of the syllable in
terms of movements. Only the outline of the coordination
(of movements, contacts and air pressures)
is given here; the system of movements will be
treated more fully later.

In the first syllable “ab” there is no releasing
consonant; the chest pulse, curve C, is chest-released.
The descending stroke, 1-3, represents the
rapid movement of the pulse; in 1 the sudden contraction
of the chest muscles occurs, compressing the
chest. During phase 2 the momentum of the chest
continues the compression and a stream of air flows
freely through the vocal canal. In phase 3 the
movement of the consonant closes the canal and
blocks the flow of air; the rapid rise in air pressure
arrests the chest movement. In this case the consonant
movement replaces the contraction of the
negative muscle-group of the chest. The actual connection
between the chest and the accessory movement
of the consonant is effected through the
column of compressed air in the vocal canal.

Between the syllables “ab” and “ba” (between
3 and 1) the chest muscles readjust for a new compression.
The lips remain in contact but the muscles
of the lips relax in preparation for the releasing
stroke of the “b” in “ba” (1 of the next syllable).
In this second syllable “ba” the rapid contraction
of the chest muscles recurs in 1, the pressure suddenly
rises because of the consonant closure, but is
suddenly released by the back stroke of the lower
lip from the upper, and the air flows freely through
the canal in phase 2 as before. In phase 3 the
inspiratory muscles of the chest itself contract to
stop the momentum of the chest, forming an open
syllabic “ba”. The syllable movement is chest-arrested
and the air pressure remains at zero (3)
because there is no consonant closure.

Between the syllables “ba” and “vak” (between
phases 3 and 1) the chest muscles readjust for the
new compression.

In the third syllable “vak”, the consonant and the
syllable movements are coordinated in phase 1.
Phase 2 shows the free flow of air with zero pressure
in mouth. The coordination in phase 3 is precisely
that of the arresting consonant in “ab”.

The maxima of the air pressure are of rounded
form when the arresting movement closes the canal,
and the fall in pressure is abrupt when the releasing
consonant opens the canal. Cf. Fig. 37, syl. 4. An
actual tracing which will give something of these
relations will be found in Fig. 35.

The project of the Bell Telephone Laboratories,
which is concerned with “Visible Speech”, is like
that of Marichelle (50), some fifty years earlier, in
that a visual representation of the acoustic patterns
of speech is the purpose of an experimental study.
Steinberg and French (82, p. 4), Kopp and Green
(43, p. 74), confirm Marichelle's finding that the
syllable is perceived as a unit in which the consonants
are apparent primarily in their effect on the
vowel. Although the acoustic recordings in wax
which Marichelle studied gave small indication of
the syllable division, possibly Marichelle's first-hand
experience in the instruction of the deaf led him to
stress the syllable and the syllable division.

The acoustic patterns are presented in “Visible
Speech” as ‘spectrograms’ in which the changing
sound components at different frequency levels are
translated into concurrent bars, each of which
concentrates on one of a series of bands of frequency.
The bands of frequency were chosen to bring out
the prominent and most obviously modulated
spectrum zones which represent the influence of
the changing cavity-orifice of the vocal organs.

While the vowels and the influence of the
consonants on the vowels are well portrayed, the
actual movements, contacts, and air pressures of
the speech process must be inferred. In the attempt
to make the main acoustic features prominent,
variations of intensity are omitted in the present
form of the apparatus, and it is therefore impossible
to catch the pulse of the syllable which is so
important for recognizing the individual syllable
and the groups of syllables.

The findings can be collated with the movement
process of a motor phonetics, as Fig. 36 shows. (See
also Fig. 114, App. XIII.)

The constriction for the “b” brings a sudden discontinuity
because acoustic conditions are abruptly
changed, but there is no “spike”, i.e. there is no
disturbance at all in the frequency levels, because
the closure is not marked by a sudden emission of
air under pressure. The “stop gap” means the interval
of consonant closure. Since the “b” is a sonant,
it will have a “voice bar”, showing that the vocal
folds are vibrating. There will be no ‘spike’ since
there is no release of air between the doublets.
When the arresting member of the double appears
there is no release of air between the doublets.
But when the releasing “b” opens in “ba”, the
‘spike’ portrays the sudden release of air under
pressure, and the slight stop-gap fill indicates the
aspiration as the air escapes in the English syllable
before the vowel sounds. Just as the “resonance
bars” of “ab” were kinked, so the resonance bars of
“ba” will be kinked down as they sound after the
vowel. But as the mouth opens and the “ae” prescribes
the normal cavity-orifice, the resonance bars
become horizontal at the usual levels.

Assuming that the rate of the syllables is 4-5 per
sec., discontinuities will appear between consonant
and vowel, and between vowel and vowel. They
indicate the very brief intervals during which the
ballistic movement of the consonant closure, or of
the vowel shaping, is so rapid that the modulation
cannot follow the cavity-orifice change. This
indication of a ballistic movement contributes to
the perception of the total syllable movement.
Both sounds and silence, continuous and discontinuous
change, figure in the motor patterns of
speech.

The third syllable “vak” shows the voiced interval
of a sonant labial (without the spike and aspiration
of the opening English stop), with the downward
kink of the “resonance bars” indicating the
labial opening. The “resonance bars” kink upward
as the syllable is arrested by the “k” because, this
time, the constriction is in the back of the mouth,
the reverse of the lip closure.

The Microphone line gives the envelopes of the
syllables. The consonants show slightly, with the
exception of the “k” in “vak”, which is a surd. The
vowels constitute the principal acoustic events.

The internal intercostals contract to release the
pulse of each syllable; the fact that the contraction
is a momentary impulse is apparent.

The external intercostals contract to arrest the
pulse of each syllable.

The Rectus abdominis contracts in anticipation
of the breath group involved; the syllables are
assumed to occur no faster than three per sec., and
an abdominal stroke (breath group) for each syllable
is possible. If the syllable rate exceeds three per sec
the abdominal muscles fixate or make a slow,
‘tense’, phrasing movement.

The action-current tracings give the muscle action
for the syllables and for the foot- and breath-group-units.
Acoustic and pneumatic recordings can give
only indirect evidence of the movements of the
foot and the breath group.

In order to make the diagram, any reference
to the foot or breath group has been omitted. The
syllables “ab-ba vak” must be considered as feet in
a breath group. Cf. Figs. 35 and 37.

The feet are produced by the abdominal muscles
which make the stress of the foot and group the
syllables.

The breath group is the movement of expiration
involving the abdomen-chest adjustment which
groups the feet in the breath group.

10. Certain differences in the syllable as modified
by the auxiliary movements of release and
arrest

When the syllable has a consonant release, the
length of the syllable movement is not affected; the
57consonant movement fuses with the syllable
movement. The syllable may be very short, a rapid
ballistic movement, with chest-arrest, or it may be
indefinitely prolonged, changing into a controlled
movement. The consonant release does not limit
the duration of the syllable.

On the other hand, when the syllable movement
has a consonant arrest, the length of the syllable is
conditioned. The ballistic stroke of the syllable is
arrested by the consonant movement and the syllabic
movement cannot be indefinitely prolonged into a
controlled movement. Its maximum length, therefore,
is that of a ballistic movement, .13 sec. at the
most. In a group of 40 cases studied (3 subjects), the
maximum is .13 and the minimum .05 sec. The
arresting consonant may be prolonged, or a pause
may occur after the syllable, but the syllable beat
stroke is of necessity brief. This means that the
vowel of the syllable, with arresting consonant, will
always be a relatively brief vowel.

Thus the records of “closed” syllables (syllables
with arresting consonants) show a short duration of
the vowel: patte, .16 sec. (including the “p”);
patelin, .10 sec. (“pa-”) (72, p. 86). Beat, .14; bid,
.14; bin, .10; bit, .06 sec. (37). These may be compared
with pa, .24; pate, .28 (pa-); bee, .50; bonjour,
.68 sec. (bon-); in all of which the syllable is
“open”. E. W. Scripture quotes Gregoire's observation
that for some reason in pâte sucrée the “t-s”
shortens the vowel and lengthens the occlusion. (76)
The “t” has become an arresting consonant, and the
syllable “pat” is now closed, therefore the vowel
shortens. The position of the tongue against the
palate is maintained from “t” to “s”, and the back
stroke phase” of the first syllable occurs during that
interval, therefore the occlusion is lengthened.

In English (and in German as well) the prevailing
form of the syllable is actually both released and
arrested by a consonant; it is consonant delimited,
CVC. This is true of the simplest possible vocabulary
of 300 words suggested for deaf children.
(22) It is true of the first 500 words, and of the
first 1000 and 2000 word lists of the most frequently
used words. (96)

The common syllable is not the consonant-vowel,
CVO, which many think of. At least 80 per cent
of the syllables are arrested by a consonant (OVC,
CVC); at least 70 per cent are both released and
arrested by a consonant (CVC). This means that
arresting consonants and abutting consonants are
very common; a heavy stress on such syllables will
produce the changes in the stressed and unstressed
syllables to be discussed later.

The number of different syllables runs very high;
at least 5000 are in use in English and German, and
more than half that number of syllables occur
frequently in French. It is not possible to handle
the individual syllables in a European language.
Therefore the syllables are made into classes according
to the characterized syllable factors. There are
the syllables which are released by a given consonant
or by chest action, the syllables which have a
given vowel shape, and the syllables which are
arrested by a given consonant, or by chest action.
Each syllable belongs to two or three of these classes.
The aspects which characterize these three syllable
factors (releasing, vowel shaping, and arresting)
prove to be limited; there may be 5-20 vowels,
and 10-50 consonants; these constitute the
alphabet of the language; they are the familiar
“sounds” or “phonemes”.

The compound consonants are very common.
They are composed of two or more simple consonant
strokes together which characterize releasing and
arresting syllable factors. The 300-word vocabulary
cited above includes some 37 compound consonants;
common words in English are as likely to contain
compound consonants as simple consonants.

As can be inferred from Sievers' discussion, not
all syllables which end in a consonant are actually
closed syllables. If the vowel is long in duration,
the arrest of the syllable movement, and possibly
the change to a controlled movement, will be well
under way before the consonant is uttered. In such
cases the consonant occurs with the latter part of the
syllable movement but is not an integral part of it.
Such consonants are often noticeable in singing
where the prolongation of the vowel leaves the
58consonant dangling. Often the consonant is given
as an explosive and really constitutes an inconspicuous,
added syllable. (88, Cf. Fig. 75, syl. 2.)

The tradition, in English orthography, which
makes a spelling like “mate” and “hate” indicate a
“long a”, while “mat” and “hat” indicate a short
a” harks back to the days when the “e” was the
sign of a second syllable, and the first syllable was
therefore open: ma-te, ha-te, while “mat” and “hat”
were closed syllables. (Appendix X: The Demarcation
of the Syllable.)59

IV
Influence of the phonetic units on each other

1. Function of consonants in the linkage of
syllables

A. Abutting (linking) consonants

The syllable is always incorporated in a breath
group. If there are two syllables or more they are
not only adjacent but connected in the breath
group. The movements of the syllable pulses affect
each other and their auxiliary movements may join.
The arresting consonant of one syllable and the
releasing consonant of the next syllable may be
linked.

If the group confined to a single syllable is called
a “compound” consonant, the abutting consonants
of two syllables may be called an intersyllabic pair.
In case of abutting consonants it often happens that
the arresting consonant of the first syllable and the
releasing consonant of the next syllable are the
same consonant repeated; this is the familiar “double
consonant”. It is of especial interest because the
syllable division must take place during this double
consonant. The process which divides the syllable
also makes the consonant “double”.

Sievers notes that the double consonant is different
from a single consonant prolonged. He points
out that when implosion-explosion occurs within
such a double consonant, there is “discontinuity of
expiration” and that the syllable division takes place
between the doublets. (78, p. 191)

F. Josselyn, using Rousselot's methods for the
study of Italian phonetics, discusses the peculiarities
of the Italian double consonant, showing the increased
length of the occlusion and the compensatory
shortening of the preceding vowel. It is evident
that the double consonant is very unlike the single
consonant in duration and in function. (40, p. 227
f., cf. 72, p. 78)

Rousselot publishes a tracing of the Swedish “pappa”
with a distinct depression in the middle of the
“p-p”, and points out the “p implosive” and the
“p explosive”. (72, p. 77)

He also publishes such tracings for the labial “v”,
but is certain that the other double consonants show
no such maxima. (70, p. 351, 1087) Rousselots'
apparatus could not respond to fluctuations at the
rate involved, so that his experimental findings
seemed to bear out the statement.

Rousselot's statement that a consonant between
vowels, if prolonged, becomes double, is not adequate.
The doubling depends on the occurrence of a
second consonant stroke with a new chest pulse. In
the phrase “I'm Ike”, the “m” may be indefinitely
prolonged without becoming double. In the case of
“I'm Mike”, the second consonant stroke and the
entrance of the new chest pulse with this second
“m” can be perceived. Neither Sievers nor Rousselot
explain the greatly increased length of the double
consonant or its influence on the preceding vowel.
(94, p. 4 ff.)

Only a few English and French words contain
true double consonants, but in many phrases the
consonant is doubled. In English, “up puppy; lob
Bobby; topic and top pick; top egg and top peg;
Otto ought to; hit him, hit Tim; I do, I'd do; this
eye and this sigh; Z is Z; unknown, un-own; I owe
none, I own none; I'm Ike, I'm Mike; I lie, I'll lie”
are phrases in which the contrast between the double
and single consonant is marked. In most cases the
meaning depends on making the distinction. A careful
English enunciation can distinguish between
“whole ode, hoe load”, and “whole load”; between
“thus E, the C, thus C”.

If records give proper details, one should expect
to find two distinct consonant strokes which arrest
and release the separate chest pulses of the two
syllables; and there will be two distinct maxima of
the air pressure in the mouth, as the first chest pulse
is arrested and as the second chest pulse is released.
This should be the form of the typical tracings of
doubling consonants. Cf. Fig. 36.

B. Tracings of double consonants (special
form of abutting pairs)

The tracings of the various words and phrases containing
doubled labials and doubled linguals show
without question that the “double” consonant is
actually two consonants. There are two distinct
maxima in the curve indicating the movements of
the lips or of the tongue. And the curve of the air
pressure in the mouth also shows the ending of the
one chest pulse and the beginning of the second.
The pressure for the “double”, consonant is a bi-maximal
curve, showing the arrest of the one syllable
pulse and the back stroke of the chest muscles,
before the increasing pressure for the release of the
second syllable pulse.

The division of the syllable is to be seen 1) in the
two maxima of the double articulation; 2) in the
two maxima of the curve of pressure in the mouth
(which mark the arrest of the one syllable pulse and
the release of the second syllable pulse); 3) in the
minimum chest pressure between the syllables (in
the recordings which include a tracing of the chest
pressure).

Fig. 34 shows the tracings of the phrase “up
puppy” in which the first “p” is doubled, but the
second “p”, in spite of its orthography, is not doubled.,
The utterance is fluent enough to produce the
complete double consonant; both the consonant
curve and the air pressure curve show the two
maxima. It is apparent that the pressure of the lips
is greater for the arresting consonant.

Fig. 38 gives the tracing of a continuant, double
and single, “Z is Z”. The two maxima of the consonant
curve are not so well defined as in an
occlusive, but they are evident. The air pressure
curve for the double shows the stress of the arresting
consonant of the double.

Fig. 39, “topic” and “top pick”, shows a sharp
contrast between the single consonant and the
double. The air pressure curve has the ordinary
doubling form, but the maxima of the “p”'s in the
double are separated, although the lips do not part.

Fig. 40, “half pay”, shows a form precisely like
61the doubles preceding; the abutting consonants in
this case are two different labials. It is plain that
the two consonants of the “double” function precisely
as do the two abutting consonants. Table I
(Appendix XI: Table I; statistics and double consonants)
gives summaries of the actual measurements
of tracings of the various phrases recorded to show
the contrast between the double and the single consonant.
The attention of the subject was not called,
however, to the fact that the same consonant appeared
as single and as double in each word or phrase;
instead he was asked to say the phrase so that the
meaning would be clear; he chose his own rate of
utterance. In most cases the two maxima show in
the curve of the consonant and also in the curve of
the air pressure. It is difficult to record the very
slight air pressure in the case of the “l” and nearly
all the records are defective.

The difference in length between the single and
the double consonant is pronounced, except in the
case of the “n”. The word chosen, “unknown”, was
not fortunate, as the single consonant occurs in the
final position where there is a tendency to prolong
the nasal consonant. Many of the recorded lengths
of the single “n” are greatly in excess of the average
value of the consonant as given by Rousselot, e.g.,
and as shown in other records. “Unknown” and
“un-own” give comparable forms, but were used
with only one subject.

It is to be noted that liquids and nasals like
“l, n, m”, and fricatives like “s/z”, give the same
type of doubles as do the occlusives “t/d” and “p/b”.
Compare Rousselot's records for “v-v” previously
cited.

The subject S. sometimes substitutes a prolonged
releasing consonant in the second syllabic for a
double. (Cf. Fig. 42.) It is probable that the series
“Otto ought to” (A Fig. 47 d), and “Topic, top
pick” (A Fig.64 b, d, e), are of this type, but they
have been included in the table.

The graph shows that some 29 per cent of all the
cases of double consonants lie between the lengths
0.2-0.24 sec. There are very few indeed whose
length is below 0.15 sec; the distribution is much
more extended above the mode .20-.24. There is
a physiological limit to the rate at which the consonants
may be spoken, and as speech always tends
toward rapid rate, the lengths of the consonants
cluster toward the lower limit. There is no assignable
upper limit to the length of a consonant if a
subject chooses to prolong it. The continuant may
obviously continue indefinitely and the occlusion of
the other consonants may be held indefinitely.

The distributions are most significant in showing
the limits of the doubles. The modes are significant
in that they show that there is a single set of causes
at work, and that they probably appear in the
several similar groups. The modes do not represent
constants which it is important to determine. The
distribution must be somewhat general; there are
variations from subject to subject, dependent on
variation in the subject's “repetition time”.

There are a number of exceptions to the typical
form of tracing indicated. Of 854 cases studied,
117, or 14 per cent, do not show the doubling form
in the tracings, either in the mouth pressure curves
or in the consonant curves; 41 per cent show the
doubling in both mouth pressure and consonant
curves; 69 per cent show the doubling in the mouth
pressure curves.

The tracings of two different abutting consonants
made by the same member, like “f-p” in “half-pay”,
“m-b” in “humbug”, show precisely the same curve
forms as do the double consonants. But it is evident
that the two different movements are managed with
more care by the subjects. Of 251 cases studied,
13, or 5 per cent only, do not show the “doubling”.

In some cases the record of the pressure in the
mouth is imperfect; in others the classification of
the curve forms is uncertain. Only those pressure
forms which show two obvious maxima have been
classed as “bi-maximal” (bm). The convex-concave
form (cc) may be the result of two chest pulses; indeed
many of the records show that it does actually
result from such a succession, but it has not been
included in the list of “double forms”. Cf. Fig. 43.

2. Influence of the rate of utterance on abutting
consonants (including doubles)

In all these cases we have seen the properties of the
“double consonant” as they appear in ordinary
speech, and the properties of an abutting pair. What
leads the two consonants to assume a more or less
continuous form? If the two chest pulses are actually
present, and if the abutting pair has actually the two
functions of arresting and releasing the syllable
movements, what causes the abutting? It is evident
that if the syllables are spoken slowly in a word like
“un-known”, or in the phrase “up puppy”, the
consonants are separate. The rate of utterance is
the thing which leads to the doubling of the consonants.
Cf. Fig. 34.

The modification of the articulations is one of the
most important aspects for study in experimental
phonetics. The study of such changes reveals the
speech apparatus at work. The two great causes of
phonetic modification are changes of rate and of
stress. And these two factors are both involved in
rhythmic grouping.

Rate forces movements to coalesce or drop because
the rate determines the duration of the individual
movement in the movement series. The stressed
syllables are most resistent to such changes. The
syllables are more resistent than the consonants
which are auxiliary movements and may be “dropped”
(i.e. actually replaced by the faster, briefer
chest movement of releasing or arresting).

The releasing consonants are more resistent than
the arresting consonants because the duration of the
releasing consonant does not add to the syllable.
The vowel is conserved with its syllable with the
increasing rate; but it changes quality; less and less
time is given in which to approximate the specific
shape of the vocal canal. In English it is possible
to note a regular series of reduced “values” of the
vowel, which ends in shwa. With the increase in
rate all vowels in unstressed syllables arrive at the
common shwa.

A. Stress

Stress must affect rate; the stressed syllable is
lengthened in the breath group. The breath group
tends to maintain its rate and therefore the unstressed
syllables within the group are shortened in
compensation for the lengthening of the stressed
syllables.

Stress affects the factors of the syllable on which
it falls; all the auxiliary movements tend to increase
in amplitude. Added breath pressure increases the
intensity; increased opening of the jaw modifies the
quality of the vowel toward the maximum opening
of “ah”. Added stress on the consonants increases
the duration of the consonants. The stops tend to
become continuants, since the increased air pressure
breaks through the occlusion. A surd phase of a
sonant appears; and the consonants tend to be
aspirated, because of the accumulating mouth pressure
due to the stress.

Extreme stress on the syllable leads to “Brechung”,
diphthongizing, common in languages with heavy
word stress.

Rhythm shows the effect of rate and stress in the
grouping of the syllables. It is fundamental to the
character of the syllable that it appears as a factor
in the rhythmic pattern. The breath groups are the
phrases of the rhythm, and may appear as the
phrases of a prose period, or as the lines (“verses”)
of poetic stanzas.

The method of gradually increasing the rate, and
gradually decreasing the rate of a series consisting
of a selected syllable repeated, shows the various
types of modification of the factors of the syllable
as the changing conditions throw them together or
force them apart.

Any utterance consists of feet and of breath
groups, and although the primary concern at this
stage of the discussion is with the syllables and their
factors, the types of foot and breath group involved
are to be noted. The foot throughout is either a
one syllabic foot in each breath group, or it is a
rapid succession of syllables in a breath group of
indeterminate length. Such an indeterminate unit-group
is common in music, as the trill or run, but
67such an indeterminate breath group of rapid syllables
occurs in speech only in the unusual form of
“patter”. The breath groups each consist of a one-syllable
foot, or of the rapid-series ‘foot’ in the
breath group of indeterminate length. Cf. Figs. 48,
49. 50, 57, 59, 60, 61, 63.

These unusual forms of foot and breath groups
are supplemented below with a detailed study of
the various types of feet in breath groups of two to
five syllables.

In order to study the influence of rate on doubling
(abutting), tracings were made of a series of
syllables like “pup, pup…” in which the rate of
utterance is gradually increased so that the syllables
at first distinct, come closer and closer together. As
the rate increases, the arresting consonant of each
syllable doubles with the releasing consonant of the
next syllable, “pup, pup…” becomes “pup-pup…”
At a still higher rate of speed it is impossible
to execute the prescribed number of consonant
movements per sec., and the arresting consonant
of each syllable drops; “pup-pup…” becomes
“pu' pu'…”. The whole series appears as:
“Pup, pup, pup pup pup-pup-pup-pup-pup-pu' pu'
pu' pu'…”.

It is possible to note the beginning of the doubling
process, the rate at which it occurs, and the point at
which singling begins when the arresting consonant
drops.

B. Tracings of double consonants produced
by increasing the rate

The full list of syllables developed into series by
increasing the rate is given in Appendix A.

The tracings of Fig. 43 show forms of transition
from the series “pup pup…” to “pu' pu'…”
produced by the increasing rate of utterance. There
are the usual phases in which a stage of doubling,
“pup-pup…”, occurs before the arresting consonant
drops and the series becomes “pu' pu'…”.
Both the tracings of the movement of the lips and
of the pressure in mouth show the distinct abutting
consonants in the doubling forms, and it is clear
that there is a definite change in the coordination
when the arresting consonant drops, and “singling”
occurs. Rarely doubling does not occur at all. Instead,
the arresting consonant drops abruptly.

Fig. 45 is added so that the doubling of a continuant
like “s” or “f” can be compared with that of
the occlusives. It is clear that the process is exactly
the same, that the continuants function in doubling
precisely as do the other consonants.

Fig. 49 shows tracings of the pressure outside and
inside the mouth for continuants. The outside pressure
runs high during the “f” or “s”, but the mouth
pressure has its maximum as usual during the consonants.

Fig. 48 shows the chest pressure tracings for occlusives
with long vowels. It is to be noted that the rise
in the chest pressure, marking the inception of the
second syllable, occurs in the middle of the doublet,
not in the middle of the releasing double consonant.

Fig. 57 shows a single doubling form and has the
striking rise in general pressure level which is usual
with rapidly increasing rate; this leads to the vocalization
of a releasing surd.

Fig. 41 gives the distribution of the lengths of
such consonants doubled by increasing the rate. A
comparison of the curve P with curve O shows that
the general distribution of the lengths of these
doubles, by increasing the rate, is very like that of
the lengths of the actual doubles.

These “doubles induced by increasing rate” show
all the varieties noted in the case of actual doubles.
On occasion, type 5 (Fig. 42, 6, p. 64), appears just
after doubling ceases with the increase of rate.

After a little practice in increasing the rate of
utterance uniformly, the subjects usually produce
tracings with the doubling stage. Like the general
group of one-member abutting pairs, only about
one series in seven fails to show doubles.

Of 444 double consonants tabulated in series of
this type, 124, or 28 per cent, do not have a clear
indication of the bi-maximal curve in either the
consonant or the mouth pressure tracings. This is
a large fraction; but the mouth pressure tracings of
liquids and nasals are often unsatisfactory, and the
consonants “ch” and “sh” (English) double with
difficulty. Of the labial forms, 80 per cent have
a clear indication of doubling in the form of the
curves. As in other cases, only those curves have
been counted “bi-maximal” in which the two
maxima are apparent.

There are two discontinuities in these series. The
first is between the separate syllables, “pup,
pup…”. and the doubling form “pup-pup…”.
This is indicated by the closing up of the interval
between the syllables, and by the double form of
the consonant- and of the mouth-pressure-curves.
The length of the vowels tends to remain the same
throughout the series.

The syllable movement is arrested by the arresting
consonant. Although the movements are not
in phase, there is ample time for the beat- and back-strokes
of the movement. Fig. 52.

This involves two consonant movements (beat-and
back-stroke) per syllable and the consonant
movements are unevenly spaced; the abutting consonants
are much closer than the consonants on
either side of the vowel. Figs. 53, 54.

As the rate increases the movement, the second
discontinuity, between the doubling form “pup-pup-pup…”
and the singling from “pu' pu'
pu'…” appears. The consonant movements will
tend to become regular by dropping the arresting
consonant; this will equalize the movements of the
consonant, and the syllable movement, now with
chest-arrest, will fall in line. The arresting consonant
drops as the syllable becomes chest-arrested.

The breath group of the separate syllables and of
the syllables with doubles is a breath group with a
one-syllable foot.

This second discontinuity is marked by the change
in the consonant curves which again show a single
maximum; the form of the mouth pressure tracing
also shows the change from the bi-maximal, arresting-releasing
type, to the simpler releasing form.
At the speed of 4.-4.5 syllables per sec, this change
must occur; the doubling of the consonant is no
longer possible, for it entails 8 consonant movements
per sec., a rate which is only possible when the
movements are evenly spaced and, is at best, near
the maximum rate for repeating a movement.
There must be fewer consonants per syllable.70

And it is now apparent that the arresting consonant
will be the one to drop. The arresting consonant
is losing its function because the syllable movement
is becoming chest-arresting. The arresting
consonant is becoming a superfluous item in the
syllable; it is off-phase, out of step with the syllable
movement in that its beat stroke occurs with the
back stroke of the syllable movement. The chest
arrest is faster than the consonant-arrest. The consonant
mechanism involves a column of compressed
air, while the muscles in chest-arrest act directly
on the ribs of the chest. At a rapid rate the movements
tend either to get into step or to drop in
order to simplify the coordination; therefore the
arresting consonant will drop while the releasing
consonant retains its position because it comes in
on the beat stroke of the syllable. This tendency
is very marked in all phonetic coordinations; the
case has already been mentioned in which the series
“at, at…” becomes “ta, ta…” as the rate increases.
Cf. Figs. 26, 27 and 28. The phenomena
are very striking in the case of abutting consonants
involving two members. Cf. Fig. 59.

It is to be said, however, that the arresting consonant
does not always “drop” without a struggle.
Often the movements of the arresting consonant
fuse with those of the releasing; it is not a matter
of mere omission. Tracings of series of writing
movements, in which the rate is so rapid that
“dropping” occurs, show that the amplitude of the
movement crowded out is reduced because the beat-
and back-strokes of the “dropping”, movement
71come to overlap; a stasis may result; in the end the
slight excursion, or the stasis, is absorbed into the
back stroke of the coming movement.

In Fig. 55, I shows the beat actually present, but
reduced in excursion, and becoming part of the
back stroke of the coming beat. II shows the more
advanced stage in which the beat appears as nothing
but a slight inflection in the back stroke of the
following beat. III shows the dropping beat represented
by a stasis at the opening of the following
beat. In IV the beat has disappeared but the back
stroke of the following beat is lengthened. Such
“dropping” of movements is very common in
handwriting at rapid rate. It has played a part in
the development of cursive forms, just as the
“dropping” of phonetic movements has played a
part in the modification of pronunciations.

There is likely to be a sudden increase in the
rate of utterance when “singling” occurs; there is
a definite physiological limit to the speed of doubling;
but when the arresting consonant drops, there
is nothing to prevent a rapid rise in rate. Cf. Fig. 51.

The breath group of these syllables with singled
consonants is a continuous group of an indeterminate
number of syllables, like the trill or run in
music; an unusual form of utterance heard only in
patter. Cf. Figs. 48, 50 and 57.

The study of these tracings of double consonants,
produced by increasing the rate, shows that the
criteria of doubling are quite consistent. The rate
at which doubling begins and ends, the form of the
consonant curve which shows the combination of
the doublets, and the bi-maximal form of the curve
of mouth pressure during doubling, are all in accord.
And the results agree with the characteristics of the
actual doubles (ordinary words and phrases)
determined above. Fig. 54.

3. Restoration of the original form when the
rate decreases

This restoration of the prescribed consonant,
when the rate permits, is of importance because
the process of elimination and restoration is constantly
going on under the ordinary conditions of
speech. It is not the case that a consonant has
actually disappeared because it is for the moment
suppressed by the rate of utterance; it is in abeyance
in the breath group, and will promptly reappear
when the rate permits. Much survives in the movements
of a language which is not heard; the French
aspirate “h” still has a place in the pronunciation,
though it is seldom heard. One may expect then
great freedom in modifying the prescribed consonant
forms. When the rhythms compel rapid
utterance the consonants may “double”, or “single”,
and they may be shifted from one syllable to another,
but as the rhythm and the rate revert, the
original form will return. The subject has not lost
the original pattern of the breath group, though
it is freely modified by phonetic exigencies.

This restoration of the consonant forms, and
later of syllable forms, will be treated when other
series at increasing and decreasing rates are discussed,
and also when the various stressed groups of
syllables are considered. This tendency to restoration
is important in all these cases; it is the great
conservative factor in pronunciation (89, p. 46).

4. One-member abutting pairs. Tracing of
abutting consonants produced by the same
articulatory member

When series like “puf, puf…” and “sat,
sat…” are uttered at increasing rate, abutting
pairs of the form “f-p” and “t-s” result. Such
abutting consonants are very much like double consonants;
the two consonants are different but they
combine under much the same conditions.

Fig. 56 shows series of the syllable “puf…”
passing through the stages of “separate”, “doubling”,
(= abutting) and “singling”. Fig. 58 illustrates cases
in which the doubling does not occur, and instead
the arresting consonant drops at once. It is significant
that the presence or absence of a consonant
does not depend on the articulatory stroke, but on
the function of the consonant. If the consonant
arrests the chest pulse, it is audible; but if the chest
pulse is chest arrested, the consonant movement
does not make a consonant. A consonant may
drop, although the articulatory movement is
vigorous. And it is interesting to see that although
it fails to function as an arresting consonant, the
“t” movement may nevertheless combine with the
“s” movement, and actually produce the stroke of
a compound consonant “ts”.

The lengths of such one-member pairs abutting
with increasing rate are comparable to those of the
doubles by increasing the rate, and to those of the
actual doubles in ordinary words and phrases.
Fig. 51, p. 69 shows the practical identity in curves,
N, O, P.

The rates at which doubling takes place are practically
identical with other abutting rates. Cf.
Fig. 51, p. 69.

An inspection of the tracings makes it apparent
that a majority of one-member abutting pairs have
the doubling form clearly indicated in the consonant-
and mouth-pressure-curves. Of 607 cases
of front lingual pairs, like “t-s, d-l, th-n”, etc., only
12 per cent do not show the characteristics of
the arresting-releasing pair; 40 per cent show the
characteristics in both chest pulse and consonant
movement; 76 per cent have the characteristic
abutting form in the mouth pressure curve. Of
142 labial pairs, “f-p, “v-b” etc. only 4 per cent do
not show the arresting-releasing form; 56 per cent
show the characteristics in both air pressure and
consonant movement; 90 per cent show the characteristic
abutting form in the mouth pressure curve.
This is a decidedly larger number than in the case
of the double consonants produced by increasing
the speed of utterance. It is possible that the two
different consonants tend to preserve the two consonant
strokes and the two pressure maxima. The
difference in the force of the consonant strokes
of the two components, and the difference in the
mouth pressure of the two components show clearly
in the tracings of such one-member abutting pairs.
Series in which the rate decreases from the
maximum show the return of the “doubling” and
finally of the original form. .This is the tendency
to reversion mentioned in connection with the
doubling series, p. 61.

5. Two-member abutting pairs. Tracings of
abutting consonants produced by different
articulatory members

This method of study may be extended to abutting
consonants of all types. When the arresting consonant
is produced by one member, and the releasing
consonant by another, the conditions are somewhat
different.

Tracings were made of series in which such pairs
of abutting consonants were produced by increasing
the rate of utterance. “Top, top…”, in which the
arresting consonant is a labial, and the releasing
consonant a lingual, will illustrate the general form.
Independent tracings of the lip movement and of
the tongue-movement were taken, along with the
curves of pressure in the mouth and in the chest.

A list of the syllables which were developed into
series with two-member abutting pairs of consonants
by increasing the rate is given in Appendix I, C.

When the arresting consonant shifts to the next
syllable, the movement of the releasing consonant
does not inhibit the movement of the shifting consonant;
instead the movements tend to slip together,
to fall into step, and it is apparent that the movement
of the arresting consonant persists long after
it has ceased to function in the syllable. It is a
striking illustration of the tendency of the movements
of speech to get in phase. The movement of
the arresting consonant in the two-member combination
can shift its position in the coordination to
the next syllabic without affecting the releasing
consonant of that syllable; it actually slides into
phase, so that the beat strokes of both consonant
movements and of the syllable movement occur
together; in reality, both consonants have the
releasing position, but one of them does not function
and cannot be heard. This persistence of the
original arresting consonant as a movement in step
with the releasing consonant is invariable. All the
records of all the subjects show it. As the arresting
movement shifts, so as to coincide with the releasing
consonant of the next syllable, the pulse becomes
chest-arresting; the form of the curve of the mouth
pressure no longer indicates the end of the one
syllable and the beginning of the next, for the
pressure falls at the end of the syllable as in all cases
of chest-arrest of the syllable movement. This illustrates
a possible method of formation of such unusual
sounds as the Slavic “b mouillé” reported by
Rousselot (70, p. 605).

It is easier to see how the speaker retains the
prescribed form in these two-member series than
in the one-member series. Although the rate has
forced the “dropping” of the arresting consonant,
the consonant stroke is actually retained, but it
coincides with the releasing consonant; they merely
slide apart again as the rate reduces. Fig. 64 shows
the restoration of the original prescribed form in a
two-member series.

A comparison of the lengths of such two-member
abutting pairs with the lengths of the one-member
abutting pairs of consonants shows that the mode
has a value slightly less than that of the one-member
pairs (including doubles). Fig. 65 shows that the
difference in the two distribution curves Q and R
is not great; but it is probable that there is a slight
difference in the play of the movements which
makes the two-member pairs slightly shorter. The
average of the 984 readings represented by curve
R is .22 ± .04 sec. The minimum lengths of the
two-member abutting pairs lie well within the
78limits of rate of two ballistic movements. Fig. 66
gives a comparison of the distributions of the lengths
of doubles and of abutting pairs recorded in the
second group with records in the first group. It is
apparent that the distributions are practically
identical.

The rates of the syllables in these series containing
two-member abutting pairs, as given in the distributions
of Fig. 51, p. 69, show that they agree very
closely with those of the one-member pairs (e as
compared with d).

There is no fundamental distinction to be made
between the two-member pairs of abutting consonants,
the one-member pairs of different abutting
consonants, and the actual doubles; they all result
from one and the same process. Combinations like
“bit-bit-” and “fub-fub” and doubles like “pup-pup-”
and “sis-sis-” are all analogous. It is true that
singling in the course of the two-member series
does not mean that the arresting consonant stroke
drops out; instead it shifts into the releasing position,
in step with the next releasing consonant, and becomes
functionless or fuses to a compound consonant.
But the elimination of the arresting consonant
in the two-member abutting pair, the shift of
the arresting consonant series, like “up-up”, to “pu'
pu'…”, are precisely like the processes of singling
which appear in the series of doubles by increasing
the rate, or in the series of one-member abutting
pairs produced by increasing the rate.

These abutting consonants, made by two different
articulatory members, cannot occur more
rapidly than the abutting consonants made by a
single member. In the playing of key-board and
wood-wind musical instruments, and in many other
forms of skilled movements, it is possible, by the
use of two or more fingers in coordination, to attain
a rate of 15-18 per sec. Nothing of the sort
appears in the case of the consonants in speech; the
number of consonantal movements which can be
made by two different articulatory members is
8-12 per sec; it is no greater than the number
of consonants per sec. produced by one articulatory
member.

In the two-member pairs the abutting consonants
seem ot stretch toward each other (Fig. 67). The
tracings of the contacts of the consonants show that
they elongate, each in the direction of the other.
This is part of the tendency of the movements to
fall into step, to coincide. The first consonant stroke
is held in anticipation of the one to come, and the
second consonant stroke occurs early in order to
fuse with the previous stroke. Like drops of liquid
they tend to run together. Of 95 series taken at
random, 72 show such “attraction” of the consonants.
The average length of the consonants before
doubling of the individual consonants is 0.17 sec.
(214 readings). The average length of the individual
consonants while doubling is 0.20 sec. (284 readings).
The difference in these averages is significant,
as the readings were made in pairs and in every case
the value of the consonant before doubling was
lower than the value of the consonant while doubling.
Any influence of the increasing rate of utterance
would work against this tendency to lengthen
the consonants as they come to abut. Fig. 67 gives
cases in which the “attraction” of the doubling
consonants is very marked.

6. Assimilation of abutting consonants

When the rate of utterance is high enough to produce
abutting pairs of consonants, it often happens that
the vocalization of the two consonants becomes
alike. It is difficult to shift the vocalization within
the pair at a rapid rate. In French the tendency is
always to weaken the arresting consonant, because
arrest by the intercostals first supplements and then
supersedes the consonant arrest. Therefore the following
releasing consonant is dominant and the
vocalization of the releasing consonant prevails,
The assimilation is said to be regressive, meaning
that the arresting consonant takes the vocalization
of the following releasing consonant. In English or
German a stress on the first of the adjacent syllables
reinforces the arresting function of the first abutting
consonant; and the two consonants may not affect
each other. Sometimes what might be an abutting
pair is separated by a definite hiatus. Cf. “obscene,
obtrude, blackboard”. In other cases the first syllable
is vigorously arrested and the vocalization of
the arresting consonant affects the following, releasing
consonant. The pronunciation of an English
word like “absolution” varies with the rate of utterance
from no assimilation to “progressive assimilation”.80

7. Assimilation in which the abutting pair becomes
a double consonant

In many languages the pair of abutting consonants
becomes a double consonant: ad-similate = assimilate,
ad-breviate = abbreviate, syn-logism = syllogism,
etc.

Where but one member (lips or tongue) is involved,
it is ordinarily said that it is easier to repeat the
same movement than to make a different though
similar, movement to the same bearing when the
conditions of articulation are different. The repetition
of one “sound” can be said to be simpler than
the articulation of the two different “sounds”.
Where two members (both lips and tongue) are involved,
it may be said that repetition of the movement
of a single member is easier than movements
of two different members. Cf. “applicate, immemorial,
osservatore”.

It is to be noted, however, in the case of one-member
assimilations, as well as in the case of two-member
assimilations, that actual experiment does
not show the assimilation of “d-p” to “p-p”, or
“b-s” to “s-s” as a stage between the abutting of the
consonants and the dropping of the arresting consonant.
As the rate increases in a series like “pad,
pad…” the double “p-p” does not always occur;
the series may pass directly from “pad, pad…” to
“pa' pa'…”. So with “men, men…”, the stage
“mem-mem…” may not appear; the change may
be from “men, men…” into “me' me'…”.

How then is the appearance of the doubled consonant
in place of the abutting pair to be explained
since it is a very common form of assimilation? With
increasing rate there is a universal tendency to
simplify by eliminating the arresting consonant;
the restoration of such an arresting consonant is a
commonplace. Cf. Fig. 64.

But such restoration does not always occur. In
the course of generations the rapid-rate pronunciation
prevails; the sense of the original arresting
consonant is lost. However, if the “short” vowel
of the first syllable, the word stress, and the phrase
rhythm persist, slow enunciation, especially with
a pronounced stress on the first syllable, will
force adventitious doubling. Cf. Fig. 95, 100, 101.
At this stage in the phonetic modification of the
word, rapid utterance has eliminated the arresting
consonant of the original pronunciation, but a
slow, careful utterance substitutes a true double
for the original abutting pair. In all languages,
rapid utterance is to be distinguished from slow,
careful utterance; as a rule, the standard orthography
follows the slow, careful utterance (61, p. 4).
It is often the case that such standard spelling prescribes
double consonants which are seldom heard.
At a later stage of phonetic modification, the double
is no longer recognized in the actual pronunciation;
it may or may not persist in the standard spelling.
The Attic Greek shows a strong tendency to the
omission of such doubles.

8. An arresting consonant occupies the interval
of a double consonant

It is possible to consider any arresting consonant as
virtually a double, although it may not be followed
by a releasing consonant. In the case of a series of
syllables with double consonants, the increase of
rate forces the dropping of the arresting consonant.
It cannot shift position and persist because the
releasing doublet of the next syllable is produced by
the same member. In the case of a series of syllables
with two-member abutting pairs, the arresting consonant
does shift position to become member of a
releasing compound consonant, or to overlap the
actual releasing consonant as a functionless consonant
stroke. Likewise, in a series of syllables with
single arresting consonants, as the rate increases the
arresting consonant shifts to the unoccupied releasing
position in the next syllable; and the shift occurs
at about the same rate as does the dropping of the
arresting doublet. The single arresting consonant
has the position of an arresting doublet; the other
doublet is represented by the zone of chest-release of
the next syllable; when the rate increases, the arresting
consonant slides into the releasing position like
the arresting doublet of the two-member abutting
pair. In all cases the arresting consonant disappears
and a releasing consonant remains.81

The graphs representing the distribution of single
arresting consonants show the fact that there is no
distinction between the “long” vowels and the
“short” vowels in the matter of doubling. The long
vowel can be shortened, but the short vowel cannot
be lengthened. Since these changes occur at a fair
rate of utterance, all vowels are actually of short
duration.

The tendency of restoration of the original form
is apparent in these series in which the single arresting
consonant is affected by changing rate of utterance.
When the rate is reduced, so that the original
arresting form is again possible, the consonant shifts
back to the arresting position; the original prescribed
form is restored; later it will be obvious that
the same thing occurs in stressed groups of syllables.

The distribution curves of the arrested syllables
(Fig. 70), when compared with the distribution
curves of abutting pairs, Fig. 63, show that the rate
at which the change from arresting to releasing takes
place is slightly higher than the rate for singling of a
double. It is evidently easier to retain an arresting
consonant than an actual double as the rate increases.
Cf. Fig. 26.82

Fig. 68 shows the mouth pressure and the chest
pressure tracings for the phrase “at E, at C”. The
doubling form of the mouth pressure is apparent in
“at C”; the maximum of the chest pressure occurs
immediately on the détente of the double; the minimum
occurs as usual during the releasing doublet.
But in the case of “at E” the maximum of the chest
pressure does not occur until .10-.20 sec. after the
“détente” of the “t”, and the minimum occurs at
the détente.

Fig. 69 “at E, at P” shows the same contrast in
the case of a two-member abutting pair, “t-p”. In
arrested syllables of this type, if the arresting consonant
and the chest release of next syllable are
measured as a single interval, the duration corresponds
to the duration of a double consonant. Fig. 70.
If the arrested syllable is repeated at increasing rate,
the shift to a released syllable occurs at the same
rate as the singling of a double consonant. If the
rate is reduced, the consonant reverts to the arresting
position at the proper rate. Cf. Fig. 51, p. 69.

9. Compound consonants: consonant groups
which function as a single consonant

Two or more adjacent consonants may be classed
as an intra-syllabic group when the group figures
as a compound consonant in releasing or arresting
the syllable movement, as contrasted with abutting
consonants each of which has a different function
in two different syllables. It is possible, of course,
that a compound consonant figures as one of the
abutting consonants.

In the group of the compound consonant the consonant
movements are as nearly simultaneous as the
nature of the movements combined will permit.
Together they function as a single, arresting or
releasing factor in the syllable and frequently they
are so brief and so close together that they must
contribute qualities, rather than distinct elements,
according to the “law of discrete succession”.

The intrasyllabic groups differ somewhat according
to whether their function is arresting or releasing,
but there are certain features common to them
all.83

Such groups appear when certain series are uttered
at increasing rate; instead of dropping, the
arresting consonant fuses with the following releasing
consonant. Thus “tas, tas…” gives rise to
“sta, sta…”. Such compounds appear in the
releasing position. Compound consonants in the
arresting position are also common, e.g. “apt, apse,
asp”, etc.

The series involving a fricative often results in
such compound consonants when the rate of utterance
is increased.

The compound consonants “st” and “sf” appear
in Figs. 71 and 72; the process is apparent by which
they are produced. Such compound consonants
often persist as a stable releasing consonant and may
be present at rapid rates, at 5-7 syllables per sec.

In the releasing form of the compound consonant,
with an initial fricative component like “v” “f”,
and especially “s” the air pressure curve is apt to
show a slightly prolonged and rounded form “convex-concave”,
cf. Fig. 42. In some cases there develops
an unvoiced preceding syllable and a bi-syllabic
form results. This shows rarely in the records of
these English-speaking subjects. But the history of
the Romance languages shows this development of
the preceding syllable, especially in the case of s,
cf. “spiritus, esprit”, etc.

The very common group of a stop consonant and
a liquid is of a different type. The liquid (l, r) is so
open a conformation that it permits the pulse of the
syllable movement to begin, and although the liquid
component involves a distinct stroke, this occurs
84during the beat stroke of the syllable. This type of
liquid might be called an “internal consonant”. Like
the conformation of the vowel, the conformation of
the liquid may be assumed before the preceding
consonant is uttered and retained while the consonant
is being executed; the conformation of the “l”
or “r” does not interfere with the enunciation of the
consonant; this is often the case with the vowel.
Releasing groups with a liquid are very common,
e.g. “glide, crow, try, blow, pry”. For some reason
“tl-” and “dl-” do not appear in most of the western
languages.

In arresting groups the liquids figure as “internal
consonants”. Sweet long ago observed that the
liquid continues the vowel, or rather occurs with the
vowel. In a form like “grilled” there is no vowel to
be heard save the two internal consonants “r” and
“l”. The nasals also appear as possible components
of arresting groups e.g. “hand, unkempt, Kampf”.

These compound consonants are to be distinguished
from abutting consonants. In the phrase “a tall
D told E”. the succession “l, d” occurs as an abutting
pair in “tall D”, and as a compound arresting consonant
in “told E”. In Fig. 73 the doubling “l-d” is
indicated in the mouth pressure tracing, whereas the
“ld” is a single arresting form; the releasing “d” of
“l-d” contrasts with the arresting “ld” in the position
of the maximum in the chest pressure in each
case.

The grouping in time of the components of a
compound consonant is very close; the group of two
or three beat strokes often occurs in the interval of
0.08-0.10 sec. They are close enough to fuse in the
syllable movement to a single rhythmic beat (32,
p. 553; 84, p. 346).

The releasing compound consonant composed of
an occlusive, followed by a liquid, “r” or “l”, is very
like a simple releasing consonant. The “l” stroke
permits considerable escape of air, but the release is
definite.

With a liquid or a nasal as the first component
of the compound consonant, the forms do not differ
from that of the ordinary simple consonant. There
may be a slight elongation and rounding of the air
pressure curve. Increasing the rate of utterance shifts
the coordination to the releasing position, and the
compound consonant may simplify, losing its liquid
component.

Diagrams and Check Lists of the Compound Consonants

Cf. Appendix XII: Check List Compound Consonants
in English.

10. Tracings of bi-syllabic consonant groups

Rousselot has shown that in the case of an elaborate
consonant group like “aptma” the movements of
the consonants “p” and “m” tend to fuse, in spite
of the intervening “t” (70, p. 957). Tracings of
such bi-syllabic groups uttered at increasing rate
show that one-member abutting pairs tend to
develop, across the intervening movement of the
different articulatory member. This is a variation
of the “doubling” process and illustrates the fact
that when the consonant movement is repeated at
a certain rate, whatever the accompanying movements,
it takes on the “doubling”, form. Fig. 77.

At the same time, such complicated groups are
important for observation of the releasing and
arresting of the consonants. It is not the case that
a bundle of consonants is tumbled in “between
two vowels”. Instead, whatever the rate, the consonants
are definitely grouped so that they act either
as arresting or as releasing factors, either singly or
as compounds. In the word “aft-pa” the “f” coincides
with the “t” forming a compound in the
arresting position; the “t” keeps its definite arresting
position until the word becomes monosyllabic; only
then does it shift to the releasing position and become
practically-coincident with the “p”. Mutatis
mutandis, the same thing is true of “s” and “p” in
“isp-da”. In the words “upspring” and “upspringing”,
the “p” is the arresting stroke and the “sp”
the releasing stroke throughout. But in the more
rapid “upspring”, the “p-p” may form a doublet
over the intervening “s”. Cf. Fig. 78.

11. The inflected air pressure Curve of the
single consonants

The inflected curve indicates a slight fluctuation
in the air pressure in the mouth and is more likely
to occur with nasals and with voiced occlusives like
“b”. It occurs in both the arresting and in the
releasing positions in Fig. 80.

The reduction of the pressure is due to a rapid
increase in the volume of the mouth, just as the
mid-sag of the outside pressure is due to the change
in mouth volume. Cf. p. 45. The experimental
study of the process of voicing shows that this
inflected curve is due to the increase of volume
of the supra-laryngeal cavity when the sonant is
produced. The mouth tube for sonants, and Rousselot's
olive, used with nasals, tap the cavity behind
the consonant occlusion and show this pressure
variation. In one case Rousselot publishes a tracing
of the air pressure in the mouth which is interesting
because it shows the inflected form of the air
pressure curve (35; 70, p. 496, Fig. 256b, p. 908,
Fig. 588, n, p. 955, Fig. 641, m, p. 956, Fig. 642, n,
p. 966, Fig. 652, general form of n, p. 967, Fig. 653,
n, 3 cases, each a different subject; 72, p. 79,
Fig. 71, m).90

V
Classification of phonemes

The development of the theory of the mechanism
of speech determines the final classification of the
phonemes. Classifications appear early, and those in
current usage are not always consistent because
they are based on various distinctions.

The classification of the phonemes must depend
on the distinctions already made between the movements
of the breath group, foot, and of the syllable.
And the syllabic is the basic movement which
defines the constituent phonemes.

The first and fundamental distinction is between
the vowel and the consonant. Although recognized
by the ancient theorists, in cuneiform and in the
early Semitic alphabet, some of the later notions
fail to make the difference clear. Rousselot limited
the opening-closing to the individual “sound”,
where Saussure had used it for the entire syllable.
This left Rousselot without any fundamental definition
of the syllable, and therefore of the consonant
and vowel. Rousselot often assumes the
syllable as an entity, and takes the contrast between
consonant and vowel for granted, but his system
of phonetics is defective at that crucial point.

Sievers introduced the notion of sonority as the
basis for the syllable, which involved classifying the
phonemes as to their “sonority” (Schallfülle,
Schallstärke) and assumed that the phoneme of
greatest sonority became the core of the syllable,
and syllable formant. This was adopted by Passy,
who supplemented actual sonority with “apparent
sonority” and so accounted for the fact that Rousselot
and others had been unable to show the differences
in carrying power on which they depended
for this sonority classification of the phonemes.
Acoustic study of the vowels has left no place for
the concept of “sonority”; it has no standing in the
physics of sound. But under the label of “prominence”
it still figures in many discussions as the
basis for the consonant-vowel distinction. In one
form and another it is recognized or assumed by
the phonemicists. Cf. Bloch, Trager.

Physiological analysis of the mechanism of speech
shows that the distinction between vowel and consonant
is a matter of the difference of function in
the syllable. The consonants delimit the chest pulse
of the syllable; the vowels shape the vocal canal
through which the chest pulse is emitted. On occasion
the vocal canal may be shaped by a continuant
consonant; a fricative which makes emission possible,
like “sh…” or “pst” may act as a vowel; and
“syllabic consonants” (e.g., “m, n, l, r, -ng”) are
common in many languages.

The group of phonemes which are often thought
of as on the border line, the “semi-vowels”, “w,
y (j)”, are quite distinct in their function. When the
closure of the vowel “u” is forced to the limit and
quickly released the result is the consonant “w”. The
difference between consonant and vowel can be
clearly noted in a series like “oo, woo, oo, woo”.
When the closure of the vowel “i” is forced to the
limit and quickly released, the result is the consonant
“y”. The difference between consonant and vowel
can be clearly noted in a series like “ee, ye, ee, ye”.

The functional difference of vowel and consonant
as constituents of the syllable, as characterized
factors of the syllable, is an essential phase of the
motor theory of phonetics.

1. Classification of the vowels

These are the cardinal vowels of English. They
do not include the “dark-e” which is the minimum
vowel opening, and is the lowest term to which a
vowel can be reduced. In a language with a heavy
word stress, like English, vowels are frequently reduced,
so that a series by graditions from any cardinal
vowel quality to this “dark-e”, shwa, is possible.

The vowel is to be thought of as identified in
part by the acoustic quality, its “value”,v and in
part by a physiological shape which is defined by
the movement of the tongue and of the jaw and
lips. These movements are reciprocal, and there
is a whole range of positions possible for the one
vowel. Cf. Marichelle (51), and G. O. Russell (74).

The range of positions for the one vowel is inevitable,
especially in languages with arresting consonants,
because the delimiting consonants determine
the position from which and to which the vowel
movement must be made. It is customary to refer
to the vowel shape as a “position”; but in the case
of most vowels in rapid speech, the vowel movement
is continuous. There is no time for stasis
in a “position”.

In many languages the vowel has a “length”
which is phonemic. Owing to the varying rates of
utterance, the distinction proves to be between 1)
vowels which may be prolonged, but which are
often short, and 2) vowels which cannot be prolonged
and which are always relatively short induration.

English has a peculiar series of “short vowels” in
which the arrest is entirely consonantal, e.g., “pap,
pep, pip, pup”. The chest muscles do not figure at
all in the arrest. The familiar Continental languages
have no such vowels.

The distinction between vowels which are arrested
wholly or in part by the consonant has led to
the familiar distinction of “tense” and “lax” vowels.
If the arrest is due primarily to the consonant, the
vowel is denominated “lax”; if the arrest is due
primarily to the chest muscles, the vowel is denominated
“tense”. The muscles of the vocal canal
are not concerned and are neither tense nor lax;
the muscular tension which is present or absent is
that of the intercostal muscles of the chest.

The tradition of a short and long contrast in
quality may survive radical historical changes of
the values of vowels. In English the popular sense
has it that “Pete” is long and “pet” is short; that
“bite” is long and “bit” is short; that “bate” is long
and “bat” is short; that “pope” is long and “pop”
is short; although the vowels are not now so related.
It is an excellent illustration of the part that conditioning
plays in determination of the relations
of articulations.

The vowels of a language with pronounced word
stress are subject to changes due to stress, and the
consequent changes of rate. The prolongable vowels
of a heavily stressed syllable tend to prolong, and
in many cases to diphthongize, as in English. If the
vowel is not prolongable when stressed, the arresting
consonant or the intersyllabic interval are
prolonged. In the case of unstressed vowels, the
increase in rate makes for the reduction of the
vowel toward shwa, and may even lead to the
repression of the syllable.

Certain changes of vowels are due to accessory
movements. The pure and nasal vowel qualities
result from the closed and open position of the
velum. In some languages whispered vowels occur
in certain cases, due to the relaxation of the glottis.
In the case of diphthongs, vowels may unite into
double or triple compounds. It is evident that the
change of the vowel-shape is slow enough so that
the components are recognized. The components
are often of unequal duration. If the diphthong
92is lengthened, the longer component is prolonged.
Diphthongs arise by the serial fusion of two vowels,
or by the effect of a vigorous stress on a simple
vowel (Brechung). The umlaut is not a serial fusion
but the modification of the value of a vowel by an
element in the following syllable.

2. Classification of the consonants

The most obvious functional classification of the
consonants divides them into the releasing and
arresting consonants. In most cases the consonant
can function in both capacities; but there are exceptions.
In English “w, wh, y (j), h”, are releasing
only; while “-ng” is arresting only, in function.

Since the consonant is a constriction, the closure
may be either tight or leaky. The complete closure
constitutes the stop (occlusive) and the partial
closure constitutes the continuant. Continuants
are often characterized as sibilant, fricative, liquid,
nasal, depending on accessory movements and consequent
noises. The complete closure of a stop
when acting as an arresting consonant is sometimes
affected by a heavy stress and tends in that case to
become continuant.

The sounding of the vocal folds makes the
familiar difference between the surd and sonant.
This is not due to the direct action of the glottis,
though this is often assumed. The subject speaking
with an artificial larynx is able to control the surd-sonant
distinction when the reed is always in
position to speak, and the surd or sonant trait must
depend on the manipulation of an air pressure
above and below the glottis. In rapid speech this
distinction in voicing often fails, but the difference
in force between the surd and sonant persists. The
distinction of fortis and lenis proves more fundamental
than that of the voicing. (35)

In some languages there is a phonemic distinction
between long and short consonants. But there is no
such distinction in English, German or French.

The phonemic distinction between the single and
the double consonant is often assumed to be a mere
difference of long and short. But a study of abutting
consonants shows that the double consonant
is an abutting pair in which the members are the
same consonant. This has been recognized by
Swadesh (94), who speaks of the association of
phonemic length of the consonant, in Finnish, e.g.,
as always associated with other movements.

Simple, compound, and abutting consonants are
to be distinguished. The simple consonant of a
language cannot be separated into components.
The compound consonant also acts as a single
characterized factor in the syllable, but it may be
resolved into two, or, on occasion, three components.
Abutting consonants are in no wise single.
The abutting pair functions, the one as an arresting
consonant in the preceding syllable, and the other
as a releasing consonant in the following syllable.
Their traits have been considered in detail.

In cases where the compound consonant occurs
between feet, and the next syllable is without a
consonant, the compound consonant separates into
its components, and becomes an abutting pair.
This may be used as a test of the compound consonant.
Cf. p. 88.

The influence of the abutting pair on each other
is often spoken of as assimilation. The voicing of
the one often affects the other; on occasion abutting
consonants assimilate as a double determined by
the releasing consonant, (common in Latin and
Greek; “assimilate, symmetry”). In some cases the
two abutting consonants assimilate to a third form,
thus the abutting pair “s: y” become the English
“sh; t: y” become the English “ch”. Cf. Hudgins
and DiCarlo (33), pp. 462-464. In many cases the
arresting consonant drops, and the abutting pair is
replaced by a single, releasing consonant.

On occasion two or more regions are involved,
and a consonant is nasalized, labialized, or palatalized.

The position and the extent of the area of contact
is affected by the stress and rate of the syllables
involved; and differs somewhat between the releasing
and arresting forms.

In spite of its pretensions and the fact that it is
the basis of “phonemics”, the I.P.A. (International
Phonetic Alphabet) is an unsatisfactory collection of
symbols. It was thrown together when the problem
of phonetics was believed to be the determination of
a fixed position for each “sound”, and speech was
thought of as a series of such “sounds”.

The breath group, rather than the syllable, has
been the unit used consistently in I.P.A. transcriptions.
As a convenience, some transcribers divide
their transcriptions into words, though aware that
there is no phonetic unit corresponding to the word.
There is no provision for the syllable, hence the
syllable factors are not distinguished. The fundamental
differentiation of vowel and consonant is
not recognized. There is no distinction between
compound consonants with a single syllable function,
and abutting consonants, which function in two
different syllables. The I.P.A. fosters the use of a
loose term like “consonant cluster”.

In actual use the transcribers have not been consistent
in making the I.P.A. symbol represent the
phoneme. They have often represented the variants
of a phoneme by different symbols; and they speak
of a “broad” and of a “narrow” transcription. An
I.P.A. transcription is worthless unless the reader is
familiar with the phonetics of the language, knows
its syllabification and the variants of the phonemes.

Rime, assonance, alliteration, vowel harmony
signalize various resemblances of the consonants and
vowels. They are all involved in prosody.

Rime is the most elaborate. It occurs on the stress
of the breath group, and depends on the difference
of the release of the two stressed syllables of two
parallel breath groups (consonant contrasts with
consonant, or consonant contrasts with chest-release)
and on the identity of the vowel of the stressed
syllables and of anything following the vowel
(arresting consonant or following unstressed syllables).
Since the breath group for rime must be of
a regular rhythmic pattern, rime figures primarily
in verse. Rime echoes from breath group to breath
group and is one of the devices for constructing the
stanza in verse.

tataʹ tataʹ tataʹ tabaʹ (t ta),
tataʹ tataʹ tataʹ tadaʹ (t ta).

The conditions of rime (including identical intonation
on the rime) are well known; but its nature
has never been fathomed. (83)

In many languages the stress on the individual
syllable is phonemic. This is the word stress, so
variable in incidence and therefore so difficult in
English and Russian. In other languages the placement
of the stress is optional and defines the breath
group, but does not have phonemic significance.

The intonation in the western languages does not
affect the syllable; but in many Asiatic and American
languages it is phonemic, and characterizes the
syllable. In the western languages it may set up
contrasts, characterized phrases, and is roughly a
type of over-all punctuation.94

VI
Stress and rate and their relation to
the syllable and to the breath group

1. Definition of stress, its relation to other
factors in grouping

The “breath group” is a division recognized by
phoneticians of all schools. In the breath group in
English the dynamic pattern which constitutes the
“group” is made by the word stresses. That the
syllables have varying degrees of stress, and that
they are subordinated and grouped in rhythmic
forms, has been a commonplace since Aristotle. In
poetry we have called the rhythmic unit-groups
“feet”, and have noted the regular patterns. Feet
may also be observed in prose, following Aristotle,
but the patterns are not regular.

There is still a difference of opinion as to what
constitutes “word accent”. There are many who
insist that pitch may be the important factor. Ever
since Mitford, in 1804 (cited in 76), there have been
those who believe that change of pitch can produce
an “accent”. Also Coleman, cited with approval by
Daniel Jones (11, 38), believes that accent is due to
a turn in the pitch. The earlier advocates of this
theory of pitch accent assumed a rise in pitch, but
the experimental findings have made them fall back
on some change in pitch, either rise or fall; so
Abas (1), so Morris (55, 16).

It is only in the field of speech that “accent” has
been assumed to be a matter of pitch change. A
comparison with the conditions of accentuation in
other rhythmic fields makes the notion very doubtful.
In music it is apparent that a change of pitch
cannot be the determining factor in “accent”. It is
possible to make abrupt pitch changes within a
musical figure without changing the stress pattern;
in fact such changes of pitch with an undisturbed
stress pattern are a commonplace. And if changes
of pitch might in any circumstances determine an
“accent”, composers for the organ would certainly
avail themselves of the device, because there is need
of every resource for indicating stress at the organ.

It is not surprising that in speech changes of pitch
should be noted at the stress; it is often the significant
point in the group; and the heavy stroke of the
accent involves the chest pressure and is apt to
change the pitch because the laryngeal musculature
is often affected by tensions in the other musculatures
of speech. But if changes in pitch were in any
way essential to word accent, it would be impossible
to train the deaf mute to speak with a word accent,
and impossible for the subject with an artificial
larynx, or for one who whispers, to mark the word
accent.

2. Measurement of intensities in speech

It is not an easy matter to measure the intensities of
speech, whether they are conceived as intensities of
sound, or intensities of expiratory force. If one
were willing to assume that speech is a matter of
acoustics, the measurement of the intensities of the
series of acoustic combinations is difficult. The
comparison of sounds at various pitches, of different
vowels with varying acoustic patterns, of vowel
sounds with consonant sounds, is the serious problem.
The question is further complicated if it is assumed
that duration and pitch somehow figure in the
“word accent”.

Neither is the measurement of the expiratory
force an easy matter. The commonest form of
record has been of the pressure just outside the
mouth; and this has been supplemented by occasional
studies of the pressure changes in the chest during
speech. The force of expiration is due to the contraction
of chest muscles, but the flow of expired
air does not depend on the contraction of the chest
muscles alone; the flow of expired air is effected
also by the conformation of the vocal canal. If the
glottis and mouth are relatively open, the flow is
large; if the glottis and mouth are relatively closed,
the flow is small. Since the pressure just outside the
mouth depends on the flow of expired air, tracings
of that pressure are subject to correction for the
articulation involved.

If the expiratory force is to be measured in terms
of the pressures produced just outside the mouth, it
is obvious that only artificial material can be used;
it is impossible to handle the varying complexities
of ordinary speech. The articulatory conformation
must be the same from syllable to syllabic if the
changes in pressure are to correspond to the force of
the muscular contractions of the chest. The pressure
just outside the mouth will vary with the chest contractions
if the conformation remains the same.
But of course this outside pressure will drop to zero
during any consonant occlusion, and such changes
of chest pressure as occur during the occlusion are
lost. The opening and closing of the mouth also
introduces an artifact, the “mid-sag”, already mentioned.
Cf. p. 46.

It happens that some important changes take
place during the consonant occlusion. The minimal
chest pressure, marking the end of one chest pulse
and the beginning of another, occurs during the
releasing consonant, whether single or member of
an abutting pair. Only in the case of a syllable with
an arresting consonant does the pressure minimum
show in the outside pressure; but it is often masked
by the “mid-sag”. To prescribe syllables with
arresting consonants would only limit the experimental
procedure; since the rate of syllables with
arresting consonants cannot exceed three per sec.
Such a rate would make it impossible to study the
common rhythmic groupings for which the syllable
stress is a primary factor.

Tracings of the pressure in the chest, taken
directly from the trachea below the glottis, give
definite indications of the chest pulses, both maxima
and minima. As the tambour must be set to record
pressures as high as 200 mm. water, the membrane
cannot be very sensitive. (Cf. p. 16.) Variations
of pressure in the mouth, due to the articulatory
movements, do not affect the sub-glottal pressures
because of the intervening glottis vent. The chest
pressure tracings show very strikingly the groups
of syllables. The continued slow expiratory movement
which makes the grouping, maintains the
pressure well above zero throughout the group.
Fig. 18, 19, 20, 21.

It is possible to get direct records of the movements
of the muscles in speech. The negativepressure
tracings show the variations of force in
the expiratory pulses. Action-current records show
the action of both the internal and external intercostals
for the individual syllables, and the action of
96the abdominal muscles for the grouping movements
of foot and breath group. (85, Figs. 5, 7; 91, Figs.
15-22)

3. Stress and the resulting dynamic form in the
two-syllable, three-syllable, and four-syllable
breath-groups

The breath group is composed of syllables grouped
into feet. It is convenient to distinguish the types
of feet in terms of the number of syllables.

Breath groups of a single one-syllable foot are
presented in the studies of syllable modification:
the one-syllable breath groups are shown at varying
rates, the maximum rate of the simplest breath
group is made apparent as well as the various connections
between breath groups consisting of a
single one-syllable foot. As the rate increases, such
breath groups are succeeded by an indeterminate
breath group (cf. “patter”).

In the two-syllable feet, the iamb and the trochee,
there are but two degrees of force. But the dactyl
shows not only small variations in the duration of
the three component syllables, but also gradations
in the force of utterance. The stressed syllable is
merely the climax of the stresses of the dactyl. Cf.
Fig. 81, 82.

Fig. 81 is an excellent example of the uniformity
with which a subject often repeats a particular
form of utterance. The extreme emphasis on the
initial consonant is unusual, but is persisted through
the three series of the sitting. There are slight
differences in rate and a few small differences in
detail, but the identity is striking.

The determining factors in such groupings are
the stresses prescribed and the tendency to preserve
the identity of the words. In English it is possible
to make a distinction between “up up up” and
“u pu pup” which would be impossible in French.
But such distinctions are made by maintaining the
arresting consonant as in “up- up- up” or an actual
double as in “up- up-pup”.

4. Four syllable groups

If a heavy stress is prescribed in a breath group of
four syllables like “te te' te tet”, the prescription is
usually carried out. In 53 cases, three subjects, the
prescribed stress is not given in 8 series. In 25 of
the 53 cases, the primary stress is clearly on the
second syllable; but it often happens that while the
second syllabic gets a definite stress, there is a
heavier stress on the third or fourth syllable. There
is a strong tendency to prolong and to stress the
later members of such a series. No uniform pattern
for the other syllables is apparent; the form is apt
to open with an iamb, and doubling, “te tet-te tet”,
often appears; but the second unit-group may be a
trochee or iamb; the iamb is the more common,
repeating the first foot, and bringing stress on the
final syllable. Such doubles, which appear when a
single is prescribed, may be called “adventitious
doubles”; they mark the division between the feet
and often precede or follow heavy stresses. Cf.
p. 81.

In the prescribed form, “te te te' tet”; (short
vowels), of 36 cases, three subjects, 33 have the
prescribed stress; and it is obviously the primary
stress in 24 cases. The exceptions are cases in which
the grouping is “te te te'/tet'” and a heavy stress
has been thrown on the final syllable. Here again
adventitious doubles often appear, usually following
the stress, marking the division. The grouping of
the syllables is usually “te' te/te' tet”, or “te' te/tet'-
/- tet'”; but variants may appear in the first foot.

With long vowels the results are similarimage
In 63 cases, four subjects, 51 have the prescribed
stress; it is easily primary in 36 cases, in others a
heavier stress appears later.

With long vowels, image, 17 cases, two subjects,
16 show the primary stress on the prescribed
syllable; the one exception has in addition a heavy
stress on the last syllable. The unit-groups are very
like those of the four-syllable groups with short
vowels. Adventitious doubles are not likely to
appear; though they are found in some cases. Cf.
Figs. 86, 87.

When the group of syllables is as long as a series
of four, the tendency to break up into two feet is
apparent; there is a tendency to put a heavy stress
on the final syllable of the breath group. Although
all of the four subjects are English-speaking, there
is a pronounced tendency to prolong and to stress
the last syllable of the four-syllable group.

A sentence like “Lil' 'll lie low” gives the same
forms as the four-syllable group. But with no stress
prescribed, the results are more varied. The main
stress falls in 11 of 15 cases on the first syllable,
3 times on the second, and once on the third syllable.
There seems to be no tendency to stress the
last syllable with the one subject uttering this particular
phrase.

The five-syllable phrase, “pop up a pop up”, as
uttered by a single subject (W.), has a definite climactic
stress on the penultimate “pop-”. But the
form is not a dactyl “pop' up a” plus a trochee
“pop' up”, instead, the form is image, with a
definite arresting consonant marking the initial syllable
“pop”; this avoids the form “po pup” and
keeps the identity of the two words “pop” and “up”.
In 17 of 19 cases the primary stress falls on the
penultimate syllable; the secondary stress occurs
sometimes on the initial and sometimes on the
second syllable of the group. No heavy stress falls
on the final syllable; the grammatical subordination
of the word prevents it. (In strictness, the
syllables which may be compared are the initial
and the penultimate syllables “pop”, and the second
and final syllables “up”; the vowels are not otherwise
comparable.) Cf. Fig. 113.

“Runnin' 'n' neighin'” also gives a five-syllable
group, but it is impossible to say more than that it is
easy to utter it in a single “breath group”. The long
Vowel in “neighin'” is not comparable with the
short vowel of “runnin'”. Cf. Fig. 111.

5. The separating and connecting of syllables as
a result of rate *7

The device of increasing or decreasing the rate of
a uniform series of syllables makes it possible to
study in some detail the types of connection and
of division which depend on rate. The rate of a
series is an important factor in grouping for several
reasons. In the first place, the absolute length of
a group has limits, so that increasing the number
of syllables in a group increases the rate of the
component syllables. This is apparent in the studies
of Rousselot, et al., who note the decrease in the
length of syllables when something is added. In the
second place, speech tends always to a rapid rate
and the increasing rate of a rhythmic series, with
long and short syllables, will modify the short syllables
while the long syllables are still unaffected
by the rate. And finally, increasing the rate of
utterance will throw together into larger unities
small groups which are ordinarily uttered separately.

As the rate of a uniform series increases, the
adjacent syllables begin to group together; this
shows in the formation of “doubles” or “abutting
consonants” between the syllables having both
releasing and arresting consonants. Cf. Fig. 90.

If the syllables in the series have only releasing
consonants, the change is apparent in the chest-pressure
tracing. Cf. Figs. 26, 28, 50, 57. If the
syllables in the series have arresting consonants,
the intersyllabic space disappears. A second type
of grouping occurs at still higher rate when the
arresting consonant drops from the abutting pair,
or shifts to the releasing position. In these uniform
series in which the rate is gradually increased or decreased,
the indefinitely long breath group developed
is artificial and unusual, and corresponds to the
trill or run in music; such prolonged breath groups
very seldom occur in actual speech. In the earlier
part of such syllable trains, the syllables have each
a breath group; as the consonants single the syllables
have a long breath group of an indeterminate
number of syllables; an unusual form of utterance
except in “patter”. Cf. Figs. 40, 41, 42.

6. Functions of doubling and singling in the
grouping of syllables

The same types of connection seen in these uniform
series with increasing rate are to be found in syllables
uttered in ordinary groups. When the individual
feet (of one, two, or three syllables) are
uttered separately, an intersyllabic interval occurs
between the feet or breath groups, during which
the chest pressure goes down to zero. When this
space closes up, and an abutting pair or an arresting
consonant remains, the linkage indicates the connection
of one foot with another in a breath group.
A two-syllable breath group composed of two single-syllable
feet connected by a double is represented by
image. A four-syllable breath group composed of
two iambs connected by a double is represented by
image. It often happens that subjects actually
regroup into larger phrases what were intended to
be separate breath groups. This shows in the development
of the abutting form between such prescribed
groups. Fig. 85 is a case in which breath
groups are accidentally connected by abutting pairs.

In such cases it is not necessary that the double
occur either at the beginning or the end of the
stressed syllable; doubling may occur between two
stressed syllables. Cf. Fig. 85. As it often happens
that the stress begins or ends the group, it is
frequently the case that the abutting pair occurs
just before or just after the stress.

“Singling” occurs when the connection is so close
that the syllables involved form parts of a foot.
When “bub -bub” becomes “bu' bub”, the breath
group of two monosyllabic feet image has become
the iambimage.

7. Function of the word stress in the grouping
of syllables

In actual speech another factor plays a part in grouping.
As already stated, the “word stress” figures
in the organization of the foot and breath group.
The stressed syllable is in general lengthened; this
prolongation of the stressed syllable may appear
as a pause after the beat-stroke, as in the iamb and
anapest, image and image. In the case of the trochee,
the pause may appear before the beat stroke. Often
the prolongation of the final member of the group
masks the lengthening due to the stress, and the
two syllables may be of approximately the same
length. This relation of length and stress has been
considered at some length by the rhythmists. (Cf.
Riemann, 65.)

One might expect that the prolongation of the
stressed syllable would naturally appear in a
“double” (abutting pair) before or after the stress.
Such doubling can occur, however, only when the
stressed syllable occurs at the beginning or end of
the foot. Thus a double may precede the trochaic
stress in a series image; and a double may
precede the dactylic stress in a series:image.
And also a double may follow the iambic
stress in a series imageBut in a mixed
series like imagethe double neither precedes nor
follows the stressed syllables. Cf. Fig. 90.

If the stress is fixed, then the feet are formed to
accommodate themselves to these stresses. In such
cases the nature of the vowels sometimes figures in
the formation of the feet and breath groups.

If the forms are trochaic or dactylic, the heavily
stressed syllable is not prolonged; the slightly increased
length of the stressed beat stroke appears
in the preparation of the blow. Therefore there
will be no double (abutting pair) following the
trochaic or dactylic stress, however heavy the word
stress. Indeed, a very heavy stress may lead to a
“reduction” of a single median consonant to be
discussed later. If the vowel of the preceding syllable
is short, doubles are likely to appear before
the stress, and so mark the division of the feet. If
the vowel of the preceding syllable is long, a single
consonant will not be doubled (a prescribed abutting
pair may appear, however).

In the case of iambic (including anapestic) feet,
the syllable receiving the stress must be somewhat
prolonged (“apogogic accent”).This increased
length is due to the heavier blow of the stress. If
the syllable involved has a “long” vowel (i.e. a
vowel that can be prolonged), the increased force of
the stress shows in the increased length of the chest
pulse, and the vowel is somewhat longer. If, however,
the stressed syllable of the iamb has a “short”
vowel (i.e. a vowel that cannot be prolonged), the
increased chest pulse meets the obstacle of the
arresting consonant. The consequent truncation
of the blow results in the prolonging of the occlusion
of the arresting consonant which provides for
the increased length of the beat stroke as well as of
the back stroke of the syllable pulse. The heavy
stress on the iamb may force the single, releasing
consonant of the next syllable into the arresting
position, and the consonant may double. This is
the “adventitious double”. If the shift from the
releasing position to the arresting position occurs
without repeating the consonant, the result is an
“arresting consonant” which marks the limit of
the foot.

Neither the feet (one-syllable foot, iambs,
trochees, dactyls) nor the larger breath groups are
necessarily words. The identity of the word is lost
in utterance. The breath groups of the phoneticians
103constitute the larger groups; sub-groups are
the feet, within these breath groups; but none of
them need coincide with words. The tracings of
chest pressure during speech show these groupings
surprisingly well. One finds not only the breath
groups indicated but also the various feet within
the breath group.

It is natural to think of the “long” and “short”
syllables as important in the construction of rhythm
forms, and the consequent organization of breath
groups. A little study of the tracings shows that
one difference between the “long” and the “short”
vowels of English is the possibility of prolonging
the “long” vowel, if the syllable is to be lengthened.
The “short” vowel is one that cannot be prolonged
if the syllable is lengthened; the prolongation must
occur in the arresting consonant or the intersyllabic
interval. Cf. Fig. 91. This distinction appears again
in the differentiation of “tense and lax” vowels.
When the syllable is arrested primarily by the consonant,
the vowel is said to be lax. But if the syllable
is arrested in part by the chest muscles, the
vowel is counted “tense”. Thus the attempt to
prolong an English “short vowel” destroys its
quality; in such case the syllable is arrested in part
by the chest and the vowel becomes tense. In calling
or in singing, the prolonged syllables are always
arrested in part by the chest, and an arresting consonant
may become superfluous (88, p. 155).

The increase of stress means that the moving
member reaches the obstacle with a higher momentum.
The contraction of the negative muscles,
therefore, will be correspondingly greater, and the
back stroke of the arresting consonant will take
a correspondingly longer interval. If the arresting
consonant is a continuant, the consonant will be
prolonged. Cf. Fig. 82. If the arresting consonant
is an occlusive, the back stroke (or the equivalent
relaxation phase) will take a correspondingly longer
time. This will appear as a “pause” (intersyllabic
interval) after the syllable. At the end of the group,
the occlusion may be without détente. Cf. Figs.
91, 92, 94.

In the phrase “To be or not to be”, Bell notes
such a long pause, i.e. the relaxation phase of the
vigorous “t” of the “not” (2, p. 78). It happens
in this example of Bell's that the long pause occurs
in the midst of a double consonant.

The incorporation of a releasing consonant, the
elongation of a “long” vowel in the open-syllable,
the incorporation of both a releasing and arresting
consonant in the syllable with a short vowel, are
all comparable to the agogic and apogogic stress
of the musical note (65).

In music the stressed note is preceded by a slight
pause and is slightly lengthened by the stress. In
the case of the stressed note, additional time is required
to prepare for the heavier stroke and additional
time is required to take up the momentum
of the stroke and prepare for the next stroke. In
the case of the stressed syllable in speech, the
preceding consonant is drawn in to assist in the
preparation of the heavy stroke, and if the vowel
is “short”, the following consonant is drawn in to
assist in the arrest of the heavy stroke. The increase
of the interval just before the chest pulse of the stressed
syllable makes the beginning of the stressed stroke
always an available place for a consonant in rapid
utterance. If the vowel of the stressed syllable is
short, the end of the stressed syllable is also an
available place for a consonant. It is not so much
that the consonant can be tolerated at the beginning
of all stressed syllables, and at the close of
stressed syllables with short vowels, but rather that
they are needed in the releasing and arresting
coordination of the syllable movement.

It is possible to offer an explanation of the influence
of the double consonant on the preceding
vowel, so patent in the study of Italian phonetics
by Josselyn. When the consonant is doubled, the
preceding vowel is closed. The first consonant of
the double becomes the arresting consonant of the
preceding syllable. The closed syllable with an
arresting consonant is short in duration — that is
the inevitable limitation of the syllable with unassisted
consonant arrest. E.W.Z Scripture quotes
Meyer's measurements of Wanger's records showing
that a consonant is lengthened after a short vowel,
which is the reciprocal effect on the consonant; the
arresting consonant is likely to be more vigorous
than the releasing consonant and its occlusion is
really part of the syllable (76).105

VII
The breath group in english dependent
on the stress pattern of the words

Other more irregular patterns are presented of
course by the word stress of ordinary speech. The
choice of the grouping will depend on the syntax,
and also on the presence of abutting consonants
between the syllables. Doubling will tend to occur
between the feet and neither doubles nor any other
form of abutting consonants are likely to appear
within the foot. A decided increase in rate may
work further changes by eliminating certain unstressed
syllables. Changes of that type will be
considered later.

The larger groups, the breath groups, and often
component feet are marked off by low points in
the chest pressure. Every foot has its stress, and
every breath group has a dynamic climax as its
primary stress. The unity of the breath group is
constituted by a slow movement of expiration of
which the minor movements of feet and syllables
form parts like the partials of a sound wave.

1. Experimental study of groups of syllables

A. The group of two syllables

Types 1 and 2 occur when there is a single consonant
between the syllables, and the stress is not
heavy. Type 3 occurs if there is an abutting pair
of median consonants.106

This occurs when the first syllable is heavily
stressed and the group is uttered at rapid rate.

1. Discussion of the iambic group

The average length of 234 iambs was 0.56 sec.
(range 0.41-0.67 sec). The stressed syllable is
usually longer than the unstressed. The stressed
syllable occupied ca. 60 per cent of the total duration
(range 55-73 per cent).

No cases were observed of “adventitious doubling”
of the single median consonant in a two-syllable
group with the stress on the second syllable;
there is no tendency of such a group to divide under
heavy stress into two single, stressed syllables.

The elimination of a syllable does not occur in
the case of an isolated iamb, no matter how rapid
the rate of utterance, nor how emphatic the stress.
In larger groups, phrases, the median iamb is subject
to reduction; the short unaccented syllable tends
to “telescope”, leaving the releasing consonant of
the syllabic in some cases. This abbreviation is
evidently the result of vis a tergo; if there is
material preceding the unstressed syllable, it is first
compressed and then forced out. We shall consider
the cases of “telescoping” and of elimination of the
unstressed syllable of the median iamb as we come
to them in breath groups of three and four syllables.

2. Discussion of the trochaic group

The average length of 207 trochees was 0.61 sec.
(range 0.37-0.81). The two syllables, stressed and
unstressed, are usually of about the same duration.
The stressed syllable occupied an average of about
50 per cent of the total duration (range 41-59 per
cent). The fact that the unstressed syllable is the
last syllable tends to prolong the trochee; this is
more apparent in final trochees in larger groupings,
but it has some influence in isolated trochees. A
heavy stress on the trochee tends to lengthen the
stressed syllable. Both these tendencies have been
noted by Rousselot.

It is interesting to note that the average length
of the isolated trochee is slightly longer than that
of the isolated iamb. In larger groupings, of which
trochees and iambs are constituents, no such difference
in the average length is apparent.

If the rate is normal, abutting median consonants
reduce to single or compound consonants. In this
the reduction is somewhat like that of the iamb.
One may say in general that all abutting consonants
between the constituent syllables of a foot
are reduced to single or compound consonants, and
in every case it is the releasing consonant which
persists. Cf. the series in which the consonants
double with increasing rate. The rate of the
elements in the foot is responsible for the elimination
of the arresting consonant of a median abutting
pair.

3. Reduction and elimination of the median consonant
of the trochee

The iamb which stands alone is not subject to
modifications which abbreviate the syllables, but
this is not the case with the trochee. If the trochee
it uttered very quickly, or if a heavy stress is put
on the first syllable, internal modifications occur.

The median consonant of the trochee, the releasing
consonant of, the unstressed syllable, is 1)
“reduced”, 2) vocalized if a surd, 3) finally eliminated.
The “reduction” means that the contact of
the consonant becomes much briefer and the stroke
to the opposing surface much ligther, and the
pressure in the mouth for the consonant becomes
much less. Much the same thing is to be seen in all
series of syllables at very rapid rate. A heavy stress
on the trochee increases the rate of the unstressed
syllable, for it leaves less time for the unstressed
syllable.

At the same time, the consonant becomes a sonant
if it is a surd. It has been assumed that an intervocalic
consonant is vocalized because that means
less effort (10). But the fact is that the vocalization
is inevitable; the consonant occlusion grows less
and less, and breaks the flow of air through the
glottis for a shorter interval. It is not a sudden
process as Fig. 96 shows. It is easy to see that the
vibration of the vowel creeps into the occlusion
and finally runs quite through it; the closure of the
consonant becomes so brief that it does not interrupt
the flow of air through the vocal folds.

When the contact has become so brief, there is
little rise of the mouth pressure, the consonant
functions almost not at all and the syllable movement
becomes practically chest released. Since the
median consonant stroke no longer functions it
108may disappear entirely, as a consonant stroke is
certain to do if it has no part to play in the syllable
movement. The median consonant of the trochee
has dropped. The two vowels tend to fuse and a
single syllable with a modified vowel results.

If the vowel of the unstressed syllable is extremely
short, and if the trochaic median is sonant, especially
if it is a continuant, the median consonant
stroke may fuse with a final consonant, while the
second syllable movement fuses with the stressed
chest pulse. Thus “debit” if heavily stressed and
rapidly uttered becomes “debit”, which soon loses
its internal consonant and becomes “de't”. “Tes'et”
becomes “tes't”, which persists because the two
arresting consonants fuse to a compound consonant,
“st”. Some 50 cases occur in a total of 207 trochees
studied.

Sometimes the reverse process takes place, and a
syllable like “elm” or “help” opens into a trochee
and becomes “elum, helup” — as often in dialectic
English.

The causes of this reduction, vocalization, and
elimination of the trochaic median consonant are
to be found in the rate at which the unstressed
syllable is uttered. If the first syllable is heavily
stressed, the arrest must occur before the release of
the unstressed syllable is possible. The releasing
consonant has to occur in the very brief syllable
at such a rate that the blow is very swift and therefore
very light. If the unstressed syllable is not
closed, the reduction process does not occur as
readily.

The peculiarities of the, “long” and, “short”
vowels are well demonstrated in forms like “totet”
in which the stress is thrown alternately on the first
and second syllables, forming a series of alternating
trochees and iambs, “to'tet, to tet'; po'pop, po pop';
debit, debate”. In such cases the “long” vowel
shortens when unstressed; but the “short” vowel
does not lengthen when stressed. Instead, the short
vowel is followed by a double or an arresting consonant
which gives the syllable its rhythmic length.
Cf. Fig. 91.

The simple grouping, the re-grouping, and the
actual position and function of the abutting consonant
pair (including the double), is remarkably
well shown in these two-syllable breath groups.
The trochee illustrates the process of dropping the
trochaic median consonant, and also of dropping
a syllable by fusion. And the trochee furnishes the
first illustration of the “adventitious double” marking
the division of the breath group into two one-syllable
feet.

4. Relative duration of the Iamb and the Trochee

The iambs and the trochees are of much the same
average length; at maximum rate the trochee can
be shorter than the iamb. The average of these
two-syllable groups, .56-.61 sec., is much higher
than the average length of the one-syllable feet
into which the two-syllable breath group is divided
on occasion. The average length of 717 one-syllable
feet, when initial, was 0.26 sec. (range 0.18-0.50
sec.). The average length of the 657 one-syllable
feet, when final, was 0.31 sec (range 0.20-0.44 sec).
While the average length of the one-syllable feet
overlaps the shorter of the iambs and trochees (cf.
ranges pp. 107-108) it is easy to see that the distribution
is quite distinct. But it is to be remembered
that temporal equality is not to be expected of
spoken rhythms. Minimum time intervals for
movements are important because they condition
the form of the movement and may compel profound
phonetic changes. But the wide variation in
time intervals longer than the minimum has no
particular rhythmic significance.

B. Breath groups of three syllables

Of these breath groups, 1, 2, 4, 5, have each two
feet, they are varying combinations of iamb or
trochee with a one-syllable foot; 3 consists of three
one-syllable feet; 6 is a dactyl; 8 is a similar form
with the stress on the median syllable. Group 8 is
a form which occurs only in isolation; a series of
such amphibrachs would reorganize into the dactylic
form 6. If all the syllables involved were closed
syllables, the feet would be indicated by the doubling
and singling of the consonants.

In breath groups of three syllables, certain characteristics
due to the length of the group are apparent.
The initial one-syllable foot of such a breath group
is shorter than the final one-syllable foot. There is
a tendency to prolong the final element of such a
breath group; and as the expiratory movement of
the breath group has rather definite limits, the
earlier members of a breath group tend to become
shorter and shorter as the breath group increases
in length. This appears in the case of the iamb and
the dactyl. The initial iamb or dactyl of a breath
group is shorter than the final iamb or dactyl.

The presence of doubling consonants and of long
vowels will be determining factors in the breath group
organization. Thus “ex'quisite” = image;
but “exquisite thing” = image. With a pronouncedly
short vowel in the final syllable, “harmony”
has the characteristic dactylic form; but
“harmonize” becomes image.

In the three-syllable breath group, both the
iambic and the trochaic unstressed syllable may be
elided as the rate is increased. The iamb has a pronounced
difference in the length of the two syllables,
and if it occurs in succession at rapid rate,
the rate may be too fast for the occurrence of the
short syllable. If the original form gives an arresting
consonant to the unstressed syllable, this will
drop early, of course, leaving the unstressed syllable
with releasing consonant only. At a higher rate the
unstressed syllable stroke is fused with the following
stressed chest pulse, but the releasing consonant may
persist by shifting to the arresting position in the
preceding syllable. Cf. Figs. 97, 98, 99. Thus
“beautiful”, becomes “beaut'ful”; “business” has become
“bus'ness”; “plentiful” becomes “plent'ful”;
“laboratory” becomes “lab'ratory”.

In case the syllable preceding the eliding syllable
of the iamb already has an arresting consonant, the
releasing consonant of the syllable becomes functionless;
it must either fuse with the releasing consonant
of the stressed syllable or drop. In a form like
“institute” becoming “in'stute”, the median consonant
of “nst” fuses with the releasing “t”. Cf.
Fig. 99. In “hospital” becoming “hostel” the “p”
does not fuse with the following “t” and therefore
the “p” drops; so the “b” in “presbyter” becoming
“pres'ter”. Cf. Fig. 77. In “abdicate” becoming
“ab'cate” the “d” drops, although the “b” and “c”
are not formed by the same member.

It is worth noting that in these cases, with a continuant
like “s”, an intersyllabic sound does not
develop. In the form “instate”, if one attempts to
prolong the syllables, one finds that the “in-” can
be prolonged by continuing the “n”, but the “st” is
given very briefly as the releasing consonant of
“-state”. Such a prolongation of a syllable often
occupies the time of the elided syllable, but it is not
an equivalent, in the phonetic sense that the continued
sound becomes a syllable. Instead the fact is
that something like the same length is maintained
for the rhythmic foot. This is the explanation of
“epentheses”, discussed by Rousselot (70, p. 989).
The transitional “sound” which develops as “teneru”
becomes “tendre” is not the equivalent of the syllable.
Actually, the releasing consonant has become
“dr-” and the increased length means that the two
syllables represent the same rhythmic foot that the
three syllables did before.

Rousselot suggests the steps whereby “cabitale”
(capitale) becomes “ca-b-tal” with “b” explosive;
then “cab-tal”, “b” implosive; then “captal”, “b”
surdifies, then “chat'el” with the implosion still
apparent and finally “chetel” in which the implosion
has disappeared (70, p. 977). The analysis is
good, save that the function of the consonant has not
been noted; there can be no “explosive” (releasing)
function for the “b” when the syllable “-tal” has
the releasing “t”. And the same causes which force
the arresting consonant out of the syllable “cha-”
must prevent the formation of a hiatus at the close
of the syllable; were there room for the hiatus there
would be room for the arresting consonant. The
problem of a single releasing (explosive) consonant
functioning as a syllabic is treated below. The steps
would be: “cabitale (capitale), cab-tal, chatal,
chetel”.

The surd between two vowels often becomes a
sonant, as noted in the example cited by Rousselot
above. This is the usual transformation of the
trochaic median which has previously been discussed.
Cf. Fig. 96.

Rousselot assumes that in the case of a syllable
with “e muet”, the consonant alone may sometimes
figure as the syllabic when the “e muet” drops; “la
femm' se l'va” shows the elision of the “e muet”
after “m” and after “l”, and yet the syllable “fe-”
112of “femme”, is not closed, nor the syllable “se” preceding
“l'va” (70, p. 981). In such cases it is of
course impossible that the consonant alone become
a syllable; a chest pulse is essential to a syllable.
But it is possible for the compound releasing consonants
“mse” and “lva” to develop as the releasing
consonant of each syllable, and the lengths of the
rhythmic feet are adjusted as suggested above.

C. The breath group of four syllables.

The group of four syllables presents a wide variety
of combinations of the feet. The dactyl may be preceded
or followed by one-syllable foot, the iamb or
the trochee may appear in various successions with
one-syllable feet, and trochees and iambs may occur
together as initial or final. Combinations of trochees
or iambs with dactyls are the only groupings which
are not represented in the four-syllable breath group;
and they would give no new arrangements of stressed
and unstressed syllables, or of combinations of
syllables with single consonants and with abutting
pairs (including doubles). All possible modifications
of the syllables are to be found in these four syllable
breath groups.

These four-syllable successions develop series of
feet between which abutting pairs are given full
value, or between which adventitious doubles
appear.

The adventitious double develops in the four-syllable
breath group as it does in the three-syllable
breath group. It usually occurs just after a heavy
stress, but it may also occur just before a heavy
stress. If the vowel of the preceding syllable is
“short”, the pause during which the heavy stress is
prepared may be occupied by a double. The alternative
would be a hiatus after the chest-arrest of the
short vowel. In 130 four-syllable breath groups,
four subjects, adventitious doubles occur in 47 cases.

D. The elision of syllables, modification and
contraction in four-syllable breath
groups

The four-syllable breath groups offer a good field
for the study of the tendency to drop syllables.

There are two processes by which the four-syllable
breath group is reduced to three syllables, reduction
of the trochee and the telescoping of the iamb. Cf.
Figs. 100, 101, 102, 103.

Since the adventitious doubles appear before and
after the stress for reasons given above, and since
the fusion of the syllables of the trochee must involve
the stressed syllable, and since the telescoping
of the iamb occurs just before the stress, it is easy
to formulate a general rule that the various modifications
of the consonants and syllables will appear
just before and just after the stressed syllable. As
the stresses may be in several patterns, this means
that adventitious doubling, persistence of abutting
consonants, telescoping, and fusion may appear between
syllables 1 and 2, between 2 and 3, and
between 3 and 4. It is only the initial arid final
syllables and consonants which are not affected.

Here, as elsewhere, abutting consonants, long or
short vowels, along with the prescribed stresses,
determine the precise series of feet developed within
the four-syllable breath group.

It is convenient to schematize the changes in both
consonants and syllables which occur in a series of
syllables grouped in various types of feet, when the
rate is increased. It is easy to see that in the case
of the trochaic series, and of the iambic series, that
all arresting consonants finally drop with the increase
of rate; and that the releasing consonants of
unstressed syllables drop also. The unstressed syllables
invariably drop with the increase of rate. The
stressed syllables and their releasing consonants persist,
whatever the rate. Cf. Figs. 104, 105, 106.

E. Comparison of the abutting and single
consonants of stressed breath groups
with the abutting and single consonants
of series developed by increasing the rate

In Fig. 108, curve K represents the distribution of
the lengths of all the doubles and abutting pairs
recorded in 3101 cases. Curve M represents the
distribution of the lengths of the doubles and abutting
pairs in stressed groups recorded in a series of
705 cases. It is apparent that the distributions are
practically identical, and that the process of doubling
in the stressed group is precisely the process in
the artificial series of syllables with changing rate.
The curve L represents the distribution of 231 cases
of arresting consonants. They are added to show
that while the process of arrest with pause is nearly
the same, there is a slight difference; the arrests with
pause are slightly longer than the actual doubles or
abutting pairs.

The rates of utterance in stressed groups, and of
syllables in series with changing rate, show much
the same common form; so do the consonant lengths
just discussed. In Fig. 109, the curves A, B, are
composed of d, e, 506 series, and of f, g, 424 series,
thrown together from Fig. 51. C represents the
distribution of 96 cases in which a prescribed double
in a stressed group has been uttered as a single consonant
(due to the influence of the grouping of the
syllables). The syllable lengths have been reduced
to rates. It is apparent that the distribution is practically
the same, i.e. the single consonant which
appears in place of a double in a stressed group is
as near to the doubling rate as possible, and therefore
is very like the single consonant which appears
just after doubling in the series at increasing rate.

These comparisons of the lengths of the doubles
and abutting pairs, and of the rates of syllables with
double consonants and abutting pairs, show that on
the one hand, the rate of utterance will determine
the form as “doubling” or “singling”. On the other
hand, the prescribed form of “doubling” or “singling”
in stressed groups of syllables will determine
the rate of utterance of the syllable in question. The
form of doubling or singling and the appropriate
rate of utterance go together.

F. Reversion to the prescribed form when the
rate and the rhythm permit

The tendency to restoration, to reversion, after a
phonetic modification, plays a part in breath groups
of syllables with a stress (arranged in rhythmic feet)
just as it does in series in which the rate changes.
Cf. p. 73. The prescribed consonants and syllables
may be profoundly modified by the rhythm
and the consequent rates; but the original form is
still with the subject, and appears whenever the
conditions will permit. Therefore stem forms remain
through the varying modifications which result from
the addition of suffixes and the shift of the stress.
In forms like “sled”, the “d” is arresting; in “sledding”
an adventitious double might appear as the
spelling indicates, but this is very seldom the case;
instead, the arresting “d” becomes the releasing consonant
of the next syllabic, and the syllables are
really “sle-ding”.

In forms like “adverti'sing”, and “adver'tisement”,
the “s” reverts from the releasing to the
arresting position when the abutting pair “s-m” is
produced by adding the “-ment”. Shifting the stress
often changes the division of the syllables, and often
leads to the elision of syllables. The American
pronunciation “lab'oratory”, image leads to
the dropping out of the short of the iamb, leaving
“lab'ratory”, with an abutting “b-r”. The British
pronunciation “labor'a tory” image tends to the
dropping of the short syllable of the iamb, “labor'at'ry”,
and finally to reduction of the trochaic
median, leaving “labor't'ry”, three syllables.

G. Breath groups of more than four syllables

The prescribed formula “pop up' a pop' up” was
uttered so as to make an interesting variation of the
five-syllable breath group. The subject gave it
“pop' / up' a / pop' up”, with a primary stress on
the final “pop' up”. In 18 of the 19 cases measured
the primary stress occurs on the final foot. In 10 of
the 19 cases, the initial one-syllable foot “pop” is
marked off by an arresting consonant, the adventitious
double “pop-pup” is avoided to keep the
verbal “up” quite clear. Cf. Fig. 113.

Of course many phonetic changes of the consonants
are not included in these processes of dropping
the arresting consonant, of doubling and singling,
and of reducing the trochaic median consonant. The
changes involving aspirates, soft gutturals, nasals,
semi-vowels, and liquids are often not of this group
of changes. The influence of vowels on each other,
and of consonants and vowels on each other, has
not been considered in detail. Syllables are eliminated
by other processes than iambic telescoping and
trochaic reduction.

It will not be difficult, however, to devise series
in which such elements are brought together, and it
may be possible by changing the rate and the rhythm
to produce in the laboratory the changes which
occur in the history of a language.

H. Influence of stress and the syllable
movement in the modification of “sounds”

The quality of the sounds of the vowels may be
affected by the stress. Even in French it is possible
that the stress may produce slight differences in the
vowels. Rousselot and Laclotte note changes of
the vowels from open and closed to median qualities
when the vowels become atonic: “á” (fermée) becomes
“a” (moyenne) in bas de soie; “é” (fermée)
becomes “e”, (moyenne) in “bonté de coeur”; “ò”
(fermée) becomes “o” (moyenne) in “cotelette”; “oe”
(fermée) becomes “oe” (moyenne) in “feu de joie”.
(72) In every one of these cases the elision of the
“mute e” leaves a consonant in the arresting position
for the syllable in question. They are all cases of the
elimination of the unaccented syllable of the iamb
with the resulting abutting pair formed when the
syllable telescopes. “Bonté d(e) coeur, bas d(e) soie,
cot(e)lette, feu d(e) joie”.

In English the shifting of the stress modifies the
vowels profoundly. Note the. “i” of “ad'vertise,
adver'tisement”; the “a” in “profane, profanation”;
the “a” in “circulate, circulatory”; the “y” in “analyse,
analysis”. The syllables which do not have the
primary or secondary accent tend toward “dark e”.
The sense of the quality of the vowels persists and
the proper vowel reappears in derivatives and in
calling and singing.

This change of vowel quality with the change of
stress is a matter of rate of utterance of the syllable.
There is not time enough for the vowel movement
to occur in normal fashion. The English “dark e”
and the “mute e” of the French is merely the vowel
reduced to its simplest and briefest form; it is the
vowel “sound” produced when there is the least
movement possible from the neutral position. The
“dark e” is even briefer than the “short” English
vowels.

If the syllables are chosen and the stress prescribed,
so that the unstressed vowels cannot reduce
to “dark e”, the maximum rate of utterance is much
slower. “Tee' too tay…” can be repeated at a
maximum rate of 4.5-4.8 syllables per sec., “La'
pee yo…” at a maximum rate of 6 syllables per
sec., while “ta ta'…” (becoming “te ta'…”) can
be repeated at a maximum rate of 8-10 syllables
per sec.

The relation of the word stress to the quality of
the vowel has been built into an elaborate system in
the Hebrew. Possibly it is rather artificial as we
have it now, but it undoubtedly represents something
of the ancient practice. Vowels in syllables
121with the stress are systematically lengthened; vowels
preceding the stress are regularly reduced to “dark
e”, the shwa. The vowel is a more variable factor
to the Semite, because of the variations of the vowels
of the stem in the course of inflection.

The effect of the actual movements on each other
has been carefully described by Rousselot (consonant
and vowel, vowel and consonant, vowel and vowel,
consonant and consonant). The inflectional changes
bring together various articulations, and the elision
and modification of syllables in the course of historical
changes subject phonetic movements to the influence
of each other. Spelling and printing can
only delay such changes, they cannot prevent them.
In general the movements of speech are shaped by
rapid utterance. Rhythm and stress will be preserved
and the phrase must be comprehensible; but with
these reservations the influence of mere velocity
will dominate. It is a general law of language that
all phonetic coordinations shape themselves by and
for the maximum rate of utterance.

Generalizations have been made on the reciprocal
influence of “sounds”. There is a “law of assimilation”
whereby the movements tend to become alike;
“law of prevision”, whereby movements are prepared
in advance as in all fields of skilled movements;
the “law of economy” whereby only the movements
necessary for the given “sound” occur; the “law of
fusion” in the case of two contiguous vowels. (72)

All such generalizations must finally be referred
to the skilled movements involved. The word stress
and the syllable movement play an important part
in such reciprocal modifications. The back stroke
of the releasing consonant must occur during the
syllable and must modify the vowel. The preparation
and beginning of the beat stroke of the arresting
consonant must occur during the syllable movement
and must modify the vowel. The patterns of
Visible Speech show the vowel modifications in
acoustic detail; the back stroke of the releasing consonant
opens the vowel cavity and produces modulatory
changes which indicate the place of the consonant
constriction and the consonant's influence in
the beginning of the vowel. The patterns of Visible
Speech show that the beat stroke of the arresting
consonant constricts the vowel cavity producing
modulatory changes which indicate the place of
constriction and the consonant influence on the end
of the vowel. The consonant movement must be
inferred and the detail of air pressures and the syllable
pulse cannot be read in the Visible Speech
recordings, but the consonant overlap and its consequences
in the vowel are well shown for the first
time (10). The back stroke of the arresting consonant
may occur during the intersyllabic interval, or
it may modify a following initial vowel. Since a
releasing consonant in the next syllable may separate
the back stroke (détente) of the arresting consonant
from the next syllable, there may be no influence.

Consonants occur in juxtaposition in two cases:
compound consonants in which the fused consonants
function as a single releasing or arresting factor;
abutting consonants in which each member has a
different function in a different syllable.

The simplest fusions in compound consonants are
those in which a continuant becomes the preparation
or the back stroke of an occlusive, e.g. “stay,
play”. The mechanism of voicing is dependent on
the articulatory organs and on the breathing apparatus,
and changes of voicing are difficult in the
very brief interval of the compound consonant;
therefore the group of components tends to be voiced
or unvoiced. (35)

With the abutting consonants the conditions are
quite different, for members of the pair belong to
different syllables. In many languages “assimilation”,
the influence of one abutting consonant on
the other, is very common.

If the differences of stress and of length of syllable
are not extreme, the releasing consonant is the
dominant movement, and the arresting consonant is
modified toward the releasing consonant. This is
the rule in Greek and Latin, and in French. As the
rate of utterance increases in a fairly uniform series
of syllables, the arresting consonant must be uttered
very rapidly just before the releasing consonant of
the next syllable. A similar movement, or an identical
movement, can be given in sequence more
122rapidly and certainly than two different movements.

But the Teutonic languages, with heavy word
stress, series of “short” vowels in stressed syllables,
and great variety in length of syllable, show much
less assimilation. The changes of the abutting pair
are not always in the direction of the releasing consonant.
This is not due, as Sweet quaintly suggests,
to an English “striving for distinctness” — there is
too much against that idea — but to the function of
the arresting consonant in a stressed syllable with
“short vowel”. Passy calls attention to the frequent
“progressive assimilation” in English and German,
as in the English “observe”. (61) On occasion the
form of the movement of the arresting consonant
holds over to the releasing consonant. But as Passy
indicates, the assimilation is not invariable. “Observe”
may be compared with “subserve”, in which
there is no assimilation.

The words “blackboard” and “flagpole” have a
secondary stress on the second syllable and the syllables
remain distinct. When the two consecutive
stresses are lost in rapid utterance, all sorts of
changes are possible, as “forehead”. (for'ed), “forecastle”
(fok'sl), “boatswain” (bos'n), “blackguard”
(bla'gard).

Nothing has been said of the influence of habit,
“association”, in modifying phonetic movements.
It is certain that this factor plays an important part.123

VIII
Rhythm and the characteristic utterance of
a language

The rhythm is certainly one of the most fundamental
characteristics of the utterance of a language,
and is often most difficult for a foreigner to acquire.
The play of the word stresses, the rhythmic grouping
of the breath groups and phrases, the differences
in the length of the syllables are all difficult, and
all-important for a good “accent”. And it is not the
case that one can first master the “elements” of the
pronunciation, the “sounds”, and then set them in
the rhythm. It is easy to see, that the rhythm has
a vital, underlying influence on the details of utterance.
The word stress often determines the function
of the consonant as arresting or releasing and also
determines the syllable in which the consonant shall
function. The rhythm at a rapid rate determines the
slurring or the full pronunciation of syllables. In
the end the rhythm guides the phonetic changes
which every language is undergoing.

Sweet takes a practical view of the subject of
stress- and pitch-accent in the early I. E. languages.
He is certain that what happened in Greek was not
that a pitch accent was replaced by a stress accent
— which would be impossible for a variety of
reasons — but that the early I. E. accent involved
the essential stress factor along with a pitch factor;
the pitch factor was lost but the essential stress
factor remained.

Since the stress increases the tension, it is natural
that the increased chest pressure, compensated by an
increased tension of the vocal folds, raise the pitch.
Thus a rise of pitch as a result of tension is a by-product.
The stress may fall, however, on a syllable
with a low pitch because of the intonation pattern,
and especially at the end of the declarative phrase.

The Romance languages and the Slavic languages
do not have the pronounced differences in the length
of syllable. (95) This is equivalent to saying also
that they are the languages in which there are no
extreme differences in the stress of the syllables.
The French also, along with the Spanish, modern
Greek, and the Slavic languages, has a decided tendency
to the open syllable, while the English and
German have many closed syllables. (78)

In French the pattern of the alternating stressed
and unstressed syllables may vary in the same breath
group from speaker to speaker, and may vary from
time to time with the same speaker; the variation
depends on a difference of emphasis or on personal
choice. As the syllables are not greatly varied in
length, the rhythm is in the main trochaic and
dactylic — the forms in which the length of the
syllable is nearly the same. The frequent “mute e”
and the occasional closed syllables make iambs possible;
anapests are very rare. As it occurs in the
breath group, the “word” stress of the French is not
fixed. As a rule, the “mute e” does not have the
stress, but it may on occasion; thus in a series of
monosyllables with “mute e” there is a stress on the
alternate syllables. The conventional stress of the
isolated word is on the final syllable, of the word
in context on the tonic syllable; but frequent “displacements”
of this stress for emphasis and for
rhythm are essential to a good French diction. Passy
and Michaelis note the possibility of pronouncing
“impossible” and “excellent” with a stress on any
one of the three syllables. (61)

In English and German on the contrary, there
are very pronounced differences in length of syllable
124and great differences in the force of the stresses.
Sweet speaks of the possibility of distinguishing
three different grades of length, and the customary
notation of “primary, secondary and nul” distinguishes
three grades of stress. With very rare exceptions
the English contrasts with the French in
having a fixed pattern of the word stresses. In
French the stressing of the breath group determines
the stress of the word; in English the stresses of the
words determine the stress pattern of the breath
group. The prevalence of syllables short in duration,
and the extreme force of the word stress
combine to give the English a prevailingly iambic
rhythm. It is not unusual to find consecutive syllables
stressed. Trochees and dactyls are common.
While there is perhaps more variety of rhythmic
pattern than in the French, the general character
of the English is iambic. A. .W. de Groot comments
on the rhythmic difference between the English
and German, and the French. (15)

This system of great difference of stress and of
length of syllable gives important functions to the
consonants which tend to maintain the difference
of stress and length of syllable. The pronounced
differences of lengths of syllables and the variety
of rhythmic grouping make a place for the arresting
consonants, whether followed by a vowel or a
releasing consonant.

In French, on the other hand, the tendency is to
evenness of stress on the syllables and to an even
length of syllable. All consonants tend to become
releasing in utterance at rapid rate, and all consonants
in the arresting position tend to shift or
drop; there are no heavy stresses and no great variations
of length of syllable which would accommodate
the arresting consonants and give them definite
function.

The tremendous changes which have taken place
in the “sounds” of the French as it developed from
the vulgar Latin must have been due to something
beside the ordinary changes to which a language is
subjected. Side by side with the Latin, the Greek
also has had a continuous history of development
into a modern language; but the Greek presents no
such extensive changes. There can be little question
that a profound change in the rhythm underlies
the extensive alterations during the transition from
Latin to French. A new rhythm has produced a
new language, reshaping the ancient words, eliminating
syllables and shifting the stress to an alternation
from syllable to syllable. Gradually the French
tended to the open syllable, probably under the influence
of a rhythm which prefers rapid and even
syllables. The tendency to equalize the syllables,
and the inevitable rate of utterance, resulted in the
elimination of the arresting consonants (final and
“appuyantes”). Other consonants brought into
juxtaposition by the dropping of the syllables affected
each other profoundly. Such a change of
rhythm and of stress is apparent in the Hebrew
over against the Arabic, and in the Czech over
against the Lithuanian, and in the French over
against the Italian.

A language mechanism is an interrelated set of
movements in which the various traits have their
function; they are not isolated. The habit of describing
the “sounds” and the rhythms of a language
as if they were collections of independent peculiarities
overlooks the fact that they are part and parcel
of each other. And changes which occur affect the
entire language mechanism.

The changes occur at first as normal, synchronous
changes which revert with the restoration of
the rate and stress. Diachronous changes are such
changes which have become irreversible because of
perturbations in the language mechanism as a
whole. (89, p. 46, 97).125