m

WA'.V
••'';.. '>,fvw

M.
'$'
'sW<S

mm, >;,
^>

^i'\.
 Digitized by the Internet Archive
in 2011 with funding from
University of Toronto

http://www.archive.org/details/introductiontosOObloo
 AN INTRODUCTION
TO THE STUDY OF LANGUAGE

BY

LEONARD BLOOMFIELD
Ph. D., Assistant Professor of Comparative Philology and German
in the University of Illinois

NEW YORK
HENRY HOLT AND COMPANY
""s^'ihi
 TO

n. ^. ^.
 PEEFACE.
Tliis little book is intended, as tlie title implies, for
the general reader and for the student who is entering
upon linguistic work. Its purpose is the same, according-
ly, as that of Whitney's Language and
Study of Lan- the
guage and The Life and Groivth of Language, hooks which
fifty years ago represented the attainments of linguistic

science and, owing to their author's clearness of view

and conscientious discrimination between ascertained fact
and mere surmise, contain little to which we cannot to-
day subscribe. The great progress of our science in the
last half-century is, I believe, nevertheless sufficient ex-
cuse for my attempt to give a summary of what is now
known about language.
That the general reader needs such information as is

here given was recognized by Whitney, who wrote, in

the preface of his first-named book: 'It can hardly admit
of question that at least so much knowledge of the na-
ture, history, and classifications of language as is here
presented ought to be included in every scheme of higher
education.' While questions of a linguistic nature are
everywhere a frequent subject of discussion, it is surpris-
ing how little even educated people are in touch with
the scientific study of language. I hope that my book
will furnish a simple aid for those who choose to make
up this deficiency in our scheme of general education
Students whose vocation demands linguistic knowledge
are subjected in our universities to a detached course or

.66'
VI PEEFACE

two on phonologic and morphologic history

details of the
of such languages as Old English, Gothic, or Old French,
— details which are meaningless and soon forgotten, if
no instruction as to their concrete significance has pre-
ceded. To method of presentation is due, I think,
this
the dislike which so many workers in related fields bear
toward linguistic study. I hope that this essay may help
to introduce students of philosophy, psychology, ethnol-
ogy, philology, and other related subjects to a juster
acquaintance with matters of language.
In accordance with this twofold aim, I have limited
myself to a presentation of the accepted doctrine, not
even avoiding well-used standard examples. In a few
places I have spoken of views that cannot claim more
than probability, of hypotheses, and of problems yet to
be solved, but I have done this explicitly and only be-
cause I think it fitting to indicate the direction in which
our study is at present tending. Consequently the matter
here presented is by no means my own, but rather the
property of all students of language. It will be found in
fullerform and with bibliographic support in the books
mentioned in Chapter Ten, and these books I may there-
fore name as my more immediate sources.^) It will be
apparent, especially, that I depend for my psychology,
general and linguistic, entirely on Wundt; I can only
hope that I have not misrepresented his doctrine. The
day is past when students of mental sciences could draw
on their own fancy or on 'popular psychology' for their
views of mental occurrence. L. B.

Of Sweet's Primer of Phonetics the first, and of Meillet's

1)
Introduction the second edition was used in compilation, but
the later editions do not, I believe, differ materially as to any-
thing here discussed.
 CONTENTS.
Pag«
CHAPTER I

THE NATURE AND ORIGIN OF LANGUAGE.

1. Expressive movements 1
2. Gesture-language 4
3. Writing 7
4. Audible expressive movements 8
5. Development of language in the child 10
6. The origin of language IS
7. Language constantly changing 16
8. Social character of language ..<..•. 17

CHAPTER n.
THE PHYSICAL BASIS OF LANGUAGE.
1. Unconsciousness of speech-movements 18
2. "Writing an imperfect analysis 19
3. The vocal chords 2t
4. The velum 26
6. Oral articulation 27
6. Oral noise-articulations 28
7. Musical oral articulations 33
8. Infinite variety of possible sounds . , ... ... 38
9. Glides and mixtures of articulation 40
10. Syllables . . . - 41
11. Stress 43
12. Pitch 61
13. Duration ,65!
14. Limitation of articulations in each dialect 63
15. Automatic variatioas • . 64
vni CONTENTS
Page
CHAPTER m.
THE MENTIL BASIS OF LANGUAGE.
1. The place of language in our mental life ...,,, 66
2. Total experiences 56
3. The analysis of total experiences 59
4. The naming of objects 63
6. The development of abstract words 65
6. Psychologic composition of the word 66
7. Grammatical categories 67
8. Psychologic character of the linguistic forms ... ,69
9. Psychologic motives of utterance 70
10. Interpretation of the linguistic phenomena 71

CHAPTER IV.
THE FORMS OF LANGUAGE.
1. The inarticulate outcry 73
2. Primary interjections 73
3. Secondary interjections , 75
4. The arbitrary value of non-interjectional utterances . 77
5. The classifying nature of linguistic expression .... 82
6. Expression of the three types of utterance 90
7. The parts of utterances 92
8. The word: phonetic character 97
9. The word: semantic character 103
10. Word-classes 108
11. The sentence, , .. 110

CHAPTER Y
MORPHOLOGY.
1 The significance of morphologic phenomena .... 120
2. Morphologic classification by syntactic use (Parts of
speech) .... 120
3. Classification by congruence 127
4. Phonetic-semantic classes IHl
6. Classes on a partially phonetic basis 136
6. Difference between morphologic classification and non-
linguistic association 139
7. Classes by composition .,,.., 140
 CONTENTS IX
Page
8. Derivation and inflection 140
9. The semantic nature of inflection: the commonest cate-
gories 141
10. The semantic nature of derivation 150
11. The phonetic character of the morphologic processes 151 .

12. Word-composition: semantic value 159

13. Word-composition not a phonetic process 162
14. Simple word: compound: phrase. . • • • 165

CHAPTER VI.
SYNTAX.
1. The field of syntax 167
2. The discursive relations 168
3. The emotional relations 170
4. Material relations 171
5. Syntactic categories 174
6. The expression of syntactic relations: modulation in the

sentence 176
7. Cross-referring constructions 178
8. Congruence 180
9. Government 182
10. Word-order 186
11. Set phrases: the transition from syntax to style. . . . 188
12. The complex sentence 190

CHAPTER VII.
INTERNAL CHANGE IN LANGUAGE.
1. Language constantly changing . 195
2. Causes of the instability of language . 195
3. Change in articulation . . 202
4. Analogic change . . 221
5. Semantic change . . 237
6. The ultimate conditions of change in language
'G" . 261

CHAPTER VIII.
EXTERNAL CHANGE OF LANGUAGES.
1. Language never uniform . , 259
2. Increase of uniformity . , 262
 .

X CONTENTS
Page
3. Decrease of uniformity does not offset the increase. . , 263
4. Inferences from historic conditions 265
6. The process of differentiation 273
6. Deduction of internal history from related forms . , 274
7. Interaction of dialects and languages , 280
8. Standard langruasres •, 288

CHAPTER IX.
THE TEACHING OF LANGUAGES.
1. The purpose of foreign language instruction . • • . • 292
2. Character of the instruction , , 293
3. Age of the pupil , 295
4. Equipment of the teacher 297
5. Drill in pronunciation . 299
6. Method of presentirg semantic material . 300
7. Grammatical information 302
8. Texts 304
9. References 305

CHAPTER X.
THE STUDY OF LANGUAGE,
1. The origin of linguistic science • . . • ; 307
2. How to study linguistics . . ....... 313
3. Relation of linguistics to other scievices . . • ... 319

INDICES.
1 Authors, etc • . • 326
2. Languages , 327
3. Subjecta 331
 CHAPTER I.

THE NATURE AND ORIGIN OF LANGUAGE.

1. Expressiye movements. In tlie animal world every
mental process is accompanied by a corresponding phys-
ical process. Some of these physical processes are express-
ive movements. Investigation has shown that the express-
ive movements are most directly co-ordinated with the
emotional element that is present in every mental process.
In man as well as in the lower animals it is primarily
the intensity of the emotional element which appears in
the expressive movements. Everyday observation recog-
nizes the intensity of emotion of monkeys, dogs, or birds
and even of such distant forms as the ant or the fly. In
man and in the animals nearer to man a mild emotion
is accompanied on the physical side by a hurrying of

pulse-beat and respiration. If the emotion is more violent,

the expressive movements extend, successively, to the
facial muscles, then to the hands and arms, and finally
to the legs and feet, embracing a set of actions well
known to common observation. As the violence of the
emotion increases, these movements also grow more ener-
getic. When a certain extreme, however, is reached, the
mental turmoil suddenly ceases and, in exact correspond-
ence with this, there is a stopping of all the physical
manifestations: the muscles grow slack, the legs often
refusing support, and heart-beat and respiration may tem-
porarily or even permanetitly stop.
Bloom field, Stady of Language 1
2 THE NATURE AND ORIGIN OF LANGUAGE

While the expressive movements are thus chiefly de-

pendent on the intensity of emotion, some of them, espe-
cially in the monkey and in man, have come secondarily
to indicate also the quality of the emotion. The quality
of the emotion shows itself in the play of the facial
muscles. The various facial expressions are probably
mechanized forms of what were once instinctive efforts
at dealing with experiences of taste. The familiar 'sweet'
or pleasurable expression brings any substance that may
be in the mouth as much as possible into contact with
the tip of the tongue, which is most sensitive to sweet
tastes. Similarly, the 'bitter' or abhorrent expression
withdraws the back of the tongue, which is most sensi-
tive to a bitter taste. Sour tastes are most felt by the
sides of the tongue: a pleasantly sour taste can be best
perceived in the position which we know as a 'smile' and
an over-sour one best avoided by the 'weeping' grimace.
These responses have, in the history of the race, become
purely reflex and hereditary, appearing even in new-born
children.
Owing, moreover, to association between these move-
ments and the emotional qualities in these taste-experiences,
the movements have come to be constant attendants of
all experiences, even other than of taste, which involve

such qualities of emotion. That is, the 'sweet', 'bitter',

'smiling', and 'weeping' expressions are now the phys-
ical concomitants of any and all experiences whose emo-
tional quality resembles that, respectively, of a sweet,
bitter, sour, or over-sour taste. Thus any pleasure is ac-
companied by the first of these expressions and any
abhorrence by the second; the uses of the smile and of
the weeping grimace are too well known to need descrip-
tion. It is not known to what extent this associational
extension of these movements' is hereditary.
 EXPRESSIVE MOVEMENTS 3

Beside these expressions we — pleasant

find tension,
or unpleasant anticipation, — expressed by the inner-
vation of the cheek-muscles, and relaxation, — satiation
or disappointment, — by their loosening. Perhaps these
reflexes originated in the use of these muscles in eating.
Another specialized type of expressive movements are
those which indicate the perceptual content of an ex-
perience. In every experience there is present, beside the
emotional elements (with which the expressive movements,
we must suppose, are most directly connected), a series
of perceptual impressions, whether of outer sensation or
of imagery. In fact, it is only by an abstraction that we

can separate the emotional and the perceptual contents

of our mental life. Just as certain expressive movements
originally connected with experiences of taste have come
to indicate the emotional quality of an experience, so
certain other movements, especially of the hands and
arms, have come to indicate its perceptual content.
Such a movement is that of pointing at things. When
a child grasps at things which it cannot reach, its mis-
judgment of distance results, in each case, in a mere
movement of the hand in the direction of the object
desired. As the child grows in intelligence it performs
this movement even when it knows it cannot reach things,
and finally also uses the movement to indicate things which
it does not want, — things which merely excite its cu-
riosity or interest, the subjects of its discourse. This de-
velopment of the deictic expressive movement, which
occurs in every child, is peculiarly human; the monkey
does not get beyond the first stage of sometimes grasping
at things which it cannot reach.
Another type of expressive movement that indicates
perceptual content is the imitative movement. Imitation
is a term that can be applied to many phenomena of ex-
1*
4 THE NATURE AND ORIGIN OF LANGUAGE

pressive movement throughout the animal kingdom. When

we findnumbers of ants or bees, for instance, congruentlj
performing some one task, we must suppose that an in-
stinctive action of some individuals called forth the same
action in all the others. The explanation seems to be that
the bodily movements have become so closely associated
with the mental processes which they accompany, that
the sight of a fellow-individual going through the former
at once awakens the same mental state in the beholder.
Thus a child, seeing another child weep, enters at once
upon the state of anguish associated with this expression,
and consequently weeps in sympathy, as we say, with
the other child. In a grown civilized man these imitative
actions are, however, usually suppressed and even the
sympathetic emotion is reduced to a minimum. This sub-
jection of the imitative movements to the will allows
them become expressive of perceptual contents. For
to
we may now accompany any chosen perceptual element
of our mental state by imitative gestures, provided —
only that this element is sufficiently charged emotionally,
for, after all, these movements are at bottom indicative
of intensity of emotion. Especially in speaking of actions
we accompany our picturing with imitative gestures. Also,
anyone asked to define the qualities 'compact' or 'spiral'
will resort to imitative movements. The prevalence of
these varies greatly as a matter of communal habit or
good form among different nations.
2. Gestiire-lnnguage. Gestures are frequently used as
the means of communication where vocal speech is im-
possible or undesirable. The systems of gesture-language
thus used by different peoples are strikingly uniform.
The gesture-language of certain of the American Indians^
used where tribes of different language wished to com-
municate, is closely like that which has been current in
 GESTURE-LANGUAGE 5

southern Italy since Roman times (and no one knows

how much earlier), or like that used by the lower classes
in Japan, or by the Cistercian monks under their vow of
silence; and all these forms closely resemble that which
a company of untaught deaf-mutes wiU, in the course
of a few years, produce for themselves.
Gesture -language is so uniform because it consists
everywhere chiefly of the universally human expressive
movements voluntarily used for communication. The origin
of the communicative use is psychologically intelligible.
An individual sympathetically taking up another's emo-
tion might yet reproduce an entirely different perceptual
content. In so far as his expressive movements indicated
the latter they would differ from those of the first indi-
vidual. This already would be rudimentary communica-
tion. It would develope into more and more deliberate
and explicit forms as the race attained to voluntary use
of expressive movements for any chosen part of one's
ideas, and as individuals, after repeated occurrence of the
divergence of gesture, should foresee this divergence and
make gestures in order to call forth divergent gestures
from their fellow, —
in other words, as the exchange of
messages became a motive. We must suppose that all
this took place in connection with vocal language, but
even where gestures are used without vocal language
they remain close to their character of expressive move-
ments.
The deictic movement is of very limited use in gesture-
language. Objects which, under circumstances, may be
absent cease to be designated by pointing gestures even
when they are present. The deictic gesture thus comes
to be used only of certain constant relations: for express-
ing the T, the *you', the 'here' or Hhis', and the 'there'
or 'that*.
6 THE NATURE AND ORIGIN OF LANGUAGE

The imitative movements ^ on the other hand, receive

a wide development in the depicting gestures. These have
been divided into three classes. The simplest kind are
the represeyitative, v^hich depart butlittle from primitive

imitative movements, as, —

for instance, when ^joy' is
expressed by a glad grimace or sleep' by closing the
^

eyes and inclining the head to one side. Like all depicting
gestures, representative •i/^^tures are either graphic as
when one draws the outli^'i* of a 'house' in the air (gable-
roof and side walls), or plastic, as in the above gestures
for 'joy' and'sleep' or when one joins first finger and
thumb in the shape of a circle to indicate 'coin' or 'mon-
ey'. Suggestive gestures depict not the thing intended
but some part or accompaniment of it that brings it up
by association. Graphic examples are the outlining of a
beard under one's chin to express 'goat' or of a hat over
one's head to express, among the Indians, 'white man'.
The plastic type appears in the gesture for 'silence' in
which the lips are compressed and a finger raised or in
that for 'hunger' in which the cheeks are hollowed and
two*^ fingers, as if grasping a morsel, are held before the
open mouth. Syynbolic gestures, finally, arise when still
further associational processes have removed the gesture
from all resemblance to the thing intended or any part
of it. Thus the deictic gestures for space may be used
for time: one points backward for the past and forward
for the future, or, as a plastic example, the suggestive
gesture for 'hunger' may be used for 'wish' or 'desire',
or the suggestive gesture for a 'bad smell', raising of
the nostrils, may be used to express anything arousing
disgust.
The transition from the immediately significant gestu-
res, the deictic and the representative, to the suggestive
and the symbolic is a process of association. The gesture
 WRITING f

IS closely associated with a type of experience, and a

new experience with the same dominant features calls
forth the same gesture, without any consciousness of a
transference on the part of the speaker. We shall meet
similar inevitable transferences or rather extensions of
meaning when we speak of y-ocal language. In gesture-
language they are limited, however, by the immediate
and apparent connection or^e identity of most gestures
with the natural expressive ,w iction to the experience.
>

Because most gestures are so immediately intelligible a

gesture not immediately intelligible is but slowly adopted,
and the number of such never becomes very great. The
main stock of every system of gesture is made up of
such original forms as the deictic and the graphic re-
presentative gestures, which are practically identical with
natural expressive movements.
3. Writing, The expressive movements so far discuss-
ed have given rise not only to gesture-language but
also to writing. Picture-writing is originally the tracing
of an expressive movement on a permanent material.
Its close kinship with gesture results in the transference
of symbols from one to the other. We find not only
delineations of objects (such as a house) made with
exactly the same strokes as are used in representative
gesture, but even symbolic gestures are indicated in the
picture. Among the Indians a hand -movement upward
from the head means 'big man' or ^chief: in picture-
writing the same meaning is expressed by a line drawn
upward from the head of the figure. Similarly, we find
transference of pictorial symbols to gesture. The picto-
rial symbol for 'exchange' among the Indians consists of
two crossed lines, —significant either of the act of ex-
change itself or of the crossing of paths at which barter
between primitive communities usually takes place. In
8 THE NATURE AND ORIGIN OF LANGUAGE

t^esture-language this symbol is used in the form of two

crossed fingers.
The further development of writing takes place, as
we shall see, entirely under the influence of vocal lan-
guage.
4.Audible expressive movements. We have seen
how the expressive movements have developed in man
into a voluntarily used set of symbols by which even
abstract meanings may be communicated. The principal
development of expressive movements in this direction
did not, however, take place in connection with the noise-
less movements which we have so far considered. These
are in several respects under a disadvantage. It is per-
haps rash to say that they are not capable of sufficient
variation to be fuRy adequate to our needs; perhaps,
if vocal speech had been denied us, they would have
shown themselves modifiable enough to serve for com-
munication in all respects. There is no question, how-
ever, but that they are laborious and slow, demand-
ing a great amount of muscular action on a larcre scale
for even the briefest utterance. They appeal, moreover,
to the sense of sight, which is not so powerful an arouser
of the attention as hearing and must, indeed, be turned,
often by movement of the entire body, to receive an im-
pression from a new direction. Opposed to all this, the
sound-producing expressive movements are performed by
a delicate machinery requiring but little muscular effort

and appeal to the attention by a channel that is nearly

always open and requires no adjustment of the receiving
apparatus.
Expressive movements producing sound occur widely
in the animal kingdom. Such insects as crickets make
noise by rubbing together parts of their bony covering;
this type of audible expressive movement has nowhere
 AUDIBLE EXPRESSIVE MOVEMENTS 9

reached a high development. The more familiar type,

in which air expelled from the lungs meets with obsta-
cles in the breath-passage^ appears in amphibians, such as
the frog, and especially, of course, in birds and mammals.
The original form seems to be the cry of pain or rage.
Under a violent unpleasurable emotion the breathing-
apparatus and trachea are suddenly contracted. The
breath, hereby forcibly expelled, sets into vibration cer-
tain elastic protuberances within the breath-passage, the
vocal chords, and is further forced through the mouth
and nose. The result is a penetrating noise.
Such animals as the mouse and the rabbit utter sound
only under extreme emotion. The development from this
primitive outcry seems to occur in two directions. Among
gregarious animals the primitive outcry becomes an in-
stinctively used call for help or for the presence of a
fellow- individual. On the other hand, the cry of anger
of the fighting males at mating- time develops into a
general vocal expression of the emotions of this period.
By a further transition this vocal expression accompanies
any emotion, as in the male song-bird.
lively pleasant
The development in this direction brings it about that
the vocal utterance is used not only under extreme stress,
but also for lesser and for pleasurable emotions. Thus
there comes about a differentiation between the utterance
of highly unpleasant emotion on the one hand and that
of lesser pleasant or unpleasant feelings on the other.
The latter, less violent expressions tend to include some
repeated movement of the mouth or some periodic change
in the production of the voice-sound itself. No better
example of this differentiation could be found than the
squeak of a bird in extreme fright or pain and, under
less emotional stress, its regular song. The less violent
kind of utterance may be modulated predominantly as to
10 THE NATURE AND ORIGIN OF LANGUAGE

pitch or as to the noise-quality of the sound. Pitch-mod-

ulation is, of course, characteristic of the song of birds,
noise-modulation of the dog's bark or of human speech. In
our song we combine the two; it has been thought that
our unmelodious speech is a degeneration from an earlier
singing habit of expression, but extended research has
shown that this is not the case, human song having prob-
ably originated in the chant of rhythmic labor. The me-
lodious quality of the bird's song due to the position
is

of its vocal chords at the very bottom of the trachea,

which leaves a long sounding-tube for the pitch-modifi-
cation of the sound; our speech, on the other hand, re-
ceives its great scope of variation as to noise-character
from the extreme mobility of our tongue and other oral
muscles. The various movements of these were, no doubt,
in their origin, expressive movements like those of the
and 'sour' or 'tense' and 'relaxed' types.
'sweet', 'bitter',
The eflfect of the sound upon the producing individual
and his fellows was, however, so forceful, as opposed to
that of the mere movement and grimace, that the acoustic
impression of the sound and not the movement itself be-
came the basis for further associational development.
5. Development of language in the child. The
diiferent stages of vocal utterance appear very clearly
during the growth of a child. The new-born child shrieks
with wide-open mouth when in pain. By the end of the
first month it yells also under other sensations of dis-
comfort and soon afterwards it croons when it is con-
tented. As these less violent emotions are accompanied
by less muscular efi'ort, theie is already some
violent
diff'erentiation in the sound produced. Gradually modi-
fications of these less violent oral movements set in and
are furthered not only by the growing practice of the
mouth-muscles, but also by the appearance of the teeth,
 DEVELOPMENT OF LANGUAGE IN THE CHILD U
which makes articulation of tongue-tooth sounds possible.
Up to about the end of the first year the child performs
an increasing variety of articulatory movements, especial-
ly during pleasurable emotion. There can be no question
that the tendency to this form of expressive movement,
and especially to the great variety of these movements, is
inherited from the past generations of speaking ancestry.
The element of mimicry —
that is, of imitation of
the speech of the surrounding adults —
becomes more
and more prominent toward the end of the first year,
until the child finally succeeds in repeating, —
with no
consciousness of their meaning, to be sure, — syllables
and words that are spoken to it.
At about the same time the child begins to understand
gestures; that is, to associate people's gestures with emo-
tional and even perceptual experiences. It begins by
connecting facial expression with states of emotion, rec-
ognizing, as we say, an angry or a cheerful countenance.
Then comes the association of deictic gestures with ob-
jects, the child's eyes following the direction in which
one points. At last words begin to be understood: aided,
at first, by pointing gestures, the child begins to associate
such sound-sequences as the nursery words for 'mother',
'father', 'good', 'bad', *bed', or 'sleep' with the corre-
sponding experiences.
As yet, however, the child does not utter these sound-
sequences to express the experiences. When it uttersthem
at all, it does so purely in mimicry. Even in a normal
child the end of the second year may arrive before the
cross-association between the sounds which it imitatively
utters and the significant sounds which it understands
when others speak them, becomes lively enough to en-
able the child to repeat words with consciousness of their
significant value. When this cross-association has been
12 THE NATURE AND ORIGIN OF LANGUAGE

formed speech may be said to have begun. To be sure,

the child's reproduction of what it hears is for a long
time imperfect. It is no simple task to associate correctl}'
a sound heard with the articulatory movements that will
produce it, even though, in the case of some, such as the
lip-closure sounds ^, &, m, the eye aids the ear. The child
is very much in the position of the adult who hears a
foreign language; its perception is often wrong. Such
mistakes as the confusion of t and A', of/* and tli are due
to the unsureness of the perceptive habit: the child actu-
ally hears the wrong sound, so far as consciousness is con-
cerned. Only after long practice do hearing and articula-
tion become accurate and closely associated with each other.
The child's associating the sounds it hears with certain
experiences is due, of course, to the fact that grown-ups
are constantly producing the sounds in connection, and
in as plain connection as possible, with the proper ob-
and actions. The association, for instance, between
jects
mama and the child's mother is presented entirely by the
child's elders. In many cases the child wiU be led to
form a wrong association, which is gradually corrected,
as when it at first calls every man papa. In no case does
the child itself invent a word, in the sense of spontaneous-
ly giving meaning to a sound-sequence. Mother or nurse,
to be sure, wiU often connect some one of the child's
meaningless sound-productions with some person, object,
or other experience and then teach the child so to connect
it: it is in this way that our nursery-words have arisen.
They are sound- groups which are uttered by most chil-
dren and have come to be traditionally connected by the
adult speech-community with certain meanings; the child,
however, learns to give them these meanings just as it
learns the value of any other words. The connection be-
tween sound and sense is in no case originated by the child.
 THE ORIGIN OF LANGUAGE 13

While we thus see in the child the development of

sound-producing expressive movements from the unmodi-
fied yell of pain to the most manifold varieties of articu-
lation, differentiated in general character to correspond
to different emotional states, the spontaneous rise of the
use of certain fixed sounds for certain fixed types of ex-
perience does not occur in the child. The significant use
of sounds is, upon the
so to speak, prematurely forced
individual, who has no opportunity of arriving by his
own powers at the goal of actual language. How the
human species arrived at this significant use of sound-
utterance is by the development
therefore not explained
of the individual under normal circumstances. There are
some accounts, most famous among them that of Hero-
dotus [^Hidories, H, 2), of children who, for the purpose
of ascertaining thedevelopment of language,
original
were left to grow up without hearing anyone speak.
The experiment is really impossible, for, to be signifi-
cant, it would have to be made with a large group of
people left to themselves for generations and even cen-
turies, since the development of language in the race can
not have been other than gradual and communal.
6. The origin of language. The question remains,
then: How did man come to associate fixed sound-se-
quences with fixed types of experience? The older an-
swer to this question was based on the individual's learn-
ing of language. According to earlier theories the place
of the child's elders was with regard to the race,
filled,

by divine care: a divinity directly gave men the use of

speech. A more but essentially identical
materialistic
notion was that man himself invented the trick of attach-
ing significance to sounds; some genius of primitive times,
for instance, may have conceived this brilliant idea.
More tenable was the view that the speech-sounds were
14 THE NATURE AND ORIGIN OF LANGUAGE

originally imitations of what they denoted (Stoics, Herder),

or the view that they were originally the natural and
inevitable emotional responses to the corresponding ex-
periences (Epicureans, Rousseau).
The evolutionary point of view has shown the falsity
of the first two explanations and growing psychologic
insight has deprived the last two also of probability.
Gesture-language is in this connection especially instruct-
ive. Gesture -language, as we have seen, is nothing but
a higher development of the expressive movements com-
mon, in their basis, to many animals. Vocal language
is not essentially different. It consists, at bottom, of ex-
pressive movements. In the case of gesture-language the
expressive movements themselves remained the means of
communication; consequently the connection between a
gesture and the original expressive movement is nearly
always apparent, as when the deictic gesture is plainly
a weakened grasping movement and the depicting gestu-
res scarcely differ from natural imitative movements. In
the case of vocal speech, on the other hand, it was not
the movement itself that attracted attention and became
the starting-point for further development, but the sound
which the movement produced. This sound is an effect
which bears only in respect to emotional intensity any
distinct and recognizable relation to the experience
calling it forth. The ^sweet^ face-gesture, for instance,
accompanied by production of the voice- sound gives a
sound in no way directly related to the experience of
something sweet or otherwise pleasant. Now, so long as
the face-gesture remained in use, the importance of the
sound could always be secondary, the gesture actually
conveying the message. The sounds themselves were
neither directly significant of the experience, nor could
they, in any conceivable way, have been imitative of most
 THE ORIGIN OF LANGUAGE 16

experiences: only the movements which produced the

sounds were the expressive correspondents and, therefore,
the indications of the experiences. After the sound, how-
ever, had entered into association with the gesture (and,
thus, with the experience), it gradually usurped the more
important place, owing to the advantages already set
forth, and finally came into independent use, without the
gesture. This use of the sound alone opened the road
for unlimited transferences of meaning of the same kind
as those which produce symbolic gestures. In the case
of the latter the predominant direct connection between
an experience and a gesture, —
a connection obvious to
all and constantly refreshed, —
forbade too divergent a
development. In vocal speech, however, where direct
connection between experiences and sounds was never
felt, the further development by means of associational

shifts of meaniDg has been unlimited. The connection be-

tween sound and meaning, thus, which cannot even in its
origin have been a direct one, is further destroyed by the
freedom of transference due to the lack of any immediately
felt connection between experience and utterance, such as
prevents too free a development of symbolic gestures.
It is clear, therefore, that even if one could survey
the whole evolution of sound-producing expressive mo-
vements from the single cry of pain to which some ani-
mals are limited, up to the present speech of man, there
would be no point at which one could say: Here lan-
guage begins. Expressive movements are the physical phase
of mental processes: whatever the mental processes, the
expressive movements correspond to them. Man's mind
and his expressive activity have developed in indissol-
uble connection. In the animal world, as we know it,
the evolution of one phase without the other is inconceiv-
able. This, indeed, is why it is impossible to set up a
 —

16 TFIE NATURE AND ORIGIN OF LANGUAGE

language as opposed to ex-

strictly logical definition of
pressive movement in general. Language is the form of
expressive movement adequate to the mentality of man.
This mentality is defined no less than man's language
in the aphorism that 'Man is a spealdng animal'.
7, Language constantly changing. The absence of
immediate connection between sound and experience
appears in the fact that, unlike gesture-language, vocal
language differs vastly in different times and places, •

a fact too familiar to need exposition.

The change of language in time is of interest in the
present connection because its phases again illustrate the
absence of any conservative relation between sound and
sense. The sounds habitually uttered under a given type
of experience are in an unceasing process of change:
those which we utter today are not like those which
speakers of English uttered a thousand or even a hundred
years ago. On the other hand, the transference of mean-
ing also is unlimited; the history of languages shows us
innumerable associational changes of meaning?, which in
gesture, where some connection between expression and
experience is upheld, would be impossible. It would be
difficult to find an English word which, if it existed at
all a thousand years ago, has not since then in some
way changed its meaning. All this is due to the fact that
there never was a stag^e in which a hearer could recoj^nize
any but an arbitrary connection between sound and sense.
The chancre of lamo-ua^e is not a mere endless shiftinc^
of sounds and meanings: we find speech rising in the
course of time to the power of more delicate and abstract
expression and to greater brevity. This development is
due to the assimilating effect, which we shall study in
detail, of experience upon expression; in return the grow-
ing power of expression, as we shall see, reacts favor-
 SOCIAL CHARACTER OF LANGUAGE 17

ably upon the mental processes. Thus the freedom in

which vocal language differs from that of gesture has
made possible a much higher development.
8. Socialcharacter of language. We have seen that
the greatest stimulus toward the development of express-
ive actions is their emergence into voluntary communi-
cative use. Language has been developed in the inter-
change of messages^ and every individual who has learned
to use language has learned it through such interchange
The individual's language, consequently, is not his cre-
ation, but consists of habits adopted in his expressive
intercourse with other members of the community. The
result of this is the individual's inability to use language
except in the form in which the community as a whole
uses it: he must speak as the others do, or he will not

be understood. As a matter of fact he does not, in nor-

mal cases, try to speak otherwise, but unquestioningly
follows his and his fellow-speakers' habits. The change
which occurs in language is thus never a conscious alter-
ation by individuals, but an unconscious, gradual change
in the habits of the entire community. The motives
which cause it are not individual reflective considerations
of the result, but new associative tendencies or new con-
ditions of innervation due to some change in the cir-
cumstances of life affecting the community. As we
examine more closely the different aspects of language,
we shall again and again find the same characteristic: as
the individual speaker receives his habits from the com-
munity, individual motives do not come into play, but only
causes aff'ecting the community as a whole. And as, more-
over, the individual, from childhood, practises his speech
until the details of itare mechanized and unconscious, he
is rarely aware of the such as the
specific characteristics,
phonetic or the grammatical, which are involved in it.
Bloom field, Study of Language 2
 CHAPTER 11.

THE PHYSICAL BASIS OF LANGUAGE.

1. Unconsciousness of speech-movements. The in-
dividual's unconsciousness of the details of his speech-
activity appears strikingly when we inquire into the
movements by which speech-sounds are produced. While
we know we speak with the mouth, tongue, and
that
larynx, the separate movements of these organs rarely
or never enter our consciousness. If we are asked to
describe them, we answer in vague, metaphoric expres-
sions or say things that are altogether wrong. In fact,
as to some of these movements not only the normal
speaker but even the scientific observer is at a loss. For,
in spite of the fact that all these muscles are ultimately
at the command of the will, the innervations which con-
trol them have become mechanized; we consciously give
the impulse for whole words and phrases, but the details
of their utterance always proceed unconsciously. The
impulse, moreover, given in terms of sound, for, in
is

the association of articulatory movements with sounds,

which is formed very early in life (p. 11) and is, of
course, constantly practised, the latter are entirely domi-
nant, the former almost forgotten. It appears, then, that
even as regards our own speech -movements of every
day, some scientific examination of the facts is necessary.
It happens, moreover, that not only diff'erent languages
but even different local variations of the same language
 WRITING AN IMPERFECT ANALYSIS 19

use different sounds. When the normal speaker hears a

foreign dialect or language, he encounters a twofold
difficulty.His perceptive habits lead him to hear sounds
that merely resemble those of his own speech as if
they were identical with the latter; and where two or
more of the strange sounds resemble one of his own, he
may fail to distinguish between them. Thus a German
who is picking up English will confuse our v, w, and wh
sounds, our d and th (as in then), our t and th (as in
think), and our sh and ^ (as in azure), for in his own
language he has but one sound resembling each of these
groups. The second difficulty lies in producing the for-
eign sounds even when their distinctive character is
heard: thus our German may in time come to appreciate
the distinctions we have mentioned, but will still be un-
able to produce the English sounds.
These difficulties usually prove fatal to the efforts of
those who try to describe languages without adequate
knowledge of phonetics. From nearly all the published
material about American Indian languages, for instance,
it is impossible to get any adequate conception of how

these languages are pronounced. So great a Chinese

scholar as Joseph Edkins was unable to describe some
of the commonest Chinese sounds. It is for this reason
that even teachers who have spoken a language from
childhood are often unable to impart their information
to others. No one can teach a foreio;ner his language,
unless he can tell his pupil exactly what to do with his
vocal organs to get the proper effect: and this, we have
seen, he cannot dp without a certain amount of scientific
study.
2. Writing an imperfect analysis. There is one
activity in the course ofwhich nearly all civilized peo-
ples have made some analysis of the sounds of their
20 THE PHYSICAL BASIS OF LANGUAGE

speech, and that is writing. This analysis has, however,

been gradual and incomplete. In its most primitive form
wiiting is simply the drawing, carving, or painting of
the visible features of an experience or of symbolic ele-
ments representing it (p. 7). When this method of
communication is frequently used, certain elements in
the pictures come to be drawn always in a certain way
and to have a fixed meaning. Gradually such elements
may come to be used as symbols for corresponding words
of the vocal language and to be arranged in the order
that these words have in speech. As the association be-
tween written symbol and spoken word becomes fixed,
the symbol may come to be drawn without reference to
its original pictorial value, and to deviate from its older

form, associating the word rather than, in a more direct

sense, the experience. When this has happened, the asso-
ciation may grow to be simply one of written symbol
and sound, regardless of the meaning borne by the sound,
until, after a time, the symbols are used purely in their
phonetic value. The number of symbols may then be
lessened to the point where there is a single character
for every syllable used in the language. Such 'syllabaries'
are a very common form of writing; examples are the
alphabets of India (derived from ancient syllabic forms
of Semitic writing), and the national alphabets of the
Japanese. It is a further simplification when these
characters come to be used not for whole syllables but
for single sounds of the language, as in the Greek, Latin,
and derived alphabets, including our own.
All this development is, of course, gradual. There is,
in most instances, at no time a deliberate and system-
atic examination of the sounds of the language and an
assigning to each of a written symbol. Accordingly, we
hardly ever find perfect consistency in the relation be-
 WRITING AN IMPERFECT ANALYSIS 21

tween sound and writing. There are two factors which

lessen even such consistency as might otherwise develop.
One of these is the use of foreign alphabets. When the
English, for instance, took over the Latin alphabet, the
sounds of English were so different from those of Latin
that consistency was impossible, —
a difficulty under
which we labor even today, for our alphabet has not
enough signs for our vowels, and none for our ^/j-sounds,
our shj our 5 as in azure, our wJi, or our ng as in sing,
and, on the other hand, contains the superfluous char-
acters c, q, and x. The second factor interferes even
more seriously with the regularity of alphabetic writing:
it is the necessary conservatism of orthography. Read-
ing and writing would be very slow processes, if, every
time we read or wrote, we actually stopped to analyze
each word into its component sounds; moreover, accord-
ing to emphasis, speed, personal habit, and so on, the
spelling of each word would then be variable, a con- —
dition which would further militate against ease. Such
a state of affairs never continues long, for the spellings
of whole words are of course remembered and become
traditional. Opposed to this necessary conservatism of
writinor there is the fact that all lancpuaoje at all times
is in an unceasing process of change, —
a process so
I gradual and subtle that no speaker, through all his life,
J,
V I
isaware of it, yet so unceasing that the orthography of
every language becomes in a few hundred years thor-
oughly antiquated even in those features which were for-
merly consistent.
This, of course, is a reason why writing, though in-
volving to a certain extent an analysis of the physical
phase of language, does not satisfy scientific requirements
in this direction. Indeed, so far as the linguistically
untrained person is concerned, writing is often mislead-
22 THE PHYSICAL BASIS OF LANGUAGE

ing, for the individual movements of writing are so much

more consciously performed than those of pronunciation,
that the naive speaker will often think that he speaks
as he writes, when this is not the case. He will think, for
instance, that passed and past or close (verb) and clothes
are pronounced differently, when actually he may never
in his life have heard or made such distinctions.
There are other reasons, too, why writing cannot and
need not accurately analyze the spoken sounds. Although
the human vocal organ can produce an infinite number
of different sounds, each language uses but a limited set.
Given, therefore, an alphabet of a limited number of
symbols, it could be used by all languages, though no
two of them would give each symbol the same value.
Now, within limits this is actually the case: thus letters
like p and t by both English and French, but
are used
with different values, v and by both English and Ger-
mans, but again with differing values in the two lan-
guages. This circumstance may be convenient, on oc-
casion, to printers; it would be absurd, at any rate, for
us to request the Germans and the French to give up
their use of these letters because it does not agree with
ours. Consequently there are differences between the
pronunciations of different languages which do not appear
in writing. The same is true, moreover, of the different
local variations of the same language. The words of
the English language are pronounced very differently,
let- us say, by a Chicagoan and by a Londoner. These
dialectal differences of pronunciation may
be so great
that scarcely a word wiU be pronounced alike over aU
the territory in which a language is spoken. In the case
of Chinese, in fact, distant dialects are mutually unintelli-
gible, though the writing is the same. It would obvious-
ly be a great inconvenience and a source of much con-
 WRITING AN IMPERFECT ANALYSIS 23

fusion, if sucli variations appeared in tlie writing: it

would mean, Chicagoan could only

for instance, that a
with difficulty read a book printed in London. X^ius we
see that much of the value of writing is actually depend-
ent on its not conveying the exact manner of pronuncia-
tion.
More than this, there are in the language even of one
and the same person many subtle and complex variations,
which do not demand notation for the practical purposes
of reading and writing. Thus we pronounce our vowels
longer before d than before ty —
the o in rode longer,
for instance, than that in wrote, —
but it would be super-
fluous to indicate this difference, for every English-speak-
ing person regularly and unconsciously speaks his vowels
longer before d than before t. An orthography which
actually indicated all the phonetic facts of speech would
be a very cumbersome even an expert
affair, difficult for

phonetician to handle, and requiring, above all, close

attention to every single utterance that one wanted to
represent in writing.
It is obvious, then, that even a regular and consistent
orthography for practical purposes would not contain a
full analysis of the pronunciation of a language, such as
is often needed by the scientific investigator and, in some

degree, by the teacher of languages. For scientific use

several such fully analytic alphabets have been devised;
today the standard one is that of the International Pho-
netic Association, which ^aU be used in this book (pho-
netic cbaraciiers being printed in square brackets). It is cus-
tomary, however, even in scientific discussions, to avoid a
constant complete analysis by describing, at the outset,
the sounds and regular variations of a language and as-
signing a simple character of the phonetic alphabet to
each typical sound. Such a simplified phonetic alphabet
 24 THE PHYSICAL BASIS OF LANGUAGE

is of course best for teaching the pronunciation of a for-

eign language, and, if it can be made to fit all the local
variations of pronunciation, would be the ideal practical
alphabet.
3. The
vocal chords. The human vocal organ is a
wind instrument which produces sounds by interfering
with the breathed air that is being driven from the lungs
in expiration. The first interference which the expired
breath meets is at the head of the trachea, in the larynx
or Adams's apple. Within the larynx, to the right and
left, are two muscular protuberances, the vocal chords,

between which the breath must pass. In ordinary breath-

ing the muscles of the vocal chords afe relaxed and the
breathed air passes freely through the aperture between
them, which is called the glottis. When one holds one's
breath with open mouth the vocal chords are stretched
so as to close the glottis firmly. Owing to their delicate
musculature, and to two movable cartilaginous hinges,
the arytenoids in which they terminate at the rear of
J

the larynx, the vocal chords can be set also in a number

of positions intermediate between that of breathing and
,that of firm closure.
Firm closure of the suddenly opened, occurs
glottis,
just before coughing or clearing the throat, also under
any strain, as in lifting a heavy weight. As a speech-
sound it is used in German initially in the pronunciation
of words that in writing begin with a vowel. The sound
\ sa produced is caUed the glottal stop, and its phonetic
symbol is p]; a German word like arm ^poor' is therefore
pronounced parm]. The glottal stop occurs also in a
great many other languages, such as Danish, where hiind
*dog' is pronounced [hu'^n], \>\xi Jiun 'she' [hun], Lettish,
Hebrew ('aleph'), Arabic ('hamza'), and some Chinese
dialects. Its frequent occurrence in such languages as
 THE VOCAL CHORDS 25

(^"Danisli produces in English ears the effect of constant

interruption by little hiccoughs.
If the vocal chords are a little less firmly closed, the
comprGSsion gives way, from instant to instant, to the
pressure of the breath, so that a vibration productive of
musical sound results. This musical sound we call voice.
The pitch of the voice is modulated by changing the length
of the chords, for this of course controls the rapidity of
vibration. The loudness or depends on
stress of the voice
the violence of the vibration, and may therefore be regu-
lated in two ways. In singing the regulation is (or ought
to be) chiefly effected by varying the breath-pressure, that
is, by expiring more or less rapidly; in ordinary speech

the less cumbersome method prevails of slightly widening

the glottis for a less loud sound and slightly narrowing
it for a louder; for, as the narrowing of the glottis allows

less breath to pass through, the accumulated breath under-

neath exercises pressure, against which the vocal chords
vibrate under tension, producing a loud sound.
The voice is not heard in every sound of speech. In
the glottal stop, for instance, it obviously is absent. Many
of the other speech-sounds, also, are unaccompanied by
the voice. If one places a finger on the Adam's apple or
stops up one's ears, the voice will be felt as a buzz or
trembling; if one now speaks, such sounds as p, t, Jc, /", s
[p, t, k, f, s] will be found to lack this buzzing accompa-
niment, while such as h, dy g, v, z [b, d, g, v, z] have it: the
former are unvoiced or hreathed, the latter voiced sounds,
as are also, for instance, our accented vowels.
If the vocal chords are so far separated that the voice
no longer sounds pure, but is accompanied by a friction
sound produced by the breath as it passes through the
glottis, we get a murmur. Most of our unaccented vowels
in English are spoken with murmur instead of voice. As
2Q THE PHYSICAL BASIS OF LANGUAGE

an independent speech-sound the murmur is heard in the

Voiced h^ of Cech and of Sanskrit, symbol [fi]. If the
glottis is still farther opened, the voice ceases and only
the friction- sound remains: this is the sound of our h [h].
Still another position of the vocal chords is represent-
ed by the whisper y in which only the cartilage-glottis,
that is, the space between the arytenoids, is open, the vo-
cal chords themselves being in contact. In what we ordi-
narily call whispering the whisper is substituted for the
voice, the unvoiced sounds remaining unaltered.
Both in whispering and in ordinary speech the unvoi-
ced sounds are pronounced with the glottis in its widest-
open position, the muscles of the vocal chords being re-
laxed and the breath passing freely through the larynx:
this, as we have seen, is also the position for regular
breathing.
The remarkable delicacy and rapidity of adjustment of
the vocal chords in passing from voice to breathing, from
either of these to murmur, whisper, or and in changing
Ji^

the pitch and stress, requires no further comment. It is

to be remembered, of course, that the details of aU these
movements, in spite of complete subjection to the will,
are so mechanized as to be unconscious anybody can speak
:

an hy but it takes careful scientific observation to deter-

mine exactly how the sound is produced.
4:. The velum. When the breath leaves the larynx it

passes, in normal breathing, through the nose. During

most of the sounds of speech it is, howeveT*, precluded
from doing so by the raising of the soft palate or velum,
which now cuts off the nasal passage from the throat and
mouth. If one stands with open mouth before a mirror,
breathing through both nose and mouth, and then sud-
denly pronounces a pure, long ^ah . .
.'
[ai], the raising of
the velum can be easily seen, especially if one watches
 ORAL ARTICULATION 27

the uvula. Most speech- sounds are thus purely oral. In

a few, such as m [m] or n [n], however, the breath es-
capes entirely through the nose, the velum being lowered:
such sounds are called nasals. There are other sounds in
which the breath escapes through both mouth and nose:
these are called nasalized (symbol "), e. g. the vowel in
the French cent [sa] ^a hundred'. Most speakers are, of
course, quite unconscious of the movements of their velum;
yet it is lowered and raised again every time they speak
an m or n.
5, Oral articulation. The mouth performs a double
function in speech. It serves, in the first place, as a re-
sonance-chamber for the musical sound of the voice or
for the whisper. By changing the shape of this resonance-
chamber we vary the tone-color of the sound: thus by
narrowing and flattening it we get the high tone-color of
the vowel-sound in fee, by hollowing it, the low tone-col-
or of the vowel in foe.
Secondly, by moving the tongue and the lower lip dur-
ing the passage of the breath, we can produce noises.
Most of these depend on the resistance of the breath- stream,
but noises can also be produced by suction (symbol [*]),
as in the sound with which we urge on a horse by
'snapping' the tongue against the palate [c*]. Such suc-
tion-noises occur as regular speech-sounds in the langua-
ges of the African Bushmen and the Hottentots. Where
the noises are produced by means of the breath, voiced
or unvoiced, there are two principal methods: either a
complete closure is made and then explosively burst, as
in our p, h, t, d, k^ g [p, b, t, d, k, g] —
stops, or explosives^
,

or the closure is incomplete and the noise is produced by

the friction of the breath passing through the aperture,
as in our v, tJi as in tJiink, th as in tlten, s, 0, sh, z as in
/",

azvire [f, v, 0, 5, s, z,
J, 3], — spirants or fricatives. Both
 .

28 THE PHYSICAL BASIS OF LANGUAGE

stops and spirants may be modified by lowering the ve-

lum; in the case of the former the breath escapes entirely
through the nose and we hear the nasals, such as m, n,
ng [m, n, g]; in the case of the spirants it escapes through
both mouth and nose, producing nasalized spirants.
6. Oral noise-articulations. The noise-articulations
can be produced in various parts of the mouth.
a) Labials. Stops produced by closure of the two lips, —
bilabial articulation, — are our unvoiced p [p] and voiced
b [b]. The corresponding nasal is our voiced m [m]. Bila-
bial spirants are not common; a voiced one [x)\ occurs in
Dutch (written w) and in Spanish (written b, v) i
Our English unvoiced [f] and voiced [v] are labiodental
spirants, in which the friction is produced between the
lower and the upper teeth and accentuated by the col-
lip
lision of the escaping stream of breath with the upper lip.
b) Dentals. Most of the oral noise-articulations are made
with theTongue. The tongae produces noises with either
the tip or the back articulating against the teeth or the
palate: articulation with the tip is called coronal, with the
back, dorsal.
Coronal articulation against the upper teeth or the gums
just behind them is called dental; it produces, of stops,
the unvoiced [t] and the voiced [d]. These occur in sev-
eral varieties, such as the interdental, against the lower
edge of the upper teeth, the post-dental, against the back
of the upper teeth (thus in Spanish and in many modern
languages of India, and, in a different variety, in French),
against the border of the upper teeth and gums (so in
German), or a little farther back still (as in the English
t and d), —
the last two variants being specifically called
alveolar. Such variations are indicated, where necessary,
in phonetic writing by diacritical marks such as [h] for
articulation with the tongue drawn back, [h] for articu-
 ORAL NOISE ARTICULATIONS 29

lation with the tongue advanced, [j.] for greater raising

of the tongue, and [t] for greater lowering; but these signs
can usually be dispensed with by stating beforehand what
varieties are current in a given language. The voiced na-
sal, corresponding to these stops is our n [n], which often
occurs unvoiced [n] in such words as mird, snoiv, where
it is spoken just before or after unvoiced sounds. Dental
spirants, more specifically interdental or post-dental, are
our unvoiced [0] as in iliinh and voiced [5] as in then.
Dental articulation is used also in the trills or r-sounds
of most languages. These sounds are produced by tighten-
ing the tongue-muscles so that they elastically resist
the pressure of the breath from instant to instant; an
example is the Slavic or Italian 'rolled' r [r], which is

u-ed also in the stage-pronunciations of French and Ger-

man. The r-sound of American English [.i] is pronounced
with the tongue relaxed, so that there is no trilling and
even very little breath- friction; in consequence the acous-
tic value of the sound is as much musical as noise-like.
An unvoiced with increased friction often occurs in
[.j]

such words as try. The friction element of a trilled [r]

reaches a maximum, if the tongue is held close to the roof
of the mouth, especially at the sides, where it touches
the upper teeth; if the friction-noise is very great, we seem
to hear a trilled [r] and, simultaneously, a spirant resem-
bling the sound of z in azure: this strongly spirant trilled
[i*] is heard in Cecil ish.
Another dental articulation is that of the ^-sounds or
latera ls. In these also friction is so slight that it would
be as well to class them with the musical sounds as with
the noises. Their characteristic resonance is due to the
fact that the breath escapes at the sides of the tongue,
the tip of the tongue being pressed tightly against the
upper teeth or guns. The tone-color of such an [1] can
 30 THE PHYSICAL BASIS OF LAN(jfUAGE

be varied by raising or lowering the back of the tongue, —

that is, altering the shape of the resonance-chamber. Very
high tone-color, due to raising of the tongue, is heard in
the 'light' I of the Slavic languages; less high is that of
German or French I, while that of English is especially
dull, owing to the lowering of the middle of the tongue.
^ c) Cerehnils. Leaving the dental position, we come to
another form of coronal articulation, the cerebral. In this
the tip of the tongue is drawn up and back, so as to articu-
late against the highest point of the palate. Many lan-
^ guages of India possess these cerebrals [t, d, n] by the
! side of the dentals, distinguishing between the two as sharp-
r-^^ *
ly as they or we should distinguish between, say, t and k»
Some of these languages have cerebral [j] and [1] which
may also be heard in the English pronunciation of many
Americans.
d) Blade sounds. We come now to the dorsal tongue-
articulations, inwhich parts of the upper surface of the
tongue (as opposed to the tip in coronal articulation) are
brought into contact with the teeth, gums, or palate. The
dorsal articulations that are made farthest forward are
produced, naturally, by the front of the upper surface of
the tongue, which is called the hlade. In these sounds the
tongue is contracted so as to form a furrow along the
median line: the breath passes along this furrow, which
directs it against the edge of the upper front teeth. Here
the narrow, strong stream of air produces a sharp, hissing
noise,whence these sounds receive the name of sibilants:
unvoiced [s], voiced [z]. They occur in several varieties: in.
French they are post-dental, the tip of the tongue touch-
ing the lower teeth, and the blade, except in the center,
where the furrow is formed, touching the upper teeth.
The English and German sibilants are alveolar: in Swedish
there is even a cerebral variety.
 ORAL NOISE-ARTICULATIONS 81

The distinctness of the sibilant hiss is lessened, if the

tongue is moved so as to displace the furrow from its
proper relation to the upper teeth, for then the narrow
stream of breath is not accurately directed against the
edge of the teeth, but, instead, eddies round, producing
a peculiar mufiled hiss. These al^m^rmal-sibikmts are usual-
ly produced by drawing the tongue back from the [s]-
position: so in our English unvoiced [JJ, as in shall and
voiced [3], as in azure ^ vision. The exact nature of the
eddying current of the breath in these sounds is not known.
The ^kettle' or ^gorge' of the eddy can be enlarged by
protruding and rounding the lips [J)], as in the German
sc/j-sound. The French varieties lower the tip of the ton-
gue and slightly raise the back.
The front articulations may be formed in more pronoun-
cedly dorsal variations. Such are our [t/h] and [d^n] as
in cheap and jump, the 'palatalized' Russian [t^], [di-],
and [sh], the Russian and Polish [tjt-], the Polish 'pala-
talized' s [/»-], the Norwegian [/h], and the German [Jh]
before consonants. These more dorsal varieties of dental
and blade sounds are often conveniently indicated, both
in practical and in phonetic writing, by placing an accent-
mark over the letter, e. g. f [f], rather than by fully in-
dicating the tongue-position, e. g. [t»-x]. It is also some-
times convenient to express them by the signs of the se-
ries of sounds to be next spoken of, provided the condi-
tions of the language do not make such expression mis-
leading or ambicruous.
Q^J^alaigls. Dorsal ai^ticulations against the hard pal-
ate are called palatal. As the hard palate is comparati-
vely extensive, they occur in several varieties. The stops,
unvoiced and voiced [j] are heard in French dialects,
[c] ,

in Lithuanian, and in Hungarian, the nasal [ji] also in

Spanish (written w), Italian (written gn)^ and French {gn),
32 THE PHYSICAL BASIS OF LANGUAGE

the French variety being pronounced farther back than

the others. The spirants of this position are unvoiced [9],
as in the German ich [^i9] 'I' and voiced [j], as in many
German pronunciations of such words as ja ^yes' and legen
'to lay'. Palatal trills cannot occur, for the back of the
tongue has not enough elasticity to vibrate. A palatal lat-
eral [X] occurs, however, in southern French, Spanish
(written 11), and Italian (written gl). Those who have not
in their native language the habit of palatal articulation
best learn it, if they produce the sounds with the tip of
the tongue pressed against the lower teeth, but this is not
necessary to the articulation.
f) Dorsal articulations against the soft palate
Velars.
again allow of a great deal of variation, owing to the ex-
tent of this region. The sounds here produced are called
velars.In English the velar stops, Jc [k] and g [g], are
produced farthest forward before the i-vowel heard in Mn
and back in can and gap, and farthest back
give^ farther
before back vowels as in coopj goosey the same habit pre-
vails in German. The velar nasal occurs in English, written
ngy as in singy symbol [g]^). The spirants are: unvoiced
[x], as in the German Bach [bax] 'brook', and voiced [g],
which occurs in modern Greek and in many German pro-
nunciations of such words as sagen 'to say'. A very open
[x-r], with little friction, is heard in the Slavic languages.

A velar trill is impossible, but a velar lateral [i] is pro-

duced in Polish by raising the back of the tongue: while
accurate median contact with lateral opening is here im-
possible, the general effect still resembles that of the ton-
gue-tip [1], as well as that of an English iv.

g) Uvular^. The hindmost of the dorsal articulations

1) In such words as finger, however, the spelling ng repre*

sents two sounds, [r)g].
 MUSICAL OKAL ARTICULATIONS 33

is that of the rear upper surface of the tongue against

the uvula, the pendent part at the back of the soft
little

palate. Of these uvular articulations the unvoiced stop [q]

occurs in Arabic and in Greenlandish; the latter language
uses also the voiced nasal [n]. In the trill [r] it is the
uvula and not, as in the dental trill, the tongue, which
vibrates. This uvular trill is the regular r-sound in Northum-
brian English (the 'burr'), in Danish, and in the city
pronunciations of French and German. In French and
Danish it occurs also unvoiced [ii]; in these languages it
is often pronounced without the trill-vibration, as a uvu-

lar spirant, both unvoiced [a] and voiced [k].

In connection with the oral noise-articulations we may
again mention the produced by the vocal chords;
larijingeal,

of these the stop p] anTlhe spirants, unvoiced [h] and

voiced [P)] have already been mentioned (p. 24, ff.). Two more
laryngeal spirants can be produced by compression of the
entire musculature of the larynx: the 'hoarse Ti* [h] and
its voiced form, the 'ayin' [q] of the Semitic languages.
7.Musical oral articulations. We
may turn now to
the musical articulations or 'vowels'. It is important to
observe that there is no definite boundary between the
noise-articulations and the musical articulations. Any spi-
rant can be articulated with varying degrees of closure:
as the pressure of the tongue is relaxed the friction-noise
decreases and the element of musical resonance becomes
more and more audible. Such spirants as the American
English [j], the laterals, and an open [j] are on the bor-
der line; if anything, the resonance- element is, in the [j] and
[1] at least, dorninant The traditional division of sounds
into 'consonants' and 'vowels', while often convenient,
is therefore untenable for purposes of exact terminology.
Instead, the sounds of speech represent an unbroken seiies
of relations between noise and resonance: the latter ele-
Bloomfield, Study cf Language 8
34 THE PHYSICAL BASIS OF LANGUAGE

ment is at a minimum in tlie unvoiced stops; then come,

ill order, the voiced stops, the unvoiced and the voiced spi-
rants, the nasals, the laterals, the r-sounds, and, finally,
the most open musical sounds, in the production of which
the mouth is merely shaped into a resonance-chamber.
In this shaping the chief factor is the tongue-position.
It iscustomary to distinguish nine typical tongue-positions,
three along the horizontal plane, front, mixed, and hack,
and three along the vertical, high, mid, and low. The for-
mer three indicate the region in which the tongue approach-
es most closely to the roof of the mouth, the latter
three, the degree of approximation. Other factors modi-
fying the quality of the resonance are the tensity or re-
laxation of the oral muscles, especially those of the ton-
gue, and the position, normal, drawn back, or rounded,
of the lips. customary to distinguish two typical
It is
states of each of these factors: ivide (that is, loose)and
narroiv (that is, tense) vowels, and rounded (lips protru-
ded and rounded) and unrounded vowels.
a) Front vowels. A high front vowel, narrow and un-
rounded, is produced, if one pronounces the spirant [j]
more and more openly, so that the friction-sound disap-
pears. This vowel [i] occurs in a very characteristic form
in French, where the corners of the mouth are drawn back
to emphasize the shape of the resonance-chamber: this
is the regular French i, as in fini 'done'. In German the
lips are not so far drawn back; the sound so produced
is the German Jong i-vowel, spelled ie or ih. In English
it is the initial sound of such words as year, yes.

The corresponding articulation with muscles relaxed

produces a very different acoustic effect, for the resonance-
chamber in a high vowel is so narrow that even the slight
increase in width produced by the relaxation of the ton-
gue-muscles is a relatively large change. This wide high
 MUSICAL ORAL ARTICULATIONS 35

unrounded front vowel [i] is the German short t, as in

bin 'am' and, in slightly lower position, the English short i,

as in bin.
If, while pronouncing one strongly rounds the lips,
[i],

the result is the high front narrow rounded vowel [y],

as in the French lune 'moon'. Decidedly lower, but still
of the same type is the German long w, as in JciiJin 'bold*.
The rounded wide vowel of this position [y], i. e. a —
rounded [i], — appears, in a lowered variety, in the Ger-
man short w, as in Hiltte 'hut'.
If the tongue is lowered to mid-position from these vow-
els, the narrow unrounded vowel is [e]. This vowel oc-
curs in German, as in geht 'goes', and in French, as in ete

'summer'.
The corresponding wide vowel [e] does not differ from
[e] so characteristically as does [i] from [i], for, what with
the greater width of the resonance-channel, the width add-
ed by the loosening of the tongue-muscles is here not
so apparent. The [e] occurs in standard German and (slight-
ly lower) in American^) English as the regular short e-

vowel, as in the English men, get.

The rounded form of [e] has usually less lip-rounding
than that of but a form with as great lip-rounding is
[i],

conceivable, since this factor is in no wise bound to that

of tongue-position, but can vary freely. The typical mid
front narrow rounded vowel [0] is the French vowel in
such words as peu 'little', jeune 'young'. A lowered varie-
ty is the German long o-vowel, as in schon 'beautiful'.
The wide form of this vowel occurs, again in a lower-
ed variety, as the short German o-sound, e. g. in Gotter
'gods'.

1) By 'American English' I mean my own Chicago pronan-

ciation, common generally to the North Central States.
3»
36 THE PHYSICAL BASIS OF LANGUAGE

Tlie low front position, which is reached from that of

the preceding vowels by lowering the tongue, scarcely ad-
mits of any distinction between narrow and wide vowels.
The unrounded vowel produced in it, [s], is the British
English vowel in men, get; in both British and American
English it is the long vowel before r in such words as
air, care; it occurs in French in such words as lait [Is]
*milk' and pere 'father'. A wide and lowered variety [ae]
is the American English vowel of such words as man, can.

The narrow rounded vowel [ce] occurs in French, as in

peur ^fear' and seul 'alone'.
b) Bach vowels. be simplest to speak next of
It will
the back vowels. In these the rear of the tongue
is near

velar articulation and the front concavely lowered, so that

the mouth is in the shape of a long, wide, hollow reso-
nance-chamber. This shape is accentuated, in most cases,
by protrusion and rounding of the The unrounded
lips.

back vowels are very hard to analyze, owing to the in-

accessibility to touch and sight and to the relatively un-
developed muscular consciousness of the back of the mouth.
The high back narrow rounded vowel [u] is typically
represented by the French sound in tour 'tower', pousse
'grows'. The German long u has a little less characteristic
lip-rounding; it occurs, for instance, in du 'thou'.
The wide rounded vowel [t] is the short sound in Eng-
lish words such as hook, foot and is the German short w,
as in Mutter 'mother'.
The unrounded vowel in this position is rare; it oc-
curs as a variant of another vowel in Russian and is said
also to be spoken in Armenian and in Turkish; its sym-
bol is [m].
The mid back narrow rounded vowel [o] is most typi-
cally represented by the French o as in rose 'a rose'. The
German long o, as in Rose, has less distinct rounding; in
 MUSICAL ORAL ARTICULATIONS 37

Norwegian and Swedish, on the other hand, there is an

[o)] with the extreme lip-rounding which French, for in-

stance, gives to [u].

The wide form [6] is the German short o^ as in GoU
'god'.
The same wide vowel, unrounded, [1], is in American
pronunciation the vowel of such words as cut, hut.
The low back rounded vowel [oj occurs in English he-
fore r in such words as hoarse, more. Within the sphere
of this symbol, though perhaps lower than the preceding
sound, is the British English vowel in got, collar, and the
like. A narrower forward variety is spoken in such French
words as mort 'death' and one still more forward, al- —
most a mixed vowel, —in such as comme 'how'. A lower-
ed variety of [o] sometimes expressed by the special
symbol [n] is the English vowel in such words as oil, law.
It exists in Swedish and Norwegian with the greater lip-
rounding normally given to [o].
The unrounded vowel corresponding to [o] is the [a^]
in the British pronunciation of cut, hut. Much commoner
is the unrounded vowel coresponding to [n], namely [a].
We may take the variety which occurs long in EngKsh
father, car as the normal type. Then the German long
vowel in Kahn 'skiff', Staat 'state' and its wider short
form in Jcann 'is able', Stadt 'city' are a little lower and
the French vowel in pas 'a step' and pale 'dough' is a
little lower and a little farther back. Higher than this

normal type is the [a±] in the American pronunciation

of such words as got, collar. A divergent variety of this
vowel is [a], pronounced much farther forward than [a];
it is the vowel of such French words as patte 'paw', part

'part', and, slightly fronted and raised, of the British pro-

nunciation of man, can, and the like.
c) Mixed vouels. The mixed vowels are less common
 .

38 THE PHYSICAL BASIS OF LANGUAGE

than tlie front or the back. The high mixed vowel, nar-
row and unrounded, [i], alternates with [ui] in the Russian

vowel of such words as [sin] ^son'; it is pronounced

somewhat back of the ideal mixed position.
Its rounded correspondent is the [ii] of Norwegian, writ-
ten Uy as in hus 'house'.
The mid mixed unrounded vowel [e], in both narrow
and wide pronunciation, is found in German unaccented
syllables where e is written^ as in alle ^alV.
The low mixed vowel, unrounded, [s] is used in the
British pronunciation of such words as heard [hsjd], nurse
[nsis]
I shallnot attempt to discuss the vowels of the un-
accented syllables of English and some other languages,
as they present many and complicated problems and have
been but imperfectly analyzed. It is customary to express
the commonest unaccented vowel of a language, such —
as in the second syllable of the English started (really
[ex]) or the German [e], as in alle, or the French 'e-mute'
(really [oe''^]), as m je T, — by the symbol which
[a],
thus has different values for different languages and is a
practical rather than a descriptive symbol.
There remain the nazalized vowels, of which French
can give us good examples. In these the velum is well
lowered, so that much of the breath escapes through the
nose, producing the peculiar nasal resonance. Thus in
French there is a nazalized [o], [5], as in ton [bo] 'good',
an [a], as in banc [ba] 'bench', an [s], as in hain [bs] 'bath',
and an [de], as in hrun [brde] 'brown'.
8. Infinite variety of possible sounds. It will be
seen that even the comparatively few of the most typical
sounds here described form a large list. By way of sum-
mary we may unite the most important of them in the
following table.
 INFINITE VARIETY OF POSSIBLE SOUNDS 39

1—4

l-H

be
a

,^//
>-»
Uvular.

> Ql3
f /
-f 9
Stops, unvoiced. . . .
q k c t P
Stops, voiced. . . . Q g J d b

Nasals, voiced. . . N g Ji n m
Spirants, unvoiced. h, H H X 9 j-sG f

Spirants, voiced. . ,
i^,Q If
g J
jjzQ V

Laterals, voiced. . . \ L 1

Trills, voiced. . . . R r

Musical sounds, high u ui ti 1 y i

Musical sounds, mid. A 6 6

Musical sounds, low. E ces

Musical sounds, lowest. D aa se

We have seen that these sounds, which may he select-

ed as typical, are only single instances from among an
infinite variety. Even the stops, which might seem fairly
inflexible, occur in a number of varieties. We have al-
ready spoken of the many variations as to point of articu-
lation and of the difference of voiced and unvoi- [b, d, g]
ced [p, t, k]. In English, standard German, and French
this difference is accompanied by another, that of energy
of articulation: the unvoiced stops of these languages are
pronounced with greater muscular tension at the point
of closure than the voiced stops. Our [p, t, kj, therefore,
are fortes, our [b, d, g] lenes. These two differences do
not always go hand in hand: in many German dialects,
for instance, there are unvoiced lenes. Another kind of
variation in stops will appear in § 9. The various possi-
40 THE PHYSICAL BASIS OF LANGUAGE

ble pronunciations of spirants, trills, vowels, and so forth,

and their variations as to place of articulation are more
obvious and have in part been mentioned.
9. Glides and mixtures of articulation. In the ac-
tual current of speech another factor of variation ap-
pears: the transition or^^?i^from one sound to another, or
from inactivity of the vocal organs to the production of
some sound (or vice versa). I shall mention only the two
most important instances. In passing from an unvoiced
stop to a vowel, we have to perform two movements: to
change the mouth-position and to begin voicing. If these
two movements are performed simultaneously, the result
is a pure stop, as spoken in the Romance languages (e.

g. French) or in the Slavic (e. g. Russian). If the stop

is opened before voicing is begun, so that a puff of un-

voiced breath first escapes, we hear an aspirated stop [p',

t', k'], as in English and German, or, even more pro-
nouncedly, in Danish. Finally, the glottis may be closed
during the stop and opened at the same time with the latter,
— this is the pronunciation in some Armenian dialects, —
or shortly after it, — this type occurs in Georgian, —
producing choice stops. The other instance I shaU mention
is the on-glide of initial vowels. Here the oral vowel-po-
sition is first taken: if voicing now begins immediately,
we hear a pure vowel American English or
initial, as in

French; if the vocal chords are gradually brought from

the breathing-position into that for voicing, they must
pass through that of an [h] (p. 26), producing the aspi-
rated initial of our words such as heel, have, hoop, etc. If,
finally, the vocal chords are first closed and then suddenly
opened into the voicing position, we hear a choke initial,
the glottal stop foUowed by the vowel: this is the way
German words written with initial vowel are pronounced,
such as arm ['arm] 'poor' (p. 24).
 SYLLABLES 41

Wliile there are a number of other instances^ notably

in connection with stop-articulation, of various glide-possi-
bilities, the glide is in the majority of cases determined
by the positions of the two successive sounds. In passing
from [a] to [u], for instance, there is only one movement
to be performed and only one path for that movement;
similarly, in passing from [n] to [d] all one needs to do
is to raise the velum, and this can be done in only one

way.
Beside glides from articulation to articulation there is

often possibility ofjnixiure of articulations. An [m], for

instance, before an [i], may be pronounced either indiffer-
ently or with the tongue-position and lip-widening of the
[i] . The latter is the habit of the Slavic languages. This
mixing in of part of the position of a front vowel, called
palatalization, is very common. In English it occurs only
in the case of [k] and [g], which are pronounced farther
forward before [i], as in Hw, give (p. 32). In the Slavic
languages almost every consonant can be palatalized; in
writing an accent mark may be used to indicate this (cf.
p. 31), for instance, in Russian [p'i Ju'] ^I am writing', —
[p'] spoken with the tongue-position of [i] and the cor-
ners of the mouth drawn back for the articulation of this
vowel. In labialization sounds are pronounced with the
lip-rounding of a rounded vowel. An instance is the Amer-
ican pronunciation of wh, as in which, whale: the vocal
chords are pronouncing an [h] while the tongue and lips
are in the [u]-position, [h)] or [hw].
10. Syllables. While much more could be said about
the different articulations and their glides and mixtures,
it must our purpose to understand how varied
suffice for
the possibilities are. Great as is this variety, everyone
who has heard a foreign language spoken will realize
that, aside from the strange sounds, the general manner
 42 THE PHYSICAL BASIS OF LANGUAGE

of pronunciation or 'accent' of a language is even more

characteristic. Here, of course, the possibilities are again
unlimited. Pitch, stress, kind of voice (e. g. full voice
and murmur), and duration (speed) are all variable
factors.
Even aside from the factors just mentioned, a sequence
of articulations never appeals to the ear as a series of
coordinate sounds. Some sounds are, in themselves and
aside from any distinction of stress or pitch we may
give then., more sonorous than others. Voiced sounds
are more auoible than unvoiced, for the obvious reason
that to the oral noise they add the tone produced in the
larynx. It is equally obvious that the more open a sound,
the greater its In a sequence of articulations,
volume.
accordingly, we hear a constant up and down of sonority.
The less sonorous articulations are heard, to speak met-
aphorically, as valleys between crests of greater sonority.
The sound- sequence between the least sonorous instants
of two such successive valleys we call a syllable. The
most sonorous sounds are the low vowels. Even in a
word like away [seuei] or [ouei], which is composed en-
rtirely of vowels, we hear two syllables, for the [u], less
I sonorous than the preceding and following lower vowels,
^s heard as a valley; similarly the [i] is less sonorous
than the preceding mid vowel [e]: we write [ae iiei] or
[a iiei]. The lower sonority of the [i] appears in a com-
bination like away again [a iiei a gen]. The most sono-
rous sound of a syllable is called the syllabic, the others
are the non-syllabics. Vowels used as non-syllabics, like
the [u] and [i] above are often called semi-vowels. The
semi-vowel [ii] is often, especially if the lips are tenser
than in the syllabic occurrence of [u] in the same langua-
ge, written [w], and, as [i], if the friction is at all above
a minimum; approaches a [j], this character is often used
 STRESS 43

to express a non-syllabic [i]. The combination of non-

syllabic with syllabic vowels is called a diphthong. If
the syllabic vowel precedes, as in the English he [hii]^

do [duii], day [dei], toe [toii], hoy [bne], die [dae], how
[hao], we speak of a falling diphthong; if the semi- vowel
precedes, as in yes [les], year [lij], your puj], wag [iiaeg],

ivaTl [unl], of a rising diphthong. A triphthong occurs,

for instance, in use [luiiz], [luus], wait [ueit], etc. One
can also write [hij, duw, dej, tow, jes, jij, juj, waeg, wnl,
juwz, juws, wejt].
Next to the vowels in sonority are the trills, laterals,
and nasals; all of them may figure as syllabics. Thus
the American pronunciation of words like sir, sJcirt, heard,
nurse is [sj, sk.it, hid, nis], and words like bottle, butter,

button, bottom are pronounced [bat}, bAt^i, bAti;i, bati]i].

In ivorh [wjk], the [u] or [w] is non-syllabic, the [i\

syllabic.
The boundary between two natural syllables is thus
always the least sonorous sound between the syllabics:
in bottle, butter, etc. it is the [t].

11. Stress, a) Syllablezsiiress The inherent sonority

of the speech-sounds is partly offset by the possibility
of speaking one sound more loudly than another, —
that is, of distributing the stress (p. 25). Thus the se-
quence can be spoken with the [u] louder, so that
[ui]
the [i] becomes non-syllabic: [uj], or with the [i] louder
and the u non- syllabic: [wi], —
for the most sonorous
sound is always the syllabic. Even an [ae] may thus be
turned into an [ae] and an [ea] into an [ea] by speaking
the [e] more loudly than the [a]. On the other hand,
stress cannot wholly offset natural lack of sonority: in
an [as], no matter how loud we try to make the [s], the
[a] will aways be the syllabic, for any voiced [a] is ap-
preciably louder than the loudest [s]
44 THE PHYSICAL BASIS OF LANGUAGE

Some languages regulate tlie stress within the syllable

in conformity with the natural sonority, pronouncing the
syllabic of each natural syllable with greater stress than
the non-syllabics. This, for instance, is the case in the
Romance and the Slavic languages. In all languages
there is some approximation to this distribution. English

and German depart from it as far as any. In these lan-

guages it often happens that a succession of two or even
more natural syllables is spoken with but one effort of
stress. While in French, for instance, one would say
[39 ma psl so ls:j], 'I am called the sun', with higher
stress on each syllabic than on the preceding and follow-
ing non-syllabics, an English word like utter [Ati] begins
with highest stress, which is maintained through the
syllabic [a], and then sinks steadily to the end of the word,
without regard to the presence of the second syllabic.
In a word like pity [piti] the stress rises through the [p],
reaches its height at the beginning of the first [i], main-
tains it through this syllabic, and then uninterruptedly
sinks. The same is true of German words like hitte
[bita] 'please' or hasse [hasa] 'hate'. There are in all
these words two natural syllables, but they consist of
only one stress-sijllable. Since stress-weakening by means
of separation of the vocal chords (p. 25) easily passes
over into the slightly wider open murmur-position (p. 25),
the unstressed parts of such words are often spoken with
murmur instead of voice.
The distribution of stress may thus conflict as to
syllable-boundaries with the inherent relations of natural
sonority. The boundary between natural syllables is, of
course, within the least sonorous articulation that inter-
venes between the syllabics. Thus in utterj bottle, butter,
button, bottom, pity it is in the [t], in the German hasse
[hasa] in the [s], in the German bitte [bite] in the [t]:
 STRESS 45

in short, the least sonorous sound belongs as much to

one syllable as to the other, — it is the valley between

the two crests of sonority. In these words there is, how-

ever, as we have just seen, but one stress-syllable and
therefore no valley or boundary of stress. The boundary
between two stress rises may, on the other hand, coincide
with the natural syllable-boundary. That is, the stress
of one stress-syllable may come to its minimum and that
of the next stress-syllable begin to rise within the least
sonorous sound. This is the case in such Italian words
as anno [an no] 'year' and atto [at to] 'act'. The effect
on the ear is that of a definite separation between the
beginning and the end of the articulation concerned.
Hence we write the symbol twice, once for each stress-
syllable, and call such sounds double or geminate sounds.
In by far the most instances, however, the minimum
of stress does not fall within an articulation: English,
German, and French, for instance, have no double sounds.
In* the Romance and the Slavic languages the stress-
boundary falls, when there is but one non-syllabic, al-
ways before the latter: the minimum of stress is reached
at the end of the preceding syllabic, and the new stress
begins to rise with the non-syllabic. Hence the division
in the preceding French sentence, or in such Russian
words as [vo* di] 'waters', [ba* ba] 'woman', [pa ta ra p'i*s']
'hurry up'. When there is more than one non-syllabic
these languages recognize certain groups of articulations
which may begin a stress-syllable: such groups are treat-
ed like a single sound; thus in French [a pie] 'to call'
or in Russian [prajtj'aj t's] 'farewell'. Sequences of
sounds which may not begin a stress- syllable must be
divided: thus in the above French sentence [Is] cannot
begin a stress-syllable, hence the division [psl so]. In
English and German, on the other hand, the conditions
 ;

46 THE PHYSICAL BASIS OF LANGUAGE

are not so simple. 1. In passing from a less Higlily

stressed syllable to one more liighly stressed, we usually
pronounce a single non-syllabic with the following stress-
syllable, as in aivaij [a wej], again [a gen], a name [9 nejm]
but we do not always do so, certain meanings demand-
ing a different division, as in an aim [en ejm], in contrast
with a name. 2. In passing from a stressed syllabic
followed by a single non-syllabic to a less stressed syl-
labic, we ignore the natural syllable-boundary, as in utteVj

pity, bottle, etc. above, — speaking but one stress-syllable.

3. If two or more non-syllabics intervene, we put the
stress-boundary between them, as in until [en til], hating
[hej tig], wholesome [howl sem]. — German differs from
English only in that in case 2 it does make a stress-
boundary (taking the non-syllabic with the following
stress-syllable),provided the preceding syllabic is long;
thus, in contrast with hitte [bite], hasse [base], spoken as
but one stress-syllable each, it says hiete [hi: te] 'offer',
Hase [ha: ze] 'hare' with two eacb. After the longer
English vowels the same distribution is often made; one
may say [fa: }Si ae p}] as well as [fa:&j[, sep}].
Within each stress-syllable also, different relations of
stress are possible. Two
forms are common: the syl-
lable either begins with highest stress, which then decrea-
ses, or it begins with less than the highest stress, rises
to the highest, and then falls off. In each case the high-
est stress may for a short time be maintained. In Eng-
lish we use the former type for syllables beginning with
the syllabic, such as all, are, utter, apple, the latter for
those beginning with a non-syllabic, in which we reach the
highest stress only at the beginning of the syllabic, as in
mid, lid, pity, hotile, etc. (It is best to try these and the fol-
lowing examples of stiess-relations in a whisper, as pitch-
variations —
see next § —
may otherwise be confusing.)
 STRESS 47

In all the preceding instances the stress

maintained is

at its height throughout the syllahic and sinks only after

the following non-syllabic is reached. This is why the
[d] in lid, for example, is so much more stressed than
that in lead (verb) [lijd], the [n] in hin so much more
than that in bean [bijn]: the former [d] and [n], respec-
tively,begin with highest stress and descend, but those
in lead and bean begin only after the stress has already
decreased during the preceding non-syllabic, [j]. This
maintenance of highest stress throughout the syllabic is
called close syllable-stress.
The decrescendo of the stress may, on the other hand,
take place within the syllabic. In the Romance and the
Slavic languages this is almost always necessarily the
case, for most of the stress-syllables of these languages
end with the syllabic. This is called open syllable- stress.
In English it is less common, occurring chiefly in our
longer vowels, which often stand at the end of a stress-
syllable and therefore must needs include the descres-
cendo, as in mama [mama:] and frequently (cf. above)
in father fa:^j], apple [sBp}]. German has open syllable-

stress in syllables with long syllabic, as in Wien [vi:n]

'Vienna', Kahn [ka:n] 'skiff', and in the first syllable of
biete [biita] 'offer', Hase [ha: zo] 'hare', and close in syl-
lables with short syllabic, as in bin [bin] 'am', kann [kon]
'is able', bitte [bita] 'please', hasse [hasa] 'hate'.
The stress may, further, rise, reach its highest point,
and fall within the syllabic; this is called compound
stress. It is found regularly in certain syllables in An-
cient Greek and in Lithuanian. In English it is heard
in a surprised, displeased What/? and in a peeved,
irritable No! (most clearly if the whispering test is used).
These differences in syllable-stress constitute one of the
chief difficulties in acquiring a foreign pronounciation
48 THE PHYSICAL BASIS OF LANGUAGE

and are perhaps tlie most important factor in tlie pecu-

liar 'accent' of a lanoruaore.
b) Group-stress. Of the several stress- syllables in an
utterance some receive louder stress than others. In this
distribution of varying degrees of stress among the syl-
lables, group-stresSf the different languages also diverge.
In French the last syllable of an utterance or of such
parts of an utterance as are fairly independent in mean-
ing, alone receives higher stress, the other syllables being
fairly equal. Thus the French sentence above quoted, re-
ceives, in contrastwith the English equivalent, I am
'called the 'sun, but one highest stress: [39 mapslsols:j],
or at most also a weaker secondary stress on [psl], the
last syllable of the part of the utterance which corre-
sponds in meaning to 'I am called': [39 ma 'psl so "ls:j].
In Japanese there is even less difference between the
syllables. English and German, on the other hand, divide
every utterance into small groups of syllables within each
of which there one highest stress,
is —
stress-groups.
But even these two languages differ from each other;
English, for instance, normally gives highest stress to
one syllable each of an adjective and a following noun^
as in a 'young 'man or 'rotten po'tato, so that such a com-
binaiion contains two stress-groups, while German nor-
mally gives higher stress to the noun: ein 'junger " Herr,
'faule Kar'toffel, one stress-group each. Russian also, in
spite of its entirely different syllable-stress, has much
the same group-stress as English or German, but differs,
for instance, in often giving higher stress to a preposition
than to a following noun, as though we should say 'at
na gu, 'po'd ge ia vu]. English,
thefootj 'under the head: ['za*
German, or Russian, with their frequent high stresses,
can distinguish different meanings by their distribution
of group stress; thus in English 'torment noun, but tor-
 STRESS 49

'mcnt, verb; in German ubersetzen ["'yrber'zstsiji] Ho set

across' but [''y : bi "zst si;i] 'to translate'; in Russian
['rau' ka] Horment' (noun) but [mu 'ka*] 'flour'. Other
languages with frequent high stress give it a uniform
place with regard to the word, — which, as we shall
see, is a division based entirely on meaning. Thus
Cechish and Icelandic have highest stress on the first
syllable of every word; short words of relational mean-
ing, such as pronouns, prepositions, conjunctions, may,
however, lack this stress. In Polish the next-to-last syl-
lable of words is similarly, in almost all instances, accent-
ed. German departs widely and English very widely
from a fundamental principle of stressing the first syl-
lable of every word. Tbere remains, however, the princi-
ple that in these languages every word is a stress-group,
containing one syllable with highest stress and, in longer
words, one or more with intermediate stress (the degrees
of stress are indicated by the varying number of accent
marks before the syllable), e. g. procrastination fpJA
'kjaes ti '"nej with four degrees of stress, highest on
/i;i],

[nej], least on [ti] and [Ji;i] and intermediate degrees on

[p.iA] and [kaaes]. Short relational words, such as a, they

he, her, in, and the like, have, however, as a rule, low
stress and stand to the preceding or following higher
stress in the same relation as do less stressed syllables
of the same word. Further, the highest stress of some
words is weaker
in certain connections than that of others,
so that which represent words fall
the stress-groups
under a higher unity of the phrase; examples are the
German combinations of adjective with noun (see above)
or such expressions as a 'great 'big "man, which, how-
ever, in nearly all cases, are really examples of sentence-
stresSy to which we shall now turn.

c) S enience-s txess. Among stress- groups the highest

Bloomfield, Study of Language 4
60 THE PHYSICAL BASIS OF LANGUAGE

stress of some is in turn higher than that of others. This

highest stress or sentence- stress is in all languages given
to the emotionally most vivid part of the sentence, —
a result of the fundamental character of speech as an
expressive movement varying in intensity according to
the intensity of emotion (p. 1). We may illustrate this
by speaking an English sentence with emotional stress
on different words, —
as though in answer, for instance,
to various contradicting statements. E. g.: "J 'saw the
'young 'man, —
/ "saw the 'young 'man, / 'saw the —
"young 'man, — I 'saw the 'young "man^ and evei; I 'saw
"the 'young 'man.
We find, thus, a threefold distribution of stress. First,
there is the up-and-down of stress within the syllable,
fixed for each language. Secondly, every language ha-
bitually gives certain syllables higher stress than others:
we have our accented and unaccented syllables and our
unaccented short words. Finally, the emotionally dom-
inant elements of the sentence receive higher stress than
all others. Usually these elements are words, in which
case the emotional stress is given in most languages to
the syllable which otherwise also bears higher stress than
the others. Thus when we say It 'wasn't dis'honesty, it

was "sheer procrasti" 'nation, the emotional stress of ^ro-

crastination is placed on the same syllable which habitu-
ally has highest stress in this word. If, however, the
emotionally most charged part of the sentence is not a

word, but only part of a word, the emotional stress may

conflict with the group -stress. Thus, while we say for-
'give, as in for'give and forget, we say "forgive in "give

and "for give. In French, where not a syllable of every

word, but only the last syllable of sentences and phrases
receives habitual group-stress, the emotional sentence-
stress is given to the first syllable that begins with a
 PITCH 51

consonant of the dominant word. Thus it is almost al-

ways a part of a sentence or even of a word which usual-
ly is unstressed, that now receives highest stress, 'It's

the same person' is [s8lams:mpsr'son] but 'It's the very

same person' is [s9la"ms:mpsr 'son]; 'It's impossible' is

[s8 ts po 'sibl] but 'It's impossible' is [satsr^po'sibl].

The rhythmic effects of stress-distribution are heightened
in English and Russian by the habit of speaking in the
less stressed syllables shorter extreme vowels
and less
than in the stressed. Other languages, like German, go
still farther and restrict the least stressed syllables to a

single vowel, — in German [e].

12, Pitch. Pitch, like stress, can be infinitely varied.

The modulation of pitch may correspond to the syllable-
division, each syllable being spoken with a unified pitch-
scheme. This is the case in Chinese; thus in Peking
['"xiia"'] with even high pitch means 'flower', [Lxua\] with
low falling pitch, 'speech' or 'picture', [Jy/] with low
rising pitchmeans 'rain' and [^ly/] with high rising pitch

means While the Peking speech has only these

'fish'.

four pitch-schemes, some dialects have as many as nine.

In Norwegian and Swedish stress-groups (words) of one
syllable are spoken with rising pitch and stress-groups
of two or more syllables (corresponding in most, but not
in all cases, to words) are spoken with either rising or
falling and then rising pitch, according to fixed habits;
thus in Norwegian [biin/] 'ground', ['biinan/J 'the groun'd',
["bii'/nan/] 'bound'. In Lithuanian, Ancient Greek, and
the oldest Sanskrit we find compound (rising -falling)
pitch belonging habitually to certain syllables.
In other languages, such as English, the different syl-
lables have no fixed but pitch is used in
pitch-relations,
the whole sentence to express emotional relations. We
use falling pitch for statements, as in He came hade or
52 THE PHYSICAL BASIS OF LANGUAGE

the answer-word YeSj rising pitch for questions that

contain no question-word, such as Did you say that? or
for question-words used alone, as What?'^ the compound
pitches are similarly used, thus rising-falling in questions
that contain a question-word, such as What was he doing?,
or in an irritated No, and the falling-rising in an angri-
ly surprised What? Compound pitch is usually, if con-
fined to one syllable, accompanied, as in these examples,
by compound stress (cf. p. 47).
13. Duration. Duration or qicantity, — that is, the
length of the different sounds, syllables, and stress-groups,
— is another important factor.
Thus English some of the vowels are longer than
in
and [d], especially, longer than [i] and [u]. All
others, [ae]
our vowels, moreover, are longer before voiced sounds,
as in hidy than before unvoiced sounds, as in hit. In one
case only have we, in American pronunciation, approxi-
mately the same vowel in two distinct quantities, namely
[a:] long, as in father, and the same sound (with a slight

difference, p. 37) short [a], as in got, collar, god; in ac-

cordance with the preceding rule the [a] before the voiced
sounds in the last two examples is, however, longer than
in got. In standard German the tense vowels are long,
the loose vowels short; in the case of [a] there is, however,
scarcely any difference except that of quantity, e. g. Stadt
[Jtat] ^city', Staat [Jta:t] 'state'
In English our non-syllabics are longer, the shorter the
preceding syllabic; thus the [n] in hin is longer than that
in men, which is in turn longer than that in man. In
other languages the duration of non-syllabics not auto- is

matic (i. e. does not depend on the surrounding sounds)

but is fixed for each word. Such long non-syUabics differ
from doubled sounds (p. 45) in that no stress-boundary
occurs during their articulation. Accordingly a difference
 LIMITATION OF THE ARTICULATIONS IN EACH DIALECT 53

exists between tbe Norwegian otte [''o\ 'tie/] ^eigbt' with

long [t] beginning tbe second syllable and tbe Italian otto
['ot to] 'eigbt' witb double [t], tbe stress-boundary com-

ing after tbe closure and before tbe opening of tbe [t]-stop.
Tbe duration of tbe various parts of a sentence is less
fixed. Certain tendencies, bowever, such as tbat to speak
a parenthetic clause very rapidly (as in This man, — who
for that matter, had very little to do ivith the affair, — . .
.),

can here be distinguished. No

doubt there are also differ-
ences between the different languages, but they have never
been ascertained, owing to the difficulty of abstracting
from factors of mood, personal habit, and tbe like, which
here have comparatively free play.
14. Limitation of the articulations in each dialect.
A language which significantly used any considerable part
of the articulations and variations of stress, pitch, and
quantity that are possible, could be understood only by
the closest application of the attention, and, if it used too
many, could not be understood at all, for the intelligi-
bility of language depends, of course, on repetition and
recognition.
As a matter of fact every language limits itself to cer-
tain sounds and to certain ways of combining them. Some,
like English and German, employ constant stress-relations
for certain syllables, leaving pitch-modulation for the sen-
tence as a whole; others, like French, use both pitch and
stress only in the sentence; still others, like Chinese, as-
sign a definite pitch-relation to each syllable and use
modulate the sentence; Norwegian and Swe-
stress only to
dish use pitch and stress both for the syllable and for
the sentence. Tbe same is true of tbe individual articu-
lations. Thus English and standard German use unvoiced
aspirated fortis [p', t',and voiced plain lenis stops [b,
k']
d, g]; the Romance and tbe SLivic languages use only

r
54 THE PHYSICAL BASIS OF LANGUAGE

plain stops^ unvoiced fortes [p, t, k] and voiced lenes [b,

d, g]; the Peking pronunciation of Chinese uses only un-
voiced stops, aspirate fortes [p^, t', k'] and plain fortes
or lenes [b, d, g]; other languages, such as the Polyne-
sian, have but one series of stops; Sanskrit had four, un-
voiced fortes and voiced lenes, each in aspirate and plain
form. Such distinctions are recognized by speakers of
the language, and forms not so recognized are interpret-
ed as standard forms by the hearer. Thus, though un-
aspirated [p, t, k] and unvoiced lenes [b, d, g] are oc-
casionally spoken in English, they are not recognized as
different from the more usual forms; such a distinction
as that between [p'] and [p] is not even noticed. Whether
we speak a [t, d, n, 1, j] a little farther forward or a little
farther back is a matter of indifference in English; it is
left to personal habit, mood, or the influence of the sur-
rounding sounds; but in Sanskrit and in many modern lan-
guages of India the difference between dental and cerebral
articulation is as important as that between any other
sounds; thus in Canarese hondu means 'killed' but IcondUy
'taken'. Whether we pronounce a h farther forward or
back depends mainly on the following vowel and is never
significant, but in Greenlandish or in Arabic [k] and [q]
must be strictly kept apart.
In other words, each language, or, better, each dialect
distinguishes only a limited number of places of articu-
lation, and in each place only a limited number of manners
of articulation, and any variations from these are never
significant.
15. Automatic variations. The variations that occur,
while not significant, may be very regular. Our English
vowels, for instance, are longer in final position and be-
fore voiced sounds than before unvoiced, longer, to repeat
our example, in hid than in hit^ in heCj head than in heat,
 AUTOMITIC VARIATIONS 65

and Englisli spoken without this variation would strike

our ear as very foreign-sounding, even though most of us
would be unable to determine exactly what the peculiarity
was Yet in spite of this universal occurrence, reaUy —
because of it, —
this difference of vowel-quantity is never

significant. It depends solely on the following sound and

can never be determined by the meaning of the word: it
is an automatic sound-variation. Before and after unvoiced

sounds we pronounce our [m, n, 1, i\ partly or whoUy

unvoiced, e. g. in try, heltj hemp, sent, snow, but we are
not even conscious of this variation: it also is purely au-
tomatic. So is the German and English variation between
[k, g] farther forward or back according to the following
vowel, as in liin, give, cap^ gap,— —
coop, goose, but in
Arabic, as above mentioned, such pairs as [ka:la] 'he spoke'
and [qaila] ^he measured' could never be confused.
Every language has^ further, limitations as to what
combinations of sounds can occur and as to where, in the
syllable, a given sound or combination may be spoken.
Thus no English syllable can begin with the combinations
(kn, gn, ts, Jpj, tsv, Jv], which are common in German,
— even though, distributed between two syllable? or at
the end of a syllable, all of these do occur in English, as
in acJcnowledge, bigness, its, cash price, ifs very cold, cash
value. The sound [g] cannot occur at the beginning of
a German or an English syllable but it does so in many
languages. In Peking Chinese a syllable can end only in
a vowel, [n], or [rj]. In the Polynesian languages no
syllable ends in any other than a vowel sound.
Each language, therefore, has a limited sound-system,
which, if only significant distinctions are counted and non-
significant variations, whether automatic or merely casual,
are ignored, is never very great.
 CHAPTER m.
THE MENTAL BASIS OF LANGUAGE.
1. The
place of language iu our mental life. Lan-
a very important part in most of our mental
S^iiage pla} s

processes, few of which, indeed, are entirely free from

linguistic elements. While it is possible, for instance, with
some effort, to picture in purely visual terms the actions
we have in mind for the morrow, we hardly ever do so,
but rather plan our day not only by visualizing but also
by wording what we intend to do. If, further, we try to
think of our reasons for these intended actions, or of
their effects, or of anything else not in immediate physical
connection with them, we must resort to language, fram-
ing our thought in words and sentences. In short, a very
little introspection shows that nearly all of our mental

life contains speech-elements. We cannot conceive of the

human mind without speech. The development of lan-
guage, accordingly, must have advanced in inseparable
connection with that of the mental powers generally. To
demonstrate in detail the role of languao^e in our mental
processes would be to outline the facts of psychology. We
are here concerned, of course, only with those mental process-
es which most immediately underlie the use of language.
2. Total experiences. The animals have in common
with us a process which may be called the formation of
total experiences. Like us, they experience the outside world
not as a chaotic jumble of sensations, but as a system of
 TOTAL EXPERIENCES 57

complex recurrent units^ as a world of objects. The per-

ceptual and emotional elements which we group togetlier,
for instance, as a rabbit, appear to a dog also coherent
and distinct from other perceptions and emotions, such
as those of the surrounding trees, the sky, other smells
and noises, the internal bodily sensations, and so on. Like
ours, the dog's apperception, — or, as we subjectively say,
his attention, — may focus the rabbit as the central ob-
ject, for the time being, of consciousness. The coherence
and unity of such a total experience are due to habits of
association formed in earlier related experiences: in our
instance the surrounding trees and the sky, the bystanders,
and those of our internal sensations and emotions that
are not connected with the present experience, have all
entered into various combinations in earlier experiences
and have thereby become familiar enough not to be irrel-
evantly confused with the present one.
Animals respond to a total experience by an expression
varying at best for a few widely distinct emotional qual-
ities; thus the doG^ barks at the rabbit as he does at a

great many other things. Man differs from the animals

first of all in that he has a distinctive sound-reaction for
each one of a great many types of experience, — e. g.
for, the type of experience which we call a 'rabbit'. When-
ever an experience of a given type occurs, the sound-re-
action connected with that type is associatively recalled
and reproduced. When we saw the rabbit, for instance,
we did not 'inarticulately' cry out, but exclaimed 'a rabbit.'
This also, to be sure, is no^ an exact way of dealing
with experiences. We react to countless experiences of
a single type (such as 'rabbit') with one and the same
utterance, while in fact no two experiences are wholly
alike. When we associate the present experience with
certain past experiences and utter with it the sound-se-
68 THE MENTAL BASIS OF LANGQAGE

quence which we heard and uttered with them, we do so

not because the present experience is exactly like the past
ones, — it is not, — but because certain elementary fea-
tures are common and each of them. These elemen-
to it

tary features are known as dominant el emenjs^ Thus a

rabbit of different size or colorpOfone running in the
opposite direction might call forth the same utterance.
We use the word 'book' for objects of many sizes, shapes,
and colors, provided they present certain features. Even
a clearly defined scientific term, such as 'triangle' applies
to an infinite variety of experiences with but a simple
common element. In short, our reaction to experiences,
though much more differentiated than that of animals,
is not just to the individuality of each experience, but
groups great numbers of experiences together under types
within each of which all the experiences are designated
by one and the same reaction.
The association of experience-types with fixed and dis-
tinctive sound-utterances represents an important step in
mental progress. It makes possible attentive and connected
thought. When we recall the experience, we repeat, ac-
tually or in imagination, the sounds with which it is con-
nected. They are a convenient means of holding the ex-
perience in the attention; by recalling the sounds (or their
visual symbols) over and over again, — at first as young
children do, aloud, but, after practice, in imagination alone,
— we can keep the experience before us much longer
than is possible in speechless picturing.
Anadvantage of the grouping together of hosts of in-
dividual experiences under one type is this, that all ex-
periences belonging to the type can be dealt with en masse
and need not be recalled one by one, if we use the lin-
guistic expression, which deals with all of them alike.
This is conceptual or general thinking.
 THE ANALYSIS OF TOTAL EXPERIENCES 50

3. The analysis of total experiences. The existence

of a fixed sound- reaction, which enables us to hold an
experience vividly in our attention, also makes possible
the analysis of experiences. Every experience is composed
of a number of elements whose individuality is due to
their having occurred in other contexts in past experiences.
Thus we have seen the color of the rabbit, other four-
footed animals, other running animals, and the like. Each
element recalls those past experiences in which it figured.
But it does this obscurely, until language has given the
experience a fixed and easily handled symbol with which
we can keep it from slipping, as it were, through our
fingers. Once language exists, however, the analysis of
the experience into these elements is bound to develop.
At least it takes place in all known languages and is in
all of them, as time goes on, being perfected by a grad-
ual but unceasing process of development, to which we
must ascribe also its origin.
This process is the assimilation of expression- relations
to experience-relations. We may
by a sche-
illustrate it
matic example. Suppose that in some language the ut-
terance connected with the experience of a white rabbit
is patilu and that connected with a white fox is melcOj —
in other words, that these experiences, of dilFerent emotional
value, are attended by two totally unlike expressions.
Nevertheless, owing to such elements as they have in
common, whenever a white rabbit is seen, not only the
past white- rabbit experiences, with their patilu, but also,
among others, the white fox experiences, with their meko,
will be awakened.
Sooner or later one of these types will
assimilate the other's expression; such assimilative pro-
cesses are constantly occurring, as we shall see, in every
language, —
rs when, in English, Chaucer s word fader
became the father of present English, under the influence
GO THE MENTAL BASIS OF LAKGUAGE '

of mother, brother. For instance, instead oipat'ilu, someone

will, under the influence of melw, say metilu. At first this
will happen occasionally, but it will be the more likely
to happen again when one has once spoken or heard the
new form. The associational circumstances are all in favor
of it. Finally the new habit will completely supersede the
old. When this has happened, there are two utterances:
me-tilii 'white-rabbit' and me-TiO Svhite-fox'. Corresponding
to the perceptual element ^white' is the phonetic element
me-. When one now utters metilu a certain amount of
analysis is involved: me- expresses the color, -tilu (or -ho)
the kind of animal. These phonetic elements may ulti-
mately attain independent use: in answer to such a question
as 'What kind of a rabbit (fox) did you see?' one may
say me ^White', and one may designate 'rabbit' in general
by tilu, 'fox' in general by Izo.
When development has taken place, such an ut-
this
terance as me tilu or white rabbit involves an analysis of
the total experience into these two elements. When we
say tvhite rabbit we more or less vividly separate the two
elements of the total experience. Sometimes we may not
attend closely to the analysis, but at others we shall in-
sist on it, as when we say 'No, a white rabbit' or 'No, a
white rabbif. Suoh^n utterance analyzing an experience
into elements we call a. sentence.—
The relation of the elements of a sentence to each other
has a distinctive psychological tone. It is called the log-
ical or discursive relation. It consists of a transition of
the attention from the total experience, which throughout
remains in consciousness, to the successive elements, which
are one after another focused by it.
The attention of an individual, — that is, apperception,
— is a unified process: we can attend to but one thing
at a time. Consequently the analysis of a total experience
 THE ANALYSIS OF TOTAL EXPERIENCES 61

always proceeds by single binary divisions into a part

for the time being focused and a remainder. In the pri-
iQary division of an experience into two parts, the one
focused is called the suhject and the one left for later
attention ih.Q predicate^ the relation between them is called
'predication. If, after this first division, either subject or
predicate or both receive further analysis, the elements
in each case first singled out are again called subjects
and the elements in relation to them, attributes. The subject
isalways the present thing, the known thing, or the con-
crete thing, the predicate or attribute, its quality, action,
or relation or the thing to which it is like. Thus in the
sentence Lean horses run fast the subject is lean horses
and the horses' action, run fast, is the predicate. Within
the subject there is the further analysis into a subject
horses and its attribute lean^ expressing the horses' qual-
ity. In the predicate fast is an attribute of the subject
run.
Constant repetition, to be sure, mechanizing these pro-
cesses, saves us the trouble of repeating the entire dis-
cursive analysis in every sentence we utter. Such groups,
especially, as are very common
are no longer felt as attri-
butions (predication always vividly discursive), the con-
is

crete relation alone remaining uppermost. Thus, in a sen-

tence such as A ivhite rabbit ran across the field, the first
three words are plainly felt to be the subject, and the
rest the predicate, and within the subject uhite, within
the predicate across the field are in vivid attributive re-
lation, respectively, to a rabbit and ran; but the groups
across (he field and a rabbit are not by the normal speaker
felt as discursive relations. He would say simply that a
expresses the *indefinitiness' and that the expresses the
'definiteness' of the thing, while across is expressive of
local relation. It is only when we give the purts of the
62 THE MENTAL BASIS OF LANGUAGE

utterance mucli more than the usual degree of attention,

that we may feel these relations as discursive, —
as, for
instance, when we say 'It was a house, but I don't think
it was the house', where a and the are plainly attributes.

In short, a frequently recurring arrangement of elements

may become habitual and not require a vivid discursive
analysis for its utterance.
As this circumstance shows, discursive analysis is not
an absolute thing: associational identification shades into
it. In most languages we find, accordingly, elements that
are but partially independent. In our schematic represen-
tation above, the stage in which me-tilu Vhite-rabbit' and
me-ko 'white-fox' are used, but neither me- nor -tilu nor
-ho are as yet used independently illustrates this. In such
an English sentence as He suddenly ran across the field
there are several such partly analyzed elements. The element
suddenly, for instance, divides itself into sudden and -ly, but
since the latter cannot be used alone, the analysis is not dis-
cursive but merely associative. The same is true of across,
where cross does, in related senses, occur alone, but not
so a-. The r-vowel-w of ran occurs also in run, and the
vowels [se] and [a] of these two forms are felt to express
the relative time of the action, but neither is an abstract
r-vowel-w, as a term for the action itself regardless of
time, in English conceivable, nor is an [se] or an [a] ever
spoken separately to express the time alone. In father^
mother, hr other, the -ther is common to all and thus ex-
presses a common element of all three; or, if we add sister,
we may say that dental-plus-r does so, but neither -ther
nor a dental-plus-r can be used alone in some such sense
as 'near relative': there is but the suggestion of an ana-
lysis.- Such imperfectly separable elements are caUed for-

mational elements, as opposed to the independently re-

current units of analysis, words. Words only and scarcely
 THE NAMING OF OBJECTS 63

ever formational elements, can be dealt with as conceptual

units of general thinking.
The naming of objects.
4, If we look into concrete
experience, we find that all of it centers round objects.
An independent (or, as we say, abstract) quality, action,
or relation never occurs. The sound-reactions, therefore,
which form language can originally have been called forth,
in so far as they refer to perceptual experience, only by
objects. Words for qualities, actions, and relations we
must suppose to have been evolved in the later course
of speech-history.
The linguistic expression of an object-experience, then,
is the simplest type, psychologically, of such expression.
It is a sound-complex heard and uttered in connection
with a number of successive concrete experiences, each
of which exhibits certain dominant elements. The words
rabbit or took are associated for each speaker with a long
having certain dominant features in
series of experiences
common, much as these experiences may have diverged
in their other features.
Even here we see a certain degree of abstraction. In
speech or thought the sound-expression may be used not
only for a given object exhibiting the dominant features,
but also as a representative of all objects exhibiting them.
In a general statement about 'the rabbit', 'books', or 'a
triangle' these words save us the task of picturing suc-
cessively all the rabbits, books, or triangles we can re-
call or imagine: we need only dwell on the word and
the associated dominant features, such as a vague visual
image of a rabbit, a book, or three intersecting lines.
Thus, to repeat, the easily handled general concept,
— the basis of logical thought, — is a product of lan-
guage.
There are numerous languages, especially on the Ameri-
64 THE MENTAL BASIS OF LANGUAGE

can continent, wliicli have not gone beyond the naming

of objects. In these Languages the qualities and actions
of objects, which in concrete experience never occur apart
from objects, are in expression also always connected
with them. Thus one cannot, at this stage, speak of
'white' or of ^runs', but only of such objects as 'white-
rabbit' or ^running-rabbit', or, at best, of Svhite-thing'
or of 'running-thing' — in terras of our diagram, of me-
tilu or me-lio, never of me. Every word
an object-ex-
is

pression; qualities or actions are never as such expressed

by separate words. One cannot say 'kills' or 'killing',
for instance, but only 'his-killing-of-it' or the like. This
state of things forbids any distinction in speech between
predication and attribution, for, as predication usually
has as its subject an object and as its predicate an action
or quality, its explicit expression depends on the exist-
ence of action-words and quality-words as separate words.
Hence in these 'nominal' or 'attributing' languages such
utterances as 'white-rabbit' correspond equally to our
predication 'It is a white rabbit' and to our attributive
"'white rabbit', and such a locution as our 'The rabbit is
white' is inconceivable: one could only say 'This-rabbit
(is a) white -rabbit' or 'This-rabbit (is a) white -thing'.
Owing to the constant possibility of use as what we feel

to be complete predications, the words of such- languages

are often called 'sentence-words'
In addition to the object-expressions such languages
have only pronominal words. These are expressions of
purely deictic value, referring to the speaker in words
for T, the one spoken to in words for 'you', the object
near the speaker in words for 'this', the object farther
away in words for and so on. Their origin is
'that',

probably to be sought in sounds uttered in connection

with deictic movements. At any rate, in most languages
 THE DEVELOPMENT OF ABSTRACT WORDS 65

they resemble exclamations:" as in English, they are usual-

ly short words, and occasionally they differ phonetically
from the rest of the word-stock, as when in Russian the
word for 'that', [V tat], is the only native word beginning
with the sound [s]. These pronominal words thus re-
semble the purely emotional responses to experience which
we shall meet as 'interjections'.
5. The
deTclopiiieut of abstract words. Language
at the nominal or attributive stage has not attained a
habit of abstraction which English, for instance, has, —
namely the habit of separating, as independent expressions,
the qualities and actions of objects. That our concepts
of quality and action are purely linguistic is evident upon
a little Experience contains qualities and
introspection.
actions only in connection with objects. If we try to
think, apart from the word, of 'white', we can do so only
by picturing an object (such as a flat surface) or a suc-
cession of fleeting objects whose white color we hold
dominantly in our attention, neglecting their other features.
Similarly, the concept of 'run', 'running', if we exclude
word-images, can be pictured only as a man or an ani-
mal or a succession of such running. This is due to the
fact that in actual experience there is no such thing as
a quality or an action apart from an object. What lan-
guage does is to furnish a fictitious object, namely the
word-symbol, by which we represent the unimaginable
abstract concept of quality or action.
The historical origin of words independently expressing
quality or action is various. In English such words as
white used to mean 'white-thing', the 'thing' being defined
as to gender, number, and
case, and such words as 'runs'
used to involve also an actor, meaning 'he-runs'. As to
the psychologic character of the expressions as we have
them today, the historic origin is, however, immaterial.
Bloomfield, Study of Language 5
 66 THE MENTAL BASIS OF LANGUAGE

In words expressive of quality the dominant element

tlie

is a single common feature, permanent in each of a

number of objects whose other elements are various.

This permanence of the dominant element allows it, in
its association with the word, to remain vivid: such a

word as white is joined to a lively image of a single ob-

ject or of successive shifting objects of white color. In
the action-words the dominant element is a feature also
common number of objects, but in all of them im-
to a
permanent. As soon as we attempt to picture the object
vividly, the action is lost: the object stands immovable,
however suggestive of action we may allow its pose to
be. Consequently the perceptual dominant element, aside
from the word, of an action- word is never vivid: as a
rule, in fact, we do not attend, in thought, to any element
except the word itself, which has thus become dominant
in the whole complex. That is why the experiment of
thinking of an action-concept without using words is
much more difficult than in the case of a quality-concept.
The psychologic character of the more abstract words,
such as in English, the prepositions (e. g. under, over,
in, hy, across), the conjunctions (e. g. if, though, because),
and the abstract nouns (e. g. cause, result, essence, being,
relation), while in. itself interesting, need not further con-
cern us here, if we remember that the principle is the
same as in the case of action-words. The dominant ele-
ment when these words are used is always the word it-
self; in any given occurrence they resolve themselves into

concrete collocations or successions of objects, which ob-

jects we do not stop to picture more than vaguely when
the word is being used.
] 6. Psychologic composition of the word. The word
'is thus psychologically a complicative association of those
perceptual and emotional elements which we call its
 GRAMMATICAL CATEGORIES 67

meaning or experience -content with the auditory and

motor elements whicli constitute the linguistic symbol.
Where reading and writing are practised the visual and
motor elements of the printed and written word join the
auditory and motor of the spoken. Disturbances of these
associational habits are the much-discussed phenomena of
the aphasias.
Among the elements constituting this complex the
dominant may, according to individual disposition, be
visual, auditory, or motor; whether the linguistic elements
alone or the experience-elements also shall be dominant,
depends, as we have seen, on the character of the word:
in object- words, and, in a different sense, in quality- words,
elements of perceptual experience may dominate, while
in action-words and more abstract expressions the lin-
guistic symboldominant, the experience-elements being
is

but vaguely imaged. This is why in absent-mindedness

or aphasic conditions the most concrete object-words
(such as proper names) are first and most frequently
forgotten, the quality-words next and the abstract words
last of all. In learning languages, on the other hand,
we succeed better in remembering object -words and
quality-words, which we can associate directly with per-
ceptual images, than action-words and abstract words
(prepositions, conjunctions, particles, etc.) which we tend
to associate only with words of our own language which
either do not correspond exactly, or, in any case, remain
dominant to the exclusion of the foreicrn o words.
7. Oraniniatical categories. In the analysis of the
total experience into independent elements and in the
partial analysis of the latter into formational elements,
certain types may become habitual and finally universal
in a language. For instance, in analyzing a total ex-
perience we who speak Engl'sh always speak of an actor
6*
68 THE MENTAL BASIS OF LANGUAGE

performing an action. Many total experiences really are

of this type, e. g. Tlie rabbit ran; in English, however,
this type has been generalized to furnish the mould for
expressing all total experiences, — that is, for all sen-
tences, — including those which really involve no actor
or action, such as The rahbit is ivliite. Here we use a
fictitious action-word, is, of whose action the rabbit is

supposedly the agent. In Latin, for instance, this would

not have to be done: one could say Cuniculus albus, liter-
ally ^Rabbit white', where no such fiction is maintained,
— and the same would be true in Russian. In short,
actor and action are grammatical categories in the English
language. Categories like this one, which universalize
certain relations between words, are syntactic categories.
In the imperfect analysis of words into formational
elements also there may be categories. These are called
morphologic categories. An English verb-form, for instance,
always contains an imperfect analysis into a formational
element expressive of the action itself and one expressive
of its relative time one can say he runs or he ran, but
:

there is no indifferent form, as, for instance, in Chinese,

where [lP'clo/] means, from our point of view, 'runs',
^ran', or 'shall run', indifferently, but, if the element of
time is vivid in the total experience, one can say also,
in two words, [lP' a 6/ la] 'ran' or [jao\LP^a6/]'will run'.
That is, just as we always express future time in a sepa-
rate word (tvill run), so Chinese also analyzes out the
past-element as a separate word. Latin, on the other
hand, has also a future category: currit 'he runs', cu-
currit 'he ran', curret 'he will run'. We say, then, that
the formational expression of present or of past time
with actions is a morphologic category in English, that
of present, past, or future time, in Latin.
The grammatical categories, then, though always based
PSYCHOLOGIC CHARACTER OF THE LINGUISTIC FORMS 69

on relations common in experience, universalize these, so

that they must be formally expressed even where they
are not actually present or where there is no occasion
for focusing them, even though they are present. We
must express actor and action in a sentence and tense
in a verb even where they are not very vivid in the total
experience, — where, respectively, a Latin or a Chinese
speaker could ignore them, just as we ignore numerous
unessential elements of every experience, and also—
where they are not present at all, as in Mount Blanc is
higK where the experience presents neither action and
actor nor any particular tense.
The normal speaker, however, blindly accepts the
categories of his language. If he reflects upon them at
all, he usually ends by supposing them to be universal

forms of thouojht. In linguistics, of course, we must be

careful to distinoruish between categories of a lanojuaoje,
be it our own or another, and the features of experience,
as apart from any particular language.
8. Psychologic character of the linguistic forms.
The categories of a language originate in the extension
of some oft-repeated type of expression. In this they
are like all linguistic forms. To the speaker they seem
fixed and universal forms of expression and even of
thought; actually they are habits of association in vogue
in a community. Owing to the similarity of dominant
elements, an experience awakens a series of past ex-
periences and is designated by the same word. Owing
to the uniformity of the process of analyzing a total ex-
perience, all such analyses, — that is, all sentences, —
may receive the form of certainnumerous past ones:
thus arise our syntactic categories. All words present-
ing certain common features, — belonging, for instance
to a certain class, — may take on formational features
70 THE MENTAL BASIS OF LANGUAGE

that corresponded to experience in only a limited part of

their occurrences, — such features as time-expression:
morphologic categories.
The best evidence of the purely associational nature
of linguistic forms lies in their change in history. The
word dog once meant ^mastiff'; it came, however, to
awaken predominantly the idea of dogs in general, with
the species, not the breed, as dominant feature, until it

became the universal expression for all these experiences.

At one time English sentences could be formed without
an actor and an action, but the process of forming a
sentence came, in the course of time, always to awaken
the process of forming actor-and-action sentences, until
became universal. Similarly, when a new action-
this type
word comes into the language, such as the German waits
or the Japanese hara-ldri, it recalls the verbs of our lan-
guage with their time-forms and unconsciously and imme-
diately submits to the morphologic tense- categories, re-
ceiving the past-forms waltzed, hara-ldried.
Thus language is not, as the sight of a grammar and
dictionary might lead us to suppose, a system of unalterably
fixed and indivisible elements. It is rather a complex set
of associations of experiences in groups, each of which
is accompanied by a habitual sound-utterance, and all—
these associations are, like all others, certain of displace-
ment in the course of time.
9. Psychologic motives of utterance. True to its
original form of an outcry under the most violent ex-
periences, language is most easily realized under emotion-
al stress. Some violence of experience must normally
be present to call forth loud expressiou. If this emotion-
al violence is the dominant cause of the utterance, we
speak of exclamation. Under the social conditions of lin-
guistic development utterance with predominantly com
INTERPRETATION UF THE LINGUISTIC PHENOMENA 71

municative motive, declarative utterance, is a natural sequel.

Likewise the question, an utterance expressive of uncer-
tainty or incompleteness of an experience, is a weaken-
dominance of the emotional motive, and a trans-
ing, as to
ference tocommunicative use, of the exclamation.
10. Interpretatiou of the linguistic phenomena. I
have troubled the reader with a psychologic description
which, though perhaps difficult, would have been aU the
more so, had there been appended to each step the ex-
amples from various languages that would illustrate the
specific linguistic phases of the phenomena in question.
The most important of these shall in the next chapter
be so iUustrated. After what foUows the reader may find
the psychologic description more intelligible, if he will
go back to it; so much however, that the phe-
is certain,
nomena themselves, without consideration of their mental
significance are unintelligible or rather, what is worse,
liable to a post factum logical interpretation which sub-
stitutes for the actual state of things our reflections upon
them.
The points of view from which linguistic phenomena
can be regarded are of course various. For those un-
familiar with them the greatest importance lies in the
realization that the categoric and other distinctions of
one's own language are not universal forms of expression
or of experience. important also to remember that
It is
the meaning of any linguistic expression is due to the
associative habits of those who use it. A deictic or a rep-
resentative gestureis intelligible at once, because it owes

itsmeaning to universal psycho-physiologic characteristics

of man. Even a suggestive or symbolic gesture hardly
ever fails of immediate understanding, for the constant
analogy of the simpler gestures predominates over associa-
tive transference. Vocal language, quite otherwise, though
72 THE MENTAL BASIS OF LANGUAGE

it has its origin in the direct reactions of our organism

to experience, is the result of a very different develop-
ment. The reactions which gave rise to it were reactions
of movement, but the effect which became of self-satisfy-
ing and of communicative value, was the acoustic effect
of these movements.
Consequently even the simplest
utterances furnished no analogy, comparable to that of
the simplest gestures, by which every kind of associative
transference and innovation might have been counteracted.
The result is that no languajye has the character of a set
of sounds in some way logically derivable from the ex-
periences which they express.
 CHAPTER IV.

THE FORMS OF LANGUAGE.

1. The inarticulate outcry. We have seen that oui
linguistic utterances are part of the expressive movements
which attend every experience. In many lower animals
also some of the expressive movements produce sound.
The bodily expression of experiences of pain, for instance,
may include not only a sudden withdrawal, but also a
contraction of the thorax forcing out breath through the
which, likewise contracted, produces the sound
glottis,
that we describe as a cry of pain. We have seen that
human language is a developed and varied form of such
vocal reflexes.
Even where language in the highest form exists, how-
ever, these most primitive reflexes occur by its side;
the inarticulate cry of pain or anger is uttered by human
beings under an extremely violent experience. As a di-

rect result of this experience, this cry has nothing to do

with any earlier experiences of the individual. It is in-
dependent, accordingly, as to its form, of the, utterer's
personal or social history: its sounds need not be speech-
sounds used in his community, and it is no more intelli-
gible in his speech-community than in any other: even
an animal may utter its like.
2. Primary iiiterjectior.s. It is only under the most
violent experiences that such purely reflex vocal utterances
are used by man. -If the experience is somewhat less rad-
74 THE FORMS OF LANGUAGE

ical, the vocal utterance completely dependent

is less
upon it alone, for, owing to the universal laws of habit,
the utterance now tends to take that form which the in-
dividual happens to have most used or heard under sim-
ilar conditions. This factor will, of course, vary accord-
ing to the earlier history of the individual. Another in-
dividual who has had, in this respect, the same history
and has, accordingly, formed the same habit of association,
will, on hearing the utterance, at once associate the same
experience: that is, he will understand. An individual,
on the other hand, who has not had the same history,
and has never heard the utterance in question, will make
no such association, and will not know what kind of an
experience the utterer is undergoing. Hearing the ex-
clamations of a Zulu or a Fiji- islander, we may be in
doubt as to whether it is joy, sorrow, anger, or surprise
that he is expressing.
Even in these less radical vocal expressions there is
some element of direct reflex. This appears, on the one
hand, in the rather extended intelligibility of these inter-
jections, as we callthem, and, on the other, in their occasion-
ally departing somewhat from the regular sound-system
of the language. An example of both features is the la-

bial trill, which is used all over northern Europe as an

expression of intense cold and of abhorrence, although
as a regular speech-sound it does not occur in the lan-
guages concerned; in writing it is usually reproduced as
hrrr! Similarly, various sound-complexes with the unusual
feature of a syllabic [s] or [JJ, written Sh"! or Fst! are
used as an urgent demand for silence. Our peculiar whist-
ling expulsion of breath, written Whew! to express ex-
treme heat as well as surprise, is another instance of di-
vergence from the usual sound-system. On the other hand,
interjections may remain within the usual sound-system
 SECONDARY INTERJECTIONS 75

and may also vary in the different communities; thus the

interjection of pain is in English ouch! and in German
an! This, indeed, by far the commoner case.
is

In the utterance of an interjection there is thus beside

the mere vocal reflex another element: the experience is
lived through as similar to certain earlier experiences,
and is accompanied by the same vocal utterances as were
these earlier experiences. We may say that these ex-
periences together constitute a class recognized by the
speech-community, in that they are always accompanied
by the utterance of these particular sounds. A certain
degree of pain might, for instance, be called in English
an 0Mc7i/- experience.
Secondcary interjections. Experiences less intense,
3.
— that is, having less predominantly emotional value, —
than those so far discussed, are accompanied by utterances
of more specific descriptive value While a person who
inadvertently got his hand into the fire might give an
inarticulate shriek, and one who o:ot his finojer blistered
might utter the interjection ouch! one who merely saw
,

a fire where he did not expect it, —

saw, for instance,
that a barn was burning, —
would utter the more delib-
erate and specific, though still exclamatory cry of FirCy
fire!
The more specific character of this utterance consists
in its perceptual value. In the inarticulate cry and such
interjections as oucli! only an emotional element of the
experience is expressed; in the utterance Fire! the sounds
uttered are associated by speakers of the language with
the specific perceptual content of fire. Exclamatory utter-
ances of this kind are called secondary interjections. There
isno limit to the amount of material detail which they
may contain. Other examples are cries of Help !^ Murder !y
Man overboard!, and thelilic; also exclamations describing
76 THE FORMS OF LANGUAGE

noises or movements, sucli as Bang!, Crash!, Snap!, Fizz!,

Puff!, JVJioop!, Rip! Here belong also utterances which
name the principal object concerned in the experience,
such as The child, the child!, Gold!, Forgery!, Mother!,
A shooting star!, A tvJdte rahhit! The calling of descrip-
tive names is, of course, also exclamatory: You thief!.
Villain!J Generous man! Of especial importance are com-
mands: March!, Get tip!, Bring me a glass of water,
'please!, or the use of people's or animals' names to call
their attention: stranger!, John!, Child!, Doggie!
The reflex element may here be present in various degrees
and find expression in modulations of pitch, stress, duration,
and the like. The modnlations so permissible are different
in different lanojuaojes: the articulations which form the
basis of the utterance, however, are in each case determined
by their association with the kind of experience concerned.
A foreigner does not understand them, because he possesses
an entirely different set of associative habits in this re-
gard. It will be noticed, also, that some of these secondary
interjections involve a considerable degree of discursive
analysis (though not, usually, a predication); in so far as
they do they are exclamatory sentences.
so,
The same articulations may be used at other times with
a minimum emotional content. A chemist, after lono; in-
vestigation of what a certain component of a preparation
was, could turn to his client or his pupils and, holding
up a test-tube, quietly say Gold. A lawyer, after some
consideration of the technical validity of a paper, could
say, with very little emotion, Forgery. The significance
of all these utterances, in other words,due not to the
is

emotional value with which they may be used, but only

to their association, in speaker's and hearer's mind, with
certain material contents of experience. This association
has to be formed by every member of the speecli-com-
 VALUE OF NON-INTERJECTIONAL UTTERANCES 77

m unity before he can speak or understand what is spoken.

It is only in the inarticulate outcry, and, to a lesser extent,
in the primary interjection, that the universal reflexes of
the human body undergoing an experience determine the
form of utterance; in the words of material content this
association is, so to speak, an external one and differs
greatly in the different speech-communities.
4. The arbitrary of non-interjectioual ut-
yaliie
terances. We saw in Chapter I how most new members
of a speech-community, namely children, are taught to
make these associations. The problem of the origin of
language, we further saw, resolves itself into the question
as to how these associations originally came into being.
The answer we found (p. 14, f.) was that the movement
which produces the sound was originaUy an expressive
movement, but, as the sound produced by the movement
was in communication the striking element, further devel-
opment proceeded from the sound and not from the move-
ment. As no essential connection between sounds and ex-
perience was felt by the speaker, transferences and changes
had free play, so that even between movement and ex-
perience there soon remained no recognizable connection.
For instance, the experience of a bitter taste produces a
very characteristic expressive movement of the facial and
oral muscles which, if the experience is violent enough,
may be accompanied by sound-production. The sounds
resulting from this expressive movement may have been,
in some time and place, the current expression for 'bitter'.
As time went on," however, there happened that which,
as we shall see, is universal in language: the manner of
articulating the sounds gradually changed until they were
very different from those formerly spoken. Even by this
time the movements which made up the articulation of
the sound-sequence were no longer those of the 'biiter'
78 THE FOIIMS OF LANGUAGE

But another, even more radical and equally

face-expression.
universal kind of change must also be considered: people
do not go on using the same expressions for ever. There
is a constant tendency, as we have seen and shall in
greater detail later see, to assimilate expressions to one
another when the experiences are at all alike. Thus our
expression for "^bitter' miglit be somewhat changed so as
to resemble the expression for "^sharp', or 'bad' or even
'sweet', for 'almond-like' or 'uneatable' or *^nasty'. Of
these processes we shall see many examples when we
come to speak of the changes of language. For the pres-
ent it is clear that the immediate physiologic connection

between expression and experience, which at some particu-

lar time must have existed in a great many expressions, can
in the case of no expression be of indefinite duration. The
English word bitter, for instance, cannot be interpreted
as an expressive movement, for we know that thousands of
years ago, if it then existed at all, it had some such form
as hhidrom and further that, whenever it began to be
used, it was not an expression arising directly from the
experience of a bitter taste, but rather a descriptive term
which meant literally 'biting', for it was originally an
adjective derived from the verb to hite. The expressive
habits of the community, in other words, are in a con-
stant process of change, and though, for language to be-
gin, was necessary that certain sound-sequences should
it

be called forth by certain stimuli, it was neither necessary,

once given this beginning, nor even possible that this
direct connection should continue to exist.
It may be asked, then, if there are in use to-day any
expressions which are still where there is a
at the stage
direct connection between experience and movement. If
we look into our own feeling with regard to certain of
our words, there might appear in English to be a great
 VALUE OF NON-INTERJECTIONAL UTTERANCES 79

many such. For instance, our words flame, flare, fliclcer,

flimmer, flash seem to us highly expressive of certain fea-
tures of the experience of fire. Other words that might seem
directly expressive are puff, fizz, hang, zip, diddle, snap,
smash, ivhach, sqiieaJc, and so on. We are very much surpris-
ed to learn that to a foreigner these words are as unintel-
ligible as any others, —
until, of course, he learns English.
Let us look a little more closely at these expressions.
In the words flame, flare, flash, flimmer, flicker we find,
corresponding to the common half-emotional, half-percep-
tual element of meaning, the common initial sound-group
fl-. In fla^re, flash, flimmer, and flicker the rest of the
meaning also seems to be directly and immediately ex-
pressed; and here again, if we look for words with simi-
lar meaning, we shall find the same sound-groups recur-
ring. Thus flare relates itself to glare and hlare. The
-icker of flicker, which expresses to our feeling the small
repeated movements of the flame, performs a similar
function in snicker. The -immer oi flimmer^ expressive to
us of a quiet, small, continued action, is similarly expres-
sive in simmer, shimmer, glimmer. In flash the sounds
-ash express to us a very diff'erent, more rapid and violent
kind of movement also conveyed in clash, crash, dash,
lash, mash, slash, smash, splash. Or, to leave our fl- words,
the articulation of h- in hang, hiff, hump, huffer, hox, heat
corresponds to a common element of meaning which, we
feel, is directly expressed by all these words. In the
common parlance of school-room and dictionary they are
'onomatopoeias'.
This peculiar feeling on the part of those who know
the language is in all probability, however, due to
nothing other than the words
existence of parallel
expressing the same shade of meaning with the same
sounds. When we utter any such word the other words
80 THE FORMS OF LANGUAGE

of similar meaning are awakened, and their similarity

of form adds corroborative strength to the impulse of
articulation. That is, if we had only flash and not the
other words in fl- and in -ash, it would not seem to us
any more aptly and immediately expressive of its meaning
than such terms as chaiVy throiv, conibustion. In short,
there is no ulterior connection between these words and
their meanings, or even between such formational sound-
groups as fl- or -ash or h- and the elements of meaning
conveyed by them. Even if it should be found with any
certainty that the movements producing these sounds
are, in a psycho-physiologic sense, the natural expressive
movements attending the experiences which they in present
English express, this would not alter the case. We might at
first wonder at the correspondence and then realize that a

selective process by which associations and assimilations

occur had favored in each case the most suitable articulations.
All this, however, would not change the fact that these
words, like others, are limited to their language and out-
side of it are understood no more than others,and that
these words have arisen and changed in the course of time
by exactly the same processes that aifect all words. The
peculiar feeling of directness of meaning which they give
us is due, then, entirely to the associative conditions of
our vocabulary and not to these words' being any such
thing as primitive reactions to experience: their history
is the same as that of other words. Aside from primary

interjections, the forms of language owe their function

entirely to their association with experiences in the speak-
ers' minds. The peculiar value in the speakers' feeling
of such expressions as the above, is colled soiind-si/mholism,
— a term which is useful, if we remember that the "^sym-
bolism' is such only within the expressive habits of the
given community.
 V^ALUE OF NON-INTERJECTIONAL UTTERANCES 81

There are still otlier cases in which there seems to be an

actual connection between the sounds uttered and the
experience, this time in the sense that the experience
contains a noise which is imitated in the expression. This
is especially the case in bird-names, such as ciiclcoo. Such

investigation as there has been shows that among the

Germans, for instance, there have been in use great num-
bers of bird-names explicable only in this way, —
that
is, as onomatopoeias. This, however, is not a general prin-

ciple, but only a special instance of the way in which

language is expressive. It happens that some birds,
— and there are probably few other such fields in human
experience, —are naturally recognizable by their calls,
and it is not surprising that, if the call became the dom-
inant element in these experiences, the expressive habit
of designating the birds by a more or less rough imita-
tion of it should have come into currency In English
this is far less the case, our bird-names being mostly de-
scriptive of the birds' appearance or habits {red-hreast,hlue-
hird, mocldng -hird) and, where an onomatopoetic name
J

seems to exist, its form is usually determined by associa-

tion with usual words of the lanojuage, as in the case of
Boh- White and uhip-poor- Will. The range of onomato-
poeia is thus at best very limited, and where it occurs
it can take I'ank only as one of the many forms of as-

sociational habit that occur in lancruaoje.

As we look then at inter-
first at inarticulate outcries,

jections, and finally at the words of ordinary speech, we

thus find a continuous gradation. The outcry is entirely
the product of the present circumstances, of the primary
and the utterance with
interjection this is not fully true,
material content depends for its form entirely on the
habits of the speaker, which he shares with his speech-
community. These habits are in a sense arbitrary, differing
Bloom field, Study of Language 6
82 THE FORMS OF LANGUAGE

for the different communities and changing gradually in

the course of time. A new member of a community must
learn its speech -habits as he would any other set of
communal habits.
5. The
classifying nature of linguistic expression.
The arbitrary nature of speech-expressions is directly due
which makes communication by
to the fruitful principle
means of any such expressions possible. If each speaker
reacted under each experience in such a way that no trace
of his earlier history affected the reaction, communication
would be impossible. No two speakers would ever react
alike and no one speaker would ever react twice alike.
Fortunately we are so constituted that our past does
unceasingly modify our present: a present experience is

inevitably assimilated by past ones of a similar nature

and is attended by the same or similar expressive actions
as were these. Thus the circumstance that an English-
speaking person and a German will express similar ex-
periences, respectively, by horse and Pferd, an arbitrary —
divergence, —
due to the very fact that each expression
is

is moulded by the past history of the speaker. The one

has heard and spoken horse when such an experience oc-
curred, the other Pferd.
The identity of the several experiences that are in each
case designated by the same expression (e. g. horse or fire)

is not actually inherent in them. This is obvious, if we

recall the psychologic truth that no two experiences,
whether belonging one person or to different persons,
to
are ever exactly alike. When we express each of a great
number of experiences by the sound-sequence fi^re, we are
associating them on the basis of an only partial similarity.
In our survey of the sounds of speech we saw that
language would be unintelligible, if all of the infinity of
possible sounds were employed, that the difficulty of
 CLASSIFYING NATURE OF LINGUISTIC EXPRESSION ^3

understanding would grow as this infinity were approacliea,

and tliat actually each comm unity uses only a limited
number of the possible sounds; that this limitation alone
makes possible well-fixed habits of articulation and hear-
ing. We are now again face to face with this principle.
If each experience, owing to its indisputable individuality,
were to be accompanied by a special utterance, no sound-
sequence would ever be uttered more than once, and
communication by means of speech would be impossible.
It is the habitual inclusion under one form of expression,
— that under one specific sound-sequence,
is, of vast —
numbers of experiences presenting certain dominant fea-
tures, which enables us to understand one another.
We are so accustomed to think and express ourselves
in the terms of our language that we are not ordinarily
conscious of the subjective character of this inclusion or
classification.Only the poet, who looks directly at the
experience and seeks for an exact expression of it, must
constantly realize this fact. Science also, on the basis of
objective analysis, can make an extended classification of
experiences and then arbitrarily determine that a given
expression shall be used whenever certain features are
present: this, of course, is the process of scientific defini-
tion. In ordinary no such analysis is made: certain
life

general, often very complex features are associated with

the expression and all experiences in which these features
are dominant are classed together and expressed alike.
Yet, even in ordinary life, there are circumstances when
the uncertain character of our classifications is thrust
upon our notice. — and that is in the face of some novel
experience. A man who for the first time confronts a
phenomenon which, let u? say, looks like fire but gives
out no heat, or one that presents a different exterior,
being, say, a liquid, but produces the same charring eflect,
6*
84 THE FORMS OF LANGUAGE

combined with smoke, as a fire, —

this man will ask 'Is
this fire or not?' —
or, if he is more philosophical, 'Am
I to call this fire or not?' The answer to the question

must come, if it be given at all, from the consensus of

the speech- community, which may or may not in turn
call upon a scientific definition to settle the usage by
determining a logically recognizable dominant feature.^)
The subjective character of our speec'h-classifications is
brought home most of all, however, by the study of
language itself; for here we constantly find that difi'erent
speech- communities make very difi'erent classifications.
There may be languages, for instance, where no such
classification as 'fire' is made, but where there is an en-
tirely different expression for each of such classes as
'camp-fire', 'cooking-fire', 'forest-fire', and so on: in such
a language experiences which we should regard as falling
into a single class would fall into several distinct classes.

In other words, a number of experiences that are classed

together in one speech -community may not be classed
together at all, or may form
but a small part of a larger
class, or may be in some other way distributed in another
speech-community. All depends on the expressive habits,
— that is, on the linguistic tradition, —
of the speech-

1) The vagueness with which these dominant features may

be defined is the motive in the anecdote of the traditional
Irishman who for the first time in his life saw a parrot. It had
escaped from its owner and perched in a tree, which the Irish-
man at once climbed. As he was about to lay his hand on the
parrot, it exclaimed 'Hands off! Hands off!' The Irishman was
dumbfounded, raised his hat, bowed, and said, 'Excuse me, sir;
I thought ye were a bird.' —
That is, speech was for him a
dominant feature of human beings, dominant even to the exclu-
sion of factors of visual appearance. General usage could have
corrected him by changing his associational habits, —
the science
of zoology , by giving him criteria of logical validity.
 CLASSIFYING NATURE OF LINGUISTIC EXPRESSION 85

community. It is especially important to remember that,

except for the case of terms of purely scientific character,
this classification is due to associative tendencies and is
not afi'ectedby any logical considerations which individual
speakers may undertake. People of a nation whose language
had no expression for ^fire' but only for ^camp-fire', ^forest-
fire', 'cooking-fire', and so on, might know very well that

all these have certain features in common, and might

even study physics and chemistry and arrive at the scien-

tific concept of combustion, — but their language would
remain the same It would provide, always in accordance
with its existing habits, some analytic expression, such
as 'camp-fires, kitchen -fires, forest-fires, and the like',
which would be used for the scientific concept of 'fire'.
This may be illustrated by a few actual instances.
In Malay the experiences which may be logically de-
fined by us as 'ofi'spring of the same parents' are classed
together, and for such an experience is used the word
sudara. In English we form no such class; we form two
classes, according to the sex, and speak of a hrotJter or a
sisler. Now, it would be manifestly absurd to say that

a Malay does not know his brother from his sister; it

would be no less absurd, however, to say that English-
speaking people are unable to form the general idea con-
veyed by the Malay word. Both languages can express
the experiences for which no single designation exists by
a compound expression which analyzes them, — the Malay
by saying sudara lalcilalci and sudara perampuwarij where
the added modifying words resemble our terms 'male' and
'female'; and the English by saying brother or sister or
child of the same parents.
There are still other possibilities. In Chinese the ex-
periences of which we are speaking fall into four classes:
['^^lurj"', ^ti^, '^tszo"', ,^mei\]. The first two denote males, the
86 THE FORMS OF LANGUAGE

second two, females; and in each of these pairs the former

denotes an older, the latter a younger member of the
family. While we make no such classes, we can analyti-
cally designate these relatives by saying older brother,
younger brother^ older sister, younger sister. The Chinese,
on the other hand, can express the idea of ^brother' by
saying ['^9iui)^ L^i\], of ^sister' by saying [""tsza^ L^ei\], and
of the Malay sudara by [^9iug'' Lti\ '^tsza"' Lmei\], — all of
which expressions are comparable to our expression of
the Malay term by brother or sister. It would be as absurd
to say that the Chinese classification shows the Chinese
to lack power of generalization or else to have a partic-
ularly strong feeling for relationship as it would be to
say that we have less power of generalization than a
Malay or more feeliug for the difference of sex; or else
that we have little feeling for the distinction between
older and younger brothers and sisters, —
when, to take
the last point, English law has from time immemorial
made much of it.

If any final demonstration were needed of how inde-

pendent linguistic classification is of logical insight, it
would be furnished by the German form of these words.
This language, when speaking of one person, makes the
same classification as English: Bruder, Schwester, but
when speaking of more than one, makes also that of the
Malay, using the term Geschwister, for experiences which in
English would have to be analyzed into brothers and sisters,
brothers or sisters, brother and sister, brother and sisters,
brothers and sister, as the case might be. It is evident that
whatever hasty conclusions were drawn from the contrast
between the Malay and English expressions would have
to be applied in turn to one and the same German, from
moment to moment, according to the number of people
he happened to be talking about.
 CLASSIFYING NATURE OF LINGUISTIC EXPRESSION 87

translations given for the German Ge-

The English
however, show that, where a classification is not
schwistery
made and an experience is instead expressed by some
analytic phrase, the analysis is constantly open to the
speaker. If the expression is very frequently used, it may,
to some extent nevertheless become mechanized, and need
not involve the entire conscious analysis every time it is

used.
Afew more instances of divergent classification may
be of value. The general word in English for locomotion
is go, in German gelien. To begin with, however, while
we can say I go, a German cannot say ich gelien^ but must
in this connection use a slightly different form, gehe: ich
geJie. Aside from this, the German word is more inclusive,
in that it is used also of the specific form of locomotion
separately classed in English as walk. On the other hand,
our word ride is more inclusive than the German terms
reiten, used of riding on the back of an animal, and fahreUy
of riding in a vehicle or vessel. A black horse is in
German Eappe, a white horse Schimmel] compare our
hay, roan, sorrel when used as nouns. The relation ex-
pressed by our on in on the tahle is in German auf, but
that in on the wall is in German an: auf dem TiscJi, an
der Wand. It will also be seen from this example how
our word the corresponds to an element variously expressed
in German. In French there are no simple expressions
corresponding to our stand or sit\ the idea must in each
case be analyzed into etre debout (assis) ^be upright (sit-
ting)', r ester debout (assis) 'remain upright (sitting)', se
tenir debout (assis) 'hold oneself upright (sitting)'.
Even pronominal expressions (p. 64), in which the simple
deictic value might lead us to expect entire uniformity,
differ greatly. Three 'persons', that of the speaker, the
one spoken to, and the person or thing spoken of, are
S8 THE FORMS OF LANGUAGE

everywliere distinguished. Some languages, however, use

material object-words instead of the first and second per-
sons; so the Malay for ^1' saJiaya 'companion', liamhaj
hetttyor pateJc 'slave', rather than the purely pronominal
aJiU,and for 'you' rather the name of the person addressed
or tuivan 'master' or datoh 'grandfather*, than anlcau 'you'
In Japanese such object-expressions are exclusively used,
no purely pronominal terms for 'you' and being known. T
Similarly, Polish uses pan 'gentleman', 'sir', pani 'lady',
'mistress', 'madam' to all but intimates and servants,
rather than ty 'you'. Other languages identify different
persons: thus the Italian uses ella or leiy literally 'she',
'it', for ^ou', the German similarly Sie 'they' for 'you';
these pronouns originally referred to such nouns as 'your
grace', singular and plural, and are thus results of the
preceding type of usage. All these forms had their origin
in polite phrases. The same was once true of the English
you: it was the plural, politely used instead of the singu-
lar thou, — a use which finds its parallel today in the
French vous instead of singular and the Russian [vi]
tu
instead of singular [ti]. In Italian, German, and French
the substitute-forms are almost universal, the old words
for 'you' (singular), —
German du^ French tu, Italian
tu, —being used only to intimates, children, and in prayer.
In the plural some languages differ from ours in distin-
guishing two kinds of 'we', one including, the other ex-
cluding the person or persons addressed: thus, in Malay,
inclusive lata, exclusive Jcami
Related to this is the expression of varieties of deixis,
such as the 'here' and 'this', the 'there' and 'that'. In
this, too, languages differ somewhat. In the Scotch dialects
of English three types of deixis occur: not only a 'here'
and a 'there', but also a 'yonder', and not only a 'this'
and a 'that', but also a 'yon'. Likewise in Latin one used
 CLASSIFYING NATURE OF LINGUISTIC EXPRESSION 89

hie for an object near one, ille for one farther off, and
iste for one near the person addressed; in German, too,
one says hier 'here', da 'there', and dort 'yonder'.
Beside the deictic expressions most languages distinguish
anaphoric reference mention of things known or spoken
:

of, as, for instance, in English: he^ she, it, they] other

languages make no distinction between anaphoric and

deictic reference. Within the anaphoric relations a single
instance may be cited of a distinction absent in some
languages (including English) but observed in others;
namely, the distinction between anaphoric reference to
an object immediately concerned and that to another
object. So in Latin: Amat sorbrem suam 'He loves his
sister', that is, his own sister, but Amat sorbrem eius 'He
loves his sister', that is, someone else's (who has been spoken
of) sister. Similarly in Norwegian 'he took his hat' is

Han tok sin hat, if the hat belongs to the one who took
it, but Han toJc hans hat, if it belongs to someone else.

The same distinction is made in the Slavic languages.

A striking example of differences in classification is

furnished by the numerals. In most languages the numbers

are divided, as in English, into series of ten, the multiples
of ten receiving analytic expression: the decimal system.
This had its counting on the fingers,
origin in an —
origin plainly apparent, also, in the quinary or fives system
of the Arowak, a Carib language, in which the expression
for 'five' is the same as that for 'one hand', aha-teJcabe,
for 'ten' as for 'two hands', hiaman-tehabe'^ that for 'fifteen'
means 'one-foot- toes' (sc. 'added'), aba-maria-lcutihihena,
while 'twenty' is 'one man', aha hilcu. Our peculig-r words
eleven and twelve (instead of oneteen, ttvoteen) mayb e traces
of a duodecimal system with which speakers of English
may have come in contact in prehistoric times. In French
one counts from sixty twenty units to eighty: 'sixty-nine,
90 THE FORMS OF LANGUAGE

seventy, seventy-one' are soixante-nenf, soixante-dix ('sixty-

ten'), soixante-oiize ('sixty-eleven'), and so on; 'eighty' is

quatre-vingt ('four-twenties'). This is a trace of a vigesimal

system, probably used by the prehistoric inhabitants of
France. At any rate, in the Basque (which probably rep-
resents the speech of prehistoric times in this part of
Europe), the vigesimal system prevails, though the dec-
imal has encroached upon it. Thus 'twenty' is hogeij
'twenty-one' Jiogei-ta-hat, 'twenty-two' hogei eta hi, 'thirty'
Jiogei eta Jiamar ('twenty and and so on, while 'forty'
ten'),
is he ogei, 'sixty' hirur-ogei ('three twenties') and 'eighty'

laiir hogei ('four twenties'). Wild peoples who have little

occasion for systematic use of numbers, often have less

extensive systems. Thus the Kham [t*kham] Bushmen in
South Africa have a trial system, with words for 'one',
'two', and 'three'; higher numbers are expressed by com-
binations: 'four people' are 'two people, two people', 'five
people' are 'two people, two people, one person', or —
else one simply uses the word for 'many'.
In short, just as each language uses only a limited set
out of the infinity of sounds possible to the human vocal
organ, so each language divides the infinitely various
experiences of life into a limited number of classes within
each of which all experiences are named by the same
expression. The classes so recognized by the different
languages are, as we have just seen, very different. It
need hardly be said that the description of the various
experience -classes and of the sound-complexes used to
express them, constitutes the lexicon or dictionary of a
language.
6. Expression of the three types of utterances.
There are, as we have seen, three types of psychic con-
ditions underwhich speech occurs (p. 70). The simplest
and most fundamental one is that in which an experience
EXPRESSION OF THE THREE TYPES OF UTTERANCES 91

by its violence forces a sound-producing expressive move-

ment. The most typical instance of this is the insuppress-
able cry of pain or rage. Almost as characteristic are
the circumstances under which the primary interjections
are uttered, and finally, the endless variety of expressions
which may be used as secondary interjections. All these
utterances, in which the dominant motive is the emotional
stress contained in the experience, are exclamatory utter-
ances. We
have seen that language must have had its
beginning in these, since it is a developed form of ex-
pressive movement (p. 9).
We have also seen that there is no fixed boundary
between an exclamatory utterance and one in which the
emotional prompting is at a minimum and the communi-
cation of a material content is the determining motive,
as in the chemist's Gold or the lawyer's Forgery (p. 76).
Most of our speech today is of the latter kind, declarative
utterance. Some emotional tone is, to be sure, present in
every experience, and the minimum of emotional tone
must be greatly exceeded before the experience wiU receive
loud expression, but the declarative utterance i» always
chiefly prompted not by the emotional content itself but
by some material content connected with sufficient value
to bring about utterance.
Finally we have interrogative utterance, unified by the
peculiar emotional tone of doubt or hesitation at the ac-
ceptance of an experience into a particular sphere. In
this form also the emotional tone may be so great that
the utterance merges with the exclamatory type, as in
What!? —
Gold!? —
Forgery!?
The constellation under which an experience receives
expression always modifies the form, though it may do
so in the most diverse ways. In English, for instance,
interrogative and declarative utterances are distioguished
 ^

92 THE FORMS OF LANGUAGE

not only bj pitcli-modulation, as in Your father has gone

out (even, then falling pich) and Your father has gone
out? (rising pitch), — see p. 51, f. — but also often by
word-order, as when the interrogative of the preceding is

Has your father gone out? as well as by particular question-

,

words: Where is your father? (rising, then falling pitch).

In Latin the three question -words ne, nmney and num
have no content except that of expressing the interroga-
tive situation, and the same is true of the Slavic li, —
e. g. Russian [zda'ro'va I'i vaja'ma't'] ^Is your mother

well?', literally 'Well {li) your mother?' and of the —

Chinese [i^mo/].
On the other hand, very much the same pitch-modula-
tion that with us is expressive of interrogation is in
Norwegian usual in declarative utterances. Similarly, ex-
clamatory sentences have in English a peculiar pitch-
modulation of greater ranore than that of other utterances,
but Italians use a very similar modulation for declarative
and interrogative speech, which makes them in our ears
seem to be excited when really they are not. The accom-
paniment of the utterance by a primary interjection may
also be used to express exclamatory value, as n stranger !
Oh, come on! The names of persons or animals used as
secondary interjections, to call them, have in many lan-
guages a particular form when so used, called a vocative'^
e. g. Latin Fill! 'Son!' (otherwise, for instance, FUius
ahcst 'The son is away') or ancient Greek Pater 'Father!'
(as opposed, for instance, to Pater apen 'The father was
away'). An action expressed exclamatorily as desired or
commanded has in many languages a particular form for
this use, an imperative, as in Latin Audi! 'Hear!' or Da!
'Give!' (as opposed, for instance, to Aiidis 'Thou hearest'
or Dare vidt 'He wants to give').
7. The parts of utterances. We have so far in this
 THE PARTS OF UTTERANCES 93

chapter been speaking of utterances as units and ignoring

the fact that most of them consist of definite parts
(p.60,ff.).
Many of the utterances of which we have spoken are,
in fact, indivisible, —
for instance, Ouch! or Fire! or
Gold! or in Malay the word sudara or each of the four
Chinese words ['^^lurj^ L^> "^tsze^, Lniei\]. They present
the simple instance of a sound-complex used in its entirety
for the expression of an experience lived through as
falling into a class with certain earlier ones.
Many other of the utterances I have quoted are, how-
ever, more complex, containing formational elements
(p. 62). The English word flash, for instance, is felt to
belong, on the one hand, to a group with flamey flare,
fliclcr, flimmeVj on the other, with clash, crash, dash, slash,
etc. This word is, to be sure, used repeatedly to express
a certain type of experience; but to this value is added

another factor: it on the one hand,

relates the experience,
to such as would be expressed by flame, flare, and so on,
and, on the other, to such as would be designated by
crash, dash, slash, and the like. It does this subtly, without
analytic consciousness on the speaker's part, and yet cer-
tainly, as is shown by the peculiar feeling of pregnant
significance (p. 79, f,). Or, to take one of several other
instances of formationally composed words that have oc-
curred, the German gchcn ^go' or 'walk' relates the ex-
perience, on the one side, to that of gehe in ich gehe U
walk' and other similar forms, and, on the other side, to
reiten, faJircn, and many others with final -en and the
meaning of general verb-forms.
That is to say, beside expressing the classification of
the experience with those past experiences with which
it is unconditionally thrown into one class, these utterances
at the same time imply that the experience is similar to
94 THE FORMS OF LANGUAGE

a number of others, — this implication being made by

a partial similarity of form. We thus obtain, beside the
total assimilation of experiences into a single class or
word, a grouping of such single classes into larger and
looser classes, the 'morphologic word-classes'. The associ-
ational character of the grouping appears in the fact that
we cannot, for instance, say -dsh for a violent movement
or fl- for an experience of fire, and so on: these are for-
mational elements, not words (p. Q2). Though the value
— especially the emotional value — of these words is

due very greatly to the associations which their formation al

elements express, the normal speech-feeling, no matter
how often it associates these words with one another,
never stops to analyze them. Such utterances as flash or
father (p. 62), therefore, though composed of parts, are
nevertheless conceptually units.
The unity of such expressions as these may outweigh
the divisibility in various degrees. In the case of flare,

flashj fldmmer, flicker^ flame the sound-complex //- is a

formational element, the expression of a similarity of the
the experiences, which can never occur alone. In fact, it
can not be added at liberty to any other utterance, but
occurs fixedly and exclusively in certain words. What is
more significant stiU, the same sounds occur in other
words, such as flow, with a different value
float, fly, flutter,

entirely, or, at any rate, if there is association with our

first set of words, in a much extended and vaguer value.

If, now, we look at an English expression, such as the

plural fires, the parts at once appear to possess a much

greater degree of independence than in the instances so
far mentioned. Even the normal speaker feels at once
that the first, lai-ger part, fire-, of the expression is iden-
tical with the singular, fii'e, and that the last part, -5 [z],
is identical with the same sound in other expressions,
 THE PARTS OF UTTERANCES 95

such as fathers, hoys, and with the similar sounds

sisters
[s] and [9z] in cats, ropes and tvatcJieSf peaches^) In fact,

even the normal speaker would not need to think long

before he could define the common element by saying
that the -s expresses plurality. Nevertheless, as the -s

cannot be used in this sense in certain words, such as

man, deer, goose, foot, —
and, further, as it could not be
used independently in the sense, let us say, of 'several'
or 'many', it is but a formational part of the expression
fires, even if a more independent part than, say, the fir

in flash.
In the possessive father's the first and larger element,
father, has as much independence as fire in the last in-
stance, but the second element, -5 [z], has more. For,
beside occurring also, with the same value, in such ex-
pressions as hoy's, Mng^s, man's, it may even occur with
some measure of independence, as in the King of Eng-
land's son and the man I saw yesterday^s father. Never-
theless its independence is not complete. One who said
's', meaning some such- thing as 'possession' or 'belong-

ing', would not be understood, nor is the speaker of

English, no matter how conscious he may be of the value
of the possessive 5 as a part of the larger expressions,
ever tempted to essay this independent use.
Another type of the same phenomenon is illustrated
by Turkish plurals, such as kullar 'slaves', evler 'houses'.
The Turkish speaker could not use lar or -ler alone in
some such sense as 'several', any more than the English
speaker could so use his -s. Moreover the vowel of this
element is a, if the preceding part of the word has a

1) Owing to the similarity of the writing and to the autom-

atism of the sound-variation, the normal speaker is not cou-
Kciv/ua of the diflFerence between the endings [s], [z], and fazj.
96 THE FORMS OP LANGUAGE

back vowel, e,has a front vowel,

if it obviously an —
indication of the unity of the whole expression in the
speaker's analysis of experience. On the other hand, side
by side with this dependence, there are features which
show the sound-sequence have a more in-
-lar or -ler to
dependent value than the English plural-suffix; most im-
portant among them the fact that, if the plurality is
otherwise expressed, the suffix is left off, as in dort adam
^four men', not dort adamlar.
Of a different character, again, is an English expression
such as thirteen. The transparency of the meaning, due
to the association with such forms as, on the one hand,
fourteen, fifteen, sixteen, etc., ten, tenth, and, on the other,
three, tJiird, thirty, makes it certain that every speaker
feels the thir- to mean the units above ten and the -teen
to mean the ten. Nevertheless, no one would say inde-
pendently, thir instead of three or teen instead often. Yet
cases like sixteen, seventeen, nineteen, where the first part,

six, seven, nine does occur independently, make the second

part so distinct in the feeling of speakers that we have
come to speak of 'a girl in her teens'. The formational
element -teen is more nearly independent, therefore, than
any we have yet analyzed out of a unified expression.
If we look finally, at an English word like bulldog,
there can be no question, from the outset, but that the
elements hull and dog are used independently. Still, there
is a reservation: for hulldog does not mean ^a bull and
a dog', but only a certain kind of a dog that may be
supposed in some way to resemble a bull. The word hidl
independently used has never this meaning; it means a
'bull' and not 'like a bull', —
it is a noun and not an

adjective. In the expression bulldog, therefore, the element

hull is not fully independent, for, though closely associa-
ted with the independent use of the same sound-sequeuce,
 THE WORD: PHONETIC CHARACTER 97

its value is not exactly the same; the expression huUclog,

consequently, retains a considerable degree of unity; as
we shall see, it is technically a compound word.
The independence of the parts is even greater in a
Chinese expression such as ["^^ui)^ Lti] for 'brother'. The
elements ['^^ug"'] and [Lti\J occur independently in the
respective senses of 'older brother' and 'younger brother';
the unity of the whole expression consists only in its

habitual use, with this order of the parts, in the sense

of 'brother' or 'brothers' — a very 'loose' compound.
8. The word: phonetic character. An expression
in which the independence of the parts is fully realized
can no longer be said to have unity in the sense of the
preceding cases. The English expression older brother for y

instance, contains two parts, older and hrother, each of

which is used to designate a class of experiences and can
recur in this capacity in the most varied connections, as
in I am older, older men and my brother, Where is brother?
younger brother. Such elements of speech, independently
recurring as expressions of experiences viewed as similar,
are, of course, words (p. 62). It will be evident from the
foregoing illustrations of less independent elements ap-
proaching the independent use of words, that the word
is by no means a mathematically definable concept; in

fact it is sometimes very hard to decide what is and what

is not a word. It may
be a puzzle, even in one's own
lan^^uage, to decide whether an element can or cannot be
independently used. Does the usage in bull terrier j bull
pup, and a few similar instances justify us in setting up
an adjective bull and calling bidldog two words? Probably
not, for all these expressions may be looked upon as
compounds of uniform type, — but the point is disputable:
a dog-fancier who spoke of 'three terriers, two fox and
one bull' would be using bull in this sense as an inde-
'Blonmf ield, Study of Language 7
9^ THE FORMS OF LANGUAGE

pendent adjective, no different from small, large, wJiite, or

hlack.^)
It is also clear tbat the unity or plurality of words
used to express a given experience must vary greatly in
different languages. We have seen hovr what a Malay ex-
presses by sudara is expressed in Chinese by a very loose
compound of ,four parts, and in English by three inde-
pendent words: brother or sister What we express by the
vvord brother, a fairly close-knit unit, the Chinese express
by a compound of two parts, and the Malays by the
two words sudara laldlah. Finally, the Chinese unit [^ti^]
would be in English two words, younger brother, and in
Malay, —
where we might expect three or four words,
— again but one: adelz. To take another example, — our
expression / am eating meat, corresponds to the German
one of three words: Ich esse Fleisch, to the Latin of two:
Carnem edo, and in Aztec to a single fairly close-knit
compound word: NinaJcaJiwa.
It is especially to be observed that the unity of such
expressions as we found above to be unified, was in no
way due to any phonetic peculiarity in these words. If
we found be a single word, this
thirteen, for instance, to
was not due anything in the immediate phonetic form
to
of the expression, but only to the fact that thir- does not
occur independently. Likewise, where an expression con-
sists of several words, phonetic observation does not reveal

any pause between them. Indeed such pronunciations as

1) The written form of the expressions gives, of course, no

answer, for the graphic (p. 20, f.) separation of the words is only
a half-conscious and unscientific attempt at answering the question
we are here dealing with; genuine compound words may be
found in good English printing as separate words, as hyphenated
combinations, or run together as one word, e. g. bull moose, bull-
pup^ bulldog.
 :

THE WORD: PHONETIC CHARACTER 99

that shown by Latin versification of sometliiiig like

[korneido] for Carnem edo, or in English of [aetju] for
at you, show this most and it is safe to challenge
plainly,
anyone who does not understand a language, be it Eng-
lish or any other, to divide the current of speech into
words. The word, in short, is a semantic, not a phonetic
unit. It is only through a process of analyzing the mean-
ing that' people can come to distinguish the word-bound-
aries, as we imperfectly do in our writing.
Secondarily, however, every language does make some
phonetic recognition of the word: but this differs greatly
in different languages.
A language which shows little phonetic recognition of
word-boundaries is modern French. In a French sentence
there is no feature which shows where one word ends and
the next begins. The stress-accent, for instance, is not
distributed according to the words, but rests on the last
syllable of the sentence, or, in longer sentences, on the
last syllable of connected word groups (p. 48). On account
of this lack of phonetic word-boundaries French has been
called, par excellence, 'the language of the pun A good .

illustration is the couplet quoted by Passy in his Petite

phonctique^, 'P^gQ 22. The two verses are pronounced ex-
actly alike. They each read:

[ga la md da la 'rsin a la tur ma jia 'ni:m];

the word-division, however, is seen in the conventional

orthography
Gal,amant de la Reiney alia, tour magnanime,
Galamment de VArene a la Tour Magne, a Ntmes.
Queen, went, brave feat, gallantly from
*Gal, lover of the
the Arena to the Large Tower, at Nimes.' For the same
reason uneducated Frenchmen have great difficulty in sep-
arating their w^rds in writmg; Passy quotes an instance
7*
100 THE FORMS OF LANGUAGE

in wbich [59 syi 'sa:^ a vsk man miia 'zsl] *I am being

good with Miss (the governess)' was written by a child:
je sidsage avecmnne moiselj the conventional orthography
(and real word-division) being: Je suis sage avec Made-
moiselle. The one- sound utterance au [0] 'to the' is two
words, for it is semantically composed of the fully ana-
lyzable elements a [a] 'to' and le [la] 'the', the substitu-
tion of au whenever they come together being a' purely
phonetic automatism.
All this is some contrast to languages like English,
in
in which nearly every word has a high stress-accent on
one of its syllables (p. 49). Certain small words which
lack this stress, —
commonly, for instance, such words
as they a, is, in, and (p. 49), — we call enclitics, if they
are semantically joined to the preceding word (hasnH, let
Hm), and proclitics, if to the word that follows {a rabhit,
in Sjieaking)] they alone can offer difficulty as to the number
of words in a sentence. This clearness is increased by the
fact that we use an almost entirely different set of vow-
els in unstressed syllables from that of the stressed. It

is only the presence of stressless words that makes half-

way possible the pun which answers the question, 'What's
the difference between a rheumatic man and a healthy
man who by saying, 'One is well
lives with his parents?'
at some times and has a rheumatism others, and the other
is well at all times and has a room at his mother's'. It

wiU be noticed, however, that the boundary between words

is sufficiently marked by certain stress-relations to rob

such similarities of their full effect: in the latter phrase

our stress begins to increase with the m of mother's, in
the other the m
is weak and stress begins on the initial

vowel oi others. There is the same difference, for instance,

between a name and an aim (p. 46).
In Norwegian and Swedish all words not enclitically
 THE WORD: PHONETIC CHARACTER Id
or proclitically used have one of two pitch-melodies, ris-

ing or falling-rising Cechish has a stress-accent

(p. 51).
regularly on the first syllable, of all but enclitically used
words, and Icelandic has a similar habit; Polish stresses
almost always the next-to-last syllable of its words (p. 49).
In Chinese the phonetic recognition of the word is es-
pecially striking. Every word here consists of only one
syllable ending in vowel, [n], or [i]], and uttered on one
of a limited number of pitch-melodies (p. 51); the only
exceptions are enclitics.
If we look beyond the single utterance, we find another
set of phenomena involving phonetic recognition of word-
boundanes. These phenomena may be described as sound-
variation in word-initial and word-final, and are spoken
of by the name which the grammarians of ancient India
gave them, sandhi. The beginning or the end of a word
often varies phonetically according to the phonetic char-
acter of the preceding or the following word. In Eng-
lish, for instance, the word you [juii] or [ju] when
coming after a final [t] is pronounced [Ju], and after
e. g. won't you, did you.
a final [d], [^u], We thus find
one word occurring with three different initials, a —
variation which does not occur within any word, and
therefore marks phonetically the word-boundary. The
most familiar example of sandhi is the so-called ^liai-
son' of French. The word vous *you', for instance, is
[vu] except before a word clusely connected in mean-
ing that begins with a vowel, where it is [vuz]; thus
vous avez 'you have' is [vu za Ve] but vous faites 'you

make' is [vu 'fst]. Such a variation without change of

meaning, as that between [vu] and [vuz] occurs only at
the end of words and is therefore a sign of active rec-
ognition of the word-boundary even in French. Another
ttf the many instances is the word a 'has' [a], which be-

\:r'
102 THE FORMS OF LANGUAGE

fore a semantically closely joined vowel-initial becomes

[at], written a-ty e. g. elle a sonne'^hQ rang' [s la so 'ne],
but a-t-elle? ^has sbe?' [a 'tsl]. The most extreme instance
of the use of sand hi, at least in writing, is Sanskrit, the
language from which the name of the phenomenon is
taken; here the end of every word has a number of forms
that appear according to the nature of the following in-
itial, which also is sometimes aflPected. Thus: devah pa-
tati 'the god falls', devas tatra Hhe god there', devas ca-
rati 'the god wanders', deva eti ^the god goes devogacchati
',

'the god walks', and, with change also of the folio wing
initial, before atra 'here', devo Hra 'the god here'. Sandhi,

however, does not imply so vivid a recognition of the

word as do those features which appear in each single
utterance; for sandhi makes itself felt only when several
utterances containing the same word are taken in view,
and under these conditions the very reappearance of the
word already constitutes such a recognition.
There always a tendency, when a word has several
is

sandhi-forms, that these may come to vary not in auto-

matic sound- variation, according to >the character of the
preceding or the following sound, (as is the case in San-
skrit), but that the difference of form may come to imply
some semantic difference. A transition to the latter type
is the French liaison, which limits the longer forms, such

as [vuz] and [at] to occurrence before words closely con-

nected in sense. An instance still farther along toward
semantic differentiation occurs in Irish. This language
has a sound-variation in word-initial which, however, does
not depend upon the phonetic character of the preceding
word-final, but arbitrarily on the preceding word; that
is, Irish words may
be divided into a number of otherwise
arbitrary classes, according to the effect they have on a
closely following word-initial. Examples are: td ha 'there
 THE WORD: SJ^MANTIC CHARACTER 103

are cows' but a va uv 'an egg', an tuv 'the

'his cows';
egg', na nuv a huv
'of the eggs','her egg\ This variation
has semantic value in that it does not depend automati-
cally on the adjoining sounds but implies a division of
words into classes, —
in this instance, however, not of
the words in which the variation occurs, but of the words
that may closely precede them. A great many other se-
mantic classifications as we shall see, are expressed by
sound- variations and affixed sounds in almost all languages:
in so far as these sound-variations and affixations affect
either the end or the beginning of words, they involve,
of course, a recognition of the word as a unit.
9. The word: semantic character. The word, then,
is not a phonetic unit, but is to be defined as a semanti-
cally independent and recurrent element which can be
dealt with as a conceptual whole. We have seen that,
in spite of this, a language may recognize within its words
a relation to other words of partially similar meaning.
This relation expresses itself, as we have seen, by partial
phonetic similarity, as in flame, flare, flimmer, flash or in
flash, crash, dash, etc., or in fathers, hoys, fires. It may,
however, receive no phonetic expression, but inhere en-
tirely in a parallelism of use, especially as to categoric
distinctions, as in the plurals fathers, men, geese, children,
or the verbs, present tense, third person singular, eats,
is, has, may, can. Here there
no phonetic similarity
is

between the forms, but their function with regard to the

English categories of actor and action, number, and tense
(p. 68) is in each group uniform. The formational elements,
as we have seen, may stand in various degrees of de-
pendence, from the comparative unity of flash, clash, and
the like, where the normal speaker is unoonscious of the
relating values, to such comparative independence as that
of the English possessive -5 (p. 95), or of the members
104 THE FORMS OF MNGUAGE

of a compound like hulldog, whicli, in slightly diyergent

use, occur as independent words. Semantically, the ele-
ments can be of the most various significance, from the
almost purely emotional tinge of /?,-, cl-j -ash, -immer to
the explicit relational value of a plural or possessive sign,
or the material explicitness of the elements of compounds
such as hulldog.
Different languages vary, of course, widely in the mean-
ings of the formational elements into which imperfect
analysis divides a word. The greatest complexes of se-
mantic elements in single words are found in the attribut-
ing languages, where every word is an object-expression
(p. 64). For here the expression of experiences of action
and quality cannot dissociate these elements from objects;
one cannot say Vhite' but only 'white-rabbit' (as a single
word) or, at best Svhite-thing', and cannot express 'runs'
or 'running' except in 'rabbit's-running' or 'running-
rabbit' (again, as a single word) or, at best, in 'he-runs',
'running-thing', 'his-running'. Consequently, any expres-
sion of quality or action must be in a word containing
these elements together with that of an object. We find
such words, therefore, as the Greenlandish [tusaRp-a-Ra]
'hearing-his-mine', that is, 'I hear him' the action is viewed
:

as an object possessed by the actor and by the object

affected, or, to put it more justly, the actor is expressed
as an object possessing the action. Similarly, where qual-
ity is to be expressed, it appears as an element of the
word which also expresses the object that has the qual-
ity, thus 'liar' (or 'he is a liar', cf. p. 64) is [sal:uto:q],

'big liar' [sahutoiqaoq]. Only an object-experience can be

independently expressed, as in [qimieq] 'dog' (or 'it is a
dog'). All this corresponds, aswe have seen, to the con-
crete facts of outer experience, where we never meet qual-
ities or actions apart from objects. The same objective
 THE WORD: SEMANTIC CHARACTER 105

character of tlie word is responsible for the inclusion of

objects standing in some relation to another object, for
these, too, are qualifying elements. Here belong such
words as [ihu-a] 'house-his', 'bis house' and [qitoRna-Ra]
"^child-mine', 'my child': the 'he' or 'I' as possessor is, of
course, not present in the actual experience as an object;
all that is tbere present is the house or the child as-
sociatively standing out as 'his' or 'my' possession: the
inclusion in the same word is, therefore, concretely jus-
tified. Thisis true also of such words as [kia-gu-n:eq]

'heat-suffering-result', 'perspiration', —
for here the heat
and the suffering are not objects figuring in the experi-
ence, but are associatively presented features of the 're-
sult'.Our abstract relational words, finally, are, of course,
by no means found in such a language, where the relation
is expressed as an associative feature of the object. Thus

our *in' appears in [nuna-me] 'land-in', 'in the land', our

'across' in [nuna-k:ut] 'land-across', 'across the land', and
our conjunction 'when' in such a form as [tuawioR-toR-
s:'uo-l:u-ne aneRlaRpoq] 'hurry-using-very-when-his he-re-
turns', i. e. 'hurrying very much, he goes home'.
In many languages which, like our own, are not con-
fined to this objective expression, we find, nevertheless,
frequent inclusion of several partly analyzed elements
under one word. It is possible that extended investigation
wiU determine that these features are always, as they
surely sometimes are, traces of an older objective habit
of expression.
The inclusion of qualities of an object in one word
with the object, as in the Greenlandish [sal:uto:-qaoq] 'liar-
big', 'bigliar', appears in 'diminutive', 'augmentative', 'pe-
jorative' and similar formations, as in the Italian sorellina
'little sister' beside sorclla 'sister', lihrone 'big book' be-

side lihro 'book', tcmpaccio 'nasty weather' beside tempo

106 THE FORMS OF LANGUAGE

'weather', or the German Mdnnchen or Mdnnlein 'little

man' beside Mann ^man', compare our manildn. Such

forms are common in many languages, especially the
Romance, Slavic, and Baltic (Lithuanian). The value,
especially as to emotional tone, of these formations is differ-

ent enough from that of the analytic expression by means

of adjective and noun to prevent interference. Much rarer
are compounds whose elements correspond to adjective
and noun, like the Sanskrit mdhd-dhanam ^big-booty'.
Such compounds are almost equal to the analytic ex-
pression, mahad dhanam ^big booty'. The only difference,
in fact, lies in the very slight tone of unity expressed
by the fixed order of the members and by the non-in-
flection of the one element, as in the plural maha-dhandni
^big-booties', opposed to the two-word mahdnti dhandni
^big booties'. The presence of genuine adjectives, which
tend to be awakened in the production of the sentence,
is the cause for the rarity of these forms. These genuine

adjectives are themselves probably sprung from nominal

expressions. In the oldest scientifically attainable stage
of English, Primitive Indo-European, the value of such
an adjective as 'white', for instance, seems to have been
'white-person' or 'white-thing' as often as the present
purely qualitative meaning. Thus in Latin, which is

another historic descendant of Primitive Indo-European,

adjectives are frequently used as substantives. So honas,
hona^ honiun mean not only ^good', but also, respectively,
'good man', 'good woman', 'good thing' or 'blessing';
juvenis means both 'young' and 'young man, youth' sa- ;

piens both 'wise' and 'wise man', and so on; this appears
also in some of the Latin adjectives borrowed in English,
as Germany Italian both noun (person of this nationality)
and adjective, — but not so the native English forms,
such as JEnglish, Danish, which are adjectives only.
 THE WORD: SEMANTIC CHARACTER 107

The verb, like the adjective, is in English an independ-

ent word, inclusive of the action-meaning only; sing or
sings or sang does not express the actor, even though the
first two of these forms can occur only with certain actors.
In Italian canta can he used as well as egli canta to include
the actor: 'he sings'; in Latin and many other languages
this is the regular usage, there being no other way of saying
^he sings' than the one-word expression cantat; the verb,
in other words, does not occur independently of an ob-
ject-element, namely that of the actor. AVhen we say in
Latin Puella cantat 'The girl sings', the latter word ex-
presses the idea not only of 'sings', but also of an actor,
'she-sings', more exactly defined by puella. This resembles
the expressions of an objective language, like the Green-
landish [takuwai] 'seeing-of-him-his', 'he sees him', which
reappears in its where the actor is specific-
entirety even
ally expressed: [qimiip takuwa:] 'to -the -dog seeing-of-
him-his', 'the dog sees him'.
Inclusion of objects in some relation to other objects
is also common. It appears most of all in compound words.

Thus 'bulldog includes in one word with the object Mog'

the other object to which the dog stands in an associative
relation, here that of resemblance; similar instances are
pay-day schoolboy, and the like. In many
sofa-cushiony ,

languages we find pronominal elements expressing these

relational object-ideas; so, especiaUy, in the Semitic langua-
ges, e. Egyptian Arabic dulab-i 'my cupboard', dul-
g.
ab-oh 'his cupboard', diddb-ha 'her cupboard', dulab-hum
'their cupboard.', and so on, like the Greenlandish [qitoRna
-Ra] 'my child'.
The abstract relational elements, finaUy, which pertain
to an object, are very extensively found formationaUy
combined with it. The extreme of this is seen in the
IJralic languages, as in F'nnish, for instance^ which has
108 THE FORMS OF LANGUAGE

twelve ^cases' expressing local relations, such as tlie *iii-

essive', e. g. silmdssd 'in the eye', the 'adessive' silmd-lld
'by the eye', the 'ablative' silmd-ltd 'from the eye', the
'comitative' silmd-ne along with the eye', and so on. The
case-forms of the more familiar languages are in part
of this type; in Latin, for instance, there are a number
of locutions in which the case-form expresses the object
together with a relational element. So especially the 'ab-
lative of means': manu 'by hand', lade vwunt 'by means
of milk they live'. Less common in Latin are
forms like
Homae 'at Rome', Bomam 'to Rome', Gallia 'from Gaul'.
The genitive or possessive case is another example: John's,
as in Mill's hat, expresses the possessive relation in par-
tial analysis in one word with the object-element, John.
But, as we have seen (p. 95), the analysis is almost equal
to that into a separate word, for we can use such turns
of speech as the man I saw yesterday's father.Even here
the use of the independent word of expressing the re-
lation ismore frequent: we say not the table's legs but
the legs of the table. The relational element of number,
also, is in most languages included in the object-expression,
as in boy: boys^ man: men. A language in which this is
not the case is Chinese. Here a word like ['^^au/] 'man,
men, people' expresses only the object, not its number;
only if the number is a vivid element in the experience,
is it expressed, and then by an independent word.

10. Word-cLiSSes. Partial analysis, such as just de-

scribed, is due to association of experiences with others
like them.Consequently, we may say that the words
containing a given formational element fall into a class.
Thus those English nouns which express, by means of
an element -s, plural number in addition to the object-
content, form a class, e. g. boys, fathers, rabbits, stones,
trees, fires, eggs, etc. Or, again, all the words containing
 WORD-CLASSES 109

a given material formational element, as hoy, hoy's, hoys,

boyish, boyishly. Where a relational element expresses a
categoric distinction, it is the basis of a class, even though
it has no uniform expression. Thus the English plurals
just quoted are only a smaller class within a larger one
containing also such forms as men, geese,
Jcnives, houses,

feet, children, oxen, etc., which have not the same plural-

formation and, in some cases, not even the final -5, but
fulfil the same function with regard to the grammatical

categories of the language.

We however, other word-classes which are not
find,
expressed by formational similarity at all, but seem to
go back, none the less, to emotional associations of the
speakers. The well-known three ^genders' of nouns in
German, Latin, and Greek, or the two of French andDano-
Norwegian are an example. To only a minimal extent do
these agree with any perceptual reality, such, for instance,
as animal sex. Thus in German two nouns for ^woman' will
be found in different genders: die Fran ^feminine' gender,
das Weib 'neuter' gender; similarly, of men: der Mann
'the man', 'masculine' gender and die Schildwache 'the
sentry', 'feminine' gender, das lldnnchen 'the little man',
'neuter'; sexless objects appear equally in all three gen-
ders; der Tisch 'the table', is 'masculine', die Tiir 'the door',
is and das Fenster 'the window' is 'neuter'.
'feminine',
Similar are the 'animate' and 'inanimate' genders of many
American languages, or the dozen and more gender-classes
of certain African languages.
Naturally, we cannot expect the associational habits of
speech-communities, which underlie these morphologic
with the results of conscious
classifications, to coincide
scientific study of the universe. The noun-genders are
an example of this. Another instance is furnished by the
English action-words. In these we make no distinction
1 10 THE FORMS OF LANGUAGE

between the performance of an action, as in I eat, I walli,

I write, and the undergoing of a sensational process, as
in / hear, I see. Our expression seems everywhere to
correspond to the former type of occurrence. In other
languages, such as Greenlandish, the second type is gener-
alized; one says, in accordance with reality, [takuwa:]
*appearing-of-it-to-him' i. e. ^he sees it', but also [tukaRpai]
'staraping-of-it-to-him', i. e. ^he stamps on it, tramples
it', where the English type of expression is more appropri-

ate. In Georgian both types exist: one very justly says

[v-t^ser] 'I-write' and, differently, [m-e-smi-s] 'me-to-sound-

ing-is', i. e. ^sound comes to me', 'I hear\ Yet, as we
must in such cases expect, the distinction is by no means
carried through with scientific correctness; seeing, for
instance, is viewed as if it were an activity, not a sen-
sation: [v-naxav] 'I-see'.
The phase of linguistics which studies these classes, —
that is, the structure of words, — is morphology.
11. The sentence. When the analysis of experience
arrives at independently recurring and therefore separate-
ly imaginable elements, words, the interrelations of these
in the sentence appear in varied and interesting linguistic
phenomena. Psychologically the basis of these interrela-
tions is the passing of the unitary apperception from one
to the other of the elements of an experience (p. 60, f.).
The leading binary division so made is into two parts,
subject and predicate, each of which may be further ana-
lyzed into successive binary groups of attribute and sub-
ject, the attribute being felt as a property of its subject.

The subject of the sentence is analyzed out of the total

experience as the substratum, more or less permanent,
and, owing to earlier experiences, the relatively familiar
element, which in the predicate receives definition {The
rabbit is an animal) or description (The rabbit is white),
 THE SENTENCE HI
or is apperceived as the substratum of an action {The
rahhif runs across the fteld. The rabbit is being chased).
The explicit predication of quality or action is impossi-
ble for languages in which every word expresses an ob-
ject (p. 64). In these languages the sentence consists of
one or more object-words. Each of these, since it can
occur alone as a sentence, is capable of expressing what
we look upon as a predication; any series of them, conse-
quently, contains no expression as to where the predica-
tion lies. These words, then, are sentence-words. The
Greenlandish [qimieq] thus can mean 'dog' or 'It is a
dog', [saliutoiqaoq] 'big liar', 'He is a big liar', or 'He
lies very much', and so on.
In contrast with this stands such a language as English,
in which the existence of independent action-words and
quality-words removes aU obstacles to the expression of
predication. Am ong such languages, also, there are, howev-
er, a great many differences. Latin, for instance, presents
some features that remind one of the nominal languages.
Itsverb always includes expression not only of the action,
but also of the acting object. Accordingly, predication
can in Latin also be expressed in one word, a sentence- —
word, —
even though only a limited portion of the words,
the verbs, can be so used: cantat 'he, she, it sings', edo
and so on. Where a quality-word,
'I eat', —
an adjective,
— forms the predicate, there is often no difference be-
tween predication and attribution. Thus magna culpa meajia
either '(a) great fault' (attributing adjective) or '(The)
fault (is) great'- (predicating adjective). Russian makes
this distinction by using different forms of the adjective;
thus, in [mu'31'k Ve'l'in] the adjective is the predicate:
'(The) peasant (is) poor', and in ['b'sdni* mu 3i'k] it is
an attribute: '(the) poor peasant'. English has gone farther.
It expresses predication only and always by means of the
112 THE FORMS OF LANGUAGE

verb; where no action is involved, the abstract verb iSy

expressive only of the relation of predication (in the form

of an action), is used (p. 67, ff.) ; we cannot say fault great
or peasant poor but only The fault is great, The peasant
is poor. Moreover, the English verb does not, like the
Latin, include the expression of an acting object: we cannot
say sings or eat, but only He sings, I eat, so that no —
sentence-word exists.
By thus contining the function of predication to our
action-words and that of subject to our object-words, we
have produced the syntactic categories of action and
actor (p. 68). If, now, it happens that the subject of a
total experience is not an object, but an action, this action
cannot be expressed by a verb, but must be put in the
form of a noun. This is the function of our abstract
nouns, such as sJcating in Skating strengthens the ankles.
Similarly, if the subject of the statement is really a quality,
no genuiue quality-word (adjective) can be used, but only
an abstract noun of quality, such as length in The length
of the ivall ivas two miles. When the predicate does not
really involve an action, we have seen that the abstract
verb is fulfils the predicative function. Attribution is

always expressed by adjectives with nouns, by adverbs

with verbs. Hence the use of nominal and verbal adjec-
tives aud adverbs when an object or an action is attrib-
utive: a boyish man, he spoJce boyishly, skating boys, he
spoke draivlingly. For attributive occurrence of objects
we have, however, also our possessive form: John's hat
The categoric distinction between these ^parts of
speech', —verbs, nouns, and so on, is by no means —
a necessary attendant of independent words for quality
and action. Chinese, for instance, also has such words,
[e. g. [lXQo/] 'good' or [Lmas/] 'buy', but the functions—
of subject, predicate, attribute are not confined to any
 THE SENTENCE 113

such classes. Thus the quality- word [lXqo/] appears as

attribute in [Lxao/'^^Au/] Vood man' and as predicate in
[''t'a'' L^ao/] *He (is) good'. The action-words [Lmas/] 'buy'
and [^masyj 'sell appear as attributes (where we in English
'

should have to use verbal adjectives 'buying'^ 'selling') in

[Lma£/ Lmas\ "^^An/] 'a trader, merchant', and so on. In the
modern speech there are also the independent words ['^ti^]
and [lJ9^] which independently express the relations, re-
spectively, of attribution and predication; thus one can
say also [L^ao/ ti '^^An/] 'good (attribution) man', and
[^fa"" ti Linao J 'he (attribution) hat', 'his hat', and [Lmas/

Lmas\ ti "^^aq/] 'trader', as well as ['^t'a"' Jb lXQo/] 'He

(predication) good', almost exactly our 'He is good'.
The process of analyzing an experience may be tem-
porarily interrupted by the associative addition of ele-
ments viewed as entering into the same discursive relation
as some one of the original elements. As well as we say
He is a good studentyWe can include other attributes of stu-
dent suggesting themselves as parallel to good: He is a
good, intelligent, industrious student. Such groups are call-
ed serial groups. It is possible that they represent a
specialized, automatized form of the discursive relation.
They are especially common in English and the languages
most closely related to it; we say John and Mart) ran

rather than John ran, Mary ran, but this condensed habit
of expression not everywhere so common.
is

While the sentence has its foundation in the discursive

analysis, other forces also play a part in determining its

form. Most important of these are perhaps the emotional
relations of the elements. The relations of emotional
stress find expression especially in the modulation of
loudness (p. 50). In addition to this, however, they affect
the sentence in various ways in different languages. A
method in English, for instance, is to place the emotion-
Bloomfiold, Study of Language 8
114 THE FORMS OF LANGUAGE

ally dominant element^) in some way out of its usual

position, preferably first or last. Thus He came last is
turned into Last came This inversion can be effected
lie.

also by making the dominant element predicate of an

introductory sentence: It was he who came lastj It was
last that he came. It was me they heat. The introductory
words are here entirely abstract and are spoken with
very low stress^ so that phonetically the dominant element
practically comes first. This construction is the regular
one, as we shall see, in some languages.
The material or between the elements
concrete relations
of a sentence may also play a part in its structure. These
relations are, of course, endless in variety, and their
linguistic expression is scarcely less manifold; we are
interested,however, only in those cases where such a
concrete relation receives some other expression than
that of the underlying discursive construction. Such
specialized expression of concrete, non-discursive relations
is of course, always a sequence of words that once stood
only in discursive relation but then became mechanized
Are the two words in Rome in
in the particular use. dis-
cursive relation? Which is attribute, which subject? We
cannot answer: for our feeling the relation is simply this,
that in expresses a local inclusion with regard to Rome.
Our feeling is due to the fact that this type of expression
has become mechanized: we without entering
reel it off
upon the discursive analysis, which, when such locutions
were first used, was vividly present.

1) The term 'psychological (as opposed to the 'logical' or

^grammatical') subject', used in this meaning by many writers,
is to be avoided as confusing. There is nothing more 'psycho-

logical' about an emotionally dominant element than about the

subject of a discursive analysis, and 'subject' is a discursive,
not an emotional concept.
 THE SENTENCE 115

Thus, as we have seen, the concrete relation of an actor

performing an action has in English furnished the uni-
versal form for the sentence. When, for instance, the
subject is not an actor but the goal (object affected) of
an action, we make it actor-subject of the abstract verb
is and use in the predicate a verbal adjective denoting
the quality of something tbat has undergone an action.
Thus we say He is hurty The rahhit was Icilled, The house
was heing huilt For such locutions Latin has special verb-
forms, the 'passive', which express an action as being
undergone, allowing the object affected to be expressed
as actor: Domus struehafur 'The house was being built'.
Where there is no actor at all, we use in English a purely
formal word: It is raining, It was four years ago.
Another concrete relation which we feel as entirely
unique is that of the goal or object affected of an action
Originally this seems to have been an ordinary attribute
of the predicate verb. In Latin, as in Old English, there
are two case-forms of nouns used in this way: the 'accu-
sative' for the object fully affected and the 'dative' for
the object less fully affected. Thus in Pater fllio lihrum
dat 'the father gives the son a book', lihrum is in the
'accusative' fllio in the 'dative' case. In present English
these distinctions of word-form are almost entirely lost.
Nevertheless the expression of these relations has remained
a thing by itself The object fully affected follows the
verb; that less fully affected either stands between the
two or viewed as in prepositional (local) relation: this
is

latter usage amounts to an analysis of the relation of

object less fully affected into an independent word. Thus
we say either The father gives the son a hooh or TJie father
gives a hook to the san. In Chinese also the object fully
affected has a constriction aU. its own: whUe all other
attributes precede their subject, the object affected foUowB
116 THE FORMS OF lANGUAGE

its action-word.Thus ^wo/ lP'oN ''t'a''] 'I fear him', pa"*

,^p'a\ lWo/] ^He fears me'. The object less fully affected
has a different expression, which, however looks like a
specialized form of the preceding: a few action-words take
it as their goal, forming a phrase which then as attribute
precedes the main action-word of the sentence. For in-
stance [^Ibd? Jiei/ L^o/ lSui)\ "^c^ie/ Ji/] 'He sends me
festival-presents' (more literally: 'He, giving me, sends
festival-presents'). Here ["^t'ci^], the subject, is followed by
the predicate, in which [ikej/ l^o/] is an attribute com-
posed of the action -word [Lkei/] 'give' followed by its
object affected, ^wo/] T. This two-word attribute, accord-
ing to the general principle, precedes its subject [lSu^ ],
an action- word meaning 'send', which is followed by its

object affected ['^C9ie/ Ji/], in which the former word

an attribute of the latter 'presents'.
'festival' is
Coming, finally, to our Eaglish preposition -groups,
with which we began as an example of crystallized con-
crete relations, we may seek their origin in older con-
structions of attribution. Local relations are always, con-
cretely, relations with regard to objects; we find them,
accordingly, in many languages expressed by case -forms
of object-words, as in the Latin Fugit Corintho 'He-flees
frora-Corinth', ('ablative' case), Fiomani venit ('He comes
to-Rome' ('accusative' case), the Sanskrit paryate tisthati
'On-the-mountain he-stands' ('locative' case) prayacchati
savyena 'He-hands-out with-(his)-left-(hand)' ('instrumen-
tal' case); compare also the Finnish case-forms on p. 108.

This purely attributive usage is still seen in a later stage,

when there come into use set phrases of certain adverbial
attributive words with these case-forms of nouns. Thus,
to take an example from Ancient Greek, we find such
sentences as Kephales dpo pharos helesJce 'From-(his)-head
off the -cloak he- drew', 'He drew the cloak from his
 THE SENTENCE 117

head'. Here the verb heleske *He-drew' has, beside the

object fully affected pharos, the attributes dpo 'off', an
adverbial word, and the case -form, of ablative value,
Jcephales 'from-the-head*. Later such combinations of ad-
verbial word and case-form became habitual and were
crystallized into a standard expression of the concrete
local relations with regard to objects: apd Jcephales 'from
the head'. The same occurred in English, and even today,
when our case -forms are practically lost, such phrases
are our regular expression for local relations: fro7n Corinth^
to Borne, from his head, into the Thus we obtain
fields.
the collocation of preposition plus noun which would be
entirely inexplicable on the basis of the purely discursive
relations from which history shows it has grown. In Chinese
similar phrases have a very different origin. One can there
say [""tfa^ Ltao\ ""t'len/ Ji/ LC9^y'\] ^He goes into the fields',
but it would perhaps be more literal to translate 'He,
entering (the) fields' interior, goes'. For the central ele-
ment of the predicate ishere [L<*'9^y\], preceded by its
attribute of three words, which consists of tbe action-word
[Lta6\] 'enter' followed by its object fully affected [""t'len/
Ji/] 'fields' interior'.
Of similar nature are our words the and a. The relation
of the to rahhit in the rabbit or of a in a rabbit is scarcely
the regular discursive one of attribution. Originally the
word the was probably a deictic word similar to our that:
it was used attributively with a noun; in time, however,
it came to be used anaphorically (of objects not actually

present, but of those which had been mentioned or were

otherwise specifically known), until today the use of the
is a peculiar and categoric expression of definiteness of

an object. Likewise, a, an was originally the numeral

'one', attributively used. It came in time to be used when-
ever only one object was meant and the definite the could
118 THE FORMS OF LANGUAGE

not be used. Today, in consequence, we have three syn-

tactic categories affecting nouns: every object must be
spoken of either as definite (at least with tlie)^ or as in-
definite (at least with a), or as collective, without the
article, as in Man wants hut little ... or Men are easily
moved hy such things. This is in contrast with most lan-
guages. In Latin, for instance, homo means 'the man',
'a man', or 'man', and such attributes as ille 'that' or
unus 'one' or quldam 'some' are used only where such
elements are actually and vividly present in the experience;
they stand, then, in regular attributive relation to their
noun.
As a last example I shall cite our 'infinitive' verb-forms.
These express the action as complement of another pre-
ceding verb, e. g I shall go, He can speak, You must ivrite
a letter, I want to forget it, They tried to deceive us, the —
relation between the two verbs being in many instances
expressed by the word to. This form of expression would
baffle all attempts at reducing it to terms of attribution
or predication. But it was not always so. In the older
forms of English the infinitive is a verbal noun, compa-
rable to those we now use in -ing (as skating), occurring
most frequently in two case-forms, an accusative, e. g.
Old English hindan 'to bind' and a dative case, Old
English hindannCy hindenne. The accusative form was used
as the object fully affected of a verb; thus I shall go, for
instance, meant originally 'Iowe going' and was parallel
to such expressions as *I owe money'. The dative form,
after the preposition to, was used like any other noun
with a preposition (see above), such a sentence as He
went to eat meaning 'He went to eating', parallel to 'He
went to London'. The scheme of these expressions has
long ago, however, become so automatized that they are
used where the original discursive relation could not be
 THE SENTENCE 119

interpreted into them, — for any feeling for this historic-

ally underlying relation has long ago disappeared. The
preposition tOj by the same token, has here become an
abstract relational element^ expressing the relation between
the two verb-forms; its independent value can be seen in
such expressions as He doesn't ivant to.
These peculiar developments^ beyond predication and
attribution, of the sentence-relations of a language, are,
of course, an important part of its syntax. The descrip-
tion of typical cases could be greatly expanded. The
principle, however, everywhere the same: the discur-
is

sive relations of predication and attribution, which are

inherent in the formation of the sentence, lie at the basis
of all set locutions in which the material content becomes
dominant. While a Chinese speaker, on reflection, will
realize that 'fields' in the above sentence (p. 117) is the
object affected by the action-word 'enter', — for the word
[Ltao] can thus occur in a sentence without any other
action-word, as the central element of the predicate: ]^^o?
^tao "^t'len/
^\l'\
'He enters the fields', — he would per-
haps have difficulty in appreciating the relation of 'enter'
to 'go' as the regular attributive relation of his language,
just as an English speaker would undoubtedly be at a
loss, were we to require him to explain in terms of attri-
bution and predication, the relations of the words in He
goes into the fields. In short, the change of language pro-
duces such relation-words as our prepositions, with the
simple and direct forms of expression which they make
possible, out of the concrete, cumbersome habits of an
older time. This change is at the same time a develop-
ment of the mind: the conceptual values of our words
of quality, action, and relation would be impossible with-
out these words (p. 65), just as the latter can exist only
as the result of a definite mental-linguistic development.
 CHAPTER V.

MORPHOLOGY.
1. The significance of morphologic phenomena.
The morpliologic cliusses of a la::guage represent communal
associative habits: they express the associative connections
which the national mental life of a people has made
between the types of experience which the language ex-
presses in words. Thus we in English find some connec-
tion between flare and flash, between father and mother,
between hoys and stones. Every formational element
common to a number of words involves a grouping
together of these words on the basis of what to the com-
munity has appealed as a common element in the ex-
periences expressed by these words. The classifications
of language are, in fact, the clearest expressions of the
associations made by the community as a whole. They
are, accordingly, of great ethnologic significance. This
significance is increased by the i'act that they are far less
subject to reflection than other communal activities (such
as religion) and are never, any but the most highly
in
cultured communities, modified by such reflection.
2. Morphologic classification toy syntactic use
(Parts of speech). The first kind of morphologic word-
class of which we shall speak, —
and it is in many
languages the most fundamental, — is really a syntactic
phenomenon. It is the division into parts of speech. This
MOKPHOLOGIC CLASSIFICATION BY SYNTACTIC USE 121

division due to the existence of syntactic categories,

is —
in English mainly to that of actor and action. The Eng-
lish parts of speech received some mention in the last
chapter, where our classification of noun, verb, and ad-
jective was spoken of. The classification may in its entirety
be described as follows:
a) The verbs express an action (I eat, they danced) or
another element viewed as such (he is). Their distinctive
characteristics are several. They always form the nucleus
of the predicate; predication cannot be expressed in any
other way in the English language. The actor, real or
formally viewed as such, must always be explicitly men-
tioned with the verb. Nearly every verb expresses by its
form the time, present or past, of the action. In the
present-time forms most verbs vary according to the
number and person of the actor, in that a third-person
singular actor requires a special form: i, we, you, they
eat, but he, she, it eats.

The only exception to all this is the non-committal or

infinitive verb-form, e. g. eat in I shall eat this apple. This
form expresses the action apart from any actor. It can
be used only in exclamatory utterance, where it serves
as a command (Eat!), and as a supplement to another
verb (p. 118). The infinitive differs from the present- tense
form in only one verb: he (present tense: we are)] it is
lacking in several others (can, may, shall, will).
Verbs are modified by adverbs (see below).
b) The nouns express an object-experience, be it really
such (stone, house, man) or viewed as such (skating, lengthy
greenness). They are distinguished by a number of char-
acteristics from verbs (e. g. they cannot express predi-
cation) and from the other parts of speech. They express
the actor or the objects afi*ected by an action (The man
gave his son a hook), as weU as that to which something
122 MORPHOLOGY

is equated or under which it is subsumed by the predi-

cation (He is a merchant, The whale is a mammal). They
can stand in attribution to other nouns only when in
their possessive form (the mavUs hat). Nearly every noun
shows by its form whether one or more than one of the
objects is intended. The nouns are modified by the attrib-
utive nouns already mentioned, by adjectives, and by
attributively used pronouns, among which the and a, an
are especially frequent, owing to certain syntactic habits
of the language (p. 117, and they are used in set phrases
f.),

with prepositions (p. 117). The constrast with verbs is

thus complete.
This contrast is, however, less in the verbal nouns
(ending in -ing) which express what is usually looked
upon as an action (e. g. skating in Skating strengthens the
ankles), for these verbal nouns can, be follow-
like a verb,
ed by mention of the objects affected: Giving them alms
is no remedy, I am tired of hearing him grumble.
The adjectives express a quality (green, large, long)
c)
or what is viewed as such (growing, burning, boyish).
They can be used to express neither predication nor action,
actor, or objects affected, but stand only in attribution
with nouns or, in the predicate, as qualities predicated
of the subject (The man is good). Beside the usual form
they have two variations which express a superior and
a superlative degree of a quality (better, best). Adjectives
are modified attributively by adverbs.
A peculiar variety of adjectives are the verbal adjecti-
ves or participles, which express as a quality what is

usually viewed as an action, e. g. a running boy, broken

toys.These verbal adjectives are used in set phrases with
the verbs is, has, to express durative and perfectic manner
of action (see below, p. 145), as in J am reading a book.
He was dreaming, I have written him a letter. Have you
MORPHOLOGIC CLASSIFICATION BY SYNTACTIC USE 123

ivritten?, He Like the verbal nouns, the

Jiad arrived.
verbal adjectives can be followed by expression of the
objects affected, as in several of the above examples.
A few of our adjectives may be said to form a sub-class
in that they can be used in the predicate only: He is
asleep, awaJce; The ship ran aground.
d) The pronominal words are unified by their relational
value, personal, deictic, anaphoric numeral, etc.; in their
,

syntactic use they can be divided into a number of sub-

classes.They are used to express actor or objects affected
(He gave me it) or, in the predicate, that relational element
to which something is equatr/id (Is it I, It is mine), —
differing in these uses from nouns in that they never are
attributively modified by adjectives, though some of them
are so modified by other pronouns. They aroused, further,
as attributive modifiers of nouns and of other pronouns
(the many this man^ three men, the other), but differ in
this use, as well as in the predicate, from adjectives in
that they always precede the latter when both are present
(three good men, the old house). Some pronouns occur in
the nominal uses only (who^ I, mine), these and a few
others (what) are never modified by another preceding
pronoun; two, the and a, never occur in the nominal, but
only in the modifying uses. While less homogeneous as
to syntactic use, then, than the other parts of speech,
the pronouns, taken together, yet constitute as distinct
a class as any, owing to their peculiar meaning and to
their resistance to material modifying elements.
e) The adverbs attributively express the circumstances
of qualities and actions, such as place (here, there, where),
time (then, yesterday, afterwards) degree (more, very),
,

manner (rapidly, slowly, kindly), and the like. They alone

have the function of modifying adjectives and, in a direct
sense, verbs. As in the case of adjectives^ their form may
 124 MORrHOLOGY

express a superior or superlative degree of tlie experience

(kindlier^ Idndliest) but no other relations.
f ) Prepositions express a relation, usually spatial,
with
regard to an object. Accordingly, they are used only
with nouns or with nominally used pronouns, preceding
them and their modifiers in a unique construction, the
set phrase of preposition plus noun (p. 117).

g) The conjunctions express relations between coordi-

nate parts of speech and between predications. Subordi-
nating conjunctions express a relation of time, condition,
cause, and the like (wJien^ if, hecause, though) with regard
to a predication. Thus they relate a predication, as a
whole, subordinately to another predication (When he saw
the house, he . . . .). Coordinating conjunctions express
serial relations of all kinds (and, or, hut, both . . . and, ei-

ther . . . or). Externally, this function involves their appear-

ing between coordinate words, phrases, and predications.
h) Interjections (of the primary type) are, of course,
opposed to all the other classes of words, in content, use,
and form.
The most striking circumstance about this classification
is that the normal speaker is utterly unconscious of it.

It requires a considerable degree of mental training and

even of linguistic habit of thought before one can by
introspection analyze these classes. And yet they are used
correctly every day by millions of speakers who would
be utterly incapable of making such an introspective anal-
ysis, and perhaps even of understanding it, if it were
made for them. With all the complexity of the classifi-
cation, confusion between the different parts of speech
^
/ never occurs. This is a most important fact, especially
in view of the unconscious nature of the habits, and one
which could be illustrated by many features of English
usage. It is attested, for instance, by the failure of homo-
MORPHOLOGIC CLASSIFICATION BY SYNTACTIC USE 125

nyms to introduce even the slightest confusion. Such

homonyms as wood (noun) and ivould (verb), for instance,
could be confused only in the dream-world of Alice in
Wonderland. The noun wood occurs with preceding ad-
jectives, prepositions, and pronominal modifiers, in the
function of actor, object aflPected, or predicate noun; the
verb woidd expresses only and always a predication, must
be accompanied by mention of an actor, is modified by
adverbs only, and is followed by an infinitive supplement.
Even where the homonymy is significant, corresponding
to some resemblance of semantic content, as in the case
of stonej noun in a stone ^ verb in they stone him^ and ad-
jective in a stone gate, the same distinctions hold true,
and a single syntactically joined word will show which
of the homonyms is present: all that is necessary to show
that the noun is meant is the modifier a, the verb is at
once identified by a preceding actor, and the adjective
by a following subject. In short such homonymy never
obscures the boundaries between these classes, as it well
might, were they less clearly drawn; thus one is never
tempted to confuse house [haos] noun and house [haoz]
verb, or gun, bullet, arrow, nouns and shoot verb, in spite
of the corresponding homonymy in stone. Likewise there
is no confusion between adverb and preposition in spite

of such homonymy as that of iw, preposition (in thehouse)

and in, adverb (He walked in)] no confusion between prep-
osition and conjunction in spite of the homonymy of
after, preposition (after the meal) and after, conjunction
' after they had For instance, the conjunction cor-
gone).
responding to in would never be expressed as in, no
matter how ignorant the speaker, but would be while,
during, or the like; and the adverb corresponding to after
would (in spite of homonymy of adverb and preposition
in such cases as in) never be after, but always afterwards
126 MORPHOLOGY

or then. In short, tTie classification into parts of speech,

though not appearing in the phonetic form of the single
word, is as distinct as any other classification in the
language. Self-explanatory and self-understood as it seems
to us when once we are made conscious of its existence,
it is by no means universal in linp;uistic expression. In
fact, the parts of speech used in English occur in only
a limited number of languages. In Chinese, for instance,
as we have seen (p. 112, f), a word, no matter whether it

expresses object, quality, or action, is externally treated

alike, and may express subject, predicate, or attribute;
thus we saw on p. 113 the word [lXOO;'] Vood' (quality)
as attribute and as predicate, and the words [Lmas/ Lma5\]
'buy' and 'sell' (action) as attributes; as predicate they
appear in such sentences as [Lmas/^ji/LpA.n/^Ju^] 'buy one
volume book', i. e. 'buy a book'. Similarly the action-
word ['Juo^J 'speak' appears as predicate in ['^t'a^ ^Juo^
Qiei/ "^cyii)^ L^iia',] 'He speaks North-capital (Peking) lan-
guage', but as subject in lP^V ^^9^^^ L^^a\ 'Jilo^ J*9
["^t'a^

xao/] 'His North-capital language speaking is good', i. e.

'He speaks the Peking language well', where we must
translate the uniform ['Juo^] by a verb in the one case,
where it is predicate, and by an abstract noun of action
in the second, where it is subject.
Chinese may, indeed, serve us as an example of a lan-
guage with parts of speech entirely different from ours.
It has no such parts of speech as noun, verb, adjective,
and adverb. 'Good' is a quality, 'man' an object, 'speaks'
an action in China as everywhere else, but the fact that
these experiences belong to these different spheres is not
expressed in the Chinese sentence. In Chinese we can
distinguish primarily two parts of speech. One, by far
the more numerous, used according to certain rules of
is

word -order, chiefly the following: subject precedes predi-

 CLASSIFICATION BY CONGRUENCE 127

cate, attribute its subject. Example: ['^t'a^ ^xao/^^AJof] *he

good man', to be taken as subject and predicate, the latter
consisting of an attribute and its subject, i. e. 'He is a
good man'. Other examples have occurred earlier in this
book (p. 113).
Within this first class of words a subdivision can be
made between intransitive and transitive words. The object
affected by a transitive word follows it (in opposition to
the usual rule of attribute-subject), e. g. ['^t'a'' 'Juo"' lP^V

•^c^ii)^L^ua\] 'He speaks North-capital language'. The word

[lXuclv], for instance, is intransitive, and could never, like

[^Juo"'], be followed by the expression of an object affected.

Other examples will be found on pages 116, 119, 126.

The words of the transitive class thus resemble some of
our verbs (or, again, our prepositions, see page 117); but
the resemblance is distant, for the Chinese transitive
words by no means, as we have seen, either occur only
as predicates or monopolize this function; further they
alone can, by definition, be followed by expression of
objects affected, whereas in English this is exactly the
feature in which verbal nouns and adjectives (p. 122, f.) com-
pete with verbs; and our verbs, on the other hand, can by
no means all of them take an object affected.
The other part of speech consists of words not subject
to these rules of word-order, but used, sometimes invaria-
bly, sometimes at will, between words of the former type
to express explicitly the relation between them. Thus
J*8
[ ] expresses predication, [""ti^J attribution, and the sen-

tence above could read ["^f a^ Ja Lxao/ ti "^^An/ ], the mean-

ing being unchanged but more fully stated (cf. p. 113).
3. Classification by congruence. A peculiarity of the
classificationsby use in the sentence, —
parts of speech,
— which we have just seen in Chinese and English, is
that the phonetic form of the word itself does not express
128 MORPHOLOGY

the classification. Tlius stone noun, verb, and adjective

are alike in form; similarly in preposition and in adverb,
after preposition and after conjunction, or in Chinese
[•^ti^ expressing attribution and ["^ti^J 'low'. It would be
impossible to find in such words as street, hotise [haos],
arrow any formal feature to show that they are nouns,
giirtj

as opposed to such adjectives as sweet, narrow or to such

verbs as heat, souse, run.
The next type of morphologic word-class that demands
discussion, has the same feature of not involving formal
identification of the classes. It also is really a syntactic
rather than a morphologic phenomenon.
We had occasion in the last chapter to notice that
word-boundaries are sometimes phonetically recognized
by the fact that word-initial or word-final may vary accord-
ing to the sounds that precede or follow (p. 101, ff.). In
some cases, as in the Sanskritexample there quoted, the
variation is an automatic sound-variation and therefore of
no morphologic significance. In another example this
was difi'erent; in the Irish td ha 'there are cows' and a
va 'his cows' (p. 102, f.), the variation of ha: va does not
depend upon the preceding sound. One says also na ha
'the cows'; further, hog 'soft' but ro vog 'very soft', hdn
'white' but ho van 'white cow', hrish 'break' but do vrish
'did break', and so on. The speaker must have a class-
feeling for words such as td, na, after which h- is spoken,
as opposed to words such as a, ro, ho, do, after which v-
is spoken. The words in these two classes possess no

distinguishing characteristics, by themselves, as to form

or meaning: they constitute a class, in each case, by virtue
of the effect they have on other, —
in our instance, the
following, —words in the utterance. These are word-
classes by congruence. That there are several such classes
in Irish appears from ho 'cow': an vd 'the cow: or md^
 CLASSIFICATION BY CONGRUENCE 129

'our cow' and from uv 'egg': an tiiv 'the egg': na nuv

'of the eorcys': a huv 'her eo-cr'. Each of these classes has
one distinctive feature, and that is simply the fact that
the words in it are felt to have this phonetic effect on
following words, as it were necessarily and as part of
the expression of the meaning. To the Irish speech-con-
sciousness nothing else seems possible: here, as always,
the morphologic classification of a word is, in the feeling
of the speakers, part of semantic value. For this very
its

reason most speakers of a language are unconscious of

their morphologic classifications, taking the classification-
element for granted as an inevitable part of the meaning.
Another example of classification apparent not in the
classified word itself, but in its effect on other words, —
classification by congruence, — is that of the German
noun-genders. There is nothing in such German nouns
as Leib 'body', Anker 'anchor', Auster 'oyster', Fran
'woman', Weib 'woman', Fenster 'window', either in form
or in material meaning, to indicate a classification. All
attributive words, however, such as adjectives and pro-
nouns, and all later anaphoric reference to these or other
nouns at once show them to fall into three separate
classes. Thus the definite article 'the' has in the nomina-
tive case singular three forms, one being used with each
class of nouns: cler Leib, der Anker, —
die Aasler, die
Frau, — das Weib, das Fenster; and similarly in anaphoric
reference, the pronoun referring to nouns of the first or
'masculine' gender is er, to the second or 'feminine', 67'e,
to the third, 'neuter', es. This German classification differs,
then, from the Irish in that not the next word, but all
the attributively modifying words and all words express-
ing anaphoric reference, even though spoken much later,
are affected.
Speakers of English can contrast this German gender-
Bloom field, Study of Language 9
130 MORPHOLOGY

classification with their own language. In English such

words as the or good are the same for all nouns, and,
though we have in the singular three anaphoric pronouns,
hCj she, it, these differ not in being assigned to different
classes of nouns, but only in actual meaning, just as any
three other words may differ. While we refer to human
beings beyond infancy according to sex as he or she, we
are in other cases free to recognize sex or not: the horse he
or the horse ... it. When we use the sex-forms to refer
to a sexless object, we are, by a genuine metaphor, attribut-
ing personal and sexual character to it, as when we refer
to a ship or a steam-engine or the moon as she. In short,
he and she differ as any other words may differ, man,
e. g.

tvoman, child, and do not, like the three German forms

of the pronoun, involve the constant presence of a classifica-
tion of nouns. Chinese, which has but one anaphoric
word ["^t'a^], does not in this differ much from English.
The difference is merely parallel to that between the
Chinese ['^9iui)"'] 'older brother', [^i\\ 'younger brother'
and our where the Chinese has two different words
"brother,

to our one: in the anaphoric pronoun we have three words

to the Chinese one. Both languages are in this respect
widely different from the German with its cr, sie, es, which
demand a complete and always present classification of
the nouns in that language.
Another type of classification by congruence is seen
in Chinese. Anything counted is expressed in this lan-
guage by the numeral-word and the designation of the
thing counted, and the latter is preceded by a modifying
word of fairly material content expressing the unit of
the object-idea; e. g. [^ji/ko "^^An/] 'one piece man', i. e.

*one man'. Now, there is number of such

a considerable
numerative words, and the choice of the numerative word
depends upon the thing counted. Thus one says:
 PHONETIC-SEMANTIC CLASSES 131

["^san"" ko "^^aii/] 'three piece man', i. e. 'three men',

[•^san^ ko '^C9'ien/] 'three piece mace', i. e. 'three mace
(coin)', but:
['^san'' LPAn/ 'Ju"'] 'three root book', 'three books',
f san"" lWCi/ "^jy/] 'three tail fish', 'three fishes',
fsan'' Lwei\ '^9ien'' 'Jaij''] 'three rank earlier born', 'three
teachers',
[•"san"" '^t'"iao/L9ien\] 'three branch thread', 'three threads',
["san"" '^t'iao/Lpei\] 'three coverlets', —
very much as we speak
of three head of cattle. Owing to
the material content of these numeratives there is a cer-
tain amount of freedom North, for
in their use: in the
instance, [Lko\] encroaches on the others, but this freedom
has definite limits; each of the forty or more numeratives
has its range of objects with which it is used. It is appar-
ent that the Chinese speech-feeling divides everything
that may be counted into a number of classes which re-
ceive distinction in the numerative word. Here again we
see a classification by congruence, though of a very different
kind from those which we examined in Irish and German.
4. Phonetic -semantic classes. The two kinds of
have this ir. common, thai
classification so far considered
the words classified in no way show the classificatioi*
in their immediate form. In the classification by syn
tactic use it is in the sentence that the classification appears^
the different classes here performing entirely different
functions; and in the classifications by congruence it is

in the form of following words or of attributively o^

anaphorically connected words or in the choice of certam
other words, never in the classified word itself, that the
classification is expressed.
Now, it is conceivabie that such classifications may
receive expression in the classified woi d itself. For instance,
there are in Italian two gender-ciasses of nouns, the classifi-
132 MORPHOLOGY

cation receiving expression, as in German, by variation

in the form of attributive and anaphoric words. In addi-
tion to this, however, very many of the masculine nouns
end in -o and very many of the feminine in -a. Although
there is one feminine in -o (la mano 'the hand') and there
are a few foreign masculines, usually, however characterized
by peculiar accent, in -a (e. g. il sofa 'the sofa'), the feel-
ing that nouns in -o are masculine, those in -a feminine
is part of the Italian speech-feeling. In German and Eng-

lish the prepositigns, pronouns, and conjunctions tend to

be shorter than other words, and similar tendencies occur
in other languages, notably in the Semitic and the Malayan.
AU this shows that a word-classification may express it-
self in more than one way, although, as a rule, the ex-
pression is in only one direction complete and regular,
and in the others imperfect and irregular, as in the Italian
-0 and -a endings.

There are, however, word-classes which receive expres-

sion in the form of the classified words alone. These are
the commonest of all classes: most of our instances in
the third and fourth chapters illustrated them. Thus
we there set up a formula by which metilu might be used
for 'white rabbit' and meko for 'white fox'. In these two
hypothetical expressions the phonetic element me- corre-
sponds to the common semantic element of 'white'. To
turn, now, to our actual instances, we may recall a num-
ber of them.
English father, mother, brother , sister: common phonetic
element, dental plus -er; these words form a semantic
class in that they are aU. nouns designating a near rel-

ative.
English flame, flare, flash, flimmer, flicJcer: the common
phonetic element is the initial fl-; the words fall into a
semantic class in that aU of them express phenomena of
 PHONETIC-SEMANTIC CLASSES 133

fire with especial reference to its peculiar moving light.

This class is of interest in the present connection, be-
cause it illustrates the emotional rather than perceptual
value and the ill-defined rather than clear-cut extent of
many of these classes. For there can be no doubt that,
in the feeling of many speakers, flicker again associates
itself with such words as flutter, fly, and even, further,
with flit, flip, flop, flap, and so on. All these words share
the initial fl- and are more or less vaguely related in
meaning indeed, the feeling for the semantic connection
;

may vary in the same speaker under dijfferent circum-

stances. In short, the extent or the existence of a phonetic-
semantic word-class may be very doubtful, and could
be determined with accuracy only for a given person at
a given time, and here only if a full insight into his
associative disposition at the moment were attainable.
The different members of such word-classes, may, more-
over, cohere in differing degrees. Thus and flash read-
flare
ily and vividly associate each other; so do flicker and

flimmer; so, perhaps, though in lesser degree, do flicker

and flutter; then, again, flip, flop, and flap form a smaller
class within the larger, their coherence being expressed
by the common final -p; or, further, flicker, flimmer, flit,

flip possess a common semantic value in the smaUness

and fineness of the movement designated, to which corre-
sponds, in form, the common vowel -^-. We are dealing
here with complex and delicate habits of association of
emotional rather than perceptual significance.
This appears, further, in the fact that these classes
cross each other. We have seen (p. 79), in our instance,
that flicker also connects itself with snicker; likewise,
flimmer with glimmer, shimmer, simmer, where the first
two words are closely associated, the last one perhaps
more loosely. The word flash belongs also to the large
 134 MORPHOLOGY

but coliereiit class of dash, crash, dash, gash, gnash, hash,

lash, mash, plash, slash, splash. To this there join quash,
squash, even among us, who pronounce the vowel in these
words as [d] instead of [se]. Further, the word sash, no
doubt belongs with lash, even though it is far from the
other words of the class, and plash, splash weakly join
to wash. Then again, crossing this class is the associa-
tion of mash with mush.
English hang, hiff, hump, form a more
huffer, box, heat
or less homogeneous class, but hed or buy or boat, which
also have initial h- surely do not belong to it. The best
,

illustration of the peculiar character of these classes is,

however, box ('to strike blows with the fist'), which de-
cidedly belongs to this class, while the homonymous box
('receptacle') surely does not. If the reader, in first read-
ing the took box in the latter sense, he no doubt felt
list,

a disturbing value when he came to it; yet, keeping the

former sense in mind, he will be able to re-read the list
without this feeling. This is comparable, of course, to
the drawings which may be interpreted as a concave or
a convex object, according to one's momentary predisposi-
tion. So heat above is homonymous with beet.
Phonetic-semantic classes are also the following, some
of which were quoted in the fourth chapter:
English thirteen, fourteen, fifteen, etc., to which come,
more loosely, ten and tenth.
English thirteen, thirty, third, three, crossing the latter
class.
English fires, fathers, boys, sisters, cats, ropes, watches,
peaches, etc., all expressing a plurality of objects with
the common phonetic element [z] and, by an automatic
sound- variation peculiar to this ending, [s] and [ez].

'English father s, hoy's, Mng's, man's, priest's, boss's, etc.;

the common phonetic element is homonymous with that

 PHONETIC-SEMANTIC CLASSES 135

in the last class, but the attributive use puts these words
into dijff'erent connections; notice, moreover, such opposi-
tions as meriy but man's
English dancej dances danced, dancing , dancer, with a
,

common material element expressed by the common

sounds [daens].
English eat, eats, ate, eaten, eating, eater, eatable: com-
mon phonetic element, front diphthong plus t, with ma-
terial content of the action ^eat'.
English danced, walhed, rocked, loved, cried, landed,
hounded, etc.; common phonetic element [t] and, by an
automatic variation, [d] and [ad], the content being the
past time of the action relative to that of speaking.
Russian ['b's* dni] ^poor' in attributive use, ['b'e' d'in]
'poor' in predicative use (p. 111). Here the common ma-
terial content finds expression in the common phonetic
element ['b'sdn .
., 'b'ed' . . n], the sound-variations being
common in the language,
Greenlandish [tusaRpaRa] 'I hear him*, [tusaRpat] *thou
hearest him', [tusaRpaiJ 'he hears him', etc.^ where the
common phonetic element [tusaRpa] corresponds to the
common meaning, 'sounding of him' (cf. p. 104)
Semantically similar are the Nahwatl (Aztec) form^
niktlamaka 'I give him something', nildemaka 'I give it
to someone', nikmalca 'I give it to him', and so on.
The crossing of the classes is very apparent in a lan-

guage like Latin, in which, for instance, words like edo

'I eat', edis 'thou eatest', edit 'he eats', edimus 'we eat', etc.

have the common phonetic element ed- and the common

semantic element of the material content 'eat', while such
groups as edo 'I eat', rego 'I rule', lego 'I read', and so on^
have a common -o expressing the actor as the speaker, edis
'thou eatest', regis 'thou rulest', icgis 'thou readest', a com-
mon 'is expressing the actoi as the one addressed, and so on.
136 MORPHOLOGY

5. Classes on a partially phonetic basis. Still othei

morphologic classes depend partly, but not entirely on
phonetic similarities.
English nouns, for instance, fall into two categoric
classes: every noun expresses an object either as one or
as more than one. Now, this classification of nouns into
singular nouns and plural nouns is, to begin with, classifica-
tion by congruence, for our present-tense verbs and many
of our pronouns vary in form according to whether a
singular or a plural noun is the actor, or, respectively,
that modified or referred to, e. g. The hoy skates. The
hoys slicite, The man smokes, The men smoke, this hoy, these
hoys my hat
J
it, my hats
. . they. In addition to this,
. . . .

however, almost every singular noun has by its side, clo-

sely associated with it and falling with it into a seman-
tic-phonetic class, a plural noun of the same material
content: hoy, hoys (belonging to the larger semantic-phonet-
ic class hoy, hoys, hoy's, hoyish, hoyhood, etc.); man, men
{manhood, manly, mannish, etc.); hat, hats; knife, knives;

and vice Our singular nouns form a class, conse-

versa.
quently, in that nearly all ofthem are related, in uniform
semantic relation, to plural nouns which resemble them
in form; and our plural nouns form a class because each
of them has by its side a similarly related singular. Now,
within these classes there are a number of sub-classes
according to the formal relation to the corresponding
word of the other class. Thus the plural nouns, hoys,
and the great majority
fathers, hats, rocks, peaches, watches
of other plural nouns form a large class in that they add
[z], [sj, or [ezj, — these three endings varying automati-
cally, — form of the corresponding singulars. A
to the
smaller class formed by those which also add a sibi-
is

lantj but at the same time substitute [v] for the final [f]
01 the singular: calves, knives, loaves (as opposed to cliff,
 CLASSES ON A PARTIALLY PHONETIC BASIS 137

hluff, etc. in the preceding class; all these plural nouns

together however belong to a single phonetic-semantic
class, see page 134). Houses, beside adding the ending
[9z], substitutes [z] for the final [s] of the corresponding
singular house, and thus forms a class by itself. Meiz,
women, form a class
geese, mice, children, oxen, sheep, etc.
in that they lack the usual sibilant ending; and within
this class the first four words belong with a few others
to a sub-class, the members of which difi'er from their
singulars in vowel only; within this, again, mice and lice
form a smaller class in having exactly the same vowel,
corresponding to the same singular vowel in mouse, louse.
Children and oxen, further, probably form a class in that
they add an -n suffix, within which class each word again
stands by itself. Sheep, fish, and deer constitute a sub-
class by virtue of homonymy with their singulars.
Owing, finally, to the close association between corre-
sponding singulars and plurals, the singulars correspond-
ing to the plurals within each of these classes, larger or
smaller, also form a class. Thus most singular nouns
belong to the large class of hoy, father, hat, rock, peach,
watch, because they correspond to plurals with the reg-
ular sibilant addition; hnife, calf, loaf, etc. form a class
because they correspond to plurals with [v] for the final
[fj; house forms a class; so do man, woman, mouse, louse,
goose, etc., within which mouse and louse are a smaller
class, and so on.
We nouns two kinds of word-
see thus in the English
classification not entirely marked by phonetic common
elements, namely:
Classes due to the association of each word with another
word uniform semantic relation to it; for instanc-e: aU
in
singular nouns; all plural nouns;
Classes due to the association of each word with another
138 MORPHOLOGY

word in uniform semantic and phonetic relation to it;

for instance: the plural nouns calves j Jcnives, lives, wives,
loaves, etc., or tlie singular pouns mouse and louse, or the
singulars or plurals, respectively, corresponding to these
classes.
We may illustrate such classes by our verbs also. Classes
by semantic parallelism are: a) all infinitives: he, have,
eat, sing, love, dance; b) all forms used of the present
tense, when the speaker himself is the actor: am, have,
eat, sing, love, dance, homonymous in all cases except am
with a) and d); c) all present-tense verbs used with a
third person singular actor: is, has, eats, sings, loves, dan-
ces, — in a few instances homonymous with b) and d),
e. g. can, shall, ivill; d) all present- tense verbs used with

actor in the second person {^yovi) or in the plural: are,

have, eat, sing, love, dance, — homonymous in all instan-
ces except are with a) and b).
As and all plurals of
in the instance of all singulars
nouns, these classes are at the same time classes by con-
gruence, for b) is used only with the actor I, c) only with
a singular-noun or singular-pronoun actor, and so on.
However, all these together constitute a large class, again
by semantic parallelism, as opposed to those that now
follow, and this classification is not supported by any
features of congruence. For, while all the preceding refer
to an action viewed as present in relative time, those
which follow express the action as past in relative time
or unreal in modal character (cf. below):
e) all verbs expressing the action as past, used with
a singular actor of first or third person: (/, he, John) was,
had, ate, sang, loved, danced, —
rarely homonymous with
a), b), d), e. g let, put, cost; f) all past tense verbs used

with an actor in the seco nd person or in the plural num-

ber, and all verbs referring to an action as not reaUy
 MORPHOLOGIC CLASSIFICATION etc. 139

taking place: ivere, had, ate, sang, loved, danced, — homon-

ymous in all instances except tvere with e).

Within each of these six classes there are, as among

the nouns, sub-classes by phonetic and semantic parallelism.
To take only one instance, class e) has one largest sub-
class within which all words have a final [d] (loved, trudg-
ed), [t] (danced, passed), or [ad] (rested, waited), — these
suffixes vary automatically, — as opposed to the corre-
sponding forms of a), b), d). Other sub-classes show other
forms *of dental-addition, e. g. sent, lent, etc. or shoidd,
would, could, where each also stands in a smaller class by
itself. Still other classes lack the dental, differing from the
present-tense forms usually by change of vowel; e. g. such a
class as sang, rang, dranlc, san'k, etc. (present: sing, ring,
drinlc, sink ). Another class is formed by the few instances
of homonymy with the present (let, hit, beat, cost, etc.). The
two past verbs which bear no relation whatever, formally,
to the corresponding presents also each form a class: was
and went. The present-forms corresponding to each of
these classes again form a class.
6. Difference between morphologic classification
and non-linguistic association. The ways in which a
morphologic word-classification may express itself, then,
are various. Nevertheless, it is always possible to recog-
nize a morphologic class, as opposed to a non-linguistic
psychologic connection. sew closely associates needle,
If
the connection is not linguistic, for the two words belong
in English to no one morphologic class, not even to the
same part of speech. On the other hand, if go is closely
associated with went, this association receives linguistic
expression, for the semantic relation between these two
words so habitually receives expression by phonetic simi-
larity (dance: danced; sing: sang) that in this one instance
the lack of phonetic similarity does not disturb the usual
140 MORPHOLOGY

feeling of coherence of past-form and present-form: go:

tuent fall into a morphologic class by their parallelism,
semantically, with dance: danced, sing: sang, etc. If the
English language possessed no other pair of words that
stood to each other in the same semantic relation as go:
iventj or if there were other such pairs, but phonetic re-
semblance between them were everywhere as totally out
of the question as in this instance, —
then, to be sure,
go and went would not fall into a common morphologic
class beyond that of verbs.
7. by composition. The most explicit ex-
Classes
pression of a classification of words is the likeness of
compounds to simple words and to each other, as when
hed, hedsheet, hed-cover, hedpostj bedroom, bedridden, etc., or
bedroom, dining-room, room, etc., fall into a class. Of this
type of word-classification we shall speak later.
Derivation and inflection. From the survey which
8.
we have just made of the principal types of morphologic
classes, it appears that most commonly, when a number
of words fall into a morphologic class, they present some
phonetic resemblance to one another, and, of course, some
phonetic divergence. That is, they differ formationally,
e. g. flame: flash: flare or hoys: stones: fathers or hoy:
hoys: boyish.
In grammatical writings about English and the langua-
ges possessing a similar morphology it has become usual
to distinguish two kinds of formational differences, accord-
ing to the semantic nature of the classification. If the
words have in common an element expressing material
meaning and differ only in an element of relational con-
tent such as is categoric in the language (e. g., in Eng-
lish, number or tense), it is customary to speak of them
as different 'forms' of one 'word' and of the relation be-
tween them as inflection. Thus boy: hoy's: hoys or eat:
 COMxMONEST MORPHOLOGIC CATEGORIES 141

ents: eating: ate: eaten are examples of inflection. If the

words have in common an element expressing material
meaning, but differ also in such an element, the relation
between them is called derivation. Examples are flash:
flare: flame or flash: crash: dash or hoy: hoyish or eat:
eater: eatable, and so on, provided always that the relation
between the words shall be not merely a difference of
categoric function. If the words have in common only
a relational element, as hoySy fathers, stones, etc., it is

common to call them 'parallel forms' of different words.

In the habits of speakers words related by inflection
are very closely associated with each other. For the naive
speaker, taking the categories of his language for granted
as the natural and inevitable forms of expression, feels
the inflectionally different words (or forms'), e. g. boy:
boys, as necessary variations in the expression of a ma-
terial content. The inflectionally related words are for
him really 'forms' of one 'word', — 'forms' made neces-
sary by the exigencies of expression.
9. The semantic nature of inflection: the common-
est categories. Inflection, therefore, could be defined as
variation between words to express relational differences
which involve appurtenance to different categories. What
is inflection in one language may, of course, be nothing

of the sort in another, where the categories are different.

It will be worth while, then, to mention some of the
relations which are expressed by inflection in different
languages, — that is, to mention some of the commoner
morphologic categories. Some of these we met in the third
and fourth chapters, where, however, we were interested
in the general rationale of relational expression, rather
than in the ground covered by the individual categories
Number. Among the English parts of speech the nouns
have the categories of singular <ind plural number. Keurlj
142 MORPHOLOGY

every noun must express whether one ohject or a plural-

ity is meant: we may say that each of these Vords' has
two inflectional 'forms', a singular and a plural.
Some languages distinguish three numbers, singular,
dual (for two objects), and plural, an instance being the
Sanskrit devah 'a god, the god, god', devdii 'the two gods,
two gods', devah 'the gods, gods (more than two)'. An-
cient Greek also had inflection for these three numbers,
and the singular-plural distinction was categoric, as in
English. The distinction between dual and plural, how-
ever, was not categoric; the dual, in the writings that
have come down to us, is used only of such objects as
usually exist in pairs (osse 'eyes', Icheire 'hands', etc.) and
even there is not obligatory, occurring, indeed, less often
than the plural. The contrast, in this respect, between
Sanskrit and Greek is instructive: the categories represent
obligatory forms of expression, the element which the
different forms express being always associatively perceived
in the experience; a non-categoric distinction receives
expression only where the element involved (here, in Greek,
duality) is vivid enough to enter into the analysis in spite
of the lack of a regular habit in this direction. It accord-
ingly can serve logical or esthetic impulses of the speaker;
the dual of Ancient Greek, as lovers of Greek literature
will testify, appears as one of the many graces of that
tongue, while in a Sanskrit utterance the use of the dual
is esthetically a matter of indifference. have seen We
(p.108) that in Chinese, for instance, the expression of
the number of an object-idea is by no means obligatory,
the category of number being absent.
Gender. It has already been mentioned that English
has no categories of noun-gender. The genders of German
appear in the congruence of the adjectives and pronouns.
Like German Latin, Greek, Sanskrit, and the Slavic langua-
 COMMONEST MORPHOLOGIC CATEGORIES 143

ges have tliree genders.The Romance and the Scandi-

navian languages have two, ani^some of the Bantu lan-
^guagex (it^Air-ica _go as high as twenty-one, e. g. the Su-
biya: here relations of number and person are viewed as
coordinate with the purely emotional gender-distinctions,
there being a category for the speaker, one for the speaker
and those with him, one for the person addressed, one for
the person addressed and those with him, one for a single
person, one for several persons, one for one small thing,
one for several small things, one for abstracts, and so on.
Case. Case appears as a category in English especially
in the personal and anaphoric pronouns, which vary accord-
ing to the function of the object in the sentence: as actors
appear I, he, she, they, as objects affected by the action
of a verb or as objects with regard to which a preposition-
al relation is expressed, me, hiniy her, them, and as attrib-
utive possessors my, his, her, their. In the nouns there
is inflection for the first and third of these relations only,
and the possessive form is limited in occurrence almost
entirely to nouns denoting living beings {John: John's;
father: father's)] nevertheless, the obligatory inflection
of the pronouns forces the speaker to make a constant
(categoric) distinction between these three relations.
German has four cases, the objects affected by a verb
being divided into two categories, that of objects fully
affectedand that of objects less fully affected: e. g. Er
gab mir das Buch 'He gave me the book', where mir, as
object less fully affected, is in the 'dative' case, das Buch,
as object fully affected, in the 'accusative' and Er schlug
mich 'He beat me', where mich is in the accusative. The
prepositions in German also vary as to the case they: de-
mand: Er legte das Buch auf den Tisch 'He laid the book
on the table', den Tisch accusative: Es liegt auf dem Tische
'It is lying on the table', dem Tische, dative.
144 MORPHOLOGY

The which English or German

local relations of objects
analyze fully out of the experience and express by inde-
pendent words (prepositions), are in many languages in-
flectionally included in the object-word, which thus varies
categorically according to them, —
so that there may be
a great number of cases. Thus, in Sanskrit, devah 'god'
corresponds to our 'nominative' (7, he), devam to the Ger-
man accusative, expressing the object as fully affected,
devaya to the German dative, expressing the object as
less fully affected by the action, devasya to our possessive,
'the god's'; but there are also a number of further cases:
the 'instrumental' devena expressing the object as a means
or an accompaniment ('by means of the god' or 'with
the god'), the 'ablative' devat expressing it as that from

which ('from the god'), the 'locative' devSj as that in or

near which ('by the god' or 'in the god'); and to these
comes also a special form, the 'vocative', for interjectional
use, as in calling, deva '0 god'. (See also pages 76, 92).
The Latin cases have already been mentioned (p. 108).
The number of cases in some languages, especially some
of the Uralic, such as Finnish (cf. p. 108), is greater, but

the principle everywhere the same.

is

Tense. We have in English two tenses of verbs, past

and present. Some European languages add a third tense,
a future, e. g. Latin canto '1 sing', cantabam 'I sang',
cantabo 'I shall sing' (cf. p. 68). Future action is in Eng-
lish, as the translation shows, analyzed into the independ-

ent words shall or willy which express futurity with a

suggestion, respectively, of obligation or intention (in the
present time), or, else, the present tense is used: Tomorrow
we die. The category of tense is in many languages, as,

for instance, in Chinese, entirely absent as such, time-

relations being expressed by independent words (p. 68).
Manner {Aspect). A much commoner basis of cate-
 COMMONEST MORPHOLOGIC CATEGORIES 145

gories than tense is manner of action. It does not exist

as a morphologic category in English, where we either
ignore the manner of an action or else analytically ex-
press it. Thus / am writing, I was writing are expressive
of 'durative' action, I have written, I had written of ^per-
fectic' (completed) action, I often write, I used to write
of 'iterative' action, / wrote it down of 'final^ terminative',
He hurst out weeping of 'inceptive terminative' action, and
/ once wrote of 'punctual' action. Many languages, how-
ever, express the manner of action in the same word with
the action itself, making categoric distinctions between
the different manners. Thus the Slavic languages distin-
guish categorically between, on the one hand, durative
and iterative (in Slavic gi-ammar called, together, 'imper-
fective') action, e. g. Russian [p'i 'su't'] 'to write' and
['p'r si vet'] 'to he wont to write, to write repeatedly',
and, on the other hand, punctual and terminative action
(in Slavic grammar, together, 'perfective'), e. g. Russian
[na p'i 'sQ't'] 'to write (once), to write down', [sp'i 'so't']

'to write off*, i. e. 'to copy', [pr'i p'i 'so't'] 'to write over',
i. 6. 'to sign away'.
Voice. Another not found in English
set of categories
are the voices or conjugations, such as 'active', 'middle',
'passive', 'causative', 'applicative', and the like. Thus, in
Latin amat, active voice, is 'he loves', amatiir passive voice,
'he is loved, is being loved', or in Greek elyse 'he freed'
is active, elythe 'he was freed' is passive (actor as suffer-
ing the action), and elysafo 'he freed himself or 'he freed
for himself (e. g.'elysato ten ihijgatera 'lle-freed-for-himself
the daughter', i. e. 'He freed his — own — daughter')
is middle (actor as acting upon himself or for himself).
In Sanskrit the active voice shows the following 'con-
jugations': pdtati, normal, 'he falls'; pdidyati, causative,
'he causes to fall, fells'; papatUi, intensive, 'he falls hard'
Bloom field, Stady of Language 10
146 MORPHOLOGY

or 'he falls repeatedly'; pipatisatiy desiderative, 'he wishes

to fall' or 'he is about to Middle and passive forms
fall'.

run parallel to the active; thus the passive of the above

causative is pdtyate 'he is being felled, he is felled'; of.
further, ydjati 'he sacrifices' (used of the priest who
sacrifices in another's behalf) active, ydjate 'he sacrifices
(for himself)' middle, and ijydte 'he (it) is being sacrificed'
passive, all three being of normal conjugation. In both
Greek and Sanskrit middle and passive are in a large
part of the forms homonymous; in Modern Greek there
is no middle voice.

The applicative conjugation is frequent in American

languages; it expresses the action as applying to some
person or thing that would not be involved, were the
action-word used in normal conjugation. Thus in Na-
hwatl the normal ni-petla-tsiwa 'I-mat-make', i. e. 'I make
mats' or ni-Jc-tmva 'I-it-make', i. e. 'I make it' (as in
ni-Jc-tsiwa se kali 'I-it-make one house', i. e. 'I build a
house') has by its side an applicative ni-lc-isiwi-lia^ as in

ni-k-tsiwi-lia in no-jnltsm se JcaXi 'I-it-make-for the my-

son one house', i. e. 'I build a house for my son', with
two objects affected instead of one. The applicative, in
applying an action normally without objects affected to
such an object, often coincides in meaning with the caus-
ative of Sanskrit, as in ni-miJd 'I-die', ni-k-mih-tia 'I

make him die, kill him'. As the English translations

show, we lack these categories, looking upon the various
forms of action either as upon totally different experiences
(die: kill), giving them an indifferent derivational ex-
pression (fall: fell), or analyzing the relation {sacrifices:
sacrifices for hiinsclf: is sacrificed, etc.).
Mode. One verb keeps alive in standard English a
categoric distinction of mode, namely the verb to he. In
he were, as opposed to the actual or 'indicative' he is,
 COMMONEST MORPHOLOGIC CATEGORIES 147

he was, we have an 'unreaP mode, and, as it is not stand-

ard English to say, for instance, If he was here, he would
help us, we may call the distinction of mode categoric.
Those dialectic ('illiterate') forms of English which do
not use the form he were have lost this category: in them
the 'past' tense-forms are expressive not specifically of
action in the past, but of any action not present, be it
past or viewed as unreal; they have a present and a past-
unreal form, merging what are in standard English the
categories of mode and tense. In older and still to some
extent in literary English a third mode, an 'optative',
also exists, and is used to express action as possible. It
is homonymous with the infinitive, from which it differs
by the precedence of a subject-actor: If he he there, Be
he live or he he dead, as opposed to he is, he were.
In Ancient Greek three modes were categorically
distinguished. An action viewed as really occurring was
in the indicative: pherei 'he carries', touto gignetai 'this
happens', while actions not so viewed by a categoric
fell,

distinction, which, however, was in part merged with

congruence-relations of tense, into the 'subjunctive' : hina
pherei 'so-that he-may- carry', phoheitai me touto genetai
'he-fears that this may -happen', or into the O'ptntiY e pheroi :

an 'he might carry', ei touto genoito 'if this should-happen'.

The English translation shows how we analyze such
modal relations by means of words like may, can, shoidd.
German also has three modes, an indicative: Er ist Icranlc
'He is sick', an optative: Er gehe 'Let bim go', Man
sagte, er sei JcranJc 'Tbey said he was sick', and an unreal:
Wenn er JcranJc ware 'If he were sick'; German grammars
call the last two the 'first' or 'present' and 'second' or
'past' subjunctive.
Both Ancient Greek and German have, like many other
languages, a special set of imperative forms for intcr-
10*
148 MORPHOLOGY

jectional use of the verb incommands: German geli! 'go*,

Ancient Greek phere 'carry', phereto 'let him carry', where
the English equivalent is either the infinitive form or,
for the tliird person, the infinitive of the verb let which
directs the command, by an analysis of the situation, at
the person spoken to.
Actor. We have seen that the English verb varies in
form according to the person and number of the actor.
The variation according to number (^The hoy shales, The
hoys shate) is properly a phenomenon of congruence with
the number-category of the noun or pronoun expressing
the subject-actor. The same may be said of the variation
according to person {I am, you are, he is), although there
can hardly be said to exist a categoric system of 'persons'
in nouns and pronouns, since there is only one first-
person pronoun (7, we), and only one for the second
person (you): the 'person' of these words is simply their
content as words.
In some other languages, as we have seen (p. 107),
words expressing action really include personal-anaphoric
mention of the actor. Thus Latin verb-forms such as
edo '1-eat', edimus
edis 'you-eat', edit 'he (she, it)-eats',
'we-eat', etc., do not vary in mere congruence with cat-
egories of an actor, but actually include mention of the
actor, who may not in any other way be expressed. The
Latin verb, then, expresses not only an action, but an
actor and an action, and just as it has categoric variation
according to tense, mode, and voice of the action, it also
varies categorically according to number and person of
the actor.
Goal (Object affected). In other languages the action-
word includes the objects or object affected by an action,
or these together with an actor. This phenomenon is
known as 'incorporation'. Here the action-word may, of
 COMMONEST MORPHOLOGIC CATEGORIES 149

course, be inflected for person and number of the object

affected. Wehave already seen such forms from Green-
landish (p. 104) and from Nahwatl (p. 135). The follow-
ing forms from the latter language will further illustrate
this inflection:
ni-mits-maUtia 'I- thee- teach', 'I teach thee',
n-amets-matst'ia ^I-ye- teach', ^I teach ye',
ni-k-maUtia 'I-him-teach', 'I teach him',
ni-kin-matstia 'I-them-teach', 'I teach them*,
ti-nets-matstia 'thou-me-teachest'^ 'thou teachest me',
ti-tds-7natstia 'thou-us-teachest', 'thou teachest us',

ti-k-matstia 'thou-him-teachest', 'thou teachest him',

ti'kin-matstia 'thou -them -teachest', 'thou teachest
them',
nets-matsiia 'me-teaches', 'he teaches me',
and so on. Two objects affected are seen in ni-te-tla-maka
'I-someone-something-give', 'Isomeone something',
give
ni-k-tla-maka 'I-him-somethinor-oriYe', 'I give him some-
thing', ni-k-maka 'I give it to him'. Here we see a three-
fold inflection: for actor and for two objects affected.
Possessor. With object-words the person, number, and
even gender of another attributive object may be expressed
(p. 107, with example from modern Arabic), and this ex-
pression may be categoric. In Nahwatl, for instance, one
cannot say 'mother' or 'hand' without expressing an at-
tributive (possessing) object, as in no-nan 'my-mother
or to-ma 'our-hand'; one can also say te-nan 'someone's,
an uncertain person's mother'. In some hmguages this
applies to every. object-word, so that one cannot say, for
instance, 'house', but only 'my house', 'his-house', 'an-
uncertain-person's-house', or the like.
An inteiesting phenomenon found in some languages
is the fusion of the categories of possessor of an object
with those of performer of an action. In the Innguage
150 MORPHOLOGY

of the extinct Lules in South America, for instance, the

following showed parallel inflection:
umue-s my mother amaitsi-s I love
umue-tse thy mother amaitsi-tse thou lovest
umue-p his mother amaitsi-p he loves.
No English translation, of course, could do justice to
this complete merging of what we analyze as two entirely
different relations. In other languages the possessor of
an object is fused with the object affected by an action;
thus in Greenlandish these and the actor also are in part
expressed alike: [il:u-a] 'house -his', 'his house', [il:u-t]
'thy house' show the same inflectional endings as
[tusaRp-a-t] 'sounding-his-thine', i. e. 'thou hearest him'.
All this illustrates the vast divergence as to the semantic
character of inflection.
10. derivation. When the
The semantic nature of
relation between words of a phonetic-semantic class is
not a difference of category, we call it derivation (p. 141);
thus the relation between flamej flash, flare, etc. or be-
tween hullj ImllocJc, we have seen, is derivation. In such
instances as the latter it is frequently said that the longer
word is 'derived' from the shorter, or a 'derivative' of
it. This mode of expression is permissible, as long as
one does not allow it to affect one's view as to the his-
torical priority: historically it is quite possible in such
cases that the longer word existed before the shorter
one, which then, in reality, may have been derived from
the longer.
As to the semantic values of derivation, it is impos-
sible to set limits, or even to quote, as in the case of in-
flection, some of the commoner relations expressed. Al-
most any material relation may be expressed by it. The
following sets of derivatives may illustrate this multi-
plicity:
 THE SEMANTIC NATURE OF DERIVATION 151

Englisli flamej flashy flare, flimmer, fliclier,

English drip, drop, droop, dribble, drabble.

English clash, crash, dash, flash, gash, gnash, hash, lash,
mash, plash, slash, splash.
English dribble, nibble, quibble,
English tend, tense, tension, tensity, tenseness, intense,
intensive, intensity, intend, intent, attend, attention, ifiatten-
tion, attentive, inattentive, etc. No doubt tend is moreover
associated, for most of us, with trend.
From theNass dialect of Tsimshian (British Columbia):
[haliiei] 'to walk along the edge of the water', [wi ts'eniei] :

'to walk back through the house', [aldaie:] 'to walk in

the dark', and so on.
Crossing this class: [aldawa:9] 'to paddle in the night',
[aldaie:] 'to walk in the dark'.
11, The phonetic character of the morphologic
processes. The formal phase of morphology includes
every conceivable phonetic variation.
This phonetic variation is to be sharply distinguished
from automatic sound-variation. Whether we say [ju] as
in Will you? or [Ju] as in Won't you? [wilju, wowntju],
has nothing to do with our meaning (and is therefore of
no grammatical significance) but depends entirely on the
nature of the sound we have been uttering when we come
to the you. The same is true, for instance, of the Sanskrit
sandhi-variations (p. 102). On the other hand, whether we
say dash or 7nash or plash or splash, and whether a Ger-
man says der, die, or das, and an Irishman bo, v6, or mo,
is of decided significance. In these English words the
material content varies with the difference of form; in
the German and the Irish the morphologic category of
other words is involved.
Pitch-variation. Pitch-variation for derivation can occur,
of course, only where the pitch-relations within a word
152 MORPHOLOGY

are fixed. found, for instance in Norwegian and

It is
Swedish; thus in the former language: ['skri: var/] write, *"

writes' (present-tense form of verb): ['skri:\ var/] Vriter'.

In Chinese there are a great many words distinguished
only by their pitch-relations, and in sonie instances such
words are derivationally connected in the feeling of the
speakers, e. g. [""nan/] 'difficult' and Lnan\] 'suffer', Lmas/]
'buy' and [Lmas\] 'sell'.

Stress-variation. The place of the stress in a word is in Eng-

lish significant and consequently can be used for morpho-
logic sound-variation. As the quality of certain vowels fur-
ther depends automatically in English on the place of the
stress, the following examples illustrate both morphologic
and automatic sound-variation: accent, noun ['aeksnt]: verb
[ek 'sent]; — address, noun ['sedjQs] or ['aedas]: verb [a
'djes]; — overthrow, noun ['owvjOjo]: verb [ovj '0jow].
Similarly, in Russian: ['ru'ki] 'the hands': [ru 'l^i'] 'of
the hand'; — [uzna ju'] 'I recognize': [u 'zna' ju] 'I shall
recognize'. Place of stress is interesting in Ancient Greek:
in the verb it varies automatically and cannot, therefore,

be of morphologic significance; in the other parts of speech

it is free and receives morphologic employment, e. g. in

tomos 'slice': tomos 'cutter'.

Variation of articidations. Vowel-variation is common

in English morphologic expression: goose: geese; man:
men; mouse: mice; woman [wuman]: ivomen
foot: feet;
[wimen]; eat: ate; sec: saw; sing: sang: sung: song; ride:
rode; read, present tense [jijd]: read, past tense [^ed]; sip:
sop: sup: seep: sap; drip: drop: droop; sniff: snuff; snip:
snap, — and so on. The terms 'umlaut' and 'ablaut' are
used in the grammar of English and the related languages
as designations for certain cases of vowel- variation: 'um-
laut' for our vowel-variation for number in the noun (niouse:
 PHON. CHARACTER OF MORPHOL. PROCESSES 153

mice)y 'ablaut' for that in the tense-inflection of the verb

(sing: sang).

Consonant variation also is common in English: we

have seen it in our example of clash: crash: dash: flash, etc.;
also in the Irish ho: v6: mo and the like. Further Eng-
lish examples are have: has: had; crash: crack; squeak:
squeal; sqiiaivk: squall; lend: hent; send: sent Another
example is the Kafir word for 'all' which varies in con-
gruence with the gender —
the Kafir is a Bantu lan-
guage, cf. p. 143, — of the word it modifies: honke: lonke:
yonJce: zonke: wonke: konke.

It is common to find sound-variation in both conso-

nants and vowels, as in flame: flash; crash: creak; ivas:
were; ivill: would [wud]; can: could; German schneiden
'to cut' ['Jnaedn]: schnitten 'were cutting' ['Jnitiji], and
the like.

Affixation. Somewhat different from sound variation is

the plus or minus of sounds seen in such groups of words

as sing: sings: singer: singing or man: manly or hull:
hullock. This kind of word- variation is called affixation,
the phonetic element that is common to a set of words
related by affixation being spoken of as the kernel, the
elements present or absent in the different words, as the
affixes. Thus group among our instances sing-
in the first
is the kernel and -5 [z], -er, and -ing are affixes. Here

again our terminology is metaphoric: there is no reason

to believe that the longer words of a group related by
affixation necessarily arose from the shortest word by
any actual process of 'affixing' or adding phonetic elements:
that is a question of historic fact which is not answered
by our use of the terms 'affix' and 'affixation'. Instead
of 'affix' the terms 'determinative', 'formative', and 'for-
mans' are current in certain branches of grammar; and
154 MORPHOLOGY

the kernels of certain word groups in certain languages

are called 'bases', 'stems', and 'roots'^).
The difference between sound-variation and affixation
is not an absolute one, but depends frequently on our
point of view. We might call Latin amas 'thou lovest':
amat 'he loves' an example of sound-variation; if, however,
we take into view some other forms of this verb, such
as amo 'I love', amavit 'he loved', ametur 'he may be loved',
it is possible to call them all related by affixation, the
kernel being am-, the affixes in the quoted forms, -as, -at,
-0, -dvit, -etur. In this instance the group does not contain
a word that equals the kernel, or, as we might say, has
'affix zero', — a condition fulfilled by the first of the
quoted forms in the example of English sing: sings, etc.
An English group like flash: flame: flare: flicJxer might
be regarded as an example either of sound-variation or
of affixation to a kernel ft-.
In aU these examples the affix appears at the end of
the word and is, specifically, a suffix or ending. Other
examples of s affixation are the German article-forms der:
die: das (with kernel d-), English hull: hulloch; man: man's:
manly: mannish; hoy: hoys: hoyish, and so on.
Frefixation borders, like suffixation, on sound- variation:
thus the forms of the Kafir word for 'all', honlfe: lonke:

1) As to the last of these terms, it may be well to warn the

reader against attributing to it any mystical character. Fifty
years ago most students of language believed that the kernels
they had abstracted from some of the ancient languages possessed
a unique validity and age: that, in fact, they or their like were
the original elements with which language had begun and out
of which its more elaborate forms were pieced together. Today
we mean by all the above terms simply certain phonetic ele-
ments common to sets of formationally related words, and do
not allow our terminology to commit us to any view as to the
histoiic origin of these related words.
 PHON. CHARACTER OF MORPHOL. PROCESSES 155

yonke: zorike: wonlce: Jconke are often spoken of as a kernel

-nke with different prefixes, —
owing to certain morpho-
logic features which prevail in the language. Other ex-
amples of prefixation are, from the English, kind: unkind;
speak: bespeak; septic: aseptic, and the like. Flash: splash;
mash: smash; lash: slash, and the like may be viewed
as sound-variation or as prefixation, —
the former be-
cause the initial group of s plus consonant is exceedingly
frequent in the language.
Infixation is affixation within the word. We may view
Latin amat 'he, she, it loves': aitiant Hhey love' as infixa-
tion, the kernel being ama-t, the infix, -n-; on the other hand,
if we include in consideration the above-quoted forms
amo 'I love', amavit ^he loved', ametur 'he may be loved',
and so on, we shall speak rather of suffixation of -at, -ant,
-0, etc. to a kernel am-. Less pliable examples of infixation
are English clap: clamp and Latin fidit 'he split': findit
'he splits', scidit 'he rent': scindit 'he rends'. Another
example of affixation may be seen in Sanskrit, e. g. in
yuktdh 'yoked, bound': yunktdh 'they-two yoke, bind':
yundkti 'he (she, it) yokes, hinds'; the kernel of the word-
group to which these forms belong is here seen as yu-k-,
with the suffixes -tah and -ti and the infix -wa-; -w-.
These Sanskrit forms illustrate also the phenomenon
of automatic sound-variation in affixes. The infix is here
-na- when accented, but -n- when unaccented, a varia- —
tion by no means universal in the Sanskrit language. We
have a few similar instances in English, where, for in-
stance, our plural-suffix of nouns, our possessive suffix
of nouns, and our third-person singular congruence-suffix
of verbs all have a homonymous form which appears as
[az] after sibilants, as [z] after non-sibilant voiced sounds,
and as [s] after non-sibilant unvoiced sounds:
plurals: watches, peaches, — hoys, fathers, — hats, cliffs;
156 MORPHOLOGY

possessives: boss's^ Madge's, — John's, father's, — counfsy

Pete's;
third-person verbs: watches, dances, — loves, hears, —
waifs, counts.
Yet the language as a whole cannot be said tohave such
an automatic sound- variation as this: the automatic va-
riation is peculiar to these suffixes.
Affixation may of course be accompanied by morpho-
logic sound-variation; instances are the German tvill Vants
to': ivollen Svant to' (suffix -en, with vowel-variation i: o);
English child: children (suffix -orn with vowel-variation
[ae]: [i]); or the different affixational processes may occur
together, as in Ancient Greek lamhdno 'I take': clahon
'I took' (kernel -la-h-, prefix e-, infix -m-, suffixes-awo and
-on; the shifting of the accent is here automatic).
Reduplication. A peculiar kind of affixation is redupli-
cation, which consists of the repetition of the whole word
or some part of it. Examples are Malay tuwan 'master':
tmviin-tuivan 'masters'; Latin quis 'who?': quisquis 'whoev-
er'; Sanskrit hharti 'he carries': hhanlharti 'he carries
hither and thither' (reduplication with infixed -^-); Latin
tendit 'he stretches': tetendit 'he stretched'. Reduplication
is often irregular, approaching the other forms of affixation,
e. g., connected with the above Sanskrit hharti a form

jahhdra [jabhaira] 'he carried' (kernel -hh-r-, prefix or

irregular reduplication ja-, vowel-variation a: a, suf-
fixes -ti and -a); Latin pello 'I drive': pepull 'I drove'
(reduplication with sound-variation and different suffixes).
The boundary between reduplication and the other forms
of affixation is thus not a sharp one: English do: did
could be looked upon as an example of either process.
noteworthy that reduplication is confined to cer-
It is
tain narrow spheres of meaning. It often denotes increase,
changing, for instance, words in the singular number to
 PHON. CHARACTER OF MORPHOL. PROCESSES 157

Malay
plurals of different kinds, as in our instance of the
tuwan-tuwan ""masters'. Other examples are Japanese Jama-
jama 'mountains' {jama 'mountain'); Nahwatl in tsatsan
O'jajalce '(into) the his-houses they-each-went', i. e. 'They
went each into his house', as opposed to in tsan o-jake
'they went (together) into their (one) house'; oldts-pil
'little man', plural okits-pipil 'little men'; Tsimshian [aljix]

'to speak', [el'aljix] 'to speak, of several people'; [Gan]

'tree', [GanGan] 'trees'. It may further denote repetition,
continuity, or intensity: Latin qicisquis 'who-who', i. e.

'everytime who', 'whoever'; Sanskrit dame-dame 'in every

house', as opposed to dame 'in the house': Greek ehen 'I
stepped': ebiben 'I made repeated steps, walked'; Sanskrit
hJiarlhharti 'he carries hither and thither'; Nahwatl Icotona
'cut': Iwkotdna 'cut in many pieces': JcoJcotona 'cut many
things' English snip-snap, fiddle-faddle, slip-slop, flip-flop.
;

Occasionally it has diminutive sense, as in these English

examples, where, for instance, flip-flop expresses a less
violent movement than
flop; the same is true in Dayak
(a Malayan language of Borneo) of hai 'large': hahai
'fairly large'; handang 'red': hahandang 'reddish'. Desire
for an action as opposed to the action itself we see in
Sanskrit jlmmi 'I live': jijivisami 'I desire to live'; vid-
mah 'we know': viviisati or vividisati 'he wishes to learn',
compare page 146. Finally, it expresses perfectic
also
action, as in the Greek leloipa 'I have left' as opposed to
the durative leipo 'I am leaving', and, rarely, but in this
use familiar to us from Latin, it expresses past tense:
Sanskrit ja/^Mra 'he carried', Latin tetendit 'he stretched',
pepulit 'he drove'.
Homomorphy and snppletion One and the same se-
mantic relation between two words, —
such as in Eng-
lish the difference of present and past tense, may find —
expression in the most various formal processes. Thus in
158 MORPHOLOGY
the present-past inflection of the Eno:Hsh verb we find
suffixation (dance: danced), vowel-variation {sing: sang),
consonant-variation {send: sent), and so on. saw in We
paragraph 5 that all words making a given inflection in
the same formal way fall into a class by phonetic-semantic
parallelism.Thus our 'regular' verbs with the suffix -ed
for the past form a large class, as do also our diff'erent
groups of 'irregular' verbs. The same can be said of the
plural-inflection of our nouns.
Among the small 'irregular' classes so singled off, the
formal relation between inflections may be that of iden-
tity. This phenomenon is called homonymy or, more
specifically, homomorphy. For instance, among the Eng-
lish verbs there are a few which are alike in the two
tenses: cost, hit, heat, put, These unchanging verbs are,
let.

however, so much in the minority that we do not ordina-

rily realize that their present and past tenses are alike.
The identity of the two forms constitutes simply an irreg-
ular kind of inflection, owing to the vast preponderance
of verbs that do vary. In the noun-inflection homomorphy
is seen in deer, sheep, fish.
The opposite of homomorphy is which con-
suppletion,
sists of the entire absence of phonetic relation in some
few members of a large class by semantic parallelism;
that is, where the whole class is so numerous that in the
few cases where the 'forms' are not formally related, we
still feel that they belong together. Among the English

verbs, for instance, we find the inflectional forms he, am,

are, is, was, were which in spite of complete dissimilarity
belong together by parallelism with the other verbs. An-
other example is go: went (p. 139, f.). In the forms of the
adjective we have, beside the regular type, Mnd: Jcinder:
Mndest, the suppletive sets good: hettcr: hed, had: worse:
worst. Similarly, in the relation between adjective and
 WORD-COMPOSITION: SEMANTIC VALUE 159

derived adverb, beside the regular Icind: hndly, rapid:

rapidly, etc., the suppletive good: well, and in that between
cardinal numeral and ordinal numeral, beside the regular
four: fourth, six: sixth, etc., and tbe irregular three: third,

five: fifth, also tbe suppletive one: first, two: second. Supple-
tion is found, as a rule, in very common words, and irregu-
larity also, though in a far less degree, tends to confine
itself to these.
12. Word -composition: semantic yalue. In para-
graph 7 I mentioned the relation of compounds to each
other and to simple words as the most explicit expression
of morphologic classification.
Word-composition consists of the use of two or more
words in a combination that has a diJOPerent meaning from
that of the simple words in syntactic collocation. This
may be illustrated by a few transparent Erglish examples.
Our word long-nose, as in I can't stand that long-nose (mean-
ing a person) differs from a long nose, for it means not
a nose of this shape, but a person having such a nose.
Shorthand does not mean what the words short hand as
separate successive words in syntactic collocation would
mean, but, instead, is used of a certain kind of writing.
A different kind of deviation from the meaning of inde-
pendent words appears in bulldog. This compound, to be
sure, does designate a dog: but hidl, a noun, could not in
English syntactic collocation, modify another following
noun, such as dog (p. 96).
The problem of any way classifying compounds is
in
an exceedingly difficult one, because the material and log-
ical relations between the 'members' of compounds are,
even within one and the same language, often well-nigh
endless in variety. Perhaps the most justifiable basis of
classification is that which distinguishes compounds which
in form resemble a syntactic word-group and those which
160 MORPHOLOGY

do not. By this classification long-nose, shorthand, crows-

foot,man-of-war^ for instance, would belong to the former
class, for, though diverging in meaning, these compounds
externally resemble such collocations as long nose, short
hand, crow's foot, man of ivar (as opposed to man ofpeace).
To the second class would belong bulldog, apple-tree, sofa-
cushion, and the like, which do not resemble syntactic
groups. The compounds of the latter class can further be
divided according to whether they describe an actually
present feature of the experience, like apple-tree, or, like
hidldog, express merely an associative element entering
from past experiences: the apples are perhaps visible on
the tree, but the dog merely reminds one of a bull.
While the specific values of compounds as opposed to
collocations of simple words vary greatly even within a
single language, yet, when we look at other languages,
we find that each one has certain limitations. As an instance
of a kind of composition not found in English, the follow-
ing Nahwatl examples may serve: simple words, naJiatl
'meat' (or 'It is meat') and nikkiva 'I eat it': compound,
ni-nalxa-kiva 'I-meat-eat'; in English we form no veibal
compounds of this type, though we freely parallel them
in our nouns, such as meat-eater. A further example from
Nahwatl is ni-sotsi-temoa '1 seek flowers': simple words,
sotsitl and niJdemoa 'I seek it'. These Nahwatl
'flower'
compounds diff'er from simple words, in that the simple
words can always express a predication a-piece: nikktva
nahatl can mean 'I eat it. It is meat'
The following Sanskrit compounds, called in the grammar
of that language 'copulative', would also be impossible
in English: crlhih yavah 'barley' (both in the nom-
'rice',

inative case, singular number): vrihiijavdu 'rice and bar-

ley' (nominative case, dual number), as though we should
say 'rice-barleys' for 'rice and barley'. This Sanskrit
 WORD-COMFOSITION: SEMANTIC VALUE 161

compound differs from a syntactic collocation in that the

case and number are expressed not for each member sep-
arately, but only once, at the end, as in any simple word;
this gives a tone of unity to the whole. Other examples
of this type are: hralimana-lisatnya-vit-sudraliy nominative
plural, 'brahmans, kshatriyas, vai9yas, and ^udras' (men
of the four castes); or vrttali ^Youndi^pmdh 'plump': vrtta-
pinah 'round and plump'
In Chinese ["^^ug^ L^i\]? ^s we have seen means (p. 86),
'brother, brothers'; a syntactic collocation of the two
words ['^Qiug''] 'older brother' and [Lti\] 'younger brother',
if it occurred at all, mean only 'younger brother
could
of an older brother'. This differs from anything we have
in English; it would correspond to a compound brother-
sister in the meaning of 'brother or sister', 'child of the
same parents'. Similarly ['X9/lSZ9'\], as a syntactic collo-
cation would mean 'ten fours' (compare, for instance
[^szal 'jb/] 'four tens', 'forty'), but it is really a compound,
'ten-four', meaning 'fourteen'.
We may illustrate now some of the varieties of com-
pounds used in English. Most strikingly different from
the simple words in syntactic succession are the so-called
'exocentric' compounds, which denote an object having
the thing named in the compound, as long-nose, short-
horn, swallow-tail. Similar to these are the forms long-
nosed, short-horned, swallow- tailed, rough-shod; these are
in total effect nearer the simple words (compare, for in-
stance, tailed monkeys), but still, in their use of adjective
modifying adjective differ from a collocation of words,
where we use adverbs to modify adjectives (roughly shod).
The compounds consisting of two nouns have already
been mentioned We may name such further examples
as dog -pound, dust-rag, schoolroom, headache, summer-
house, man-nurse. The semantic relation between the
Bloomfield, Study of Language 11
162 MORPHOLOGY

parts is, it appears, of practically unlimited variety. We

have mentioned also the type in which an ohject affected
by an action and the actor or instrument (in the form
of a verbal noun) are joined together: meat-eater, lion-
tamer^ tennis-player, screw-driver, coat-hanger.
13. Word-composition not a phonetic process. It
will be noticed that nothing has here been said about
any formal difference between compounds and syntactic
collocations. If we recall the fact that the word, —
and,
therefore, by inclusion, also the compound word, is—
not a phonetic unit (p. 99), it wiU be clear that there
need not necessarily be a phonetic difference between a
compound and a succession of words. The difference be-
tween a single word, simple or compound, and a succes-
sion of words lies in the semantic value and not in the
sounds. Not even the stress -accent of a language like
English or German necessarily distinguishes a compound
from a succession of words. This is especially clear in
German, where such compounds as durchmessen 'to mea-
sure through, to cross' [dur^ 'mssn] do not differ in ac-
cent from such collocations as sich messen 'to vie, to
measure oneself [zi^ 'mssn]. In English we have the
same similarity under certain conditions of stress, thus
my mother with an emphatic my, resembles in stress-re-
lations the compounds godmother, grandmother, and simi-
larly mother of mine resembles mother-in-law.
Furthermore, a compound need not be phonetically
continuous. In Latin ne quidem 'not even' is a compound
word which may be separated, as in ne Caesar quidem
'not even Caesar'. Similarly, in German, wenngleich 'even
if, as in Wenn ich gleich schreie 'Even if I cry', and a great
many compound verbs like vorlesen 'to read aloud', as in
Ich ihmjeden Abend vor 'I read to him every evening'.
les'

In French we find such compound words as ne pas 'not'

 WORD-COMPOSITION NOT A PHONETIC PROCESS 163

in Je ne lui donne pas . . . *I don't give him . .

'
In English
we have such verb-compounds as bring out in the sense
of 'emphasize, make clear' (the simple words appear in
collocation in Bring out your golf-sticks), which are sepa-
rated in such sentences as You don't bring that out very
clearly.
All this shows us that the concept of a compound,
like that of a word is not absolutely definable. Is stand

off in Stand off, compound? It differs from the

there! a
ordinary use of stand, which excludes the idea of move-
ment; on the other hand, in view of stand up and stand
aside we might say that stand means not only *to he in
an upright position' but also ^to assume an upright po-
sition'. (Cf. p. 97, f.). That is to say, then, the differeuce
between compounds and sets of simple words is, like that
between derivation ally formed words and compounds
(p. 96), a matter of the speaker's associative disposition
which may vary from person to person and from hour
to hour.
While phonetic between compounds and
differences
simple words are thus by no means necessary, they are,
on the other hand, not uncommon. A number of examples
deserve mention.
In English compounds usually differ in stress from
successions of simple words. In general, our syntactic
groups of simple words tend to be evenly stressed, with
a highest stress on each word, while our compounds, like
all other single words, have a high stress on one S341able,
usually the first. Thus, for instance, we distinguish
phonetically between 'hulldc^ and the simple words in
'hull, 'dog, and between a 'crowsfoot and a 'crowds
'cat,

'foot, between a 'longnose and a 'long 'nose, and between

'bloodshed an I 'all the 'blood 'shed in the 'Civil War. '

It is evident that in languages that have a regulated

164 MORPHOLOGY

pitcli within the word, difference of pitch may figure il

phonetic divergence between compound and simple word.
Thus in Norwegian we find land [Ian/] 'country', mand
[man/] 'man': landmand [lan\man/] 'farmer, peasant,
countryman'.
It is common also to find compounds differing in
vowel and consonant articulations from the simple words,
e. g. fore and head, but forehead^ —
[fO'-i]? [bed] ['faiad];
:

German die Sonne 'the sun', genitive case der Sonne 'the
sun's', das Licht 'the light' das SonnepMcht 'the sunlight',
:

with an added -n-. Thus Sanskrit vrlhiyavau Vice and

barley' is in the genitive vrihiyavayoh 'of rice and barley',
where the lack of inflection of ^;n/^^-'rice', —
a form
which, moreover, never occurs as a separate word, —
distinguishes it both semantically and form from
in
vrihih 'rice', genitive vriheh. So mahadhanam as opposed
to mahad dhanam (p. 106). Likewise, in Ancient Greek
hylotomos means 'wood-cutting (adjective)' and is inflected,
for instance in the accusative, for congruence with a noun,
hylotomon: the first element is here uninflected: as an in-
dependent word, moreover, it is in the nominative hyle,
in the accusative hylen, and in no form hylo-.
The greatest divergence between simple words and
compounds appears in those languages which, owing to
an objective habit of expression, compounds are in most
frequent use (p. 104). Thus the Nahwatl compound above
quoted ni-naJca-kwa 'I eat meat' corresponds to the simple
words naJcatl 'meat, it is meat', and nikhwa 'I eat it'.
Thus, aside from the semantic divergences, there are the
following in form alone: -tl in simple word is left off in
compound; ni- at beginning of simple word appears at
the beginning of the compound; -Ic- (semantically 'it')
left off in compound.
Like the English forehead (where the spelling preserves
 SIMPLE WORD: COMPOUND: PHRASE 165

our feeling of connectioa with simple words), such Na-

hwatl compounds approach the boundary where the com-
pound would cease to be felt by the speaker to resemble
any simple words, and, consequently, would no longer
be a compound. This line is, of course, not sharply trace-
able. Those who know the word ['hAzef] 'sewing-bag'
from speech alone will scarcely feel it to be a compound,
as will those who know it from its written form house-
wife or in its 'spelling pronunciation' ['haoswaef].
14. Simple word: compound: phrase. This, then,
is the second direction in which compounds approach
simple words. On the one hand, we have seen instances
where it was doubtful whether a certain element was
merely an affix or a member of a compound word: In
fourteen, sixteen, seventeen, etc. (p. 96) the element -teen
may be a suffix: in that case the words are simple; if,
however, in view of the usage in She is in her teens, teen
is an independent word, then fourteen, etc. are compounds.

Similarly, in Italian there are a number of words with

a suffix -accio, -accia, expressing the idea of unpleasant-
ness, e. g. roha 'stuff, goods': rohaccia 'trash', te^npo 'weath-
er': tempaccio 'nasty weather', Alfredo 'Alfred': Alfre-
daccio 'naughty Alfred', vecchio 'old', 'old man': vecchiaccio
'unpleasant old man', etc. In view of the locution Quanta
siete accio! 'How unpleasant you are!' all these may, how-
ever, be looked upon as compounds. On the other hand,
we now find compounds departing so widely in form from
the corresponding simple words, that the compositional or
merely formatiomil structure is again questionable. It is, of
course, really in both cases the same phenomenon seen from
a different point of view: in cases like fourteen our feeling
inclines to take the word as simple, in those like housewife
to recognize a compound of two members: both approach
the boundary between simple words and compounds.
166 MORPHOLOGT

We tave seen, also, that compouncls may approach

the value of syntactic collocations, until, in cases like
hidldog (p. 97) and stand off (p. 163), we may hesitate
before the alternative of speaking of composition or of
setting up the apparent first members as independent
words.
Thus we see, in our survey of morphology, the most
varied types of expression: first, the unit word, which,

if classed at all with any other, must be suppletively

classed, in the manner of go: ivcnt; then the inflected

word, grouping itself with others that express the same
material content with a difference of category, such as
eat, eats, ate, eaten, eating; then again, the derived word,
bordering on a compound, such as iinhind, fourteen; and
finally, the compound, bordering, in its turn, on a syn-
tactic collocation of words.
It iswith this syntactic collocation that we shall have
to do in the next chapter.
 CHAPTER VI.

SYNTAX.
1. The of syntax. Syntax studies tlie inter-
field
relations of words in the sentence. These interrelations
are primarily the discursive ones of predication and at-
tribution (pp. 61, 110, f.), to which may be added the serial
relation (p. 113). These are modified by the emotional
dominance of individual words (p. 113, f.) and specialized
into set forms designating material relations of objects
(p. 114, f.).

Syntax cannot be sharply separated from morphology.

This is apparent when we find that it is not always pos-
sible to determine what is one word, what a combination
of words. There is, however, another more essential
point of contact. To the extent in which the words of
a language include relational content, to that extent the
morphology of a language involves questions of syntax.
In an objective language, where relations are included
with the material content of every word, there is com-
paratively little left to say of the syntax. The sentence,
indeed, is often a single word, as in the Nahwatl nina-
hakwa 'I-meat-eat': the syntax of such a sentence is, of
course, the morphology of a word. When we come to
languages like Latin or Sanskrit, in which the noun, for
instance, appears in a number of case-forms with each
its relational content (pp. 108, 143, f.), it is the task of mor-
1 68 SYNTAX

phology to define the values of these cases. These re-

lational values must be carefully studied, before syntax,
in the strict sense, can be begun: but in defining them
we are already entering upon the interrelations of words.
The value, for example, of the dative case in Latin (p. 115)
involves the use of dative-case nouns in the sentence. In
English the syntax more extended, for relations are
is

here more frequently expressed by independent words:

the dative case of Latin, for instance, often by the prep-
ositions for or to (p. 115). Chinese has no inflection
and scarcely any derivation: its grammar is almost en-
tirely a matter of syntax.
The forms of syntax are less fixed than those of mor-
phology, because the utterance of a sentence is a more
complex process and one more easily displaced than that
of a word. In exclamation, especially, the usual syntactic
habit is often disturbed, the elements of the experience
effecting expression in other than the accustomed form.
Thus we may exclaim A rabbit — white! instead of A
white rabbit! 'Anacolouthon', the breaking off one's con-
struction, is common where the emotional charge is

considerable: I wonder if he — but of course he has not

done it!

2. The
discursive relations. The substratum of the
interrelation of words in a sentence is formed by the
binary discursive groupings of predication and attribution
(p. 61) and by the serial grouping (p. 113). We saw in
the fourth chapter (p. Ill) that these relations are not
always expressed, that in Latin, for instance. Magna
ctdpa could express a predication 'Great is the fault' or
an attribution 'Great fault'. The discursive relations may,
as a matter of constant habit, fail to receive syntactic
expression. In a Latin sentence such as Cantat 'He (she)
sings' we have an actor-subject and an action- pi odic.ite.
 THE DISCURSIVE RELATIONS 169

but, as both are included in one word, the predication

does not receive syntactic expression, but only morpho-
logic. In other languages there is no distinct expression
of predication at all. In Nahwatl piltsin means, from the
English point of view, either 'He is a son' or 'a son'
Almost every word here can be used to express a pred-
ication or, especially within a larger utterance, as part
of a predication. The utterance nikmaka tlaslcaUi in no-
piltsin can be looked upon as *I give it him. It is bread.
It is that one. It is my give my son bread'.
son.' or as 'I

The difficulty due entirely to the

of interpretation is

English idiom from the Nahwatl point of view there is

;

no difference between the two forms of expression. This

is true of many American languages: their words are
'sentence-words' (p. 64) with the power of expressing
a predication, and there is no formal indication in any
utterance as to whether a given word is or is not ex-
pressing predication. This sentence-word quality, indeed,
is what gives to the Latin verb, such as cantat, the

power of expressing an entire predication, both subject

and predicate. In Latin this quality is confined, however,
to the verb. In contrast with this is English, in which
no word can, by itself, express a whole predication: not
even our verb, for it, unlike the Latin verb, does not
include an actor-subject; we cannot say Sings, but only
He sings, She sings.
Thus we may, by contrasting two such languages as
Nahwatl and English, see the different degrees of rec-
ognition which the discursive relations may recrive. In
Nahwatl predication is formally a matter of indifference:
any word may express it or not, and the Aztec speaker
does not need to decide whether it is doing so or not.
In English at least two words are needed to express a
predication: a subject and a predicate. The sulject must
170 SYNTAX

be a word of the noun or pronoun class, and tte pred-

icate must be a verb. Verbs, moreover, except for one
special form, the infinitive, always constitute a predicate,
and such a predicate is not permitted to appear without
a subject. In short, predication is precisely defined and
serves as the basis of our word -classification (parts of
speech) and of our sentence-structure.
In viewing such phenomena as these one must guard
against confusing a sentence-equivalent with a sentence.
If one asks, in English, Bid he bring it? and gets the
answer Yes, this answer is in communicative value equal
to a predication. He brought it. Linguistically, however,
it is not a sentence, but only a sentence -equivalent. If
one asks What are you doing? and receives the answer
Writing, this is in communicative content equal to a
sentence, I am writing, but as language it is only a
word, not an utterance presenting discursive relations.
3. The emotional relations. The emotional sub-
stratum of sentences is to some extent independent of
these discursive relations. The
elements in a
different
sentence usually vary as to the place they hold in the
emotional interest of the speaker; in the whole sentence
there is often unity in that some one element markedly
exceeds the others in emotional value. The natural ex-
pression for these relations seems to be greater stress
(loudness) for the more highly charged words (p. 50);
thisseems to be a universal habit among languages,
though the exact decrree of stress-differences varies. We
can imagine a sequence of words spoken with varying
distribution of emotional value, for instance in opposition
to a series of contrary propositions: the dominant ele-
ment will in each case receive highest stress (pp. 50,
113) e. g.: To' day is my birthday. Today 'is my birthday,
Today is 'my birthday, Today is my 'birthday. There may
 MATERIAL RELATIONS 171

be two dominant elements: To day is 'my hirfliday (and

tomorrow is my mother s). In languages that express more
than one semantic element in one word, different parts
of a word may be thus stressed. Even in English, with
our rigid stress-accent on certain syllables, we can speak
not only of 'hulldogs and lapdogs, but also of hull'dogs
and hull'frogs. This may be more pronounced where
more semantic elements are included in one word, thus
in Mesquaki (one of the Algonquian languages) [Vo ba
mi nu] 'Loolx, at me', [wo ba 'mi nu] 'Look at me% [wo ba
mi nu] 'Yoii look at mo'.
Differences of word-order, also, may be used to express
the emotional relations. 'Julius loves Julia' is in Latin
Julius Juliam am at, but Mulius loves Julia' is Juliam
amat Julius. This appears to some extent also in Eng-
lish, as in the contrast between I bought a hat yesterday

and Yesterday I bought a hat^ He came last and Last came

he. Where the word-order is not free, a similar effect
can be obtained by altering the logical structure of the
sentence (p. 114), e. g. It is Julia whom Julius loves. In
the Celtic languages and in French this mode of expres-
sion is very common; we may recaU its use in Irish
English.
4. Material relations. Another factor entering into
the syntactic structure of the sentence is the perceptual
content. Certain word-sequences or types of word-sequen-
ces become habitual as the expression for certain rela-
tions of the objects of the perceptual world; thus in
English we have studied the peculiar construction of
words designating spatial relations (prepositions) with
words denoting the object with regard to which the re-
lation holds true (p. 116). I shall not here recall the
instances that have been given of such syntactic forms
(pp. 114, ff.), but shall speak in greater detail of only one
172 SYNTAX

of these, the relation of actor and action, which in

English has been identified with that of subject and
predicate.
This generalization not in accord with experience, for
is

actually a sentence may not express an action, as does

He ate an apple^ but an equation, e. g.: That tall man is
my father^ a subsumption: The whale is a mammal, the
assigning of a quality: The peasant is poor, a condition:
He is sleeping, or the process of undergoing something:
He sees; He is getting drenched^ and so on. These differ-
ences of material content are not given expression in
the English sentence: in English and many of the related
languages the sentence-type of action has been generalized.
As a French scholar has said, these languages present
the sentence in the form of a little drama in which the
subject is always acting.^)
In Russian the sentence-structure same
is in part the
as in English, but the action-type has not been so com-
pletely generalized; the assignment of a quality still has
a dijfferent form: [mu'^rk'b'e* d'in] ^peasant poor', *The
peasant is This form of locution would be im-
poor'.
possible in English, because we cannot express predi-
cation except as an action, by means of a verb, in —
this case the abstract verb is, whose actual content is

only that of predication, put into the form of an action,

and by the speaker felt as such. Latin allows in these
cases of both types expression: Magna culpa or Magna
est cidpa.

All three of these languages, English, Russian, and

Latin, express as an action by the subject sentence-con-
tents in which actually undergoing some-
the subject is

thing. They view the undergoing process directly as an

1) M. Br^al, Essai de Semantique*^ p. 86.

 MATERIAL RELATIONS 173

action, as, for instance, in saying He sees instead of

'Light-vibrations strike him'. In part they use special
expressions which reverse the occurrence; for this Latin
has a special voice, the passive (p. 145). Amatur 'He
is being loved, is loved', for instance, represents the sub-
ject as actor of a verb, the content of which is 'to be
the object of affection'. Russian uses for such sentences
a manner of expression frequent in languages of our type;
it represents the subject as acting upon itself, -when, re-
ally, other actors are acting upon it, e. g. [ga 'zs ta tj'i

'ta'jitsa] 'The newspaper reads- itself, 'The newspaper

is being read, is read'. Similarly in French: Cela se
raconte partout 'That itself tells everywhere', 'That is be-
ing told everywhere'. In English we use a circumlocution
which represents the actor as being (cf, above) in the
condition of being acted upon: It is being read, It is read,
He is getting drenched, exactly like He is being impudent,
He is poor, He is getting old. This form of expression
occurs also in French: 11 est aime de tous 'He is loved
by all', like II est Ion 'He is good'. German uses a cir-
cumlocution with the word 'becomes', e. g. Das Lied
wurde gesungen 'The song became sung', i. e. 'The song
was sung', exactly like Es wurde Icalt 'It became cold';
so also in Scandinavian (which also has a real passive,
comparable to the Latin, in restricted use), e. g. Nor-
wegian Sangen blev sunget 'The song was sung', like Det
blcv Tioldt 'It grew cold'.
The distinction between an action performed by the
subject and a sensational process undergone by it finds
expression, inconsistently, —
always the case in
as is

linguistic expression of such distinctions, in Georgian. —

There, as we have seen (p. 110), we find the contrast of
[v-t'ser] 'I- write' and [m-e-smi-s] 'me to-sound-is', i. e.

'I bear'.
174 SYNTAX

Just as in English we generalize the type of occurrence

in which the subjectis performing an action, so some

languages generalize that in which there is assigned to

the subject some quality or condition. So, for instance,
in the language of the Wolof in western Africa one says
sopa-na 'loving-he', 'He loves' just as one says haxe-na
'good-he', 'He is good': the action of loving is expressed
as a predicate of exactly the same kind as the quality of
goodness.
In still other languages the occurrence in which the
subject undergoes an action is generalized (p. 110), and
one says, for instance, not 'I kill him' but 'He dies for
me'; so in Greenlandish, not only [tusaRpai] 'Sounding-
its-his', i. e. 'There is sounding of it to him', 'He hears

it', but also [tikipiai] 'Reaching-its-his', 'There is reach-

ing of it to him', 'He reaches it' or [qaja-toR-lu-ne]

'Boat-using- in-his', that is 'It was when he used his
boat'.
As show, Greenlandish illustrates
these translations
another generalization as well: it expresses aU these oc-
currences as statements of possession. Thus [tusaRpai]
'sounding-its- his' and [tikipia:] 'reaching-its-his' are
parallel to [il:u-a] 'house-his', 'his house'; and [qaja-
toR-lu-ne] is parallel to [il:u-ne] 'house-his (own)', 'his
(own) house'. The same generalization of a possessive
sentence-t} pe we saw (p. 150) in the language of the
Lules, where one said amaitsi-s 'love-my', 'I love', amait-
si-p 'love-his', 'he loves', exactly as one said umue-s
'mother-my', iimue-p 'mother-his'.
Syntactic categories. We may thus contrast sever-
5.
al languages as to the freedom or rigidness of their
sentence-structure. As opposed to English, for instance,
Russian and Georgian are comparatively liberal, in allow-
ing of two sentence- types. In English the emotional re-
 SYNTACTIC CATEGORIES 175

may, without altering

lations receive free expression: one
anything speak with greater stress any part of the
else,

sentence that happens to be emotionally dominant. To

some extent in French and completely in Old Irish, the
emotional relations of the sentence also are forced into
set form: the dominant element must stand as the predi-
cate of an abstract deictic subject (p. 171). Thus in
French the regular way of saying 'I did that' is C'est
moi qui at fait cela 'It's I that have done that', and in
Old Irish glosses we find always such expressions as is
oc precept sosceli atto 'It's at preaching of-the-gospel
I- am'.

In short, just as there are morphologic, there are also

syntactic categories and the existence of the
(p. 68),
two is often involved in one and the same mental habit:
so with the English parts of speech and the English
action-sentence. The English language has the syntactic
categories of actor and action; that is, it has identified
subject with actor and predicate with action. Our parts
of speech serve these syntactic demands: the verbs, for
instance, express an action (or some other content viewed
as such) and must therefore be present as the central
element of the predicate in every sentence. Old Irish
has the dominant-category: the emotionally dominant
element has to be the central element of the predicate
Other syntactic categories affect single word-classes. We
have in English three syntactic categories affecting the
noun. An object (or several objects) must be explicitly
described as either (1) definite, (2) indefinite but individ-
ual, or (3) collective or unindividual (p. 118). In the first
case some pronoun must modify the noun, in the
deictic
second case some indefinite pronoun, and only in the third
case, that of collective or unindividual use, may the noun
ne unmodified. We can speak of man or of mm only if
 .

176 SYNTAX

we mean either all men: Man needs hut little . ., Men are
easily moved hy such things, or men, regardless of identity:
Men were shouting. If we do not mean this, we must i^ay
either (2) 'an indefinite man', 'a number of indefiuite men',
e.g some man, any man, one man, some men, six men, se-
man, that man, your
veral men, or else, (1) deictically, this
man, Smith's man, these ?nen, tliose men, etc. This formal
demand is so insistent that we have two pronominal words
of abstract meaning which serve no other purpose than,
with the least possible amount of incrimination, to pro-
vide this description: (1) the 'definite article' the and
(2) the 'indefinite article' a, an. These categories are ab-
sent, for instance, in Latin, where one could say homo,
whether one meant 'man', 'the man', or 'a man', and only
when such elements were actually vivid needed to say ilte
homo 'that man' or homo dliquis 'some man'.
Another syntactic category in English is that of strictly
transitive verbs, that is, of verbs which demand expression
of an object affected. Thus one cannot say He hroJce without
adding an object affected: He broke the hoivl^ He hrohe it,

or, at the very least, He hrohe something. This peculiarity

is shared by the verbal nouns and adjectives derived form

such verbs, e. g. JBreaJcing stone is hard ivork; Breaking the

shell, he examined the contents.

6. The expression of syntactic relations: modula-

tion in the sentence. We may now turn to the formal
means of expressing syntactic relation. At the basis of all
such expression lies the fact that the words of a sentence
are spoken consecutively, in an uninterrupted sequence.
Although within this sequence there may be pauses, these
cannot be extended at liberty.
The unity and the word -interrelations of the sentence
may be further expressed by modulations of pitch and of
stress. This modulation is limited by the habitual wurd-
 MODULATION IN THE SENTENCE 177

accent and the habitual syllable-accent. In a language like

ours,where certain syllables of words are habitually pro-
nounced with greater stress than others, the stress-modu-
is not entirely free, but will always
lation of the sentence
be a compromise between these habitual stressings and
the 'ideal' sentence-stress, — that is, the stress-relations
that we might conceivably use, were we not bound to
stress certain syllables. Thus in Today is 'my hirthdayy
where my, the emotionally dominant element, receives
highest stress, we are also bound by the convention of
our language, to give the syllables -day in today and
hirth- in birthday higher stress than we give to- and -day.
When we speak of huWdogs and hidl'frogs, or of 'aseptic
and 'antiseptic, the sentence -modulation of stress is at
odds with the conventional word accent and carries off the
victory.
In languages with fixed pitch-relations for certain syl-
lables, words, or groups, the ideal sentence-pitch is forced
similarly to compromise with the syllable-pitch and group-
pitch. Thus, in Norwegian, the three words Jian 'he', heter
'is called', and the name Hjalmar all have, when spoken

alone, simple rising accent, e. g. [han/]. In the sentence

Han Hjalmar^ however, the pitch-modulation of the
Jieter

sentence overcomes that of the first two words^ and one

says [(^han he: tar jal/'^mar/]. In other languages with
word-pitch, such as Chinese, the conditions are similar.
In contrast with this, English sentence-modulation of pitch
has almost no obstacles to overcome, for, beyond the cir-

cumstance that we pronounce the stressed syllables with

higher pitch, this factor is left entirely to syntactic em-
ployment. Consequently we can maintain such habits as
the use of even and then falling pitch in statements {He
came had:), rising pitch in sentence -questions {Bid you
say that?)y and rising falling pitch in word-questions ( WJtat
Bloomf iold, Study uf Language 12
178 SYNTAX

was he doing 9)^ not to speak of a number of modulations

for more delicate emotional shadings. In our ears such a
language as Norwegian sounds, accordingly, like a mono-
tonous series of questions or enlivening exhortations.
Such modulations, then, are of decided significance. In
English words of less material content receive weaker
stress, so especially abstract and purely formal words:
tJie 'man, a 'man, It is 'cold, He 'ate it, He 'gave the 'hook
to his 'brother. These words receive greater stress only
when they are used inmore than mere satisfying of syn-
tactic requirements: It 'is cold. The sequence of attribute
and subject shows in English even stress: the 'young 'man;
in He 'failed com'pletely to make his meaning clear, the ad-
verb completely, in even stress with failed, modifies but
it,

in He 'failed completely to make his meaning clear, it modi-

fies the less stressed make.
7. Cross-referring constructions. In general, however,
these sentence-modulations of pause, pitch, and stress ex-
press the emotional rather than the discursive or the ma-
terial relations between the words. These are expressed
in various other ways.
Most obvious and most cumbersome among these means
is the double expression of an element, first relationaUy
in connection with another word and then explicitly in a
separate word expressing the material content. Speakers
of English can best feel the value of such double expression
by calling to mind dialectal locutions like John his knife
or John he tvent home. Here John gives the material con-
tent of the element which in he, his is expressed anaphori-
caUy over again in relation with the other elements of
the sentence; —
the use of John here is caUed 'absolu-
tive'. These constructions are unnecessarily cumbersome,

for in English John can figure as subject -actor, or, in

possessive inflection, as attribute -possessor {John tvent
 :

CROSS-REFERRING CONSTRUCTIONS 179

homey John*s Tcnife)-^ and it is safe to suppose tliat educated

writers' and speakers' disuse of this construction is due
to tlie feeling that the added word is superfluous. Sup-
posing, however, that it were impossible in English to
say went home without an anaphoric actor he^ and impos-
sible to say linife without an anaphorically expressed pos
sessor: then such constructions as John he went and John
his-knife would be unavoidable.
These conditions are fulfilled in numerous languages
In Latin, for instance, where the verb always includes
personal-anaphoric mention of the actor, and a form such
as cantat means *he-siiigs' or 'she-sings', one cannot say
The girl sings' but only puella cantat literally, then, '(The)
girl she-sings'.
The same may be true of objects affected. In Nahwatl
nihnaka means 'I-it-him-give'. To express concretely the
objects affected one may form a compound with one of
them, e. g. For the sec-
niJctlasJcalmaJca 'I-him-bread-give'.
ond of the objects affected, however, and most commonly
also for the first, one adds specific mention of the object
*1-
in an absolutive form: nihnaJca tlaskalli in nopiltsin
him-it give, bread, (the) my-son'. Just so in Greenlandish
[qame tukaRpa:] 'His-boat he-tramples-it' (more literally,
— cf. above, — 'It is his boat. There is trampling of it

of him').
The parallel to John his knife is seen in the Greenlan-
dish [qim:ip neqa:] '('^^®) ^^o his-meat'.
This method of syntactic expression is subject to some
ambiguity. The relation of the words for 'bread' and 'my-
son' in the Nahwatl example above can be deduced from
their content, but it is not expressed otherwise than by
the material absurdity of the subject's giving his son to
some bread. When we learn, further, that the word ex-
pressing the actor may also in absolutive form follow the
12*
180 SYNTAX

action- word, we see tliat this method of syntactic expres-

sion, with all its explicitness, is far from being propor-

tionately exact or clear.

8, Congruence. Another expression of syntactic rela-
tions is the phenomenon of congruence, which we have
several times met (pp. 128, ff.). Congruence is syntactically
expressive because it is limited to words in certain rela-
tions with each other, —
such as subject and verb in Eng-
lish. When we say I am; you, tve, they are; he, she, it is,

the form of the verb shows in the first case, for instance,
that the verb has as its actor the speaker. To be sure, the
congruence is not needed, for even without it, as in J, you,
he, she, it, we, they can {shall, tvill, did, gave, etc.), the po-
sition of the actor- word immediately before the verb ex-
presses the relation between the words: our congruence
of verb with person and number of actor is logically su-
perfluous.
This, however, is not true of most cases of congruence.
In those European languages which divide their nouns
categorically into gender-classes and express this classifi-
cation by congruence of and rela-
attributive, anaphoric,
tive words, locutions constantly occur in which congruence
alone expresses the syntactic relations. In German such
expressions as das Fremden unzugllngliche Hans, literally
Hhe to-strangers unapproachable house', are clear because
congruence ('agrees') only with Haus,
das, the article, is in
not with Fremden, which would require, by congruence,
another form of an article attributive to it. So in relative
reference,Die Maus im Keller, welche nafi ivar, (literally . . .

'The mouse in the cellar, which wet was, .') is clear: the. .

form welche shows that the relative pronoun refers to the

'feminine' noun Maus; if it is the cellar which was wet,
one says Die Maus im Keller, welcher nafi war, ., with . .

the 'masculine' congruence form of the relative pronoun

 CONGRUENCE 181

Sucli congruence of attributive words with gender, num-

ber, and case of their subject nouns may serve to relate
words in what would seem to us the most puzzling jumble.
This is true, especially, in the older stages of the European
languages. Thus, in Latin, Horace was able to write:
. . . Me tahuld sacer
votlvd paries indicat uvida
suspendisse potentl
vestimenta maris deo.
'By shows me to have hung
a votive tablet the sacred wall
up drenched garments (as an offering) to the mighty god
of the sea', —word for word: 'Me by-a-tablet sacred vo-
tive wall shows wet to-have-hung-up powerful clothes of-
the-sea to-the-god'. The sentence includes four adjectives
and five nouns, which are grouped, adjective with noun,
only by the congruence of the former with the latter as
to gender number, and case. That is, sacer 'sacred' agrees
only with paries 'wall', votiva 'votive' only with tahuld
'tablet', uvida 'wet' with vestimenta 'clothes', and potentl
'powerful' agrees only with deo 'god', leaving maris 'the
without an adjective. In short, the application of
sea's'
the adjectivesis expressed entirely by their congruence.

To be sure, not every instance is as clear as this one.

Suppose that instead of 'sacred wall' a Roman had said
'sacred temple', sacra aedes. Now, owing to homonymy
between the adjective forms for the nominative singular
feminine (sacra) and the accusative (and nominative) plural
neuter (uvida), sacra could belong with vestimenta ('sacred
garments') and uvida with aedes ('drenched temple'), and
only the material content or the order in which the words
were presented to the hearer could determine the syn-
tactic relations between these adjectives and nouns.
But even when there is no homonymy congruence can
never be unambiguous. For, if ambiguity is to be avoided
182 SYNTAX

in every instance, the number of categories according to

whicli there is congruence must approach infinity. In
the Bantu lan^uaojes of southern and central Africa con-
gruence plays a very large role in syntactic expression,
and we find, accordingly, a large number of categories
of the object- expressions with which other words must
agree. The Subiya language, above mentioned, (p. 143),
as
so distinguishes in all twenty-one categories. In the Kafir,
which is closely related to this dialect, one says, for in-
stance (p. 154):
honlie ahazdlwana ^all the brothers*,
Ipnlce ilizwe 'all the land',
yonlie indlu 'all the house',
eonke isilo 'all the creatures',
wonJce umhlaha 'all the earth*,
Iconlx uliutya 'all the food',
where the attributive 'all' varies in congruence with the
noun-genders. Similarly, an attributive noun varies in
congruence with that modified, which precedes it:
umfazi gowomtu 'the wife of the man',
iJiase lehmtii 'the horse of the man',

umfazi goivenlwsi 'the wife of the chief*,

ihase lelenJcosi 'the horse of the chief.
The numeral words similarly agree:
immini zamasumi tnane 'days tens four, fourty days',
uhusuke hamasumi mane 'nights tens four, forty nights'.
So also the action- word:
umntu uyadla innyama 'the man goes to eat meat',
ahantu hayadla innyama 'the men go to eat meat'.
9. Government. Another means of expressing syn-
according to the relation
tactic relations is word-variation
the word bears to some other word. With such word-
variation we are familiar from the discussion of morphol-
ogy, where we saw (pp. 104, ff., 143, f.) how relational ele-
 GOVERNMENT 183

meiits are in most languages included in the same word

with concrete elements to which they pertain. In so far as
such inflection includes relations to other words of the
same sentence, it is of syntactic force. When the English
verb includes in its form an indication of time, this in-
dication is, generally, of no syntactic importance. In the
sentence He hought many doilies the form bought express-
es the time, relative to the speaker's present, of the oc-
currence, but not the relation of the verb to the other
parts of the sentence: this relation is no different when

the other tense-form. He hugsmany clothes^ is used. In

the noun, on the other hand, we have a special form,
the possessive, which indicates that the noun in this form
is an attribute (possessor) of some other noun. As oppo-
sed to Father has a new hat^ Who has seen my father P^
with my father ^ the form father^s in Where is father's new
hat? shows the experience -element father to be neither
actor nor object affected nor point of view of a local re-
lation,but attribute of another noun. In our pronouns,
where we have a third case-form, this goes farther. The
forms I, he, she, we, they, ivho are used only as subject
of a verb or in the predicate of a sentence expressing
equation, e. g. It is he. The forms my, his, her, our,
their,whose are attributive, and the third set of forms,
me, him, her, us, them, whom, show the content of the
pronoun to stand in the relation of object affected to a
verb, or of point of view to a preposition: He saiv me,
I Inow him. Come to us. As between It is he and It hurts
him, for instance, the variation in the form of the pro-
noun shows the relation to the verb: him is object affected,
he not so, but 'complement', that is, genuine predicate
following the abstract equation-verb is. He loves her
shows by the form of the pronouns who is the actor,
who the object affected. Reclion or government, then, is'
184 SYNTAX

tlie process by wHcli a word has a different form accord-

ing to its relation to other words in the sentence.^)
Government is in very active use in such languages
as German, Latin, Greek, or Sanskrit, where the nouns
(or words modifying them) have different case-forms foi
different relations. In Latin Pater fllio lihrum dat in any
order of the words is equally clear, for pater is in the
nominative form, as actor, fllib in the dative, as object
less fully affected, librum in the accusative, as object fully
affected ^The father gives the son a book' the case-forms
: ;

of the words express the syntactic relations.*) In the

quotation from Horace,

• . . Me tabula sacer
votiva paries indicat uvida
suspendisse potenti
vestimenta maris deo,

we have seen how congruence shows the grouping of

adjectives with nouns, to wit: votiva tabula *
votive tablet',
sacer paries ^sacred wall', uvida vestimenta 'drenched
garments', potenti deo 'powerful god'. Government lays
clear the syntactic relations of these nouns. Namely,
{sacer) paries^ in the nominative case, must figure as sub-
ject of the verb {indicat 'shows'); we, as accusative form,
is object fully affected by this verb; tabula {votiva), an
ablative form of instrumental value, expresses the means

1) Those dialectal speakers who say It is me, him, her^ ws,

them^ using the accusative form wherever a pronoun appears
after a verb, make no use of government, but instead use the
nominative and accusative forms merely in congruence with the
position of the pronoun: in position before the verb, one set of
forms; in position after it or after a preposition, another.
•2) Between pater 'father* and dat, literally 'he-gives*, the re-
lation is, however, really expressed by cross-reference, cf. above-
 GOVERNMENT 185

of tlie action ('by means of a votive tablet*); (uvida)

vestimentay again accusative, is tbe object fully affected
by tbe infinitive suspendisse 'to have bung up'; maris is
in tbe genitive case, expressing, somewhat like the Eng-
lish possessive, an attributive relation, 'the sea's'; (po-
tentl) deOj finally, in the dative case, is the object less
fully affected by suspendisse, 'to the mighty god'. The
meaning is, accordingly, as we have seen, 'The sacred
wall, by a votive tablet, shows me to have hung up
drenched garments to the mighty god of the sea*.
But, as a matter of fact, some of these forms are by
no means unambiguous. Partly this is due to homonymy:
me could be ablative ('by means of me') as well as ac-
cusative, and the accusative uvida vestimenta could also
be nominative (subject of a verb). The ambiguity is re-
moved only by the fact that indicate the verb, must have
an object affected, for which me, in the beginning the
only form available, is accordingly taken, as an accusa-
tive: and later, when uvida vestimenta appears, the indu-
bitable nominative, sacer paries has already occupied the
place of subject-actor.
Moreover, even aside from homonymy, the sentence,
so far as government is concerned, is left ambiguous, for

government tells us neither which accusative, me or uvida

vestimenta^ is the object affected by indicat and which by
suspendissey nor does it show which noun maris 'of the
sea' Like the ambiguities already mentioned,
modifies.
these also are r-emoved by that very natural means, the
order of the words: when we hear indicate me alone is
available as object affected, and the position of maris
between potentl and deo suggests the concretely obvious
connection. In other sentences the material content alono
can remove the ambiguity: Uvida vestimenta me suspen-
disse indicat, —
a good Latin sentence, could mean —
186 SYNTAX

either shows me to liave hung up wet garments* or

'It

'It shows wet garments to have hung me up'.

Government may extend to aU kinds of relational ex-

pression. Thus in German one says Er ging in den Wold
'He walked into the woods', but Er ging in dem Walde
'He walked in the woods'. The two relations which we
express in English as into and in are expressed in Ger-
man by the same word in, government alone distinguish-
ing: the object with regard to which the relation is
viewed stands in the former instance in the accusative,
in the latter in the dative case.
10. Word-order. In the discussion of our quotation
from Horace it appeared that some of the ambiguity left
by government as a syntactic method was naturally re-
moved by the order in which the words were presented
to the hearer. This function of the word-order was by
no means due to any convention of the Latin language,
but entirely to the natural reproductive processes of the
listener, who would, for instance, at once take up me as
the object fully affected by indicat, because, when he
heard the latter word, no other accusative had yet reached
him. The speaker analytically expressing his total ex-
perience will naturally follow some connected order, ex-
cept for such emotionally dominant elements as thrust
themselves, regardless of logical or material connection,
at once into expression. The hearer, for his part, re-
produces as he hears and uses each element as it comes,
unless habitual processes of the language force him to
hold them apart, as when the Roman, hearing tabula
sacer, is kept by the congruence-habits of his speech,
from even tentatively applying the adjective to this noun,
but waits for another with which sacer will agree. In
short, quite independently of any fixed habit, the order
of words in a sentence cannot but be to some extent
 WORD-ORDER 187

indicative of their syntactic relations, just as all the

words of a sentence, to begin with, form an uninterrupted
sequence.
In many languages, such as our own, the order of words
has a habitual syntactic significance. Where a Latin, for
example, could pronounce in any one of the six mathe-
matically possible orders the words Julius (nominative),
Juliam and amat ('he, she loves', verb), ex-
(accusative),
pressing by cross-reference and goverment that Julius is
the actor, by government that Jftlia is the object affected,
there we can say only Julius loves Julia. The noun pre-
ceding the verb is in English the actor, that following
it, the object affected.
This use of word-order as a syntactic method is doubly
significant. From a logical, post factum point of view, it
makes use of what is inevitably present, for in some order
the words must be spoken, and dispenses with any further
encumbrance for the expression of relations. Psychologic-
ally, it involves a fixed habit as to the order with which
one in speaking analyzes and in hearing reproduces a
total experience. The fixing of this habit cannot but save
energy: the English sentence can be understood, — that
is, the experience reproduced by the hearer, — with less
effort of the attention than the Latin.
In English word-order is the prevailing method of syn-
tactic expression. We have already seen that the actor-
subject precedes the verb. The objects afifected follow it,

the object less fully afi'ccted preceding that fully afi*ected

(He gave John a hooJ,), unless, indeed, the former relation
is analyzed into a separate word (lie gave a hooJc to JoJin),

when it falls under the head of preposition with its noun,

which always follows the objects affected. The abstract
verb is precedes the true predicate {TJns man is my bro-
ther j You are good). Attributive modifiers precede nouns
188 SYNTAX

(a good friend, father's hat), and prepositions precede tbe

noun with regard to wliicli they are used {in the house).
Only as to the attributive modifiers of verbs is there some
freedom (QaicMy he ran, He quickly ran. He ran qidcldy).
Thus not only the discursive but also material and emo-
tional relations are expressed by word-order.
In Chinese word-order is even more exclusively the
expression of syntactic relations. The subject precedes
the predicate, as in [""fa^ l^olo/] *He (is) good' or ['^fa"'

L^9y\] 'He goes', and attributes precede their subject:

[^xao/ "^^An/] 'good man, good people', [Lman\ lC9 y\] 'Slow-
ly go', 'Go slowly'. The material relation of objects affected
is also recognized: they follow the transitive word, e. g.
[lWo/ lP a\ '"fa"'] 'I fear him', ["fa"' ,p a\ ^wo/] 'He fears
me'. Other examples have been explained on pages 115, ff.

and 126, f.
11. Set phrases: the transition from syntax to style.
In spite of this simplicity of Chinese syntactic expression,
an English-speaking person who had access to information
as to the meaning of every individual Chinese word and
knew these rules of word-order, would still fail to under-
stand many sentences of this language. He would be baf-
fled by set combinations of words, 'idioms', deviating from
the meaning of the simple words and thus approaching
the value of compounds. In such idiomatic phrases Chinese
is very rich. They exist, however, in every speech-communi-

ty and not least in English and its various local forms.

A stranger would have no success, if he attempted, word
by word, to understand such usage as that of at in We
sat at the table, He threw it at me, He is at work. Don't
he angry at me. Not at all, and so on, or of do in He did
it, Did he go?, Did he do it?. Do to death, It does him

credit, He did me a service, Hoiv do you do?, Do the roast

well, I have done six copies, You would do wisely to go, of
 THE TRANSITION FROM SYNTAX TO STYLE 189

have in I have your haty I have written^ I have to do it. You

had better go, of about in We walhcd about the garden, We
talked about you, About three mileSy and so on.
When the idiomatic set phrase deviates too considerably
from the individual meanings of the words, we have, of
course, a compound word: the boundary is by no means
clearly traceable. Is at all a compound? In Chinese we
have seen ['^9iug^ L^i] in which the first word means 'older
brother' and the second 'younger brother'; together the
two, were this not an idiomatic phrase, would mean 'younger
brother of an older brother'. The phrase, however, may
be looked upon, from the English standpoint as mean-
ing 'older brother(s) and (or) younger brother(s)', i. e.
often 'brothers, brother'. Shall we call the Chinese ex-
pression a compound or a two- word phrase?
On the otlier hand, the boundary between set phrase
and syntactic expression is never clearly traceable. What
is one of a number of equally favored possibilities of ex-

pression in one generation, may in the next become a

habitual phrase, and a few generations later be the only
correct expression. Some time ago one said indifferently
in English He gave me it. He gave it me, or He gave it
to me. Today we use, in America at least, the last of these

far more frequently than the first two; tomorrow it alone

may be correct, —
or else an idiomatic use here prevalent
in the sense of 'He scolded me, punished me' may spe-
cialize it. In such a sentence as He sees me the syntactic
relations are expressed by congruence (he sees), govern-
ment {he sees, sees me), and word order. The expression
was therefore just as intelligible in times when the third
factor was absent and one could say Me he sees, He me
sees, and the like. The order He sees me here became,
successively, the favorite phraseologic order and then the
only correct one.
190 SYNTAX

12. The complex sentence. This crystallization of word-

groups is psychologically most natural. Once a certain
expression has been heard and used^ it is, by the principle
of habituation, more likely to be called up and used in
the future than some other less familiar form. Ultimately,
if enough used, it becomes mechanized, in the sense that

a single initial impulse is sufficient to start off the utter-

ance of the entire word-group, so that the relations be-
tween the single words, like the single articulatory move-
ments, need no longer be conscious. Most of our speak-
ing, in fact, done in this way. We speak by whole
is

phrases, even by whole sentences, and, at the very least, by

certain well-practised schemes, — sentence-skeletons that
require but the variation of a few words from utterance to
utterance. If we stop to analyze even a moderately long
sentence of ordinary speech into single words and stop to
determine the relations between them, the result is a very
complex structure and the process always strikes us as
unnatural: we are certain that in speaking the casual
sentence we built up no such maze of variously inter-
woven attributions, serial groups, and predications. Our
feeling is justified, for most of the word-groups within
which such relations ultimately subsist are in actuality
so practised that no analytic activity of the attention is
necessary for their utterance. Unless this were the case,
it would require a master mind to construct any but the

briefest and baldest utterances.

The utterance of longer, more complex sentences is
due, therefore, to the mechanization of greater and greater
groups of words. The most typical illustration of this is
the attributive subordination of one predication to an-
other. In languages that have no specific expression of
predication this subordination does not formally appear.
When a Greenlander says [tuawioRtoRsf uol:une aneRlaR-
 THE COMPLEX SENTENCE 191

poq] we may interpret it as 'It-is-in-liis-liurrying-very-

much. He-goes-home' or as 'Hurrying very much, he goes
home'. Which interpretation is the juster depends on the
mental process of the speaker in each instance, unless it
happens that the expression is knoAvn to he a very common
one, when we may be sure of subordination. This simple
succession of two predications one of which is psychologic-
ally, but not in expression, subordinate (i. e. attributive)
to the other, is called parataxis. Another example of it

is the locution so frequent in German fairy tales: JEs ivar

einmal ein Mann, der hatte drei Tocliter 'There was once
a man. He had three daughters' or 'There was once a
man who had three daughters'; exactly as in the Green-
landish example, both English renderings are wrong: the
German expression could, formally, be two sentences, but
in the feeling of every German speaker the second pred-
ication is subordinate to the first. An English example
is He tvrites me he is sicJc. Modulations of stress, pitch,
and duration (pause) may, of course, enter as expression
of the subordination.
however, the expression of subordination goes be-
If,

yond this, We have no longer parataxis, but hyiiotaxis. Such

a phenomenon as hypotaxis is conceivable only when
large word-groups constituting parts of the whole utter-
ance have become specifically or in plan mechanized.
Hypotaxis is attained by means of special attributive forms
for larger subordinate elements. Such a form is the Eng-
lish verbal adjective (participle) in -ing followed, like the
verb, by expression of objects affected: Giving his friend
the letter, he turned to go. This form of expression is in
some languages so grout a favorite that, of a number,
say, of occurrences, only a few receive independent pred-
icative expression. In English the participle is generally
used only where it may be taken ap an attribute of the
192 SYNTAX

subject of the sentence: in other languages the subordinate

elements may modify various parts of the sentence. In
Turkish, for instance, there occur such constructions as
[xod^a mnanmaz, biri daxra gelip, oile s0iler, xod3a gertjek
eanmp, kuzuju bogazlar], literally: 'Master believing-not,
one-of-them again coming, similarly speaking, master true
believing, the lamb slaughters', i. e. 'When the master
did not believe this, another of them came and spoke
similarly; whereupon the master, now believing it, slaugh-
tered the lamb'. In Sanskrit, which has generally a struc-
ture similar to that of Latin, we find in certain writings,
perhaps owing to the influence of other languages of India,
which construct very much on the Turkish principle, an
extended use of such participial expression. For instance,
Ity alocya, tena, grdmq gatvd, DadhiJcarna-ndma hidalo,
mqS'ddy-dJtdrena sqtosya, prayatndd dniya, sva-landare
dhrtdhj literally: 'So having- reflected, by-him, to-the-village
having-gone, Dadhikarna-by-name (a) cat (nom.), by-means-
of-meat-and-other-food having-satisfied, carefully having-
br ought, in-his-cave was-kept'. All the participial ex-
pressions (ending in -ya and -tvd) here apply not to the
subject (the cat), but to the instrumental ('by-him'); here,
as above in the Turkish example, these expressions are
reaUy attributes not of any object-word or other element,
but of the predication as a whole. Hence we may here
translate: 'After he had so reflected, by him, when he had
gone to the village, a cat, Dadhikarna by name, when he
had given it its fill of meat and other food and led it
carefully home, was kept in bis cave', or, changing to the
active and coordinating construction favored in English:
'After these reflections he went to the village, and, hav-
ing satisfied a certain cat named Dadhikarna with meat
and other food, brought it carefully home and kept it in
his cave'
 THE COMPLEX SENTENCE 193

In these examples the subordinated elements deviate

considerably in form from independent predications. In
English, however, we may, by means of a subordinating
conjunction express one predication, word for word, as
attributive to another. Thus, paratactic: It looks like rain;
he had better carry an umhrella ; hypotactic: As it looks
like rain, he had letter carry an umhrella , where as, the
subordinating conjunction, expresses the attributive char-
acter of the first predication. Beside using such subordinat-
ing conjunctions as as, ivhen, if, how, since, because, ivhile,

after, although, and the like, we express hypotaxis also by

our relative pronouns, e. g.: I dont know
whom, the man
you mention, where the peculiar word-order (whom you
mention, as opposed, for instance, to you mention him) and
the double function of the relative pronoun whom as both
member (in this case, object affected) of the subordinate
sentence and expression of the subordination, carry the
hypotaxis. A peculiar form of hypotaxis is the English
construction in which a noun figures as predicate or as
object affected in the principal sentence and at the same
time as object affected in the second; the subordination
is expressed by this double function of the word and by

a word-order differing from the normal: thus the last

sentence could be put: 1 don't know
you mention the man
or He isn't the man I mean. This type of hypotaxis, is
called, by its Greek name, the construciion apo koinou.
In German subordination is expressed by subordinating
conjunctions or relative pronouns and, most strikingly,
by the word-order: the verb stands second among the
elements of" the declarative sentence, except in case of
when it stands last. Consequently, although
subordination,
German has much homonymy between adverbs and sub-
ordinating conjunctions and between relative and demon-
strative pronouns, hypotaxis is always clearly distinguished.
lUoo raf eld, Study
1 of Langaag* 13
194 SYNTAX

Thus in Da Icam er, 'Then he came', da is adverb and

we have an independent sentence, but in Da er kam, Iconnten
u'ir ilin fragen 'Since he came, vre w^ere able to ask him',
da is a subordinating conjunction and da er Icam, a 'de-
pendent clause'. The paratactic Es ivar einmal ein Mann^
der hatte drei Tochter 'There v^as once a man; he had
three daughters' (der anaphoric pronoun) is distinguished
by word-order only from the hypotactic Es war einmal
ein Mann, der drei Tochter hatte 'There was once a man
who had three daughters' (der relative pronoun). By such
means the sentence may in literary languages like San-
skrit, Latin, German, or English be expanded until it is

simply a kind of logical puzzle. In natural speech, how-

ever, no matter how complex a sentence may be from
the logical point of view, it really never consists of more
than a very few elements, each of which, even if discursi-
vely divisible, is in the mind of the speaker nevertheless
a single associatively mechanized element, which he is
not compelled to analyze, unless some circumstance should
particularly draw his attention to it.

Here again, the actual conditions of language are not

mathematically definable. It is impossible to determine
exactly how far every speaker goes in the analysis of the
total experience and how much of what he says is a
matter of practised combinations. Here again, moreover,
the constant change in language makes itself apparent:
new phrases ..ud methods of construction come into favor
and old ones lapse into oblivion.
 CHAPTER VH
INTERNAL CHANGE IN LANGUAGE.
1. Language constantly changing. The speech of
former times, wherever history has given us records of
it, differs from that of the present. When we read Shak-

spere, for example, we are disturbed by subtle deviations

from our own habit in the use of words and in construc-
tion; if our actors pronounced the lines as Shakspere and
his contemporaries did, we should say that they had an Irish
or German brogue. Chaucer we cannot read without some
grammatical explanation and a glossary; correctly pro-
nounced his language would sound to us more like Low
German than like our English. If we go back only about
forty generations from our time to that of Alfred the
Great, we come to En£flish as strans^e to us as modern
German, and quite unintelligible, unless we study carefully
both grammar and lexicon.
2. Causes of the instability of language. It is by
no means surprising that language changes. As a physical
phenomenon it consists of certain finely graded habitual
movements, which, we know, cannot always be performed
in exactly the same way. There must be endless infinitesi-
mal variations, smaller even than those which scientific
observation unmistakably reveals, — not to speak of such
as can be heard by anyone who listens for them. Thus,
such a sentence as Going to the university? is often pro-
196 INTERNAL CHANGE IN LANGUAGE

nounced ['gowi t^js Vasti]. To be sure, we are so macli

in memory of the sound that these words have in more
deliberate speech (and, in our state of culture, of their
form in writing) that we ordinarily fail to notice what
sounds we or our acquaintances have actually uttered;
yet, as the language is spoken by generation after genera-
tion, such tendencies cannot but have their lasting effect.
The same is true, mutatis mutandis, of the other phases
of language. The experience itself is always new: shall
the speaker class it under this particular word or that?
A member of a happy family points to his house and
says: 'This is our little home.* What more natural than
that, just as his interlocutor might now tell people that
'N. has a beautiful little home,' other people should gradu-
ally come any house intended for dwelling
to speak of
purposes as a homey until we read in our newspapers that
there is for sale 'a fine new ten-room home/ and realize
that for many English speaking people the word home
has no longer the meaning which it used to have.
In the morphologic word-classes we need only look at
the assimilative process by which inflection takes place
(p. 59, f.), to see how unstable it must be, from its very
nature. Preterite tense and unreal mode are expressed by
only one form in every English verb except teas: were.
It is a natural consequence that many speakers use was
for were, saying, for instance, 'If he ujas here, he wouldn't
allow it.' Owing to the identity of form of other verbs
{came, said, had, thought) such a speaker, without con-
sciousness of innovation, says was where were is in our
feeling alone correct.
Gradual change in the manner of performing articula-
tion-movements, the inclusion of new experiences under
new words, and the occurrence of a few unprecedented
morphologic assimilations among the many previously
 CAUSES OF THE INSTABILITY OF LANGUAGE 197

usual ones, — these are inevitable attendant occurrences

of aU language. On the otlier side are the no less inevi-
table conservative forces. The speaker hears others who
are not making the same innovation and either realizes
that he spoke in violation of custom and perhaps un-
intelligibly, or, far more conimonl}-, has the correct form
reawakened or strengthened in his speech-predisposition
without ever growing aware of the temporary divergence. His
interlocutor's answer, for instance, contains the forms [juni
If he were here, and, inadvertently, these
'vjsi ti], house,

instead of the new associations are uppermost when next

he speaks. But when our interlocutors, too, have formed
the new association, so that it is in them strengthened
by our use of the new form, they will notice nothing
unusual in our words and will utter the like. Finally the*
new association may become practised and vigorous, and
the old fall into such desuetude that, when we hear its
forms, they seem strange and unusual.
an innovation, in order to spread,
It is evident, then, that
and not to lapse into oblivion as a once-made slip of the
tongue, must be such as to fall in readily with the other
habits and associations of the speakers. The use of was
for uere has spread because one form for all the numbers
and persons of preterite and unreal, —
that is, the absence
of these distinctions, —
is customary in all other verbs.

No change takes place in a language unless there is a

predisposition for it number of speakers. Fashion,
in a large
to be sure, and the conscious desire to be like some ad-
mired person or class of people, may help to spread or
to check the spreading of an innovation. The young boy
who wants to seem virile imitates the speech of the 'tough*,
and the snob affects the manner of speech that happens
to be natural to the aristocrat. More legitimately, the pub-
lic speaker, the teacher, and, above all, the writer, exert
 INTERNAL CHANGE IN LANGUAGE

a wide influence over the speech of others. In spite of

Y*all this, however, every innovation, in its beginnings, is

. w^ lT the result of a psychic predisposition not only on the

yr\j^ ptart of those speakers who independently originate itj

vV I . but also on the part of those who

unconsciously take it
^ V J^P" I^ ""^0^1^? indeed, be impossible to determine who
Lio
jT '^
spoke a given innovation, who spoke it only after
I
first

^-n^ieariiig it from others. Neither speaker is conscious of

/»<>/^ saying anything novel. If the innovation fails to find accept^

—"'V^^i^ce, both kinds of speakers lapse back to the old forms
of speech, without ever realizing that they onG6 or a few
times spoke differently, and if the innovation spreads, it
remains usually for scholars who long afterward look
back at an earlier form of speech, to realize that a change
'has taken place. No London-English mother, in the forty
generations from Alfred's time to ours, has realized that
gvlKK
her children were not learning exactly the same English
that she had learned in her infancy; and, indeed, had she
been able to hear the two forms of speech in close suc-
cession, she could not have detected any difference, unless
she were an extremely careful observer. It is only under
the most favorable conditions that linguistic study has
been able to perceive the deviation of speech from genera-
tion to generation in a small community. It is evident
that we are, all of us, contributing, through all our lives,
to the change of our language, but neither do we from
any direct consciousness of the process know this, nor
could we, though our lives were at stake, tell how or in
what respects we are altering the language we learned
from our parents.
To say all this, is, of course, only to repeat that the
facts of language are facts of social, not of individual
psychology. Could we definitely mark out the speaker
who first spoke a given innovation, trace the forces wliicli.
 CAUSES OF THE INSTABILITY OF LANGUAGE 199

impelled him to make it to certain features of his mental

situation at the time, and similarly lay clear the motives
of all the other speakers who propagated the new form,
then we should be accomplishing the interpretation of
a social development into terms of individual psychology.
This, of course, could he done only by an omniscient
observer. It is, for that matter, immaterial who first spoke
a given innovation. Both he and those others who, in-
dependently, produced the same form, as well as those
who spoke the new form only after hearing it, were un-
conscious of any change. The expressive habits of the
community whole were ready for the innovation,
as a —
or rather, were in such a form that what was, physically,
an innovation, was psychologically no change at aU.
Here lies, of course, the great difficulty of historical
language study. In descriptive study we can, in the worst
case, confine ourselves to the phenomena in a limited
number of utterances or speakers. To tell with historical
correctness the story of a single change, however, we
should need not only an exact knowlege of when, where,
by whom, and under what circumstances the change was
first made, and of exactly how, occurrence by occurrence,

hour by hour, speaker by speaker, house by house, village

by village, it spread, but also an insight into the entire
mentality of each speaker, so as to see what favorable
predispositions the change met, and what obstacles, how
it became strengthened by hearing and speaking, and

weakened and strenorthened again, —

and all these occurren-
ces, we must remember, belong to a phase of activity so
mechanized that the details of it are never, except in the
rarest instances, sufficiently focused by the attention to
come into vivid consciousness.
Another our records of
difficulty lies ia the fact that
past speech are always, in the face of such an ideal, ex-
200 INTERNAL CHANGE IN LANGUAGE

iremely scanty, and that they are representative of writ-

ing, an activity in which the single actions are performed
much more slowly and much more under the spotlight
of the attention than are those of ordinary speech, and,
finally, that writing is capable of but a very imperfect
representation of the phonetic facts.
Historical language study is thus at best imperfect.
Its imperfections can be partly repaired by certain tech-
nical means, which enable us to correct and supplement
our records of past speech, especially as to phonetic form,
and also to ascertain facts about the speech of certain
periods from which no records have come down to us.
Of these technical means, called the 'comparative method',
I shall be able to speak more fully in the next chapter;
for the present a single instance may suffice. Although
the orthography of Shakspere and of Chaucer's 'Middle
English' does not materially differ from that current to-
day, we can determine with considerable precision how
the English of those times was pronounced. We are
further able to interpret into phonetic terms the orthog-
raphy of the 'Old English' of Alfred's time. Beyond
however, we can arrive at a great many facts
all this,
and many more probabilities about the English that was
spoken before the time of our written records, at the
time, for instance, when the 'Angles, Saxons, and Jutes',
the bearers of English speech, stillon the European
lived
mainland. This prehistoric English, back to a certain
time, we call 'pre-English'. Back of pre-English lies a
point in time only relatively determinable, about which
also we can state a good many linguistic facts: the lan-
guage at this point is called 'Primitive West- Germanic'.
Back of this point lies another period, which we call
'pre- West-Germanic'; and back of this period another
point in time, where the language is called 'Primitive
 CAUSES OF THE INSTABILITY OF LANGUAGE 201

Germanic*; then, through the 'pre-Germanic' period, we

come, once more, to a specific though not absolutely de-
terminable point in time, thousands of years ago, when
'Primitive Indo-European' was spoken. Beyond this stage
of the language we cannot as yet penetrate. To return,
however, from our example to our point: in spite of aU
this supplementation, our historic study cannot go be-
yond the crudest outlines as compared with the ideal
demands of the situation. We can tell, at best, that a
specific change, beginning at such and such a time, —
often determinable only within a century, if at all other-
wise than relatively, —
and spreading in a certain gener-
al direction, had become, by the expiration of such and
such a period, the universal form of speech. In very
many cases we cannot determine what the predisposition
was that made the innovation successful; at other times
we can understand the predisposition (as in the case of
was for were) but then, as a rule, we fail to see why
the change succeeded in spreading at this and no other
time, and no other place. This last is, indeed,
in this
the greatest difficultywe have to encounter. It is almost
always left a mystery why a given change occurred where
and when it did, even though the motives of the change,
when it does occur, seem clear. ^)

1) This drawback is one that attends every phase of investiga-

tion neither amenable to expenment (as is, for instance, physics)
nor of universal validity (mathematics). That determination of
an event which the physicist or chemist can make in his labora-
tory iB due to the artificial simplification of the conditions which
is at his disposal. In language such experiment is impossible,
becanse we cannot make and unmake communities, govern all
essential factors, and observe through any length of time. No
g^ven occurrence, moreover, —
not even in the domain of natural
science, — could be fully accounted for by anything less than
omniscience: the distinction between an immediate cause and
202 INTERNAL CHANGE IN LANGUAGE

3. Change in articulation. Phonetic change consists

of change in the articulatory habits, independently of the
semantic content of speech. In Shakspere's time, for in-
stance, one of the English vowels was a long open [oi],
which occurred in stone [sto n], hone [bo n], rode [ro d],
: : :

and a great many other words. By the eighteenth century

the pronounciation of this sound had gradually changed,
until it was a long closed [o:]. Today we further close
the end of it, so that it has become a diphthong [ow],
and we pronounce [stown, bown, jowd], and so on. In
many others, the cause of the change
this instance, as in
of habit is unknown to us. What this statement means
will appear, if we consider in detail how such changes
take place.
The Elizabethan [o:]-sound, like all other speech-sounds
at all times, was pronounced in a great many variants.
Even the same speaker would pronounce it now more open,
now more closed, now longer, now shorter, in an infinity of
unconscious, minute variations, — just as we today, and
all people, cannot possibly perform a repeated action, no
matter how habitual, with unfailing accuracy. So far as
the variations of quantity were concerned, the equilibri-
um was maintained, and has been to this day. If a speaker
at one time shortened the sound, he lengthened it at an-

an underlying condition is here forced upon us. If the physicist

were asked to tell why a certain electric fuse 'burned out' at a
certain time, he could tell us that such and such a current of
electricity had passed through it, but how, ultimately, he had
come to perform this experiment at this and no other time and s

place, he could not tell. Or, if a stroke of lightning had caused

the destruction of the fuse, no meteorologist could tell why the
lightning struck where and when it did. So in language, the >

immediate causes of phenomena are generally, except for sound-

changes, known, but the underlying conditions are too complex
for any known methods of investigation.
 CHANGE IN ARTICULATION 203

otlier, and if one speaker tended more often to shorten

it, his neighbor counterbalanced him in the other direc-
tion. With the variations of closure, however, it was
different. While in the speech of many individuals there
may, in this respect also, have been an equilibrium; while
with other individuals the more open variations may
have predominated, the more closed variations, whenever
they occurred, seem to have struck more fertile ground
in the hearers, —
to have better stimulated sound-mem-
ory and movement- memory. Gradually the more closed
variants predominated, until the average of articulations
was shifted more closed sound, let us say [oix]. By
to a
this time the most open variants of [o:] no longer occur-
red, and, on the other hand, more closed forms began to
be spoken than had ever been pronounced when [oi] rep-
resented the mean. Why all this took place in this
direction and not in the opposite (in which a similar
change has, in fact, occurred at other times and in other
places), and why equilibrium was not, as in the case of
quantity, maintained, no one can say. Since then Eng-
lish pronunciation-habits have leaned toward the closed
variants, until in the eighteenth century the mean of
pronunciation, as we have seen, became
and today[oi],

we speak [ow]. Thus the same variations have been

favored by eight to ten generations, if we do not look
back of Shakspere's time; if we do, we shall find that
the process had been going on since long before, as the
sound in Old English times was [a:]. Although theories
have been proposed, attributing such movements as these
to influences of climate, food, occupation, and the like,
none of them have been more than mere surmises, con-
tradicted by the next best set of facts that presented it-
self. Thus, it has been suggested that phonetic change
is due to increasing speed of the process of articulation,
204 INTERNAL CHANGE IN LANGUAGE

v\^hich, it is accompanies the increasing rapidity of

said,
the mental processes due to the advance of culture and
the general development of man's intellectual powers.
This explanation is in itself plausible, because, as we
shall see, sound-change does tend to produce more and
more rapidly pronounceable and psychologically more
appropriate forms. It is true, moreover, that a prescribed
complicated series of delicate movements to be rapidly
performed wiU be incorrectly gone through. In sound-
change, however, we see the movements being performed
in approximately the same way sometimes for centuries,
before certain variants come to be favored; and when the
mean of articulation has been shifted, stability may again
ensue. It has never been proved either that there is an
increase in the rapidity of articulation or that an increase
of rapidity would account for such changes as have
all

occurred; — the vowel change of English just described,

for instance, does not seem referable, directly at least, to
increased speed of articulation. So much we can safely
say: that sound change, altogether, is an evolutionary
process, a phenomenon of the historic change of man's
physical and mental organization and habits.
While the causes of sound-change, then, are obscure,
its general manner of action and, in many cases, the trend

of its results are plain.

The action of sound-change has often been described
as a process suffering no exception and comparable, con-
sequently, to the laws discovered in physics or chemistry.
In reality sound change is a change of habit in a com-
munity of people and is due, ultimately, to some change
in the organization or environment of these people. And,
as mankind is undoubtedly active in the shaping of en-
vironment, the description of 'phonetic laws' as 'natural
laws' cannot be correct: a sound change is not a law of
 CHANGE IN ARTICULATION 205

nature, but a historic occurrence. Those who, otherwise

than metaphorically, have subscribed to the above defi-
nition, have been misled by a certain salient feature of
phonetic change, — namely, its unsemantic character.
From the definition of phonetic change, it is clear, that,
once certain variants are being favored, it matters not
what the semantic content of the word or of the sen-
tence may be, the result is the same. The speaker has,
of course, no suspicion that he is making any variations
at all or that he and his fellows are favoring one or the
other type of variants. He is intent on expressing his
meaning, and for all he knows, is expressing it in the
same sounds, words, and constructions from one end of
his life to the other. The favoring of phonetic variants
in this or that direction, — that is, sound change, —
has nothing to do with the meaning of the particular
word or sentence that is being uttered. Owing to its
uniformity, then, throughout words of the most various
meaning, a historic change in the manner or place of a
given articulation may be called a 'phonetic law'. There
can be no objection to this term, provided we do not
allow it to mislead us. When we say, as is customary,
that by a phonetic law 'Elizabethan [o;] >
modern [ow]',
we mean that the average of articulation at the former
time was [o ;], that the variants in the direction of [ow]
were favored, unconsciously and regardless of the mean-
ings of words, and that today, as a result, [ow] represents
the average of pronunciation.
The metaphoric term 'phonetic law' is very useful, be-
cause it emphasizes the phonetic, articulatory character
and the regularity, no matter what the semantic content,
of the process of sound-change. Linguistically untrained
observers will often claim that a possible sound-change
did not take place because, if it had, it would have ob-
206 INTERNAL CHANGE IN LANGUAGE

literated some important semantic distinction, or that a

given soimd-cliange did take place because a certain se-
mantic distinction, which it obliterated, was no longer felt
to be necessary. There is no need of referring to the
concrete details of the process, to show the impossibility
of such motives. For all we know, we are today in the
act of making a sound-change that will obliterate the
most clear-cut or the most universal distinctions in the
English language. At least that is what has happened
again and again in every language whose history we
know. Thus, a categoric distinction in the oldest stages
of English was that of the cases of the noun; yet, by the
time of our earliest records the nominative and accusa-
tive of many nouns had come to be spoken alike: for in-
stance, 'stone' in these two cases was in the singular
staUj in the plural stanaSj although many other words
still had distinct forms: for instance, 'care', nominative

singular carUy accusative singular cara. In primitive

Germanic the nominative singular of 5^aw had been *6'^a«2O0,^)
the accusative ^stainon, but a phonetic change had led
to the dropping of these semantically important final
syllables. Meanwhile the dative case was still distinct:

e. g. singular stdne; likewise, the genitive singular stdnes

was from the nominative-accusative plural stdnas.
different
We know how little phonetic change has spared these
categoric distinctions: the dative case has been lost,
coinciding today with the nominative -accusative in the
form stone [stown]; cara and caru are now both care
[kejj]; the genitive singular stdnes and the nominative-
accusative plural stdnas are now stone's stones [stownz],
only the written language making a distinction. As to

1) The asterisk means that the form does not occur in our
historical records.
 CHANGE IN ARTICULATION 207

words, TiYiigJit and niglit were in Chaucer's time different:

[kni9t] and [ni9t]; today tliey are indistinguishably [naet].
In present English [hiz'maet] may mean either 'his great
power' {might) or 'his trifle' (mite), as in 'He contributed
[biz 'maet] to the effort',but in Middle English the former
meaning was pronounced [mi^t], the latter [miite]. If we
wish to distinguish, we must do so by what we explicit-
ly say, for instance by using the adjectives great and
little. In Chinese there has been a great deal of such loss

of distinctions through phonetic change; the language

is consequently so full of homonyms that, though on
paper the writing may, as with us, show which sense is
meant, in speech phrases or compounds must be used
whenever homonymy might otherwise make the meaning
obscure. Thus yu' ] 'father', as in [Lfu\Lmu/] 'father and
mother', 'parents', cannot be used alone in speaking,
owing to the homonymy with a number of other words;
one must use the phrase [fu '^C9'in^] 'father-relative'. Sim-
ilarly [T] 'garment, clothing' cannot be used alone:
one says [T'^fu''] 'clothing-atensil', for the word by itself
is homonymous with a number of others, e. g. ["^i^J 'phy-

sician'. A similar phenomenon appears in English dic-

tionaries: our verbs have so largely become homonymous
with nouns that they have to be quoted with the word
to, e. g. in a French dictionary ^dormiVj to sleep', where

in German, for instance, one could write 'dormir, schlafen',

— for in German the noun 'sleep' is ScJilaf.
Not only does, a sound-change always extend over all
the occurrences of a single sound, but it may extend over
several sounds. The sound-change by which Elizabethan
[stoin] became eighteenth-century [sto:n] was a closing
to mid-position of the long open vowel which took effect
regardless of the point of articulation; for the front-
Towel counterpart of [o:], namely [s:] as in [ns: m] 'name'
208 INTERNAL CHANGE IN LANGUAGE

shared iD the change, being in the eighteenth century

[ne:m]. Similarly, the diphthongization that followed
affected both of these vowels: we say [stown] and [nejm].
In fact, the sound-change has gone even beyond this: we
may say that there has been, since Chaucer's time, a
constant closing and diphthongization of long vowels,
all
— as will appear from the following examples:

Middle Early Modern .g^h Centurv

^^ Century. Present
Present.
English. English.
[naima] [nsim] [ne:m] [nejm] 'name'
[drsim] [dreim] [d.ii:m] [daijm] 'dream'
[sto^in] [stoin] [stoin] [stowu] 'stone'
[moid] [moid] [muid] [muwd] 'mood'
A similar parallelism appears in:

[wiin] [wejn] [wsjn] [waen] 'wine'

[huis] [hows] [hows] [haos] 'house'.

The reason for such parallelisms is plain, if we recall

(p. 53, f.) that the various sounds of a language consist of

a number of manners of articulation practised at a
number of points of articulation. In phonetic change it
is usually one of these factors rather than the production
of any one sound that changes; this involves, of course,
allsounds in which the changed factor occurs. Thus, if
the articulation of fortis stops changes to spirant arti-
culation, [p, t, k] will become, respectively, [f, 0, x], —
a change that took place in the pre- Germanic period in
the history of English. It was followed by another
example in point: the voiced aspirate stops [b', d', g']
became, in complete parallelism, voiced spirants [id, 5, g].
Later the plain voiced stops [b, d, g] became unvoiced
[p, t, k]. These three changes together are known, by
the name of their discoverer, as 'Grimm's law'. So, much
later, in German, [p, t, k] became, between vowels, [f, s, x],
 CHANGE IN ARTICULATION 209

whence such correspondences as English grope: German

greiferij English water: German Wassery English make:
German machen.
There are many phonetic changes which, though the
causes that brought them about are no clearer than in
other instances, allow at least of classification as to their
immediate tendency. In speaking of sound-changes that
obliterated semantic distinctions, we saw a sound-change
by which knight and night became homonymous: early in
the seventeenth century initial k before n was dropped.
As this did not happen after vowel, as in acknowledge, it

was probably after certain preceding final consonants

that the new variants were first favored. These new va-
riants involved a simplification of the required articula-
tory movements. To pronounce kn- one had to begin
with voiceless, non-nasal, velar articulation [k] and then
to change to voiced, nasal, dental articulation [n], — in
other words, simultaneously (1) to start one's vocal chords
into voice- vibration, (2) to lower the velum, and (3) to
move the tongue from velar to dental articulation. It is

evident that a variant which saved, for instance, the

lowering of the velum (2), by beginning the word with
nasal articulation (or retaining it after a final nasal of
a preceding word, as in yon knight) simplified the whole
performance by lessening the number of required simul-
taneous movements. Such a variant would be [gnejt].
A further variant which began at once with tongue-arti-
culation at the n- position (or kept that position after a
preceding dental consonant, as in that knight), namely
[nnejt], brought another simplification by saving the ne-
cessity of a change of tongue-position (3). A variant
that saved the adjustment of the vocal chords into voi-
cing position during the n- sound, was to voice the initial
ound (or to continue voicing after a preceding voiced
Bleomfi«ld, Study of Lanyuago 14
210 INTERNAL CHANGE IN LANGUAGE

sound, as in the Tinight), giving [niejt] and [nejt], present-

day [naet]. The variants that were favored resulted,
thus, in a simplification of the movements necessary to
produce the word. They may be said to have resulted,
further, in a lessening of the total number of different
combinations of movement occurring in the language, —
since [nn], for instance, did not otherwise occur. Thus,
while we do not know the actual causes, we see here the
typical results of many sound-changes. But we cannot
follow the tempting course of arguing directly from
these results to the causes, for the results do not indi-
cate why the change took place where and when it did.

Thus, if the simplification in the above instance were

looked upon as the cause, would be inexplicable why
it

it did not take place earlier: by the seventeenth century

people had been speaking initial Izn- from time imme-

morial and always, it seems, maintaining an equilibrium
of habit. Or, if we look to German, how does it come
that the simpler variants have never gained ground there,
Ten- being still spoken, as in Knecht [kns^t] 'servant',
Knie [kni:] 'knee', Knahe [kriaibe] *boy'? We are face
to face with the same difficulty that we met above, with
regard to the closing and diphthongization of the English
vowels. In this second case we can see that the favored
variants brought a simplification, — involved a lessening
of the labor of speech, but do not know why
we still

they were favored here and not elsewhere, now and not
sooner.^)

1) It seems possible that the new variants in the change of

kn- tended to occur rather after non-syllabic than after syllabic
sounds (cf. acJcnoivledge), and that the increasing diphthongization
of long vowels, which would often precede the kn- words, as
in I know; thou knowest etc., led the n- variants to be favored
ttft this particular time. If this explanation should prove correct,
 CHANGE IN ARTICULATION 211

Very many of the sound-changes known to have oc-

curred in different languages show this same result of
simplification of movements. In JcnigJitj night, mighty etc.
we have dropped the palatal spirant of Middle English
[kni^t, ni9t, migt]. This was a simplification not only
because it saved an articulation, but also because it less-

ened the number of different articulations in the lan-

guage, which after the change contained no more [9] 's.
In general, the successive sound-changes in a language
often result in shortening the words. Where there is
highest stress-accent on one syllable, we find sound-
changes taking place, which shorten the unstressed syl-
lables. For instance^ in English, which has since pre-
historic times had such an accent, the primitive Germanic
^stainoz 'stone', nominative singular, and ^stainorij accu-
sative singular, were shortened, by the historic time, to
Old English stan; the Old English stanej dative singular,
is in modern English, like these other forms, monosyl-

labic stone [stovvn], and the Old English genitive singular

stdnes and nominative and accusative plural stanas also,
are today monosyllables, stone's, stones [stownz]. The Old
English cam 'care', nominative singular, and aU the
other forms of this word, in Old English disyllabic, are
today limited to one syllable, that which was in Old
English stressed, —a phenomenon so universal, in fact,
that our language, which in earlier periods had almost
no words of less than two syllables has now few uncom-
pounded native words of more than one. So, to quote
a classical example, pre-English ^'haheda and *hahedun
became in Old English hcefde and hccfdon and are in pres-
ent English (i, ue) had

onr sound-change would be due to the closinf^ and diphthong-

ization of the long vowels, which in turn, however, ia unex-
plained ag to motive
212 INTERNAL CHANGE IN LANGUAGE

In addition to tlie general facilitation involved in the

reduction of words from several syllables to one, it is

especially to be noticed, that, given a powerful stress-

accent, indistinct, shortened articulation of the unstressed
syllables represents an immediate simplification, because
of the particular voice and breath conditions. The strbng
stress on the accented vowel means that during its pro-
duction the vocal chords are vibrating under high tension
(p. 25); after this tension is released there follows an
outrush of the breath which was shut in under the vocal
chords during voicing: it is easier, consequently, to pro-
nounce them with shortened, weakened articulations and
with murmur instead of voice, lessening as much as pos-
sible all interference with the breath-stream. Thus, in
the history of English we find such Latin-French loan-
words as contemporaneous prohibition changing from the
full values of the vowels as written to the modern forms
[kijitempe 'jejnjes, pjhe 'bijn], with the unstressed vowels
often murmured. One might weakening an
see in this
immediate result of the stress-accent on one favored syl-
lable, were it not that in Italian, for instance, the same
words, with a similar accent, retain their full form con-
temporaneOj proihizione. The same change as in English
has occurred in other languages with high word-stress.
The contrast in Latin, for example, between facio *I do,
I make' and its compound
conficio due to the
'I finish' is

pre-Latin conditions, in which all words were spoken with

stress on the first syllable: at this time the second syl-
lable of conficio received its divergent weakened form.
The stress-accent of Russian has similarly changed an
older [po to ro 'p'i s'i] 'hurry up' to [patara V^'^']? ^^
older ['kia- n aj t e s'i] 'give greeting' to ['kia* n'i t'is'],

and so on.
In other instances the simplification brought about by
 CHANGE IN ARTICULATION 213

a sound-ctange is apparent upon closer inspection. For

instance, it has happened in the history of a number of
languages that such combinations as nrj Ir became w^r,
Idr^ and such as mr, ml became mhr, mhl: in Middle
English, for instance, jAinres 'thunder's' became liindreSy
aire 'alder-tree's' became Hhimble' became
aldre, J,ywle
lywhle, and *geiimre 'carpentry' had in Old English al-
ready become getimhre (cf. modern timber). The apparent
addition to the required series of articulations is really
a simplification. In the transition from n to r, for in-
stance, two changes had to be made simultaneously: (1)
the velum raised, to stop nasalizing, and (2) the tongue
moved from the contact of n to the position for r. The
change consists in raising^ the velum a little before the
tongue-movement, — an w, 7n with raised velum is of
course a c?, &, — so that only one movement need now
be made at a time: addition to the number of movements
there is none.
Some of the sound- changes so far discussed affected
certain sounds wherever they occurred in the language;
such, for instance, was the English vowel-shift, also the
loss of [9] in English. Other changes, called conditioned
sound-changes, occur only in certain phonetic surround-
ings. Thus hi in English was not simplified to n be-
tween vowels, as in achiowledge, the English, pre-Latin,
and Russian vowel-weakenings occurred only in unstressed
syllables, and the pre-Germanic change of [p, t, k] to
[f, 0, x] (p. 20^) did not take place when these sounds

were preceded by a spirant: spoon and stonCj for example,

have preserved their p and t since before this pre-Ger-
manic change. The High German change of [p, t, k] to
[f, s, x], similarly, did not occur after spirants: the Ger-
man words Spalin 'splinter' and Stein 'stone', German
representatives oi spoon and stone^ have retained the stops.
214 INTERNAL CHANGE IN LANGUAGE

Initially the same sounds changed in German to [pf, ts,

kx], the former two, for instance, in Pfund 'pound'
and ^ehn 'ten', as opposed to the corresponding English
words.
Such conditioned sound-changes are usually transparent
as to the simplification they afi'ord. We find, thus, that
the velar spirant [x] in German became the palatal [9]
wherever it followed a front vowel, as in ich [H^] T,
from older ['ix]. As in this case, the result is often an
assimilation of one articulation to the other, and the
term assimilation has come to be regularly used in this
connection.
When, as in thisGerman example, a sound is assimi-
lated to one preceding, we speak of progressive assimilation.
The modern English automatic sound- variation in the
plural-suffix between [z] and [sj is the result of a pro-
gressive assimilation: when in Early Modern English,
the unstressed vowel of forms like stones [stoinez], leasts
[beistez] was dropped, the [z] was assimilated, as to voi-
cing, to a preceding unvoiced sound, whence modern
[stownz] but [bijsts].

Regressive assimilation is in the history of most lan-

guages commoner. A widespread type of it is 'palatali-

zation', the assimilation of a velar or dental to a follow-

ing palatal. Thus pre-English [k, g, g] became before
front vowels Old English [c, j, j]. The former two have
since become [tj, ^7^. Instances are our words childy
ridge, yield, which were in Old English cild, hrycg,gieldan,
pronounced [cild, hryj, jeldan], and go back to Primitive
Germanic *hildi3, ^hnigjoz, *geIdonon. Latin cinque (with
initial [k], a late form for quinque 'five') and generum
(accusative, 'son-in-law') have had their initial sounds
palatalized in the modern forms, such as Italian ['t/igkwe,
'd^enero] and French [ssik, 3a:dr]; in these languages we
 CHANGE IN ARTICULATION 215

can see also the palatalization of dentals, as when Latin

pretium 'price' and radium (accusative) ^ray' give in
Italian ['prstso, 'rad^o]. The tendency to palatalization
is especially active in the Slavic languages, where velar
and dental stops have heen repeatedly subjected to it,
and all other consonants at least once; thus a pre-Slavic
H'ensii- 'part' gave in Cechish [tja* st*] in Russian [tja* st'],
and in Polish [tJs'J't'J'].
Another example of regressive assimilation is the vowel-
change which took place during the first millenium of
the Christian era in all the Germanic languaores, assimi-
lating stressed back vowels to following front vowels,
probably through the medium of palatalization of the
intervening consonants. Thus primitive Germanic *harjo3
gave Old English here 'army', *fdti;2 gave fet 'feet', the
loan-word (from Latin undo) *uyM- gave ynce 'inch', and
*musiz gave mys 'mice'.
A total regressive assimilation is the development of
Latin p^ and ht into Italian tt, as in septem 'seven' >settCj

octo 'eight' > otto.

Assimilation may be to both the preceding and the

following sounds, as when in pre-Latin s between vowels
became voiced, 2\ this z later changed to r, whence the
inflection of genus 'race, kin, sex', genitive generis^ from
older *geneses.
The assimilative tendency may
be counteracted in cer-
tain connections. In English there has been a strong
tendency to voice unvoiced spirants; this tendency seems
to be assimilative and due to contiguous voiced sounds.
Thus in the early period of Modern English 5/on^5['sto:n8s]
became ['stoinaz], luxurious [luk 'Ju: ri us] became [lug
'ju: ri us], tvith [wi0] became [wicJ], is [is] became [iz],
was [was] became [waz], of [of] became [ov], and so on;
but this tendency was counteracted, after an accented
216 INTERNAL CHANGE IN LANGUAGE

vowel, by some other factor, — perhaps by the outrush

of breath after the stress. Hence in such words as geesBy
pencCf luxury ['lAkJeii], and in the accented uses of o/*,

now in such cases written offj the spirant remained un-

voiced. Initially, also, it was kept, owing, no doubt, to
the lack of preceding voicing, as in select, forget. At an
earlier stage of the language, probably in pre-English,
the older [9] of the, this, that, then, etc. became in the
same way, owing to unaccented use of these words, [cJ].

At a still earlier period, in pre-Germanic, there was a

change of exactly the same kind, by which for instance,
*wdsa 'I was' remained unchanged, but Hvesumun 'we were'
became Hvesfumun, —
which difference of consonant is
preserved in the present luas: were. This pre-Germanic
voicing is known as 'Verner's law*.
It is customary to sound-changes as
set off certain
'sudden' in opposition to the majority which, likeall our

examples so far, are called 'gradual'. These so-called

sudden sound-changes are changes in pronunciation which,
in part, could not have been arrived at through a series
of variants. Two types are comparatively common, meta-
thesis and dissimilation.
There are but few indisputable cases of metathesis in
the known history of languages. Part of the supposed
cases are really gradual sound-changes. Thus pre-Eng-
lish *rinnan (as in the Old English compound gerinnan
'to coagulate') gave Old English iernan 'to run', and pre-
English %rinnan gave Old English hiernan 'to burn'. In
these examples the 'metathesis' was probably a gradual
process, the r first becoming syllabic to the exclusion of
the vowel, which left only a palatal affection: [r'nan,
br'nan]; later a new vowel, determined by the palatal
coloring, arose before the Other cases of metathesis,
r.

like the English dialectal ax for ask or the pre-German

 .

CHANGE IN ARTICULATION 217

*atik 'vinegar' (modern Essig) for older *akiio are prob-

ably not phonetic changes at all, as we shall see.
Dissimilation is due to the tendency which appears
when one tries to articulate such series as Feter Piper
picked a peclz of pickled peppers. When the vocal organs
are to be placed repeatedly into the same position, it is
hard to keep in focus the exact part of the prospective
movement-complex at which one has arrived: the ten-
dency is to mistake the quicker movement of the atten-
tion for the slower one of actual articulation, to con- —
fuse an earlier for a later stage of the series; thus one
might say: Feter Fiked for Feter Fi{per pi) eked
. . . . .

Another tendency is to confuse the unwonted repetition

of the same movement with some more practised succes-
sion of diverse articulations, — to say, for instance,
Feter Piper ticked .Both of these tendencies have in
.

rare instances brought about permanent phonetic changes.

To the former tendency are due the so-called haplologies,
such as Latin stipendium 'stipend' for older *siipipendium
or Ancient Greek amphoreus 'amphora' for earlier amphi-
phorals. As well as a repeated syllable a repeated sound
may be omitted, as in the colloquial Latin cinque [kirjkwe]
'five' for earlier qiiinque [kwiijkwe]. The other phase
of the dissimilative tendency appears in such changes as
Late Latin pelegrlnus 'pilgrim' for earlier peregrimis.
Here again most of the quotable examples, including
probably this, are really not cases of phonetic change in
the strict sense,, but rather of assimilative mispronuncia-
tion of words of a foreign language: pelegrlnus was prob-
ably originated by people whose native language was
not Latin, or at any rate did not contain this word. A
genuine dissimilation occurred in pre-Greek, where two
successive syllables beginning with aspirated stops were
dissimilated, the former losing its aspiration: thus, *lhc-
218 INTERNAL CHANGE IN LANGUAGE

thnake 'he has died, is dead' became tethneJce, *Jchekhena

'Ihave gaped' became JceJxhcna. The same dissimilation
took place in pre-Indic, such forms as *dhadhati *he puts*
becoming in Sanskrit dadhati.
A similar process may on the other hand, to
lead,
assimilative repetition of the same sound in the place
of two articulations originally different, as when a pre-
Latin *piheti 'he drinks' gives Latin hibit, with h- for p-.
Such sudden changes as all these are, however, rare,
compared to the gradual changes above described.
The change in the complex conditions of hu-
specific
man existence which brought about a given sound-change
is, then, as a rule, hidden from our view. We have reason

to believe that, if we knew the underlying change in the

conditions of life, we should find it affecting not the
particular sounds which we see changing, but rather
some more general factor of articulation, such as the
rest-position of the vocal organs, or the manner of voi-
cing and breathing. The preponderant tendency of sound-
change to simplification of word-forms, and the harmony
of various sound-changes in a language (such as change
of stop - articulations in pre-Germanic, spirant -voicing,
shifting of vowels, and weakening of unstressed syllables
in English, palatalization in Slavic), all point in this
direction. As a result of some such fundamental change
there take place the various sound-changes found to oc-
cur in a given period. The individual sound-changes
bear, each of them, accordingly, the character of lessen-
ing of the labor of speech by means of adaptation to
the prevailing rest-position of the vocal organs or to the
prevailing manner of speaking. We may
suppose that
every sound -change is assimilative in nature, changing
some discordant element in the habits of pronunciation
into an articulation harmonious with the total speech ac-
 CHANGE IN ARTICULATION 219

tivities of the time. Each change, however, probably in

turn displaces this total habit-complex, so that further
adjustments become necessary and an endless series of
sound-changes results. Accordingly, when we establish
that a given sound-change is simplification, we are not
naming its cause, but merely describing, in part, the
general nature of sound-change. The cause is probably
in every case a change in the speech-habits as a whole,
due, in turn, to a change in the underlying conditions of
life. As soon as we try to determine these we are on
the open sea of surmise.
Only under one set of circumstances do sound-changes
bid fair to be thoroughly explained. We know that it

is only under the most favorable pedagogic conditions

that people ever learn to pronounce a foreign language
correctly. As a result of a simultaneous association, one
hears a resembling familiar sound where a foreign one
has actually been uttered, and even when one hears the
latter correctly, imitation is usually impossible (p. 19).
We see this in the speech of the foreign people who
learn English. In some parts of Wisconsin, for instance,
a German even in the speech of the
'accent' is audible
younger generation that does not speak German, but
learned from its parents the English (with German sounds)
which they spoke. In this case the growing intercom-
munication with people who speak purer English wiU
no doubt in time efface the peculiarity. There have been
however, instances where a comparatively small number
of conquerors have forced their language on a people of
alien language. In such a case we may expect to find
substituted for the sounds of the new language the cor-
responding nearest sounds of the old. The clearest in-
stance at present known of such sound-substitution is in
the Indo-European 1 inguagcB of India. In prehistoric tinu-H
220 INTERNAL CHANGE IN LANGUAGE

the bearers of the Indo-European pre-Indic language enter-

ed the country and forced their speech upon certain sections
of the inhabitants, from whom, in the course of time, it

spread farther and farther. As some of the older languages,

however, escaped and are spoken to this day, for in- —
stance, in the Dravidian languages, —
we know something
about the phonetic habits of the people upon whom the
pre-Indic language was forced. They had two series of
tongue- tip one purely dental, like the modern
stops,
French series [t, d] and one pronounced even farther
back than our alveolars, namely the cerebrals [t, d]
(p. 30). The language of the Indo-European invaders
had only the former series, but in all the historic lan-
guages that represent it, such as Sanskrit, we find the
latter series also: it was substituted, evidently, for some
sounds not familiar to the older inhabitants, probably
for the dorsal articulations [c, j] produced in the same
place. The task of tracing such sound-changes has been
barely undertaken by students of language; it may be
expected that the phonetic change in the history of such
dialects as South German, the Romance languages, and
the Slavic languages will receive light from this treatment.
When a conditioned sound-change occurs a new multi-
formity of sounds may be created Thus pre-English
had no palatal stops until there occurred the change of
velar stops to palatals before front vowels, as in the
words childy ridge. Similarly, Old English had no ;8?-sound,
but, owing to changes in which 5, under certain condi-
tions, came to be voiced, we have now both 5 and z. As
long as the conditions remain undisturbed the result of
such a chancre is an automatic sound-variation. In form-
ing the genitive of nouns, for instance, we use the suffix
[ez] after sibilants, [s] after unvoiced sounds other than
sibilants, and [z] after voiced sounds other than sibilants
 ANALOGIC CHANGE 221

The automatic in this suffix. The same is

variation is

true of the preterite suffix of our regular verbs {waitedy

waded, —
passed, hoped, —
turned, roived). The variation
is automatic, further, in the sibilant plural suffix of our
nouns, in so far as they are regular. In our colloquial
pronunciation [j] varies automatically with [J] after [t]
and with [3] after [d], for we say have you, don't you,
did you with three different initials of you. There was
a time when of [ov], witJi [wi^] were used in unstressed,
off [of], tvith [wi9] in stressed position (p. 215).
This automatic sound -variation may in several ways
be disturbed. Further phonetic change may do it The
pre-Germanic spirant-voicing after unaccented vowel, for
instance, left such automatic variations as ^wdsa T was':
*wezumun Ve were' (p. 21G); when, however, the stress
was later everywhere to the first syllable, the
shifted
variation was of course no longer automatic, but purely
traditional, as still in the modern forms, luas: iiere. So,
by a pre-English vowel-assimilation (p. 215) ^-fotiz, the
nominative plural oi*fdt 'foot', became *fetiz, a variation
whose automatism was destroyed by the phonetic change
which dropped the second syllable of *f"etiz, giving Old
English fit: here, as in the Modern English foot: feet,
the variation is, of course, no longer automatic.
The other processes of change in language mentioned
at the beginning of this chapter (p. 19G), which we shall
now discuss, may bring about the same result.
Analogic change. When change in the form of
-4.

words is in any sense due to their meaning, we speak

not of phonetic, but of analogic change.
We have seen that partial formal similarity between
words, when it expresses a corresponding semantic simi-
larity, is due to the psychic factor of assimilation (p. 59, f.).
We have seen, further, that all such correspondence be«
222 INTERNAL CHANGE IN LANGUAGE

tween meaning and form maintained by repeated pro-

is

cesses of assimilation in mind of every speaker

the
(p. 93, f.). If the same assimilations were always made, the
form of words would, except for phonetic change (I use
the word here, as always, in the strict sense),
remain
stable. There however, a number of factors which
are,
constantly displace the conditions under which the assimi-
lations of speech take place.
Such factors are changes in mental organization, in
culture, in surroundings. Not only changes in habitat
and the progress of civilization, but also the vicissitudes
of all the individual lives that go to make up the com-
munity, make it impossible that the same topics should
always be spoken of, by the same peo-
or be spoken of
ple. The frequency, absolute and relative, with which
any sentence or word recurs is constantly changing. As
a result, words well practised a few centuries ago are
now rare, and words then rare or unknown are today in
constant use.
Another factor is the effect of phonetic change. Forms
that where once nearly alike may become very different,
owiug to a conditioned sound-change. In pre -English
there was a large group of nouns that formed their
plural in -as, e. g. Old English Jiringj plural hringas
(modern ring, rings)'^ stan, stanas (modern stone, stones).
When one spoke the plural of such a noun the others,
of course, gave assimilative support. Among them was
also ^kmf 'knife', plural Vinlfas. During this period,
however, /"between voiced sounds became voiced, so that
we have to this day singular Imife, plural knives, with
change of consonant, and this plural is now, of course,
irregular: when it is being spoken the regular plurals, such
as rings, stones, cliff's, no longer lend their full support. In
other words, the conditions of word-formation are altered.
 ANALOGIC CHANGE 223

Another highly potent factor is the change of speak-

ing individuals. Every child has to learn all the habits
of association which form the language, including of
course the assimilative habits of inflection and derivation.
Years elapse before the child's experience with speech is
anything like that which may be called normal in the
community. Insufficient practice, to take a common in-
stance, in the preterite forms brought came will allow
j

an assimilation by the regular preterite forms, such as

lived, played to become efi'ective: the child says hringedy
corned. The child is soon cured of the most striking of
these false assimilations, but no speaker and no genera-
tion of speakers ever succeeds in reproducing entirely, in
this respect, thespeech of those who went before. Now-
adays the great prevalence of printed speech lessens our
divergence from earlier forms, especially as it allows of
compendia (grammar and dictionary) of what is 'standard'
speech; yet the obvious fact that we cannot and do not
speak as we write is a confirmation of what has been
said. The 'ungrammaticaF speech of the classes less
familiar with books, is a further witness. Thus we find
that the plural of cow was formerly Jcine, of hook a form
that would now be leech (Old English hec). These
changes scarcely differ from those of the child when it
says hringedy corned, — only, in fact, in that the condi-
tions of the speech-community as a whole were such as
to produce the innovation independently in many indi-
vidual speakers and to make it so natural for their hear-
ers,that these for the most part accepted the new forms
without being aroused by their novelty. Where a hearer
of an older generation, who had used and heard the older
form of the word in question too much not to notice the
innovation, would correct a spesiker, there people might
become conscious of the change, but for the most part
224 INTERNAL CHANGE IN LANGUAGE

the process is as unconscious as the child's innovation

of hringed. Every one of us is taking part in such move-
ments, unconsciously, unless corrected by someone of an
older generation or someone more practised in the word
concerned, — and even then, if the change is well under
way among our fellow-speakers, we usually forget the
correction or at least the direction which it took, —
whether it was form A or form B which we were told
was wrong. In communities without written records or
with but little use of writing, —
that is, in all but a
few modern communities, —
the authoritative force of
literary usage, grammar, and dictionary is, of course,
wanting on the conservative side.
This process, then, is called analogic change. We
ought rather to speak of innovation due to the assimi-
lative influencing of articulation by semantically associated
words, — that is, of ^associative sound-assimilation'. The
term 'analogic change' or ^analogy' is, in fact, retained
only because it is conveniently brief. The modification
'false analogy' is better, because it conveys at least the
idea of innovation, as opposed to the regular assimilative
processes by which all speech is formed.
The term 'analogy' is most out of place in the simplest
instances, those which most clearly show their character
of assimilations; in speaking of these many scholars ac-
cordingly prefer the name 'contamination'. Contamina-
tion is said to occur when the articulation of a word is

assimilatively modifiedby that of another single word

of related meaning. Genuine cases of such contamination
are hard to find; as a rule, when the conditions are
minutely studied, it appears that some further factor
beyond the single semantically associated word was ac-
tive in inducing the change. The English pronoun ye
is supposed to have been a contamination-form due to
 ANALOGIC CHANGE 225

tbe influence of we upon you. The numeral four began

in Primitive Indo-European with a velar or uvular con-
sonant with lip-rounding, qu, and probably got its initial
under the influence of five^ but there may have been
also an element of dissimilation in the change: the word
seems to have been at one time *quequor-^ where dissimi-
lation to *pequor- would be possible. Even aside from
this possibility, the process of counting is so common
that we might speak of a regressive assimilation, ^five'
affecting 'four'. The initial of the Slavic word for *nine',
originally no-, seems to have been assimilated by the
word for *ten': Old Bulgarian deveU 'nine' (we should
expect *nov§{i\ deseti 'ten'. The Latin nodu 'by night'
is due to an assimilation of node (ablative of nox 'night')
by diU 'by day'.
In most cases the assimilative effect is due not to
a single word, but to a whole series of words in which
the assimilating sound goes hand in hand with the com-
mon semantic element. Thus the word squauh may be
looked upon as a contamination of sqiieaJc by squall, but
talk and caw and even matv may also have been effective:
in such cases the term 'adaptation' has been used. Latin
gravis 'heavy' appears in Italian as greve. The change
of a to c may be looked upon as a contamination with
leve 'light' (Latin levis),or as an adaptation, if one sup-
poses breve 'short' (Latin hrevis) also to have been effect-
ive. English render is a loan from the French rcndre,
which owes its nasal to an adaptation of Latin reddere
to prehendere (French, prendre) 'to take' and other words
in Latin -endcre (French -endre), such as vender e (French
vendre) 'to sell'. an adaptation of
English egotism is

egoism to such words as deapotism, nepotism. English

shimmer appeared by the side of such words as Old
English scima 'a light' and the verb shine primarily under
Bloomfl«ld, Studj of Lan(fu»g« ]5
226 INTERNAL CHANGE IN LANGUAGE

the influence of glimmerj but other words in -immer were

no doubt also involved. A certain case of adaptation it
is, when, much later, the word flimmer appears by the

side of flatne and glhnmcr, shimmer. In this way arise

such sets of words as dasJij crash, dash, flash, etc. (p. 133, f.).
For instance, flash is on the one hand due to flame and
perhaps others of the fl- words, and on the other to
those of the -ash group. Jounce is due to jump, jolt and
to bounce, trounce.
The third and commonest kind of analogic change, to
which the term best applies, is called also 'proportional
analogy'. When, normal speech, we wish to form the
in
plural of girl and the number-element, owing to such as-
sociated words as hoys, curls, and, in fact to aU regular
plurals, is at once presented as [z], the result may, meta-
phorically, be viewed as the solution of a proportional
equation: 'hoy: hoys as girl: x\ This way of stating the
thing is, to be sure, misleading, — it is characteristic,
in fact, of a post factum way of viewing linguistic oc-
currences as if they were results of deliberate individual
action, — nevertheless, it makes possible a diagrammatic
indication of the place a new form holds in the morpho-
logic system. Thus when the form cows for Mne arose it
could be looked upon as the result of a similar equation
to that which gave girls, e. g. dog: dogs as cow: x. Hence
the name ^proportional analogy'; needless to say that in
reality the result is due not to any such mathematical
comparison, but to a number of complex and variously
graded psychic forces in each of the many individuals
that make up the community.
'Proportional analogy' is not separable from the pro-
cesses of contamination and adaptation. The English
dialectal squench, similar in meaning to quench, may be
looked upon as a contamination of this word and squelch*^
 ANALOGIC CHANGE 227

but, as the association of squirt and even of otter words

with initial S-, such as splash, spout, souse, etc., must also
have contributed to the assimilation, we may also speak
of an adaptation; finally, we may look upon the initial
s- as an element entering into articulation because of its
semantic tone, due to occurrence in squash, as opposed to
quash, splash as opposed to plash, smash as opposed to mash,
and even souse as opposed to douse: as this last was surely a
phase in the psychic process resulting in squench, this word
may be looked upon as a 'proportional' formation (plash:
splosh as quench: x). In short, any attempt at classifying
assimilative formations by the number of inducing words
is frustrated by the complexity of the processes involved.

There remains the classification of assimilative pro-

cesses according to the semantic character of the ele-
ments involved. It appears at once that these elements
may be, from the point of view of the language concerned,
either conceptual or relational.
The conceptual elements are involved, for instance, in
the origin of the English ye (p. 224), in the changes of
initial which produced the English four and the Slavic
word for 'nine' (p. 22Q>)'^ further, in the origin of squawk,
of squench, and of the Italian greve. The same may per-
haps be said of shimmer, flimmer, and flash, above; as
long as we can not with any freedom add -immer or -ash
to other elements, we can hardly call these formational
sound-sequences anything but material. Let us suppose,
however, that these phonetic elements -immer and -ash
should become extended to more and more words, until
it became customary to use them in a given signification

with any initial-element, then they would have become

freely usable derivational elements. It is, thus, analogic
change which gradually gives 'life' —
i. e. morphologic

mobility —
to derivational and inflectional elements.
16»
228 INTERNAL CHANGE IN LANGUAGE

A few more examples of analogic change in the con-

ceptual parts of words may be cited. The Latin word
for 'your' was vester: in the course of time its vowel be-
came assimilated to that of noster *our', under the addi-
tional influence, no doubt, of the parallelism of vos 'y^^'
and nos 'we': the word in the modern forms of Latin
corresponds phonetically to a voster^ e. g. French votre\
Italian vostro. The German verb liigen 'to lie, tell false-
hoods' was formerly liegen and owes its vowel to the as-
sociation of Luge 'a lie', with the parallelism of fliegen
'to fly': Fliege 'a ^j\peitschen 'to whip': Feitsche 'a whip',
and the like. English neither instead of older nawtJier,
j

owes its form to the influence of either and the parallelism

of nor: or, never: ever.
Analogic change in the grammatical elements of words
is even commoner. If we recall that the partial phonetic

similarity between the different inflectional forms of a

word is due to assimilative development (p. 59), it will
be apparent that this kind of analogic change is one of
the chief shaping forces of language. The historical in-
stances, as this leads us to expect, are frequent in which
divergence between the forms of a word is thus obliter-
ated. Psychologically, the closely associated forms of
the same word are, of course, powerful factors in bring-
ing about such assimilation. Of the examples of analogic
change so quoted the child's error of hringed, corned
far
for brought, came belongs here: just such analogies as
these transform suppletive and irregular into regular in-
flection. They have changed, for instance, the plural of
cow from kine to cows, that of hooh from beech (i. e. Old
English hec) to looks (p. 223). In these changes the other
forms of the word are active together with the parallel-
ism of the regularly inflected words. Here belongs, further,
the change in dialectal English of the preterite and un-
 ANALOGIC CHANGE 229

real of tbe verb to he. The form was entered into arti-
culation instead of Standard English were because its

assimilative influencewas supported by the entire volume

of habit represented by the remaining verbs of our lan-
guage I had it, They had it, I wish I had it,
: I saw —
it, They saw it, I wish I saw it, and so on, all confirm

the habit of articulating the same form in the plural

and unreal as in the singular of the real preterite; hence
They was there, I ivish I was there. As the plural is
commoner than the unreal, and in the present tense real
partly distinguished from the singular {he has: they have,
etc.) the form were is in this value better retained than

in the unreal.
The forms was and were dijffer by vowel and consonant
variation. The vowel-variation goes back to Primitive
Indo-European time; it is known among linguistic students
by the German name 'ablaut'. The consonant variation
arose in pre-Germanic time through the spirant-voicing
after unstressed vowels, 'Verner's law' (p. 216). In pre-
Germanic the two forms were at first ^iidsa and ^ivesume\
the spirant-voicing changed the s of the latter form to z:
*wezume\ later the accent came words on
to fall in all
the first syllable, whence Primitive Germanic ^we2ume\
in pre- Wpst- Germanic, finally, the <? became r: *iccerume\
then, what with certain pre-English changes, we find Old
English lices^ tv^ron. Owinof to these same causes a number
of verbs in Old English had sound-variation in the pret-
erite. Thus one said rdd '1 rode' but ridon 'we rode',
wrdt 'I wrote' but tvriton 'we wrote', seah 'I saw' but
sduon 'we saw', and so on.While phonetic change is re-
sponsible for the loss of the plural-ending in Modern Eng-
lish, the assoriution of verbs tli;it lacked the sound-vari-
ation, such as Old English [call '1 fell', \)\\xx\i\ feollon 'we
fell', woe 'I awoke', plural wocon 'we awoke', impelled
230 INTERNAL CHANGE IN LANGUAGE

the assimilation of the two stem-forms: rode, wrote, with

vowel corresponding to that of rady wrat, are now used
in the plural also; saw corresponds in form to the Old
English plural. The same levelling has taken place in
the other lanffuaojes that are modern forms of Primitive
Germanic. Of these the Scandinavian languages have gone
farther than English, all verbs, including var 'was, were',
having the same form for singular and plural. Dutch agrees
with English, except for the retention of the plural-suffix,
having assimilated everywhere except in was Vas', plural
waren Vere'. German has regularized this verb {war,
plural waren), but retains ward (more commonly, howev-
er, wurde), plural wurden 'became'. It is interesting, further,

to see how some of the other languages which historically

represent Primitive Indo-European have made the same
levelling of the preterite ablaut. Thus the Latin 'perfect'
tense, which is partly the same in origin as our preter-
ite, has everywhere the same vowel for singular and

plural, e. g. tutudl 'I beat', tutudimus 'we beat', where

Primitive Indo-European had, respectively, Hetoud- and
Hetud- (for instance, in Sanskrit stiU iutoda and tutudimd).
So in Ancient Greek, beside the singular pepoitha 'I have
placed confidence, I trust', we find the plural pepoithamen
instead of an older *pepithmen})
Another example of the regularizing force of grammat-
ical analogic change is the development of the English
noun. In pre-English there occurred a sound-change which
turned back vowels of accented syllables into front vow-
els, if there followed a front vowel in the next syllable

(p. 215). The resulting sound-variation, as seen, for in-

stance, in the Old English nominative singular f'ot, but

1) Here the change in Latin is due chiefly to a substitution

of old middle-voice forms for active, that in Greek to direct
grammatical analogy.
 ANALOGIC CHANGE 231

plural fei from older ^fotiz (modern foot: feet), is called

by linguists 'umlaut'. The inflection of the word 'foot'
was in Old English as follows:

Singular. Plural.
Nominative-accusative: fot fit (older *fdtiz)
Genitive: fotes fota
Dative: fet (older *fdti) fotum.

In the singular the lack of vowel-variation in most nouns

(e. g. 'stone': sfdn, sfanes, stane) led in time to a new da-
tive form where the nominative and
fote; in the plural,
accusative, the most-used cases, had e, the same influence
of regular nouns led to an analogic assimilation of the
vowel to e in the other forms, whence the modern in-
flection of singular foot^ plural feet. In the case of hoohy
modern plural hooks (p. 223), the plural form was assimila-
ted to the singular in vowel and consonant and also to the
regular plural inflection with sibilant.
This kind of grammatical analogy is called 'material'
as opposed to the 'formal', in which the assimilation brings
greater unity not to the forms of one word, but to the
corresponding forms of different words. Thus the change
in the vowel and consonant of the plural of hoohj by which
this form became more like the singular was 'material'
grammatical analogy, but the assumption by this plural
of the regular sibilant suffix was 'formal'. The 'formal'
process diff'ers from the 'material' in that the inducing
factor in the assimilation is not an element of the other
forms of the same word, but the total impulse of elements
of the desired meaning in numerous other words. In Old
English, for instance, only a limited part of the nouns
formed their genitive singular and nominative plural with
s-suffixes. The nominative plural of hus 'house', word
'word', teoru 'tar', treo 'tiee' was homonymous with the
232 INTERNAL CHANGE IN LANGUAGE

nomiuative singular: hus, word, teoru, treo-^ the genitive

singular of cane 'care' was care, the nominative plural
cara] spere 'spear' was in the nominative plural speru,
tunge 'tongue', mona 'moon' were in the genitive singular
and nominative plural tungan, monan, and so on. When
in these cases the corresponding forms of regular nouns,
such as stdn 'stone' (genitive singular stdnes, nominative
plural stdnas) induced assimilation to the 5-inflection, the
different forms of any one word were often made less
alike than they were before, but the total inflectional
habits of the language gained in unity. The French or-
dinal numbers formerly differed extensively from the car-
dinals: 'one' was ww, but 'first' premier 'two' deuXy 'se- -^

cond' second', 'three' trois, 'third' tiers-, 'four' qiiatre, 'fourth*

quart", 'five' cinq^ 'fifth' quint-, but above these numbers
the ordinal was regularly derived from the cardinal by
a suffix -ieme, e. g. six 'six', sixieme 'sixth', sept 'seven',
septieme 'seventh'. Both factors, the material and the form-
al, of grammatical analogy brought about a cinquicme,
quatrieme^ troisieme, and although second is still heard,
deuxieme is commoner, premier alone being undisturbed.
While as a rule grammatical analogy is thus regulari-
zing, this is not always the case. German feminine nouns
have all their singular forms alike; the only exception
is the genitive form naclits 'at night', influenced by the
masculine genitive tags 'by day': the force of the usual
feminine declension is, however, sufficient to retain, in
all but this somewhat isolated adverbial use, the regular
genitive Nacld', compare Latin noctu (p. 225).
It is needless to add that, just as old forms are assimi-
lated, so new ones by the analogic process.
are created
The regular assimilations of all speech are no different
from the creation of an analogic new form, such as hoolis
or the French deuxieme when they first were spoken, and
 ANALOGIC CHANGE 233

this process, in turn, is no from that by which

different
an entirely new form or word may Thus, such nouns
arise.
as telephone^ automohilej aeroplane were used in the plural
in English with the regular plural- suffix as soon as they
were introduced, without the least consciousness of an
innovation (p. 70). In short, the articulation of some
new analogic form is, in the process, no different from
that of an old form. The action of the speaker who first
used the plural automobiles was no different from that by
which on the same day he may have said wagons or
horses.
Certain morphologic habits, like the English s-suffix
for the plural, extend to an unlimited number of words
and spread freely to new words, while others are limited
to certain words and do not freely spread to others, —
such, for instance, as our suffix -hood for derived nouus,
as in hoyhoodj manhood, motherhood, priesthood, which
cannot be extended, e. g. to form undehood or friendhood.
We speak, accordingly, of ^living' and ^dead' morphologic
processes. Even a 'dead* morphologic process, as long as
the formal-semantic relation between the words affected
is still felt, —
that is, as long as it is still really a morpho-
logic process, —
can occasionally be extcjided by analogy.
The amount of 'life' in a morphologic process may be
of any degree, and may constantly change. Most of our
verbs form their past tense with a denial suffix; compared
to this very living process vowel-variation for tense is

dead: yet even it has, in certain cases been extended. The

preterite of the verb dive in Standard English is dived,
but is in dialectal speech frequently assimilated to that
of drive (strive), becoming dove. In German there is a
plural-suffix of nouns, -eraccompanied in certain cases
by vowel-variation (e. g. Lamm 'lamb', plural Ldmmer),
which has never been very living, yet, word by word, it
234 INTERNAL CHANGE IN LANGUAGE

has spread from use a thousand years ago in a dozen

words (and some of them not exclusive use) to exclu-
in
sive use in ahout seventy-five today. The -sh of such Eng-
lish verbs as clash, crash was probably in Primitive Indo-
European and perhaps even later a living suffix; from
this time we have the verb thrash and perhaps also mash.
By the Primitive Germanic time the suffix was dead. In
the historical period of English, however, through assimi-
lation due to the meaning especially of thrash y verbs in
-ash have been multiplied (e. g. flashfrom flame) until
-ash (not -sh) is today a fairly important morphologic
element and may, for all we know, become living. At
least, if it did, this would be precedented, to take one of
many instances, by the development of our living suffix
-en by which we derive verbs such as fatten j shorten,
moisten from adjectives and occasionally from nouns
(hearten). Just as there was in Primitive Indo-European
a living suffix which today would give -sh, so in Old Eng-
lish there was a living suffix -ian (phonetically this gives
today zero) by which verbs could be derived from nouns
and adjectives, e. g. wundian (modern to wound) from
ivund 'a wound' (modern wound). From the adjective
fcegen 'glad' there was thus derived fcegenian 'to rejoice'
and from the adjective open 'open' the verb openian 'to
open'. The influence of these words caused the innovation
of deriving from fcest 'firm, fast' not only fcestian but also
fcestenian (our fasten) , and later, under the added impulse
of this and each new formation, other verbs in -enian^
modern -en, which at last made this suffixation a living
one.
If we had as complete records of the history of lan-
guages as we could wish, it would thus be possible to see
in detail not only how old forms changed phonetically,
but also how new forms, whether inflectional or other,
 ANALOGIC CHANGE 235

came by analogic change into the language. Only a small

minority of English, German, or French words and forms,
for instance, could be traced back by merely observing
phonetic changes, to the earliest known stage of these
languages. In by far the most instances we should find
that our word or form ('phoneme'} had been created by
analocric change. As to the exact methods of this chanoje
the internal conditions in each language are of course
decisive. Thus
in French most morphologic classes are
due to the use of suffixes, hence it is by shifting these,
as in deuxieme, that new words arise. In English we have
some suffixation, and new words, for example in -er, such
as teacher preacher bicycler, advertiser can always be form-
y
^

ed, but since in English vowel and consonant variations

also are used, these too can underly assimilations, so as
to produce new words of the squaivJc or squcnch type. It
is scarcely necessary to add that compound- words, too,

in languages that have them, are formed on the analogy

of others; if, today, we can form in English a compound
like automoh'de-driver, in which the relation of the parts
is peculiarly compositional and would not be allowed in
syntactic collocation, tbat possibility is due only to the
analogy of such older compounds as carriagC'driver, mule-
driverJ giant-killer, and so on.
would be natural, had we not this knowledge of
It
analogic change, to see in many phenomena, such as the
rise of new words or the peculiar relations allowed be-
tween the members of compounds, some mysterious force
which presided over the origin of speech and now in these
instances bashfully shows itself to the degenerate present.
This mistake, indeed, has often been made, and one may
read in many places lists of 'primitive creations' (such
words and references to a time when
as squawk, flash)
'mere word-stems' could be joined, in any semantic rela-
236 INTERNAL CHANGE IN LANGUAGE

tion to eacli other, to form compounds. But whenever

the facts in a specific case are brought to light, it appears
plainly that, apart from phonetic change, analogic change
is the only power that shapes or creates the forms of
language, — that we have no right to assume that there
ever was a golden age when this was otherwise. To say
this is only to say that the semantic value of language
is always and exclusively dependent on the habits of ex-
pression of the community. If the associational habits of
speech in the community are such as to call forth the
utterance of a given form, that form will come, whether
it has ever been spoken before or is an innovation, and

will be explicable only on the basis of these habits and

never on the strength of any ulterior connection, real
or imagined, between the sound and the sense. The as-
sumption, for instance, that words like squeeze, squawk^
or flap are in some way inherently significant of the ac-
tions they designate and owe their origin to this signif-
icance, is unfounded. They are significant to English-
speaking people because, in the first place, we use them
in the sense that they have, and, secondly, because the
sound-groups they contain {squ-, fl-y -awlCf -ap) occur in
other words of related meaning. These sound-groups have
gradually come, in the history of the language, through
a series of analogic formations, to carry their present
meaning. Should it appear that they are in some other
way than by mere custom representative of the experien-
ces they designate, then we should know only one of the
factors that contributed to their spreading from sporadic
and innovational to regular use.
Even were a speaker consciously to set out to invent
a word, he could not escape the influence of his earlier
expressive habits, —
that is, to keep to the accepted term,
could not escape the process of linguistic analogy. In the
 SEMANTIC CHANGE 237

actions of an individual there come into evidence unique

factors of causation which in a communal process are
subdued and compensated by the conflicting factors of
other individuals, —
where they are not, indeed, from the
start overcome by the unquestioning submission to commu-
nalpractice. Innumerable individual tendencies are suppress-
ed in the speech, and, to a lesser extent, in the religious,
artistic, cultural, political, economic, and other activities
of every person. When
an individual invents a word, these
factors come into action and, though their complexity
and singularity make the result harder to analyze than
the result of a communal process, — often, in fact, with
the data we have, impossible to analyze, — it is certain
that the laws which produce it are ultimately the same.
The most famous individual formation is the word gaSy
invented about 1600 by the Dutch chemist Van Helmont.
He believed that gas was a phenomenon related to the
idea which the Greeks expressed by cJiaos, in Dutch
which
receives nearly the same pronunciation as gas, and he
used also a term bias (a fairly regular derivative from
the Dutch verb hlazen 'to blow') for an aerial radiation
from the stars. 'Lewis Carroirs' famous poem of The
Jdbherwocky in Through the LooUng- Glass contains a num-
ber of individual creations, together with the author's
explanation. In most cases even words known to have
been invented by individuals are regular derivatives, e. g.
radium.
6. Semantic change. The third process of change in
language alters not the form of words, but only the se-
mantic content with which they are associated This pro-
cess is called semantic change. We have seen that, as no
two experiences are ever really identical, no word can
ever be used twice in exactly the same meaning. When
our attention analyzes a total experience into elements,
238 INTERNAL CHANGE IN LANGUAGE

we constantly assimilate these elements to earlier ex-

periences and express them by words used for these earlier
experiences. The due to a partial similarity
assimilation is

between the and the present one, a

earlier experiences
similarity inhering in some uniform component of both
experiences. This component common to all of the ex-
periences designated by the same word is dom-
called the
inant element (p. 57, f.). When we say John humped his
heady for instance, we assimilate the experience of John's
head to that of the other heads we have seen and known,
even though as to size, shape, color of hair, and many
other features, his head is by no means identical with
these other heads or with any one of them. In fact, as
to certain details, both objective and subjective, the present
experience of this very head is different from any earlier
experience of the same head. In short, the experience is

assimilated,by virtue of certain dominant features, to a

host of past experiences, and is designated by the same
word. If we take into consideration the points of individu-
ality of the present experience, it is clear that even here
the word has been used in a new meaning, that there
has been here a semantic change. From this to other
changes more striking to an observer after the fact there
is a gradual and by no means definable transition. The

speaker, for instance, who first spoke of a head of cab-

bage was no more conscious of having made an innovation
than the utterer of John humped his head. At the time
of this new utterance the dominant elements that brought
about the new use of the word, were only very general
ones of shape and size: the cabbage-head was to the speaker
as much a head, —it called up the word head as immedi-

ately, — as any human or animal head. By far the most

semantic changes, that is to Sriy, are unconscious shift-
ings of meaning directly resulting from the earlier uses
 SEMANTIC CHANGE 239

of the word and the dominant element with which it is

associated, and are made independently and simultane-

ously by numerous speakers. It is only the observer after-
wards looking back over history who sees that a change
has taken place.
Our example of a head of cabbapje illustrates a 'partial'
semantic change: one in which the older meanings have
so far remained in use side by side with the newer. As
long as this is the case and no intermediate meanings
have been lost, the normal speaker is not usually conscious
of any extension. Not only were the people who first spoke
of the head or the wing of an army, the leg of a table,
the foot of a mountain, the nose of a cliff, the heart of a
cabbage or of a country, of books lying, glasses standingj
and rivers running^ unconscious of making any innovation,
but we also are normally unconscious of any deviation
from what, upon deliberation, seems to us the more original
use of these words.
When we speak at different times of John's head, the
head of an army, and a head of cabbage, the different
momentary associations in which the word appears may,
however, involve a difference of dominant elements, even
if there was no such difference when these uses first arose.

In the first case, let us say, if we stood behind Johu, the

dominant element may have been the vision of a shock
of hair, in the second case of soldiers riding at the front,
and in the last case of a cluster of green cabbage-leaves.
The same mobility of dominant element appears, for ex-
ample, in the successive uses of law in law and medicine^
law and order, law and chance, law of gravity, and so on.
If it should occur, now, that some ot these uses should
lapse, — if, let us suppose, people stopped speaking of

a person's head and instead used some other word, such

as occixmt, and stopped speaking of the study oi' hue and
240 INTERNAL CHANGE IN LANGUAGE

said in this connection only jurisprudence ^ — then the

dominant element could be said to have definitely and
permanently changed: —
'total' semantic change.
This, in fact, is a frequent occurrence. Thus disease

(dis-ease) once meant discomfort of any kind; but, as the

uses of the word aside from that represented, say, by
health and disease, were forgotten, the dominant element
has taken permanently one of the forms among which it
wavered in the earlier use. If, as is probable, the word
ness for 'cliff' once meant 'nose', it is an example of the
same process: the lapse of certain uses has left the domi-
nant element changed. So meat once meant 'food' (cf. sweet-
meat), but the value of the word in such uses as meat and honey
meat and hide, where 'meat', as the edible part of something,
was contrasted with the inedible part, has alone survived, so
that the dominant element ('flesh') of these uses has become
fixed. To spill was once to 'destroy'; the uses other than
that in spill the milk and the like lapsing, the dominant
element changed to that of pouring out a liquid. The
adverb hardly Qiard-ly) once meant 'firmly, vigorously'
They hardly folloiued the enemy then meant that they follow-
ed close upon them. The other uses lapsed until only
this of 'closely' remained, and of this use the locutions
where 'closely' had the sense of 'just, barely' alone re-
mained to fix the dominant element of present speakers.
The history of nearly (near-ly) is similar. In all these
instances the change is at root a change in the habits of
association. In the case of the word meat, for example,
what has happened is really that English-speaking people
in such experiences as that of someone's having enough
to eat and to drink have ceased to associate the experience-
element in question with the sounds meat and associate
it instead with the word food] the sounds meat, on the

other hand, they have come to associate with the ex-

 SEMANTIC CHANGE 241

perience of edible flesh. In some cases formal causes

prompt these changes of association. Thus mash originally
meant 'to mix', —
of. sour mash, bran mash, but owing —
to the association of the phonetic element -ash with vio-
lent action (dash, crash, dash, etc.) and of the Avhole word
with smash (ci. plash: splash, lash: slash, etc.), the present
meaning of Ho crush' became associated with it.

In the examples so far given the change in habits of

association was due to the prevalence, for reasons mostly
too complex to allow of our tracing, of certain semantic
connections. A few examples can be cited where the prev-
alence was due to connection with actually uttered words
of a sentence rather than with semantically related words.
The French word for 'nothing' rien originally meant ^a
thing, something' (Latin rem); it came to be used, how-
ever, so prevalently in the connection ne . . . rien 'not ... a
thing', e. ne dit rien 'He says not a thing, He says
g. II
nothing' that it came to be associated with a dominant
element that included the negative idea, until now a
Frenchman answering a question 'What did he say?' can
answer i^/ew 'Nothing'. The same is true o^ jamais 'never':
formerly meant 'any more, ever' (Latin jam ma(Js)'^
it

the negative element has come to be associated with it

owing to prevalence of the usage in II ne lient jamais
*He does not ever come. He never conies'. The commonest
French negative is ne pas, originally 'not... a step'
(Latin passum), but today one can say, for instance Fas
moi 'Not r, {or pas has come to be associated with ne-
gation as its dominant element. Meanwhile the word ^rzs
in the older sense of 'step' is still in use, but the normal
speaker of French is not conscious of the historial connec-
tion between the two words. The same cannot be said
oi perSonne, used in both the negative sense of 'nobody'
(due, of course, to ne personne) and the original of
. . .

Bloomfield, Stady of Language 16

242 INTERNAL CHANGE IN LANGUAGE

'person*. The use of hut in such sentences as He had htd

one child is an example of the same process. But here
used to have the usual meaning of 'except' (as in All hut
Jolin ivent) He had not hut one child;
and the sentence was
only after the association of the negative element with
hut was the not^ now superfluous, omitted. This, indeed,
is the process by which cross-referring constructions be-
come simplified. In Latin amat meant 'He (she, it) loves',
but, owing to the necessary frequency of such sentences
as Pater amat '(The) father he-loves', the dominant ele-
ment associated with amat ceased to include an actor:
today in French one cannot say aime, but only il aime,
elle aime 'he (it) loves, she (it) loves'.

The adjective capital was used so frequently in the

connection capital city that it came to be associated, in
spite of the conflict of its other uses (cf. capital punish-
ment, a capital story), with the object-idea of 'chief city',
and is today so used. The same is true of general for
general offlcery of glass for glass tumhler, of lyric for lyric
poem J
and so on: the other uses have examples
in these
continued by the side of the new 'condensed' meaning.
The change in morphologic and syntactic value of the
words is apparent: adjectives become nouns. Similarly
our conjunction while used to be the accusative case of
the noun while, as still appears in the archaic the while, —
the accusative expressing extent of time. Today, owing
to the loss of such case-constructions, the connection is
lost, the value of the conjunction tvhile being totally dis-

tinct from that of the noun. In expressions such as He

ivas frightfully angry the value of frightfully may come
to be felt as merely intensifying, until people say I am
frightfidly glad; the German word for 'very' sehr is the
result of such a process, for it originally meant 'sorely,

painfully*.
 SEMANTIC CHANGE 243

Our prepositions are an instance of an entire word-

class that has received its present value from this process
of 'condensation'. In an earlier stage of the language the
relations of nouns were shown by their case-endings "only,
— not only relations of actor and object affected, but
also the local relations now expressed by prepositions
At thistime there were also a number of common ad-
verbs which came to be used regularly with nouns in
certain cases. This state of things was illustrated above
(p. 116, f) from the Ancient Greek by such a sentence as
Kephales dpo pharos lieleske. The first word is here the
genitive of the word for 'head', this case being used to
express separation: 'from the head'; dpo is an adverb,
meaning 'ofi'', 'away': 'He-drew (the) cloak off from-the-
head', 'He drew the cloak from the head'. The constant
occurrence of dpo with genitives of separation finally gave
rise to a semantic change by which the genitive forms
were no longer felt to express the idea of separation, but
to stand simply in government or congruence with the
adverb. So, in Old and Middle English, expressions like
He heom stod witS 'He them stood against, He withstood
them' show the transition: is the case of heom in itself
expressive of the relation, or is it merely the habitual
accompaniment (congruence, rection) of w'liS 'against'?
In the modern form He stood against them the case-form
of them is of course purely in syntactic government of
against and by itself expresses no local relation. Originally,
however, our prepositions were adverbs like the Greek
dpo in its early stage, and the form of the noun by itself
directly expressed the local relation.
There are a number of instances in which the change
of mental habit underlying a semantic change is a tan-
gible alteration in the external conditions of life. The
ancient Romans were originally an agricultural and cattle-
16*
244 INTERNAL CHANGE IN LANGUAGE

raising people; the standard and also, no doubt, tiie medium

of exchange for larger values was among them cattle.
Hence pecunia, the Latin word for 'money', and peculiumy
that for 'property', are nouns originally meaning 'that
pertaining to cattle'. Under the dominant element 'me-
dium of exchange', the former word was later applied to
the coined metal money used in historic times. The Eng-
lish word fee also once meant 'cattle' (cf. the German
VieK) and has passed through a change similar to that
of the Latin word. Fen originally meant 'feather', includ-
ing the quills used for writing; then, when the latter were
being superseded by steel points, the dominant element
of 'writing-implement' mediated the present use. German
Feder and French plume have still both meanings, 'feather'
and 'pen'. The same is true of such words as marsliall
which meant 'horse servant' (of the king), constable which
meant 'attendant of the stable', and similar words in many
languages: the conditions of court life changing, these
offices gradually lost their old significance and attained
to higher dignity, until the old dominant elements in the
words faded. The history of such words as arrive^ which
once meant 'come to shore', and eqiiipj which meant 'fit
with a ship', is probably due to similar changes in the
conditions of life. At bottom all this is no different from
the changes of meaning in such words as housCj street,
carriage, car, light, liai, coat, shoe, gun, and so on, which
every advance in civilization and shift in fashion brings
about.
Thus the history of words, etymology^ is interesting to
the student of civilization and culture. Often the only
trace of chancres in a nation's mode of life is in semantic
changes; for instance, no better testimony for the use of
cattle as a medium of exchange in ancient Rome and
ancient England exists than the 'etymology' of pecimia
 SEMANTIC CHANGE 245

and of fee. Often, on the other hand, the history of a

word can be understood only if one knows the cultural
or material history that underlies it. The German Wand
Vail' for a long time puzzled etymologists; its obvious
connection in the language seemed to be as a derivative
of the verb winden 'to wind', but this seemed semantically
improbable, — until it was pointed out that houses
with basket-work walls are still to be found in parts of
Europe.
The development of expressive material which keeps
pace with the general mental progress of a community, —
or rather, which forms an integral and vital part of this
progress, — is largely mediated by semantic change. The

processes most favorable to this growth are semantic

changes transferring a word into a new sphere of ex-
perience, especially, of course, from a sensational to an
intellectual value.
Leading up to this process are those instances in which
a word originally belonging to one sphere of sensafion
comes to be used for others also: this, of course, con-
word. The transi-
stitutes a subtilization of the value of the
tion is immediate: after speaking of a dear liquid one
feels no 'transference' at all in speaking of a clear sound
or tone. Thus we speak of sharp sounds, tastes, or smells,
using a word that referred originally only to a touch-
experience. We speak of warm and of cold colors; when
we use the word tone-color we let the value of colory origi-
nally visual, apply to auditory sensations, — with no
difficulty, for thedominant element of 'gradations of a
peculiar emotional value' ap])lies to one as well as the
other. The opposite extension is, of course, equally natural:
we speak of tones in a painting. In all these instances the
emotional value of the word becomes dominant, allowing
it to be applied to experiences of similar emotional, if
246 EXTERNAL CHANGE IN LANGUAGE

very different perceptual value. The power, as instruments

of thouglit, — in these instances of esthetic thought, —
of such words thereby greatly increased
is

These sensation-words themselves, for that matter, seem

universally to have developed, through a similar process,
out of stillmore concrete words. Modern examples that
point to such a conclusion are our color-words orange and
violet: the quality is expressed in terms of the object with
which associated. Thus hitter was once a derivative
it is

of and sour of a word meaning Ho scratch'; salty and

bite

salt are derivatives of the noun salt; stvcet, however, has

not been traced to any earlier meaning. The word tone
in Ancient Greek (from which language it has come to
us) was a derivative of a word meaning Ho stretch' and
meant originally the stretched string of a musical instru-
ment. To feel was originally a derivative of a word for
'palm of the hand'.
Purely subjective terms have often the same history
of transference on the basis of an emotional dominant
element. Thus we speak of a cold reception, hitter hatred,
hurning or glowing anger, and the like. Anxiety meant a
'narrow place', just as we speak of someone's being in a
pinch y having a close shave or a narrow escape , or being
in straits or straitened circumstances. Distress, similarly,
was originally 'destriction', i. e. 'constriction*.
Finally, our words denoting intellectual processes lead
back in their history, to physical actions. Thus to under-
stand meant to 'stand in the midst of (that is, to be in
a position to judge), to define meant to 'give bounds to',

to conceive meant to 'take up', just as we speak of catch-

ing on to a thing or grasping To refute once meant
it.

literally to knoch out an argument. To thinh seems to have

meant 'to handle'; feel, which we have seen meant to
'handle with the palm of the hand' and thence came to
 SEMANTIC CHANGE 247

be used of the sense of toucli, then of sensation generally,

now is used also of emotion, subjectively.
In all these instances the transition was an immediate
one, a psychic resultant: and at the time of its occurrence
really no change at all. Yet, once the new meaning was
current, it required no unusual analytic gift to realize
that a transference had taken place. ^) Individuals recog-
nize in the use of many words an original and a trans-
ferred meaning, and good speakers and poets have in all
times, now more, now less consciously, refreshed and in-
tensified these transferences, or imitated them. Thus poetic
metaphor is an outgrowth of the natural transferences
'of normal speech. It was a general transition, no doubt,
when people spoke of ruffled or of deep or of stormy feel-
ings; this general usage was revived and deepened when,
to quote a very well chosen example, Wordsworth wrote:

The gods approve

The depth and not the tumult of the soul.

The usual poetic metaphors, then, are individual creations

on the model of the regular linguistic transference. The
picturesque saying that language is a book of faded

metaphors' is exactly the reverse of reality, where poetry

is rather a blazoned book of language.

Individual semantic change appears also in a great many

namings. Most towns in the United States, for instance,
are named after English and other places: Boston^ Lynriy
Plymouth, New York, Cairo , Troy, or after people: St. Louis,
Bismarch. The magnet is named from a place in Asia
Minor, and copper after the island of Cyprus. The word
money meant originally the 'mint', which was named in

1) It does not concern hb here that such popular realization

always distorts the process, looking upon it as a deliberate in-
tellectual action, rather than %n unconscious development.
248 INTERNAL CHANGE IN LANGUAGE

Rome from the close-by temple of Juno Morieta, this last

being perhaps the family name of some old Roman clan
for whom the temple was named. So we speak of Ohm's
or Grinmi's law. Probably a great many other words are
the result of individual transference: electricityy for in-
stance, was derived from the Greek word for 'amber', a
substance used in producing it, by some individual. This
is true of most precise scientific terms. People's names

furnish words, aU probably by way of individual trans-

ference; thus, in English to hurle, to boycott: Burke was
a certain Edinburgh murderer, Boycott a hated Irish land-
lord who suffered this form of persecution. We speak,
similarly, of a Xanthippe, of a Quixotic action, of Homeric
laughter, of tantalizing a person (Tantalus),and so on.
It is not alwa\s easy to say, however, whether a given
transference was originally individual or communal. The
use of names of animals for people seems, for instance,
to be a general tendency, but the individual cases are so
characteristic tliat we must suspect individual mots to
have started such locutions as calling a man an ass^ a
hog, a fox, or an ox, a child a monlcy, a woman a mouse
or a goose or hen Similar are such expressions as wliite-
caps or, in French, moutons 'sheep' for the white crests
of waves. The Romance lauouao^es have number of clever
derivations of this kind that make the impression of in-
dividual creations: French (and English) caprice is derived
from a word for 'goat', se pavanner 'to puff oneself up,
to strut' from a word for 'peacock', the Spanish moscar-
don 'bore, insists it, bothersome person' from moscarda
'a gadfly' and calahazada 'a blow in the head' from cala-
haza 'a gourd, calabash'. Transferences embodied in meta-
phoric expressions extending over more than one word
are equally common. The origin of fighting windmills is
thus well known. Tempering the wind to the shorn lamh,
 SEMANTIC CHANGE 249

often attributed to Sterne, thougli it seems to have occurr-

ed before him, no doubt also an individual creation.

is

Such expressions as to be hand and (in) glove seem also

to be individual, for different languages differ in the met-
aphor employed; thus the German sa3's 'to be one heart
and soul', the French Hwo heads under one cap', the
Italian 'one soul in two bodies', kSpanish 'nail and flesh'.
We have seen how transferences that were never fully
conscious and have become almost entirely mechanized
may be revived by a poet who gives them a new turn.
When Shakspere speaks of talcing up arms against a sea
of troubles, the violence of the expression rouses our appre-
ciation of the more literal values of the words.
This more literal value of words, especially where the
transfer is an individual one, may be obscured or for-
gotten. The transferred word is thus left in an inexpli-
cable meaning and may come to be associated with some
other that is historically not connected at all. We shall
meet this process again as 'popular etymology'. Thus
Welsh rabbit a jocular individual creation, has failed to
J

meet understanding aud been assimilated by many speakers

to rare-bit. In German there are a number of unintelligible
proverbs due to popular etymologizing. Sein Schdfc/icn
ins Trockne bringen 'to bring one's little sheep into the
dry place' is used in the sense of our 'looking out for
A number 1*. It is due to the failure to understand the
dialectal form, in which the thing brought to the dry
place is SchcpJcen Similarly Maidafjcn feil-
'little ship'.

halten, used in the sense of 'to loaf around, to stand gap-

ing', means 'to have for sale Maidaffeii*-^ what
literally
these are no one knows: the word looks like a compound
of Maul 'mouth, maw, snout' and Affcn 'monkeys'. The
expression is an assimilated form of a dialectal Mul apen
halden 'to hold one's mouth open'.
250 INTERNAL CHANGE IN LANGUAGE

Where phonetic change causes an automatic sound-

variation, semantic change may destroy the automatism
by transferring one or both forms to different or limited
uses, — two words or forms instead
gi'^ing rise, thus, to
of one. The English spirant-voicing after unaccented vow-
els produced a by-form of the word off with voiced
spirant, namely of i^. 215), which was used in unaccented
position. The frequent use of the accented form as ad-
verb and of the unaccented as preposition brought about
an association in this sense. One came now to use off
even where the adverb was unstressed (He did not fall
off, he jumped off), and o/*even where the preposition was
stressed (of and for the people). The same development
may take place where the two forms are the result of
analogic change. In an older stage of German there was
a verb which today is in the infinitive gedeihen 'to thrive';
it had a participle which would be today gedungen 'thriven'.

A more regular analogic participle ^e(?/e^e>^, however usurp-

ed this use, and the old participle came to be used only
in the transferred sense of 'excellent, strong'. It has since
been lost in this meaning also. Today a still more regular
analogic formation gediehen serves as participle of gedeihen,
and gediegen has in its turn passed over into the meaning
'solid, excellent'. Further examples of such 'correlative'
semantic change, as it has been called, are the English
plurals clothes [klowz] with transferred meaning, and the
analogic cloths [klocJz] or [klo9s] in meaning directly corre-
sponding to the singular; or such formations as the ana-
logic unpractical 'not practical', used where the shifted
sense of the older negative form impractical is not intend-
ed; so also unmoral beside immoral.
Thus, as the cultural and intellectual life of a people
grows, new experiences, assimilated at first to the old,
are designated by the old words or analogic formations
ULTIMATE CONDITIONS OF CHANGE IN LANGUAGE 251

from them, until in time the word-stock that once ex-

pressed only the most concrete and simple experiences is
available for philosophic and scientific discourse of any
desired refinement.
G. The ultimate conditions of changein language.
Change language in thus due to the inevitable shifting
in
of the conditions under which speech is carried on. This
is most obvious in the case of semantic change: any new

experience is assimilated to the old and expressed by the

old word, which thus has changed its meaning, —
a change
which may become fixed by the lapse of the original use.
Similarly, analogic change alters the form of a word by
an assimilation to another word or set of words that is
semantically associated^ and the conditions underlying this
process, —
the weakening of the supplanted form and
the strengthening of the inducing elements, — are again
conditions which must constantly arise as the subjects of
thought and discourse, the beliefs about the interrelations
of phenomena, and the material interests of a people de-
velop. Phonetic change is also, no doubt, the result
of changes in the conditions of speech, even though
here the alteration in the conditions is not, as a rule, tra-
ceable.
In accordance with all this, we are frequently able to
recognize the outer conditions which bring about a seman-
tic change, —
a recognition which is one of the chief
aims of etymologic study, —
we are, further, able to see
in analogic change a constant adaptation of the speech-
habits to more and more harmonious relation with ex-
perience, and even in the little-understood processes of
sound-change the total result, at any rate, is a lessening
of the amount and complication of articulatory movement
that is connected with a given element of experience.
Two questions at once suggest themselves. To what extent
252 INTERNAL CHANGE IN LANGUAGE

does the history of languages, in accord with all this,

show advance in mentality? And: Is it possible to trace
in any language the mental characteristics of the nation
which speaks it?
As to the former question, there is no doubt that the
changes of language are not a chaos of haphazard and
conflicting alteration, but an evolutionary process. Just
as no individual, to the day of his death, ceases learning
to speak, so mankind is ever altering its speech to more
suitable forms. We have seen that phonetic change bears
the appearance of a constant adaptation and re-adaptation,
that analogic change tends to bring about the expression
of similar semantic content by uniform instead of varied
articulations, and that semantic change even more directly
leads to the apt expression of what is at any time the
content of thought. We must keep in mind, however,
that lanoruaoje is traditional and social: tendencies toward
alteration are constantly stifled by the conventional articu-
lations, forms, words, and constructions which the speaker
hears from others, —for the hearing of the conventional
forms reawakens the older impressions, as opposed to the
innovation. It is only when a tendency to alteration corres-
ponds to the mental predisposition of a large part of
the community that it gradually gains ground. Language
is thus not extensively subject to conscious change: its

development is by a gradual selective evolution, uncons-

ciously made by the speakers, who merely use those
forms of speech which present themselves most directly
for articulation. As the correspondence of a new form of
speech with the remaining speech-habits, and with the
relations of experience, must, if the new form is to find
acceptance, outweigh the mere habituation of an other-
wise less adapted older form, the change, when it does
take place, will usually be a step in advance. We may
ULTIMATE CONDITIONS OF CHANGE IN LANGUAGE 253

accordingly expect to see in every language a slow but

certain progress in adaptation to the forms of experience.
This is clearly apparent in the broader outlines of the
history of languages. The older stages show us compli-
cated formations, by dint of which a fairly complex ex-
perience without analysis or with but partial analysis
is

expressed in a single word. An action with its actor,

mode, manner, tense, circumstances, and objects affected
may, for instance, appear in a single word. Such words
as the already quoted Nahwatl ninalvalcwa 'I -meat -eat'
or oJiiketskofonke 'They-him-neck-cut (past)', ^They
i. e.

cut his throat', or the Tsimshian [aldaie:] Ho walk in

the dark', or, for that matter, the Latin amavisset *he
(she, it) would have loved', illustrate this. We may
consider such words as indicative of an older state of
language, in which the purely emotional responses of
pre linguistic times may be conceived as having barely
developed into a repetition of the same sound-sequences
under similar experiences. It is an advance when all but
one of the more material elements receive only anaphoric-
personal expression in the single word and are more ex-
plicitly mentioned by cross-reference, as in the Aztec
inJckwa in naltatl 'I -it -eat, the meat' or in the quoted
Latin word, e. g. Pater amavisset 'The- father, he-
would -have -loved'. It is a further step in advance
when the value of the cross-reference elements is lost by
semantic change ('condensation') and the words bear each
a separate meaning, though the formerly cross-referring
forms are still conventionally used together in congruence,
as when in English we use loves with an actor in the
third person singular, but with the other persons say
love. When at last these habits also have been removed,
we come to more unified words; thus we use a form like
may without regard to person and number of the actor.
254 INTERNAL CHANGE IN LANGUAGE

Even here the word expresses the tense as well as the

action itself: we should, perhaps, call it a step in advance,
if the tense also came by a separate word,
to be expressed
as in I am writing I was writing. Word-composition
^

may thus be an heirloom from the days when what we

now call a compound represented the regular type of word,
— that is, when words regularly contained two or more
material semantic units. It is conceivable, under that
condition, that some much-used member of compounds
with a very general sense, like that of *thing', could lose
its specific meaning, until compounds of which it formed

part represented but one material element, as opposed

to others that stiU represented two or more. Thus the
Nahwatl 'absolutive', as in nalcatl 'meat' or 'It is meat',
may originally have been a compound, the semantic fad-
ing of whose final member (~tl) first allowed people to
express the idea of 'meat' outside of compounds, such
as ninalcahwa 'I- meat-eat'. The same may be true of
the Primitive Indo-European nominative-suffix which
-5,

may have been originally a final member of compounds

and, by losing its material value, have become the means
of liberating an idea like ^ekwo-s 'horse' from exclusive
use in such compounds as ^ekwo-domo-s 'horse-tamer'.
However aU this may be, it is certain that we find in
all languages a constant diminution of the unanalyzed
content of single words, a lessening of cross-reference,
congruence, and government in favor of explicit discur-
sive expression, or, to look at the same thing from an-
other point of view, a growing constancy in the form of
words, as opposed to morphologic variation. In the older
stages material elements are viewed either in connected
groups (compound words) then only in some
or, if alone,
particular relation as to time, space, number, manner,
and the like, and as to each other (infiected words); it
ULTIMATE CONDITIONS OF CHANGE IN LANGUAGE 255

is but gradually that the speech of man attains a foller

analysis of experience, an analysis into simpler independ-
ently recurring elements.Thus the earliest scientifically
attainable stage of English shows us eight cases and
three numbers of nouns in three genders, most variously
and irregularly with adjectives agreeing in full
inflected,
congruence, a cross-referring verb containing mention of
the actor and inflecting by cumbrous, complicated, and
highly irregular prefixation, infixation, suffixation, and
sound-variation in three persons and three numbers in
two voices, all in a variety of modes and tenses, the
latter based principally on manner of action, secondarily
on relative time. If we contrast this with our modern
brief forms and comparatively regular inflection and our
simple sentence-structure, in which congruence and gov-
ernment play but a small part and cross-reference none,
the advance is unmistakable.
As all such development is gradual and unconscious,
we must not be surprised, on the other hand, when we
see, alongside the progressive simplification, an occasional
formation of the old kind arising, or a sound-change
complicating what was formerly simpler. English has
been rapidly losing derivational complexities, yet we find
old compounds denoting manner by a second member
Old English -lice 'in the manner of, becoming the reg-
ular means of deriving adverbs from adjectives, and this
second element becoming phonetically reduced to an
otherwise meaningless suffix -ly, as in quicfcli/y slowly,
sharply- The same thing has happened in the history of
the Romance languages, where the Latin mente 'with a
mind' came to be used as a suffix in the same sense,
e. g. French lent 'slow' lentenient 'slowly' (would be Lat-

in leiitd mente 'with a slow mind'). While in these in-

stances the process Liay still be looked upon as a liber-
256 INTERNAL CHANGE [N LANGUAGE

ation of the single word from the compound through

semantic fading of the other member, that is hardly pos-
sible when in the Romance langu;iges we see a new
from an older syntactic collocation:
tense-inflection arising
Late Latin amdre haheo 'I have to (am to, shall) love'
became the modern Romance future, the second word
being reduced to a suffix, e. g. French famerai ^I shall
love', — a counter-development which well illustrates
the complexity of linguistic progress. It is only by dint
of innumerable changes and readjustments and after the
most various tendencies have conflicted and come into
harmony, that simplification can occur, —
and it is, in
consequence, only by careful examination of the histor-
ical details that we can ever obtain a just idea of the
growth of language.
To what extent languages are adapted to the national
character of the speakers is a far more difficult question;
but the difficulty lies in the vagueness of the latter term.
At the present our knowledge the character of
state of
a nation is very much what our personal bias makes us
wish to think it. The most completely known of national
activities is language; it is very difficult, for instance, to
decide what of a nation's art or religion is truly com munal
and what individual in origin.
It is possible, where we know that a nation has chang-
ed its language, to trace characteristics of the earlier
language in the nation's peculiar use of the newer, and
these stable characteristics, as it were, of a nation's
speech, have the first right to be called national. Such
a characteristic are the [t, d] sounds of the languages of
India (p. 220). Another possible instance is the following.
Irish is like English a modern form of Primitive Indo-
European. Unlike this language and unlike English and
the other sister- languages (with an exception to be noted),
ULTIMATE CONDITIONS OF CHANGE IN LANGUAGE 257

ithas a few remarkable characteristics. One is the toler-

ance for phonetically divergent forms of the same word
without corresponding semantic variation beyond what
is implied in the mere existence of the forms (p. 102, f.),
— as in the example td ha 'There are cows' but a va
'his cows', and the like. In fact, there is a general lack
of stability of the word-unit. Another peculiarity of
Irish is the tendency to identify the emotionally dominant
idea of the sentence with the central element of the pred-
icate of an abstract subject and verb: It's his brother
he's cheating; the latter feature appears also in the Eng-
lish spoken by Irish people. It is possible that the Indo-
European speech of Ireland received these peculiarities
from an earlier language which it superseded.
At any rate, it so happens that the Latin which re-
placed in France a sister-language of Irish and developed
into Modern French, shows some of the same peculiari-
ties. It gives little phonetic recognition to the word-
boundary (p. 99, ff.), containing even such forms as dii 'of
the', au 'to the* (masculine) which, while plainly felt as
two words, de (a) and le (compare, for instance the fem-
inine de la, a la and the form used before vowels, de T,
a V)y are phonetically indivisible, [dy, o]. French is toler-
ant of double forms of words, used, as in Irish, not au-
tomatically and yet without genuine semantic differentia-
tion; such doublets, for instance, as [vu] and [vuz] 'you'
or [a] and [at J 'has' are distributed not entirely by the
occurrence of the .longer form before vowel ([vu hi] 'you
do' but [vuz ave] 'you have'; [el a] 'she has' but [at el]
'has she?'), for this form occurs only before words close-
ly connected in certain relations of meaning. The iden-
tification of emotionally dominant element by peculiar
syntactic position is also prevalent: Cest eux qui I'ont
fait 'It's tliey who have done it', Cest Id que je Vai via
Bloonif lel d , Stud/ of Language 17
258 INTERNAL CHANGE IN LANGUAGE

'It's there tliat I saw him', Cest moi quails ont hattu 'It's
me they beat'
Thus future researchin what may be called compara-
tive phonology, morphology, and syntax may reveal na-
tional linguistic habits to which any language a people
may come to speak is subjected. It will then remain
to compare and with such other characteris-
relate these
tics of the nation as ethnologic study shaU have ascer-
tained.
All this, then, brings us to the question of the relation
between language and race, to the question of what people
speak alike and what differently, and to the consideration
of the various changes in this distribution, —
in short,
to the external history of language.
 CHAPTER Vm.
EXTERNAL CHANGE OF LANGUAGES.
1. Language never uniform. We
have repeatedly
seen that language, far from being an object or an in-
dependent organism of some kind, is merely a set of
habits. Such similarity as there is between successive
utterances is due, therefore, entirely to the psychic assimi-
lative effects of earlier utterances upon later. The assimi-
lative predisposition is in every individual constantly
changing, for, if nothing then at least the utter-
else,
ance last spoken will alter the conditions of the next one.
We may say, then, that the language even of a single
individual is never exactly the same in any two utteran-
ces. What unity there is is due to the assimilative effect
of earlier upon later actions.
In this regard the effect of the speech one has heard
from others is the most important factor. In early child-
hood the individual's language is entirely in imitation
of it, and even later, when one's own habits are reliable,
one hears much more than one speaks. This, of course,
is the link between the speech of different individuals

which makes language a communal or social, not an in-

dividual phenomenon. Nevertheless, the predispositions
of any two individuals wiU never be identical. They will
differ more, as a rule, than the successive states of one
and the same individual because, in addition to constitu-
.^ - 17 •
260 EXTERNAL CHANGE OF LANGUAGES

tional differences, the past Ian gufige- experience of tlie

two speakers is always different. P^ven more truly than

the language of every utterance may be called unique,
it may be said that every speaker has his own peculiar
linguistic habits. These, in fact, are in everyday expe-
rience often noticed as idiosyncrasies of pronunciation,
construction, and vocabulary: 'The style is the man'.
In spite of these individual divergences, the circum-
stance that language our means of communication and,
is

as such, is learned both in the beginning and all through

life from our fellow-speakers, assures an extensive uni-
formity. The associative processes which produce an
utterance are the effect of other people's utterances which
we have heard from infancy to the present. Consequent-
ly a close-knit social group in which communication takes
place frequently between all members possesses a rela-
tively uniform set of speech-habits.
It would be very difficult, if not impossible, to find
such a community. Everywhere there are groups of in-
dividuals among whom there is more communication
than between members of the group and outsiders. One
need think only of the family, the neighborhood, the
trades and professions, the pleasures, games, vices, creeds,
parties, and the social, economic, and educational strata
The result of the more lively communication within such
groups is, of course, in every case, a relative uniformity
which is at the same time a divergence from the speech
of those outside the group. Even superficial observation
shows us family dialects, neighborhood phrases, trade
and professional vocabularies, jargons such as those of
the race-track or the base-baU field, speech of the slums,
of the middle class, and so on,
of the aristocrats, —
without end. Any one speaker's habits present a com-
bination of those different dialects which he has heard
 LANGUAGE NEVER UNIFORM 261

and spoken, a unique combination modified, further, by

individual factors.
The most important of these dialect -divisions have
always been the local. These are of various degrees.
Where there are several local groups communicating
freely with each other, each group will have its dialect,
but the differences between these dialects will not be
great enough to destroy mutual intelligibility. This condi-
tion is found, for example, in European countries, where
own dialect. When such
often every village speaks its
connected groups cover a very large area and communi-
cation between members of those at the extreme ends is

rare, these extreme dialects may be mutually unintelli-

gible, although, as each dialect of the whole group un-
derstands those near it, they are connected by an un-
broken chain of communication. Within what would
otherwise be such a group there may, however, be some
partial barrier, a political or tribal boundary, a river or
a range of hills, and the like, —
which lessens intercom-
munication of the two sides without preventing it. There
we may find a decided break in resemblance, even though
the dialects on the two sides are still intelligible to each
other. Finally, a barrier of the kinds described, or one
more impenetrable, may divide mutually unintelligible
languages.
Where a barrier of the last kind exists, reflection and,
especially, scientific research may some resem-
discover
blance between the languages. A Norwegian and an Eng-
lish sailor who learned each others' languafi^es would re-
alize that they presented, in spite of being mutually un-
intelligible, considerable similarity, as opposed, for in-
stance, to Greek or to Malay. Scientific study shows
English and Greek to possess great morphologic and
•yntactic similarity and an original, though phonetically
262 EXTERNAL CHANGE OF LANGUAGES

obscured resemblance of vocabulary, which mere reflec-

tion on the part of ordinary speakers of the languages
would not discover; between these languages and Malay
on the other hand, no such similarity has been discovered
by science. We speak, then, of delated' and 'unrelated'
languages, according to our lights.
2. Iiicroiise of uniformity. Wherever history shows
us anything of the past, we find barriers to intelligibility
decreasing. Our continent, north of Mexico, once harbored
a few million Indians speaking over a hundred, perhaps
several hundreds, of mutually unintelligible languages;
today this area contains, more than a hundred million in-
habitants, nearly all of whom speak English.
Such increase of uniformity occurs in various ways.
Conquest may, as in America, partly annihilate the con-
quered und partly assimilate them to the language of
the conquerors. Where the latter are less numerous the
assimilating process is commoner, but seems to fail, if
the vanquished are culturally superior. The Romans,
who conquered Italy, Iberia, Gaul, Dacia, and Greece,
were able to impose their language on the people of all
these countries except the culturally superior Greeks.
Differences of language may disappear, if one language
is politically or culturally supreme: this is often the last
phase of a preceding conquest, as in the gradual spread
of English in Wales and Ireland, or of Russian in Siberia.
Languages of large communicative value may spread as
second languages of speakers for commercial and similar
purposes: so English, French, Spanish, Hindustani, Malay,
and others are spoken more or less inexactly by large
numbers of people whose native language is less widely
known. In this way arise trade jargons, the various
forms of Lingua Franca, Pidgin English, 'Chinook' and
the like.
 DECREASE AND INCREASE OF UNIFORMITY 263

Decrease of uniformity does not offset the

3.
increase. This growing uniformity is in part and —
only in part —
offset by a constant process of differen-
tiation. A language spreading over a large area does
not remain uniform. The various barriers to communi-
cation result gradually in a differentiation at first into
dialects, then, often, into mutually unintelligible lan-
guages.
Thus, at the dawn of history we find Greece, many of
the surrounding islands, and a strip of the coast of Asia
Minor speaking numerous, main mutually intelli-
in the
gible dialects. The
and commercial supremacy
cultural
of Athens and the districts of Asia Minor resulted in the
spread of a uniform Greek speech, based chiefly on their
dialects and called the Koine ('common language'), over
all of this territory except, it seems, a small district
around Sparta. By the early centuries of our era this
language was uniformly spoken, but dialect-differentia-
tion soon set in, and by the nineteenth century the differ-
ent communicative conditions had resulted in a set of
dialects as unlike one another as were those of ancient
Greece. It now seems that the speech of Athens will
aorain become the common lano;uao;e of all Greece.
In Italy earliest history shows -us a welter of the most
various languages and dialects, many, so far as science
can wholly unrelated to others, some mutually un-
tell,

intelligible, though somewhat similar, and still others

existing in groups of mutually intelligible dialects. Through
military and political supremacy the Romans gradually
extended their language, Latin, over all of Italy; their
later conquests carried it over what is now Spain, Por-
tugal, France, Latin Switzerland, and Roumania. It may
be that the inhabitants of these countries who learned
Latin spoke it from ^.he first in a form so much assimi-
264 EXTERNAL CHANGE OF LANGUAGES

lated to their earlier speech-habits that, say Portugal and

Roiimania never could have understood each other. At
any rate, dialectal differentiation at once set in, and, as
the Roman power decreased, the link of communication
which had connected these dialects failed. Today we
have five or six mutually unintelligible languages, each
broken into a number of continuous dialects, such as
those of Italy or France. These dialects correspond, often,
to political, tribal, or geographic barriers: they are being
superseded, at present, by languages of important centers,
as, for instance, those of France by the speech of Paris,

and those of Spain by Castilian. Spanish and Portuguese,

in turn, have spread by conquest over the southern part
of the Americas, where they have superseded numerous
Indian languages. The unification has here vastly out-
weighed the differentiation, for the modern Romance
languages, divisible into five or six mutually unintelligible
groups, within each of which dialectal differentiation is
limited and is rapidly disappearing, represent the extinc-
tion of dozens, probably hundreds, of languages of Europe
and America. These old languages were spoken each by
a few thousand people, the Romance languages are spoken
by many millions.
At the time of our earliest records, from the seventh
what is now Holland,
to the tenth century of our era,
Germany, and part of England was a territory of some
dialectal differentiation; yet it appears that an English-
man could then understand a North German. The dia-
lectal break that there was between the English dialects
and those of the mainland was due, of course, to the
emigration of the English tribes in the fifth century.
This differentiation went on until English and the con-
tinental speech became mutually unintelligible. At the
same time the differentiation within each of these groups
 INFERENCES FROM HISTORIC CONDITIONS 265

oecame so great that the speakers of the Eng-

different
lish dialects could often not understand each other, and
the same was true of the dialects of the continent. To-
day all of these dialects are, however, disappearing before
the spread of three favored forms of speech: Standard
English, Dutch, and German. While the speakers of the
old dialects are not succeeding in speaking these stand-
ardized dialects without some assimilation to the forms
of their local speech, —
Standard German as spoken in
Bavaria differing much from that, say, of Mecklenburg,
Standard English of Scotland from that of Kent, yet —
they can understand one another's forms within each
group. Meanwhile English, for instance, has superseded
most of the Celtic speech of England, German much of
the Baltic and Slavic of what is now Germany.
must We
not forget, also, the increase in population: the England
of King Alfred's time had perhaps two million inhab-
itants, only part of whom spoke English. Meanwhile
Standard English has spread to Ireland, North America,
and Australia, and has become the uniform speech of
many millions. The differentiation which there is in
Standard English will probably never rise to the point
of unintelligibility, for printing, rapid travel, and com-
mercial intercourse are constituting communicative bonds
more close, probably, than those which two-hundred
years ago existed between the north and south of the
little island of Britain.
4. Inferences' from liistoric conditions. These histor-
ic instances allow of certain general conclusions. Where
we find an area in which a number of mutually intelli-

gible dialects are spoken, we infer that these are the re-
sult of differentiation of an older uniform speech. We do
not hesitate to suppose, for instance, that the ancient
Greek dialects were differentiated from a uniform pre-
26Q EXTERNAL CHANGE OF LANGUAGES

historic speech^ which we call Primitive Greek', just —

as the modern Greek dialects are divergent local forms
of the Koine. Similarly, we conclude that the earliest
and the continental dialects of
historic forms of English
Holland and Germany were differentiated, during a pe-
riod that we call 'pre-Eiiglish', 'pre-Frisian', ^pre-Saxon',
'pre-Franconian', 'pre -Bavarian', etc.,from a uniform
prehistoric dialect, which we call Trimitive West Ger-
manic'.
Related languages we accordingly look upon as results
of differentiation. we see Portuguese, Spanish,
Just as
French, Italian, and Roumanian diverging from Latin,
we conclude, in every case where languages are, beyond
the possibility of mere coincidence, alike, that their dif-

ference is due to gradual differentiation from a uniform

speech. Among the Indians of North America we find,
for instance, related languages spoken over three consid-
erable pieces of territory. A large part of the northwestern
interior from the Pacific coast to east of the Rocky
Mountains; a few small bands in British Columbia and
Washington and a strip of villages four-hundred miles
long in Oregon and California; a large area of Arizona,
New Mexico, western Texas, and Mexico; these three —
districts, each embrace a number of mutually unintelli-
gible languages, which, however, all present features of
similarity that lead us to call them related (the ^Athapas-
can' family of languages) and to suppose that they are
all divergent forms of a prehistoric uniform language
('Primitive Athapascan').
We have, of course, no right to suppose, in such cases,
that the same number of people spoke the 'primitive'
uniform language, or that it was spoken over the same
area, as its later forms. It is possible, as mentioned, that
the people of Portugal and those of Rou mania could
 INFERENCES FROM HISTORIC CONDITIONS 267

never have understood eacli other; it is certain tliat the

people of these countries who learned Latin from the
Romans never spoke it correctly enough to reach uni-
formity over the entire district. We can scarcely imagine
a prehistoric Athapascan state of uniform language all

over the West of our country: probably a comparatively

large tribe broke up into parts v^hich separated and then,
after communication had ceased, became differentiated in
speech, grew, linguistically assimilated other Indians, and
again split into independent speech-communities
It is very important, when we make these deductions,
thus to keep in mind the exact meaning of our results.
When we say that the West Germanic languages and
dialects, — English, Frisian, Dutch, and German, — are
differentiated forms of a uniform prehistoric language,
which we call Primitive West Germanic, we have no
right to assume anything about the exact manner in
which the differentiation took place. For instance. Prim-
itive West Germanic may have become differentiated
by certain barriers in its territory, —
by a religious con-
federation of certain of the clans, let us say, to which
the other clans did not belong. A later splitting of the
West Germanic group may not have coincided with this
earlier division. Thus, before the emigration of the Eng-
lish there was a dialectal differentiation: some of the
dialects changed an older [a] to [s], saying, for instance,
dcet 'that' instead of Oat. The English who emigrated
were part of those who had made this change; the Frisi-
ans, who had also made it, remained behind. There follow-
ed, of course, the great divergence of English from
Frisian and all the other continental dialects, due to the
overseas separation. Thus, in spite of the difference to-
day between English on the one hand and the continental
dialects on the other, we know that there was a time
268 EXTERNAL CHAKGE OF LANGUAGE

when English and Frisian belonged together, as opposed

to all the others, — that the divergence of English from
the mainland dialects was not the first differentiation to
break the Primitive West Germanic unity. Plad history
not in this case favored us, we might be led to the wrong
assumption that the first differentiation was the separa-
tion of English.
Primitive West Germanic, so far as its forms can be
determined, and also the various historic West Germanic
dialects, all show a decided resemblance to the languaores
of Iceland, Norway, Sweden, and Denmark. These lan-
guages, with their dialects, furnish another instance of
differentiation from an earlier language. At present we
find the dialect-division separating Iceland sharply from
the rest. The remaining dialects are differentiated by
lines running chiefly east and west, so that a dialect-belt
will run, for instance, across a stretch of Norway and
Sweden, regardless of present political boundaries. Had
we no older records, we should, to be sure, deduce 3
Primitive Scandinavian or Primitive North Germanic
parent-language, but the surmises which we might make
on the basis of the modern dialects would be wrong.
For our oldest records show us Norwegian and Icelandic
almost alike: the divergence of the latter did not progress
very far nntil some centuries after the settlement of Ice-
land by Norwegians a thousand years ago. Opposed to
the almost uniform Icelandic-Norwegian or West-Scan-
dinavian of the medieval records, we Swedish and
find
Danish closely alike: East- Scandinavian. This older
division has, then, been superseded by developments in
an entirely different direction. Thus, while it is safe to
set up a uniform 'primitive' language, the process of
differentiation itself may be obscured by repeated changes
in various directions, as when more modern changes have
 INFERENCES FROM HISTORIC CONDITIONS 269

crossed and in part obliterated the old division between

West-Scandinavian and East-Scandinavian in favor of a
new north- and-soatb division on the continent, opposed
to a divergent Icelandic.
Of Primitive North Germanic (Primitive Norse, Primi-
tive Scandinavian) a small amount is historically preserved
in some of the runic inscriptions. Primitive West Ger-
manic and Primitive North Germanic both closely re-
semble the language of a fourth -century Gothic Bible-
translation used by the Goths in Italy. From this three-
fold relationship we deduce an older uniform language,
Primitive Germanic, from which, in a period called pre-
West-Germanic, pre-North- Germanic, and pre-East-Ger-
manic ('pre-Gothic'), the three languages became differ-
entiated. Here we must guard against the mistake into
which we might in the other cases have fallen, had we
lacked, — as here we do, — historic records. It is pos-
sible, for instance, that the threefold division which we
know was preceded by an entirely different dialect-cleav-
age in Primitive Germanic. This would mean that such
a division as Primitive West Germanic contains some dia-
dating from the Primitive Germanic time,
lectal differences
and was therefore never wholly uniform after the original
cleavage of Primitive Germanic. In so far as we insist
that English, Dutch, Frisian, and German go back to an
absolutely uniform older speech, that speech would then
be Primitive Germanic; in so far as we considered only
those features which today appear and ignored a possible
but unauthenticated older dialect-cleavage, it would be
Primitive West-Germanic.
The Germanic languages more distantly, though un-
mistakably resemble a number of languages of Europe
and Asia. This rosomblance is increased when we com-
pare not the historic forms, but the various 'primitive'
270 EXTERNAL CHANGE OF LANGUAGES

languages, such as Primitive Germanic and Primitive

Greek, which we deduce from closely related groups.
Thus the closely interrelated Baltic languages (Lithuanian,
Lettish, and the now extinct Prussian) point to a Primi-
tive Baltic, which, with Primitive Slavic, appearing histor-
ically differentiated in the modern Slavic languages
(Russian, Polish, Bohemian or CecL, Servian, Bulgarian,
etc.), points to a Primitive Balto-Slavic. The languages
of Persia and Iran generally we derive from a Primitive
Iranian, those of India that here come into consideration
from a Primitive Indie; Primitive Iranian and Primitive
Indie resemble each other so closely as to point un-
mistakably to Primitive Indo-Iranian ('Primitive Aryan')
from which both are descended. Similarly we deduce a
Primitive Armenian, Primitive Albanese, Primitive Italic
(from Latin, Oscan, and Umbrian), and Primitive Celtic.
All these 'primitive' languages, including Primitive Greek
and Primitive Germanic, show so much similarity that
we conclude that they are differentiated forms of a Prim-
itive Indo-European, an ancient uniform language.
We must, however, again keep in mind all the limita-
tions that require observation, if our conclusion is to
have scientific value. In the first place, our conclusion
does not justify us in supposing that the same number
of people or the same districts that now speak the various
Indo-European languages spoke Primitive Indo-European.
We have, for instance, seen English spread from a mil-
lion or less speakers to many millions, some of whom
gave up another language for English, and, geographic-
ally, we have seen it spread from a part of England
over almost all of the British Isles, most of North
America, and Australia, not to speak of smaller colonies
of English-speaking people all over the world. In so far
as we insist upon Primitive Indo-European being a uni-
 INFERENCES FROM HISTORIC CONDITIONS 271

form state of speech, we must, in fact, assume that it

was spoken by a homogeneous and therefore limited
community, —
a community of not more than a few
thousand speakers.
Furthermore, we have no right to assume that Primi-
tive Indo-European was carried bodily, as it were, into
all the countries where Indo-European languages now

exist English, not Primitive Indo-European was carried

to America. The branches of the Primitive Indo-European
parent community surely altered their speech while mi-
grating to those countries upon which they were to im-
pose it. The people who had to learn
in these countries
the language of the dominant Indo-European- speaking
immigrants surely spoke the new language in some ap-
proximation to their own, —
of one such change at least
we have good evidence (p. 220). These speakers of an
implanted Indo European may then have been instrumen-
tal in the further spread of the language. The de-
duction of a Primitive Indo-European speech does not,
therefore, make probable any such improbabilities as that
there was a time, say, when a man from the north of
Europe could bave understood a Greek or a Hindu.
Again, it is too common changes of
in history to see
language or culture in a people, to allow of our assuming
that the present speakers of Indo-European languages
are all descended, physically, from speakers of Primitive
Indo-European. To all questions in this direction it can
only be answered that anthropologists and ethnologists
have found that language, culture, and physical descent
are not coordinate in history. It even bids fair to ap-
pear that physical descent and physical characteristics
are not coordinate. In other words, while we have knowl-
edge of Indo-European languages, of a Primitive Indo-
European language, and, to lome extent, of the linguistic
272 EXTERNAL CHANGE OF LANGUAGES

history which produced the former out of the latter, we

know nothing about the people who spoke Primitive
Indo-European, —
nothing about their habitat, appea-
rance, descent, or descendants, and of their culture only
so much as is involved in our knowledge of their speech.
As to the process of cleavage of Primitive Indo-Europe-
an also we must draw no hasty conclusions. There are
certain phenomena in which the historic Vestern' lan-
guages, namely, the Greek, Italic, Celtic, and Germanic,
are apparently opposed to the 'eastern', Balto-Slavic,
Indo-Iranian, Armenian, and Albanese. We find certain
velar sounds in the former corresponding to sibilants in
the latter, — in Primitive Indo-European they were prob-
ably palatals; — thus the word for 'hundred' is in An-
cient Greek he-lcaton, in Latin centum, Old Irish cet, Gothic
hund (English hund-red\ for the h- see Grimm's law,
p. 208), but in Lithuanian szimtas {sz is [/]), Avestan
(an old Iranian language) satdm^ Sanskrit satdm. It was
supposed, accordingly, that this divergence represented
the oldest dialectal cleavage of Primitive Indo-European
into an eastern and a western dialect; the eastern lan-
guages were called the 's- or 'sat9m' group, the western
the 7c-' or ^centum' group. More careful observation,
however, makes it probable that the cleavage into 'centum'
and 'satdm' languages was not a dialectal cleavage of
Primitive Indo-European, but that the languages have
separately arrived at the historic forms. Aside from the
peculiar position of Albanese between the Greek and
Italic 'centum' languages, there have recently been dis-
covered in Central Asia (East Turkestan) manuscripts in
an Indo-European language (to which has been given the
name Tocharic) which has velar, not sibilant sounds in
the corresponding words, (e. g. leant .'hundred'). Lithu-
anian, further, contains a number of words with velars
 THE PROCESS OF DIFFERENTIATION 273

instead of the sibilants which such words as that for

'hundred' and the close resemblance of Baltic to Slavic
would lead us to expect. Finally, investigation has shown
that Indie, in spite of its close resemblance to Iranian,
never had sibilants in most of the words in question.
Thus it appears that the line of cleavage between velar

development and sibilant development of the Primitive

Indo-European palatals does not coincide with the other
lines of dialectal differentiation, and would perhaps stiU
less do so, had we records of intermediate dialects that
have been lost.
5. The process of diiferentiation. The uniformity
of linguistic habit in a community is maintained by the
common expressive predisposition of the speakers, due
to their having heard since infancy approximately the
same set of words, forms, and constructions. In so far as
this predisposition, owing to the necessarily divergent
experience of individuals, varies from speaker to speaker,
we find individual peculiarities of speech; in so far as it

varies for families, social strata, occupations, and the like,

we find the stratification oflanguage mentioned at the
beginning of this chapter. The concurrence of the mem-
bers of a community is known as usage. Usage, we know,
is constantly changing: sound-changes, analogic changes

and semantic development never cease; and the changes

of usage are never the same in any two separated communi-
ties. The differentiation of a uniform speech into dialects

and into separate languages takes place wherever there

is any interruption, absolute or relative, of communication.

Where geographic or social barriers have lessened commu-

nication we find the usage of the separated communities
diverging more and more, until at first well-defined dia-
lects and then mutually unintelligible languages are found
to exist. This process of divergence is outweighed, as we
Bloomfltld, Stady of Language 18
274 EXTERNAL CHANGE OF LANGUAGE

have seen, by the constant replacement of uniformity due

to warlike, economic, or cultural domination of single
communities.
6. Deduction of internal history from related forms.
The case is frequent that we find historically a set of re-
lated dialects or languages but lack records of the uni-
form speech from which we must suppose that they have
become difi'erentiated. If we had no records, for instance,
of Latin, we should speak of it as 'Primitive Romance',
that is, as the uniform parent-speech of the Romance lan-
guages. The divergences of usage which differentiate these
languages would have to be reconciled in this 'primitive'
language. If we found, for instance, the word for 'father'
in French pere [ps:r], in Spanish and in Italian padre, we
might be doubtful as to what was the form in the common
parent-speech, from which these forms by divergent phonet-
ic and analogic changes had become differentiated. We
might perhaps set up a Trimitiv^e Romance' ^padre or
^pedre. The Latin forms, accusative patre(m) and ablative
patre show us that the ^ became d and was in French
finally dropped, and that this language also changed the
old a to e; furthermore, we find similar developments in
many parallel forms.
Where the older uniform language is not accessible,
our reconstructions are, correspondingly, most uncertain.
Nevertheless, they have a great value as formulae. The
word 'father' is in Old English feeder (for the d see p. 59, f.);
in the oldest Frisian (eleventh century) we find feder;
the north German show the oldest form (ninth
dialects
century) fader; the south German (ninth century) fafer.
As the common prehistoric form from which these were
differentiated we set up a 'Primitive West Germanic' '-^fader,

supposing the English and Frisian to have changed a to

e, and the South German d to t The starred form thus
 DEDUCTION FROM RELATED FORMS 275

set up is a formula in the sense that other words also

show the same correspondences of sound. It means, there-
fore, that the word 'father' in the West Germanic speech-
group is composed of the sounds indicated, to wit: (1) that
which everywhere appears as /", (2) that which appears
in English and Frisian as e and in the other dialects as
a, — symbolized by a in our formula, (3) that which
appears as d in all but the south German dialects, where
it is tj —symbolized by d^ (4) that which appears every-
where as an unaccented e, (5) that which appears every-
where in our group as r. Thus, to illustrate sound (3), we
set up a Primitive West Germanic ^daudo for the Old
English dead (modern dead), the Old Frisian dad, the Old
Low (i. e. North) German dod, and the Old High (i. e.
South) German tot. Here the symbol d recurs; the au is
a similar token for Old English ea. Old Frisian a, north
German o, and south German o before dentals or h (other-
wise south German aii)\ the final -o is due to considera-
tions which we may here overlook. Our Primitive West
Germanic forms, then, are mere formulae until they find
some kind of corroboration. If they find this, it will appear
that in the Primitive West Germanic speech-community
d was spoken and that in the pre-South German develop-
ment there was a change from this d to t. Or, if further
facts were to appear showing that our Primitive West
Germanic form was, in absolute phonetic value, wrong,
then a Primitive West Germanic ty changing in all the
dialects but South" German to d, would be indicated.
Tbe existence of North Germanic (Scandianavian) and
East Germanic (Gothicj forms gives us the i)0ssibility of
testing our West Germanic results. The Old Icehmdic and,
it is supposed, also the Primitive North Germanic form

of our word 'father' is fader the Gothic fadar (with the

^

d pronounced f), as certain internal conditions in Gothic

18*
276 EXTERNAL CHANGE OF LANGUAGES

conclusively show). These forms indicate that our Primi-

tive West Germanic ^fader was in all probability (not
with absolute certainty!) correct as to the a and the d:
the latter because the voiced d is nearer to the b of the
other languages than the alternative of t The comparison
of the three Germanic branches symbolized in the for-
is

mula of a Primitive Germanic ^fdcJer, —

with accent on
the first syllable, as in all the historic dialects. This form,
if literally change of d to d in pre-
correct, indicates a
West Germanic and a change of 6 in the unaccented syllable
(as other words show, only before r) to a in pre-East
Germanic. If wrong, our formula would still express the
general correspondence of d to b, of -er to -ar in these
languages.
Our Primitive Germanic form is again tested by the
Primitive Indo-European correspondences. We find the
word be in Sanskrit p^Ya (accusative ^^7dram),
'father' to
in Avestan pita (accusative pitardm)^ in Ancient Greek
pateTy in Latin pater, in Old Irish athair, and in Armenian
hair.
The correspondence of initial p of other languages in
this and other words to Germanic f makes it extremely
probable (but not certain!) that p was the older sound,
changed in the pre- Germanic development to f; for a
change of an older/* to ^ independently in Sanskrit, Greek,
Latin, and the other languages (which in other than ini-
tial position also in part show p) would be a very im-

probable coincidence; —
as would also the origin of all
these sounds from some sound not represented in any of
the historic languages, an m. Nevertheless, should
e. g.

we in spite of this be wrong, the p in our Primitive Indo-

European formula would still be a convenient symbol for
the general correspondence of Germanic Sanskrit, Greek,
/*,

and Latin p (Irish initial lost, Armenian initial h). This

 DEDUCTION FROM RELATED FORMS 277

correspondence reappears in other words, such as that

for ^cattle', where we set up, —
on the basis of Gothic
faihu (pronounced feJni), Old Icelandic fCy Old English
feo (modern fee; for the change of meaning see p. 244),
Old Low German and Old High German fihUy
fehu, a —
Primitive Germanic ^feJiu, which stands beside Latin ^ccw,
Samskrit pasu (with sibilant, cf. p. 272), Lithuanian peJcus
(with velar stop, cf. p. 272, f.). Hence, be it with literal value,
or, what is less probable but also possible, with only
symbolic value, we set up the first sound of our Primitive
Indo-European word as p-.
To return to the word 'father', the a of the different
languages, as opposed to the i of Sanskrit and Avestan
would appear as the more probable earlier form. We find,
however, that in other cases an a of the other languages
is found also in Sanskrit and Avestan, as in Old Icelandic

alia 'to ride, drive', Primitive Germanic ^aJcebi 'he drives',

Old Irish (ad-)aigy Latin Ancient Greek dgeij Arme-
agit,
nian atsem 'I lead, bring', corresponding to Sanskrit djati
'he leads', Avestan ozaiti. Consequently we suppose that
Primitive Indo-European had two vowels, represented in
most languages, owing to sound-change, by a, but distinct
inIndo Iranian as a and i. This supposition is by no
means certain and has been disputed; it receives corrobora-
tion, however, from certain conditions of vowel-variation
in the different languages. The Primitive Indo-European
vowel which preceded the i of Indo-Iranian and the a
of the other languages we represent by the symbol 9.
The third sound of our word has caused much trouble.
After Grimm's law (see p. 208), by which the old p- of
our word, for instance, became Germanic /*-, had been
established, it was expoctcd that the t of the other lan-
guages should corres])ond to a Gcjiiiiauic 7^^), not ad, —
1) ]) is the CJerinafiic hi;^!! for [0j.

h
278 EXTERNAL CHANGE OF LANGUAGES

as, for example, in the word 'brother': Sanskrit hliratd

(accusative hlirataram), Avestan hrata (accusative hratardm),
Ancient Greek phrater ('fraternity brother'), Latin frater
Old Irish hrathir, Old Bulgarian hratuj hratruj Lithuanian
hroterelis (with diminutive suffix), and Primitive Germanic
*hrdj,er (seen in Gothic hrodar, Old Icelandic brotJer, Old
English hrojor, Old Frisian hrdther, Old Low German brother,
Old High German hruoder). This difficulty was at last
solved by Verner (p. 216): after the Primitive Indo-Euro-
pean unvoiced stops (e. g. t) had in pre-Germanic become
spirants (p) these spirants became voiced, if they followed
an unaccented vowel (as in the Primitive Indo-European
word for 'father', where the Sanskrit and Greek accent
shows the second syllable to have been stressed). It was
not till after this spirant- voicing that the accent in pre-
Germanic was thrown universally on the first syllable;
whence the Primitive Germanic ^fdder.
The next sound again causes difficulty. Sanskrit and
Avestan show an a, the other languages an e, and for a
long time it was believed that the former was the Primi-
tive Indo European sound. It was discovered, however,
that the Indo-Iranian lanf^uajxes also once had an e. This
appears in the fact that velar sounds are palatalized (p.214)
in these languages before those a's to which e corresponds
in the other languages. For instance, the Primitive Indo-
European enclitic word for 'and' *q^e, appearing in Latin
as quBj inAncient Greek as te (from Primitive Greek ^qiie),
in Gothic as -h (from Primitive Germanic *hwe), is in
Sanskrit ca and in Avestan ca. Thus the e of the European
languages is in such cases assured as the more original,
Primitive Indo-European sound, which in pre-Indo-Iranian
first palatalized a preceding velar and then changed to a,
coinciding there with the Primitive Indo-European a.

This probability of a Primitive Indo-European e corrob-

 DEDUCTION FROM RELATED FORMS 279

orates our supposition that the vowel in Primitive Ger-

manic was e, not the Gothic a. The long quantity of the
vowel in Sanskrit and Greek in our word also appears
original from a number of comparisons.
As to the final r, various considerations have led to
the conclusion that in Primitive Indo-European there was
an automatic sound-variation by which the r was kept
before certain following sounds, especially vowels, and lost
before others. In the different languages one or the other
of the resulting forms was analogically generalized.
Hence we get, all in all, the formula of a Primitive
Indo-European ^pdte or *pdter. In part the absolute phonet-
ic value of this formula may be doubtful, but it serves
none the less well as a brief symbol for the various corre-
spondences between the Indo-European languages: corre-
spondences which could not be otherwise succinctly ex-
pressed. The correspondences so symbolized, moreover,
aid us in shaping our Primitive Germanic formula: thus
they assure us of an e rather than an a in the second
syllable of this word. In the word for ^dead' the Gemanic
forms alone would lead us to set up a formula of one
syllable; it is the Indo-European relationship of the word
which shows us that in Primitive Germanic and in Primi-
tive West Germanic it must have had two syllables (cf.
the formula above, p. 275). Beginning with the Primitive
Indo-European formula, then, with its rather relative value,
the history of the word 'father' can be traced, with more
and more certainty as we go on, to the present time; and
the same is true of all the lexical, phonetic, morphologic,
and syntactic features of the language.
The method of thus tracing the history of languages
wherever a number of related languages are given, is known
as the comparative method. The vista which it opens to
ns for English presents the development from a primitive
280 EXTERNAL CHANGE OF LANGUAGES

to the modern by any criteria we have of such

state. For,
things, 'Primitive' Indo-European was really a language
of decidedly primitive aspect. It had three genders, three
numbers, and eight cases of nouns, adjectives, and pro-
nouns, and several voices, modes, manners, and tenses of
verbs, each in three persons and three numbers, according
to the person and number of the actor. AU these were
inflected with great complication and irregularity, by means
of suffixes and intricate sound-variation, especially of vow-
els, together with some infixation and prefixation, in-
cluding highly irregular reduplicated forms. The deriva-
tion, also, was complex, different suffixes, as also in in-
flection, demanding shifts of accent and sound-variation
in the kernel of the word. Composition was frequent and
was accompanied by changes of form in the members of
the compound as opposed to their independent form as
simple words. The syntax identified, as today, actor and
subject, but in the predication of a quality the abstract
verb could be omitted, as in Latin (p. 111). As the verb
included pronominal mention of the actor, cross-reference
as well asgovernment related actor to action. The cases
of nouns were used in government, — that is, the differ-
ent case-forms expressed relations in which the noun stood
to the verb or to other nouns. The adjective varied in
congruence with the noun which it modified. The develop-
ment from that time to this can be traced with increasing
certainty as one approaches the testimony of narrower
and narrower ranges of comparison.
7, Interaction of dialects .and languages. The re-
sults of the comparative method do not extend to a set
of phenomena which, accordingly, must be set aside wher-
ever this method is used. As the comparative method is
based upon the unity of sound-change with regard to
different words within any dialect, it cannot be applied
 INTERACTION OF DIALECTS AND LANGUAGES 281

to words and fonns which have come into one dialect

from another dialect. Such words and forms, however,
are very common. Wherever there is communication be-
tween different speech- groups, one or both come to use
words heard in the language of the other, — words, usu-
ally, which have been associated with the appearance of
some hitherto unknown object or idea introduced by the
foreign people: as when we speak of chiffons and ruches
by their French names, or of Sprachgefiihl, Ablaut, Um-
laut, Zeitgeist, Wanderlust, Pretzels by their German. If
the two speech-groups that are in contact are mutually
intelligible dialects, the borrowing may be quite general;
if they are unintelligible, the mediators are those who,

more or less perfectly, have learned the foreign speech.

The disturbance of usual phonetic conditions appears
in such a word as street. High German Strafie, which,
normally, would point to a Primitive West Germanic
^str'oetu. This should then correspond, by Grimm's law,

to a Latin word with d for the second t (the first, stand-

ing after spirant, is unaffected); but the Latin word we
find is strata (via) 'paved road'. The explanation is, of
course, that the West Germanic people received the word
and the knowledge of paved roads from the Romans, —
at a time, needless to say, long after the sound-change
known as Grimm's law had ceased to act.
Language-mixture, where the historical conditions are
known, often determines the absolute date of a change,
which the comparative method alone can, naturally, never
fix. Thus the German Strafie shows that the change of

postvocalic Mo a sibilant in High German (cf p. 208)

occurred after the Romans had made their appearance in
Germanic territory.
As loan-words are usually of cultural significance, the
study of etymology (p. 244) receives from them an added
282 EXTERNAL CHANGE OF LANGQAGE

interest. The well-known contrast between the Old Eng-

lish ('Anglo-Saxon') stock and the French-Latin borrow-
ings of English need hardly be mentioned; though many
of the French words are today as common as those of
older currency in the language, e. g. teefj change, place,

adoption by the English can always be

chair, table, their
reduced to a cultural cause. Similarly, if to a smaller ex-
tent, we have words from every nation with which speakers
of English have come into contact; from American Indian
languages, for instance, the vegetable squash, succotash,
tobacco, not to speak of totem, papoose, squaw, wampum,
wigwam, tomahawlc, pow-wow, which are still felt as foreign.
Many other loan-words have come to us through a series
of languages, as banana, hammock (originally from Carib-
bean languages, through Spanish or Portuguese), candy,
sugar, pepper, ginger, cinnamon (originally from oriental
languages, whence they came through Arabic, Hebrew,
Greek, Latin, etc.).
Culturally significant words are not only thus bodily
taken over, but are often imitated. Thus the Latin word
conscientia 'conscience', a compound of con- 'with' and
scientia 'knowledge', was imitated Germanic lan-
in the
guages; thus German says Ge-wissen, Swedish sam-vete,
Danish and Norwegian sam-viitig-hed ; English has direct-
ly taken the Latin word in French form. The same is
true of, such Latin words as con-cipere 'to conceive', from
capere 'to grasp', German be-greifen\ Latin ob-jcctum 'ob-
ject', literally 'thing thrown before one', older German
Vor-wurf, and so on. The Slavic languages similarly imi-
tate abstract words from German, Latin, and Greek; thus
Russian ['so'v'isV] 'conscience', [pen'i 'mo't'] 'conceive,
understand' are modelled on the German compounds,
[pr'id m'stj 'object', also, is literally 'thing thrown be-
fore one*.
 IXIERACTION OF DIALECTS AND LANGUAGES 283

Wlien the foreign word is taken bodily into the lan-

guage, it is subjected to assimilative influences. Its sounds
are replaced by those of the borrowing speech, and its

structure is often assimilated to that of the native words.

This may not happen while the word is still used by
those who know the language and know that the
foreif]:n

word is foreign; as soon as it becomes genuinely popu-

lar, — that is, part of the universal usage, — it is sure
to be assimilated. Most examples of sudden sound-change
(p. 216) really belong here. Latin peregrinus 'pilgrim'
became pelegrimis and pilgrim in the mouths of people
whose native language was not Latin. The Latin acetum
Vinegar' became ^aiiJco in the mouth of Germans, whose
language had at that time no closed e, but, as the nearest
sound, only ^, and no suffix -Ito but a common one -tko.
Hence the modern High German form J^ssig ['ssik] or
['£si9]. This form shows us, moreover, that the German

change of a to e before i ('umlaut', see p. 215) and of t

after vowel to s (cf. Slrafte above and p. 208) and of k
after vowel to [x, 9] (the form with -k in modern German
is analogic) occurred since the first contact with the Ro-

mans.
So the Old French sillahe has become in English syl-
lahle in approximation to our suffix -able. Hammock was
introduced into English from the Spanish hatnaca, itself
a no doubt assimilated form of a Carib word. In Eng-
lish it was little changed, because it happened to resemble
the native words in -ockj such as hassock, hummock. In
German, however, where it resembled nothing in the na-
tive stock, it was assimilated into the form of a compound
Hdngematte 'hang- mat'. Such complete change of an ob-
scure word into a semautically orgiinized form is called
'popular etymologj^' It changed in German the Graeco-
Latiu arcuhallista 'cross-bow' into Armhrust, literally
284 EXTERNAL CHANGE OF LANGUAGES

'arm -breast', in English the Graeco-Latin asparagus into

sparrow-grass {aspa^'agus being meanwhile constantly re-
stored by those who know Latin). Old French crevice (it-
self a loan-word from Germanic) has become crayfish and
craivfish.
When words pass from one dialect to another, mutual
intelligibility modifies the assimilating process, in the
sense that the phonetic differences are often correctly
compensated. Thus we should naturally and unconsciously
put into the equivalent American sounds a new word we
heard from a Londoner. Nor need the words so borrowed
necessarily be of cultural significance. Dialect-mixtures
are as a rule recognizable only if some phonetic inconsist-
ency is retained. Thus in the speech of the northern
central part of the United States the vowel of such words
as hath, glass, laugh, path is [ae], but many speakers who
haye grown up in this pronunciation will, when on their
dignity, use the British [a], — often inconsistently. In
this way arise 'hyper' forms, where the affectation of a
foreign pronunciation is carried beyond its scope in the
imitated dialect itself; as when one speaks also [man],
where the English pronunciation itself has [ae]. The same
phenamenon appears in German: speakers whose dialect
has [i] for Standard German
[y] will affectedly substitute
[y]'s for their natural [i]'s even where the standard Ian
guage has [i], saying [ty:r] not only for Tur 'door' but
also for Tier 'animal'. Standard German [tiir].

Where words are permanently borrowed from one dia-

lect by another, they may betray themselves, like loan-
words from foreign languages, by their phonetic habit.
London English, for instance, has no native words with
initial v-; such as are not Latin-French are borrowings

from a dialect south of the Thames, which regularly has

initial v- for /*-, e. g. vat, vixen. High German has no
 INTERACTION OF DIALECTS AND LANGUAGES 285

words with h, d, g after short vowel, except those like

Krahhe 'crab', Dcgge 'mastiff, which are loans from Low
German. Phonetic investigation has shown that certain
Latin words, such as lupus 'wolf, hos 'head of cattle',
popina 'cook-shop' are borrowings from neighboring dia-
lects, such as, perhaps, the Sabine.
It is a phenomenon of dialect-mixture when we find in
English a number of the commonest words bearing un-
mistakable North Germanic character. The northern and
central English dialects of Alfredian times and the Scandi-
navian speech of the Norse invaders of that period were
not only mutually intelligible, but so much alike as to
seem only different forms of one language. When the in-
vaders settled by the side of the English, each dialect
came words of the other. Ultima-
to be interspersed with
tely the English, spoken by greater numbers and also in
the south, where there were no Scandinavians, carried
off the victory but retained, for ever after, a number of
words in Scandinavian form. Such words are egg, give,
guest, kettle, oar, they, sldrt, shy. The word egg, for instance,
is the Scandinavian correspondent of the German Ei,
Primitive Germanic *ajjon] North Germanic, but not West
Germanic changed jj to gg Give and guest would have
been palatalized in pre-English, like yield (p. 214); kettle
similarly, like child. The case of skirt is especially inter-
esting. In Old English sk had become sh [J], and the cog-
nate of Norse skirt was in English shirt. People were led,
as a result of such doublets, —
scrub and shrub, skirt and
shirt, and the like, —
where sh was spoken, to speak sk
also: consequently sk occurs by the side of sh in many
words that were not Scandinavian at all, —
as in scatter
by the side of shatter. Since that time words with 67»; have
been multiplied, until those actually brought in by the
Scandinavians are in the minority.
286 EXTERNAL CHANGE OF LANGUAGES

The transition from dialect-mixture to the unevenness

of individual speech is, of course, a gradual one. It may
be illustrated by several intermediate phenomena. Forms
from slower and from more rapid speech exist side by
side, e. g. cannot: can't, does not: doesn't. The same is

true of morphologic doublets: a speaker who ordinarily

says If he were here, he would help us may, after speaking
with members of less conservative strata, occasionally say
If he ivas here
Another such phenomenon is the mixture of older with
younger forms of speech. In most instances the older
form has been preserved in some set phrase, subject to
phonetic change, of course, but growing lexically or morpho-
logically antiquated. Thus tbe Old English sam-hlind
'half-blind' remained in use after the prefix sam- ^half
had ceased to be mobile or even to occur in any words
but this; the word consequently, became assimilated into
sand-blind and associated with a meaning 'totally blind'
Another striking instance is the German expression mit
Kind iind Kegel in the sense of 'bag and baggage', which
today means, word for word, 'with child and ninepin';
— Kegel is really here an otherwise lost word meaning
'bastard'. The English You had better go preserves an
otherwise lost use of had which troubles some speakers.
Ultimately, of course, the syntactic development which
crystallizes certain forms of discursively joined words
into the various set forms of materially specialized con-
structions (such as preposition plus noun in English) is
a process of preserving in set use what was formerly a
flexible manner of speech. Thus, to add an example to those
already given (pp. 114,ff., 171, ff.), our perfectic expression
with have originated in such sentences as I have written
a letter J
which meant originally 'I have a letter written,
in written condition' (literally or in the sense of Hiave
 INTERACTION OF DIALECTS AND LANGUAGES 287

to my credit'). Tliis turn of speech has been preser-

ved as an expression of perfectic action, and, as its ori-
ginal meaning lost dominance, has been extended to
such forms as / have slept, I have lost a hook, where it
could not at first have been used. This mixture of lin-
guistic strata is thus a factor in the regular linguistic
development.
Where alphabetic writing exists, older phonetic stages
may be preserved and borrowed by later times. This has
occurred most universally in the Romance languages,
spoken in communities extensively familiar with written
records of the older stage of their language, Latin. Thus
the Latin causa 'cause, affair' has become the French
chose 'affair, thing', but is preserved through writing in
the French cause 'cause, lawsuit'; the Latin securitatem
(accusative of securilds) 'security' has become surete 'sa-
fety, security' (whence the English loan surety), but has
had written existence, whence the French took securite
'assurance, unconcernedness' and the English, through
the French, security. The Latin separdre became French
sevrer 'to wean, deprive' (from the French English borrow-
ed sever), but was preserved in writing and borrowed
as Frencb separer and English separate. These examples
could be multiplied in great numbers. They are not essen-
tially different from the 'spelling-pronunciations', in which
an archaic spelling leads to the revival of phonetically
divergent ancient forms. Thus the old t of often, softer^
has been long lost by phonetic change, but the influence
of the orthography leads many speakers, some consciously,
some unconsciously, to pronounce it. Indeed, there may
arise in this way forms that were never spoken at all,
such as ye for the, due to misreading of the old character
y> (for th), and author, where learned orthographic ped-

antry alone is resp )n8ible for the h^ which, however,

288 EXTERNAL CHANGE OF LANGUAGES

long ago has led to the substitution of [0] for the old-
er [t].
There can be no doubt, in fact, that the existence of
written tradition has, by constantly demanding the associa-
tion of fixed and conservative forms, impeded phonetic
change. If we had no alphabetic writing, or if only a few
of us could read, such forms as [juni'vjsiti] would long
ago have given way entirely to such as [juVasti] or even
to such assimilative reformations as [Vjsti] or [Vajsti].
The written form thus tends to preserve the phonetic
form of the language; though of course, it can do so only
to a comparatively small extent. Our conscious control
over the forms of writing is not yet extensive: the ob-
stacles which the various attempts at improving English
spelling have met are an example; nevertheless, as these
attempts themselves show, not to speak of the successful
governmental regulation of orthography in European coun-
tries, consciousness and systematic reasoning in this sphere

are gaining ground. When the community wiU consciously

and deliberately shape orthography a great step to-
its

ward the conscious influencing of language will thus have

been taken. It is possible, in fact, that, very gradually,
language, like religion, government, and other once purely
communal processes, is developing into a conscious activ-
ity.

Stiindard languages. How fast and ultimately how

8.
far this development will progress it is, of course, impos-
sible to say. To it belong, however, a number of characteris-
tic features in the rise of the so-called standard languages.
These are favored dialectswhich, either in written form
alone or also in oral, are used all over a dialectally differen-
tiated territory. At first they are used for communication
between members of different dialects, the speaker whose
 STANDARD LANGUAGES 289

dialect represents the less cultured community using, as

well as may be, more civilized neighbor. Gradu-
that of the
ally it comes that members of two dialects that are per-
haps with difficulty intelligible to each other, will use, in
speaking together, the same favored dialect, though it is
native to neither of them, until at last it may become a
second language for formal and non-local discourse all
over the area. Soon there will be speakers in many parts
of the country who can speak only the favored dialect, —
such, for instance, as the upper classes of English, Ger-
man, or French society, who rarely can speak the 'patois'
of their native locality, but know only the 'national' or
'standard* language. The latter may ultimately crowd out
the local dialects; this happened in ancient Greece and
in the Roman Empire (where both related dialects and
foreign languages gave way to Latin) and is rapidly happen-
ing in modern England.
Meanwhile the favored dialect is used as the language
of literature and is learned by many out of books: the
individual writer has considerable power to influence it.
Cicero, Dante, Chaucer, the translators of the King James
Bible, Goethe, and other writers of great and enduring
renown have permanently moulded their language: in
Shakspere we see the origin not only of many quotations,
but also of some set forms of speech, which have come
to us by virtue of his baving used them. As the lan-
guage of books, the standard language is subject to fixed
canons of correctness: what good authors do not use is
wrong. This consideration, as well as the necessity of
teaching the standard speech to people who first learned
a local dialect, leads to the compilation of grammatical
descriptions of the language and to lexical summaries,
dictionaries. Tlius we arrive finally at a conscious standard
of correctness, which modifies linguistic growth, especially
Bloom field, Study of L&nga»ge 19
290 EXTERNAL CHANGE OF LANGUAGES

in checking the spread of the easily recognizable morpho-

logic innovations of the was for were type. New forms
or words are usually recognized as such, — as dialectal,
vulgar, incorrect, — and consequently associated with
etc.

a peculiar emotional tone. If a good author uses them,

they may become part of the standard language, although
they will long be felt as lacking in dignity. Such words
are in English sloh, slohher, whang, thump, thwack, squunch,
piffle, and the Sometimes they gain ground rapidly;
like.
thus moh, the assimilatively shortened form of mobile vul-
gus was fifty years ago frowned upon as a barbarism.
Entirely unchangeable are of course the literary languages
which exist only in written form, such as Latin in the
Middle Ages, Sanskrit, Classical Arabic, Hebrew. Al-
though these may be occasionally spoken, the great pre-
ponderance of use is in careful writing according to the
rules of the grammar and lexicon and on the model of
classical authors.
The standard language may be the dialect of the capital
in conservative form, as in France, England, and Russia
(Moscow), or a mixed dialect as in ancient Greece, where
the Koine was composed of Athenian and Ionic (Asiatic
Greek) elements. It has happened in a number of cases,
now, that in the determination of the forms of such a
language, individuals have been of influence. Modern
Standard German, of complex origin, was, after all, brought
into shape more by the careful work of Luther in his
Bible-translation than by any other one factor. Modern
Servian was molded by Karadjic upon older forms of the
language, and the ^Landsmaal', one of the two competing
standard languages of Norway, is in great part the crea-
tion of a nineteenth-century linguistic student, Ivar Aasen,
who founded his work on the southwestern dialects of
his country.
 STANDARD LANGUAGES 291

Thus it may be said that language also, even if

fairly
in smaller measure than any other social activity, has
shared in the human progress from unconscious evolution
into conscious shaping of conditions. In this phase of
linguistic development two features are of special impor-
tance: the conscious teaching of languages, for the pur-
pose, of course, of establishing communicative bonds, and
the conscious observation of language, linguistic science

19'
 CHAPTER rX
THE TEACHING OF LANGUAGES.
1. The purpose of foreign -language instruction.
In communities whose culture is undeveloped no lan-
guages are taught. The beginning of language-instruction
comes always when ancient writings of artistic or, especi-
ally, ethical and religious importance are to be handed

on to suceeding generations. Thus the Hindus study the

Vedas and the Sanskrit epic and classical literature, the
Mohammedans classical Arabic and the Koran, the Parsis
Avestan, the Jews Hebrew, the Chinese the old literature
of their country. In Europe the ancient Greeks of histor-
ical time studied Homer, whose language was even for
them highly antiquated, the Romans Greek, the medieval
and modern nations Ancient Greek and Latin.
To these studies are added, as the consciousness of
nations increases, the languages of important feUow-na-
tions. This is a deliberate widening of the bonds of com-
munication (p. 291): it is desired that a large element of
the nation understand the writing and speech of foreign
contemporaries. Just as the study of ancient languages
is to preserve the cultural tradition, so that of modern
is community abreast of modern progress.
to keep the
The latter study is prompted also by material motives,
such as the need of foreign languages in commerce and
the desirability of promptly utilizing foreign inventions
 CHARACTER OF THE INSTRUCTION 293

in science and industry. One may say that today the

nation which contains no large class of people who un-
derstand foreign languages dwells in pitiable seclusion.
Finally, as the idea of humanity takes form, there
comes the wish not only to be acquainted with the char-
acter and history of one's own nation, but also, in part
as an elucidation of these, to understand the motives,
achievements, and ideals of the sister-communities. At
which the European nations more fully than
this stage,
America have reached, the school studies include not
only instruction in foreign languages, but also a suitable
introduction to the life, culture, and ideals of the foreign
nations.
2. Character of the instruction. It is only in the
last twenty-five years and in the European countries that
success in modern-language teaching has ever been at-
tained. Of ancient languages this cannot be said: it is
true, however, that where here also success has been
won, it has been by the same general methods as are
today used for modern-language instruction in Europe:
by a conscious or unconscious accordance with the fun-
damental processes of language-learning and, for that
matter, of speech in general. Where, as in our own
practice, this accordance is wanting, failure is inevitable.
Of the students who take up the study of foreign lan-
guages in our schools and colleges, not one in a hundred
attains even a fair reading knowledge, and not one in a
thousand ever learns to carry on a conversation in the
foreign language. This is due to the fact that almost
every feature of our instruction runs counter to the uni-
versal conditions under which language exists. While a
growing number of our teachers have acquainted them-
selves with the modern methods, their efforts are largely
checked by the antiquated outer circumstances, such as
294 THE TEACHING OF LANGUAGES

the late age at whicli pupils begin the study and the
smallnumber of class hours, coupled with the reliance
on home assignments, which are of little use in language-
instruction.
Our fundamental mistake has been to resjard
O Do
language-
teaching as the imparting of a set of facts. The facts
of a language, however, are, as we have seen^ exceedingly
complex. To explain to the student the morphology
and syntax of a language, be it his own or a foreign one,
would require a long time, and, —
even if it were done
correctly by linguisticaUy trained teachers, —
would be
of little or no value. To set forth the lexical facts would
be an endless task, for not only does each word of the
foreign language differ in content from any word of the
native language, but this content itself is very difficult
of definition. The greatest objection of aU, however, is
that, even if the pupil managed somehow to remember
this immense mass of facts, he would scarcely be the
more able, what with it aU, to understand the foreign
language in its written or spoken phase. Minutes or
hours would often elapse before he could labor out the
value of a sentence by recaUing the facts concerned.
Language is not a process of logical reference to a con-
scious set of rules; the process of understanding, speak-
ing, and writing is everywhere an associative one. Real
language-teaching consists, therefore, of building up in
the pupil those associative habits which constitute the
language to be learned. Instead of this we try to ex-
pound to students the structure and vocabulary of the
foreign language and, on the basis of this, let them
translate foreign texts into English. Such translation is

a performance of which only people equipped with a

complete knowledge of both languages and with consid-
erable literary ability are ever capable. As a method
 AGE OF THE PUPIL 295

of study, moreover, it is worthless, for it establishes as-

sociations in which the foreign words play but a small
part as symbols (inexact symbols, of course) of English
words.
The excuse usually given for this practice is that
American conditions make only a 'reading knowledge' of
the foreign language, — especially, if ancient, — of im-
portance, — that it is not our purpose to enable pupils
to order a meal in the foreign language. Reading, how-
ever, is no from the other phases of using a
different
language: the expressions of the language are not the
given members of mathematical equations or puzzles,
but must enter into a set of rapidly and easily function-
ing associative habits. Coirect methods of language-
teaching differ from those which we are at present un-
successfully using not in aim, —
any aim can here be
attained by good missed by bad teaching,
as surely as it

— but in adaptation to the mental conditions underlying

the activities of speech. In what follows I shall natu-
rally speak of American conditions and assume that the
ability to read rather than to speak is aimed at: needless
to say that even here the desired associations cannot be
formed without much oral and auditory practice. I be-
lieve, moreover, that American conditions are coming to
make a 'speaking knowledge' more and more desirable
and that the time is not far off when here as well as
abroad the ability to converse in one or two foreign
languages will be looked upon as one of the ordinary
marks of education.
3. Age of the pupil. The best age at which to begin
a foreign language is that between the tenth and twelfth
years. begun (earlier, the progress is usual-
If the study is
ly so slow that uothiug is gained, the pupil who begins
later soon overtakiig him who began younger. If the
296 THE TEACHING OF LANGUAGES

study is begun at the age indicated, further languages

may be taken up at intervals of a few years; as the
student accumulates experience^ the later languages will
be learned more rapidly and with less effort than the
earlier, until a facility may be acquired which astonishes
those who have bad less practice. It is worth while to
say this, because there exists a superstition to the effect
that languages are acquired by some special power of
the intellect which wanes in maturity.^) If the first for-
eign language is begun later than the twelfth year or
so, — and here we see, perhaps, the source of the bit
of popular psychologizing just mentioned, — we find a
growing disinclination on the part of the pupil to go
through the constant practice by which alone success is
attainable. Older students who have never before studied
a language are too exclusively practised in conscious,
logical grouping of facts to accept the repetition of what
is already understood but not yet assimilated; when they
have grasped the ^meaning' of a text in terms of the na-
tive language, they are disinclined to go on using the
text with attention to the foreign expression. The nec-
essary simplicity as to content of the elementary texts
also bores them. At the age of ten or twelve, on the
other hand, the pupil by the novelty of what
is attracted
he learns, enjoys the growing power of expression and
understanding in a new medium, and the playing at being
something strange (e. g. an ancient Roman, a German,
or a Frenchman), nor is he intellectually too superior to
the simple content of the earlier lessons. Once the habit
of foreign-language-study has been at this age set up,

1) It actually happens that students in our universities are

excused from language requirements on the plea that they are
''too old' to leain languages.
 EQUIPMENT OF THE TEACHER 297

the student finds no difficulty in going on to other lan-

guages even when he is more mature, forhe knows from
experience the necessity of the processes involved and
the fruits which they so soon bear.
4. Equipment of the teacher. As to the prepara-
tion of the teacher, a prime requisite is, of course, mastery
of the lanojuaore to be tauo^ht, — in modern lanoruaores
a knowledge comparable to that of an educated native
speaker and in ancient a fluent reading ability and some
facility in writing. This is so obvious that it needs no
elaboration, yet we constantly find in our schools and
colleges teachers whose knowledge falls far short of this
demand. Such teachers are from the beginning incapable
of successful instruction, for, though they may vociferr
ously explain (in English) the abstract grammatical facts
of the foreign language, they cannot give the pupil
practice which will form and strengthen in him the asso-
ciative habits which constitute the languacre. If the services
of a teacher approximately possessing these qualifications
cannot be obtained, the instruction should be given up,
as it is only a waste of time.
The same may be said, though not so universally, of
teachers possessing this but lacking anotlier qualification;
namely, the knowledge and experience of how a lan-
guage must be taught. Next in uselessness to a teacher
who does not know the language is the teacher who, to
be sure, does know it, — he may be a native speaker
of it,— but has not the linguistic and pedagogic knowl-
edge of how to impart it. In English, — and, if he is
:i foreigner, often in broken English, — he indulges in
tlescriptions of the beauty, conformity to logic, etc. of
the language, and when the pupils, on the strength of
this, fail to learn anything, he attributes the failure to

their sloth, stupidity, or narrow minded dislike of what

298 THE TEACHING OF LANGUAGES

is foreign. Ultimately such teachers become either in-

different easy-goers or irascible cranks.
School language -teaching has been successful only
where thorough knowledge of the foreign language and
training in the necessary linguistic and pedagogical prin-
ciples, supplemented by experience in practice- classes
under supervision, are demanded of all candidates for
teaching positions. Even then centralized control, for
instance by a government bureau, has been found desir-
able, as every one will understand who has heard at our
teachers* meetings the grotesque ^methods' which un-
controlled and isolated teachers, innocent of the most
fundamental principles of the subject or of any accepted
writings about it, have developed during winters of
teaching. While all instruction, to be worth anything,
must be moulded by the teacher's personality, his whim,
conceit, or lack of information must not be allowed to
ignore the results of generations of labor and experience.
In short, the language-teacher must be a trained pro-
fessional, not an amateur. The postponement of much
elementary language-teaching to our colleges brings,
aside from the unfortunate age of the students, the great
disadvantage that it practically excludes such teachers.
In accord with the true purpose of college arid university,
the instructors there employed are not pedagogues but
people who have found their calling in the handing on
of culture or in scientific teaching and research. The
professional language-teacher who occasionally finds his
way into these institutions soon learns that he can ex-
pect neither honor nor advancement for excellence in his
vocation: he must exchange it for more purely cultural
or scientific studies or be content with a minor position.
Nearly all of the elementary language -teaching in our
colleges is done, accordingly, by doctors of philosophy
 DRILL IN PRONUNCIATION 299

who have no training and no ambition in this direction,

but find their interest and seek their advancement in
linguistic or literary teachingand research. Their in-
struction is directed, with ulterior evil effect on that
of the secondary schools also, by men who have come
to the front in some special branch of linguistics or in
literature and have often no understanding of the prob-
lems and conditions of foreign -language teaching. As
long as this work is inappropriately left to colleges,
these institutions should give employment and promotion
to teachers who make it their business, and allow liter-
ary and linguistic scholars to stick to their last, for they
are no more capable of this work than are grammar-
school and high-school teachers of conducting graduate
seminars.
5. Drill in pronunciation. Instruction in a foreign
language must begin by training the pupil to articu-
late the foreign sounds correctly and without difficul-
ty or hesitation. We have seen in Chapter II that the
teacher's ability to pronounce these sounds does not in-
volve ability to tell others how they are pronounced.
This information must be given in terms of movement
of the articulatory organs. The instruction must begin,
therefore, with the elements of phonetics as applied to
the pupil's native language and, by contrast, to the for-
eign one. Description alone is, of course, of no avail:
the pupils must be brought to practise the foreign arti-
culations until they have become automatic. This prac-
tice should be enlivened by the subject-matter, but it must
remain practice in articulation, an unidiomatic articulation
being in no case allowed to pass muster. Overgrown
pupils, especially if unused to accurate and painstaking
study, will cont<'nt themselves with noting certain gener-
al resemblances to native sounds and inter])roting the
300 THE TEACHING OF LANGUAGES

examples into the nearest corresponding native articula-

tion. The phonetic driU must be based, in the case of
languages that are unphonetically written, such as French,
on a transcription into a phonetic alphabet. After pro-
nunciation has been mastered, the irregularities of the
standard orthography will cause much less difficulty
than if they were at the beginning presented in inextri-
cable confusion with the foreign pronunciation.
6. Method of presenting semantic material. As
time goes on, the pronunciation will require less and less
of conscious attention on the part of the learner. From
the very beginning, however, the significance of the ex-
pressions that are practised should be made use of. The
very first phonetic examples should be characteristic
words and phrases. The signification of these cannot,
as we have seen in Chapter IV (p. 85, fi".) be taught in terms
of the pupiFs native language. This would involve either
false statements or, if these were to be avoided, lengthy,
complicated, and easily forgotten explanantions. The
foreign utterance must, instead, be associated from the
very first, with its actual content. The beginning should
be made, therefore, with expressions concretely intelli-
gible: formulas of greeting, short sentences about objects
in the classroom, and actions that can be performed
while naming them.
As the work goes on to connected narrative and de-
scriptive texts, this method must be continued. The texts,
therefore, must at first be confined to very simple dis-
course about concretely illustrable matters. Pictures are
here of great use. Any new text must be explained in
terms of what has already been learned, not in English.
Translation into the pupiFs native language or other ex-
plicatory use of it must be avoided, for two reasons. The
terms of the native lancruao^e are misleadinf^, because the
 METHOD OF PRESENTING SEMANTIC MATERIAL 301

content of any word or sentence of the foreign language

is always different from any approximate correspondent
in the native language. A pupil taught that the German
Icocn means 'read' will say ich lesen instead of ich lese.
If he is taught that wenn means Svhen', he will confuse
it with alSy and if he is taught that oh means 'if, he will

confuse it with wenn. Once such associations are formed,

— and their fictitious simplicity makes them compara-
tively easy to fix, — no amount of explanation or insist-

ence on the part of the teacher will overcome them.

The second reason for the avoidance of translation is that,
in the word with the native
association of the foreign
one, the latter will always remain the dominant feature,
and the former will be forgotten. The learner will know
that he has met the foreign word for "pencil', but the
sound and spelling of the foreign word will be very hazy
in his mind. Where continued translation has given
facility in these associations, the pupils scarcely look at
the foreign text before the English word, right or wrong,
becomes conscious. The result is that their foreign vo-
cabulary remains small; they are forced to look up in
the glossary over and over again the same common word,
and, whenever they look it up, their habit leads them to
fix only the native interpretation and to go on with the
text. Every teacher has known students who have read
hundreds of pages in a foreign language and yet have
to look up dozens of the commonest words in any page
of a new text -^ or even of the old, if they are asked to
re-read.
Instead of translation the work with a text should
consist of repeated use of its contents in hearing, read-
ing, speaking,and writing. The beginning is best made
before the pupil has even seen the text. The teacher
explains in the foreign language the new expressions
302 THE TEACHING OF LANGUAGES

whicli are to occur and leads the pupils to use them in

speech over and over again. Then the pupils are required^
first, to read the new selection correctly after the teacher,

later, to answer, with the book, then without it, simple

questions ahout it, to converse about its subject-matter,
and to retell it in speech and in writing. The text
should not be left until every phase of it has been thor-
oughly assimilated: no text should in the beginning be
used whose linguistic contents are not important and com-
mon enough to deserve such assimilation.
The range of work that the pupil can do outside the
classroom is here very small. The danger that he will
practise false pronunciation or usage must make the
teacher very cautious in the assignment of outside lessons.
Copying the text and preparation of lists of words and
sentences taken directly from it are least dangerous. As
the work must thus be done almost entirely in the class-
room, eight hours a week of class-work are not too much
in the first year or two.
It is only after the pupil has mastered for speaking
and writing as weU as reading a good central stock of
words, forms, and constructions, that more rapid reading
should be undertaken. Without a nucleus of expressive
material over which the pupil has full and accurate con-
trol, the necessary analogies even for that degree of
understanding which we call a reading-knowledge are
lacking.
7. Grammatical information. The amount of text
covered in the first year or two cannot be large. It is
to be measured not by the page, but by the amount of
new material introduced. Beginners will do well, if they
learn a thousand words in the first year of the first
foreign language. A hundred pages of carefully pre-
pared easy text wiU contain this amount of material.
 GRAMMATICAL INFORMATION 303

The texts need not be arranged in terror of introduc-

ing new grammatical features before they have been
systematically — i. e. theoretically, — explained. Gram-
mar, as such, is not necessary for the use or understand-
ing of a language: the normal speaker or reader is not
conscious of the grammatical abstractions. In foreign-
language teaching grammar is of use only where it def-
initely contributes to the ease of laarning. When a new
text appears the learner should be able to tellwhere he
has met the words and phrases it contains and others
like them. Now, when he meets, let us say, a new in-
flectional form of a known word, the differences in the
use of the two forms should be carefully illustrated and
practised. After a time, when a considerable number
of such collocations has been made, — when a number
of singulars and plurals, for instance, have been com-
pared as to use and form, —
the grammatical statement,
if simple enough to be of help, may be given. In fact,
it will be unnecessary, for the pupil will with consider-
able have formulated it for himself. On the
interest,
other hand, the grammatical statement must often be
kept temporarily incomplete. The German dative case,
for instance, is of so heterogeneous use that a statement
of its value would take a long time and would be unin-
telligible to any but a linguistically trained learner. In-
stead, we may collect our accumulated examples of datives,
observe the forms, and their occurrence after certain verbs
and certain prepositions and independently in the sentence.
All this need not be done at once: the dative with prep-
ositions, especially, in its contrast with the accusative,
may, as the most definitely recognizable use, be collected
and observed long before the other types. In every in-
stance the forms themselves in their natural connection
jhould be practised to the point of thorough habituation
304 THE TEACHING OF LANGUAGES

before the abstract statement isConsequently tbe

given.
grammatical features of a new text are of secondary im-
portance, provided tbat it is easily explained and under-
stood. Grammar should be used only as a summary and
mnemonic aid for the retention of what has been al-
ready learned. Where it cannot be so used, it should be
omitted.
8. Texts. While the matter read should, of course,
be characteristic of the foreign nation's life and culture,
the selections should not hasten to teU too much at the
cost of simplicity. Selections of literary value should not
be introduced before the pupil can understand them: if
he cannot, their literary qualities are lost to him. The
transition from the mere learning of the foreign language
to the study of its literature and culture must be gradual,
especially in the case of the first language studied. This
language, however, should by the end of the secondary-
school period, have become so familiar that the last years
are spent entirely in the study of works of ethical, ar-
tistic, and generally cultural interest. In the languages
later begun the practice in acquiring languages will make
up for the shorter time of study. All reading, no matter
of what nature, should be within the pupil's immediate
range of understanding o^ the foreign language. The
great bulk of the time must be taken up in fixing in the
pupil's mind the value of the foreign expressions, until
these, when seen or heard, are automatically understood.
It is only on the basis of such knowledge as this that
reading can go on at a rate which makes an ideal
effect upon the pupil possible. The premature reading,
or rather pottering through foreign literature in our
schools (e. g. Wilhelm Tell in the second year of iU-
taught German) is a mere working-out of senseless
puzzles.
 REFERENCES 305

The interpretation of what is read must always be

pedagogic rather than scientific in purpose. The aim of
foreign-language instruction is to acquaint the pupil
with the foreign language, through it with the foreign
culture, and generally, as in all other school studies,
to him to a higher mentality, in every sense of
train
the word. The scientific study of the foreign -language
or literature is entirely inappropriate for a school for-
eign-language course. By postponing this course to the
hicrh-school and collecre we have brought about confusion
of elementary foreign -language learning with the aims
of scientific linguistics and scientific literary history.
These studies belonn; to a later stage of education, in
which, to be sure, both should be represented; but an
exposition of Grimm's law in the elementary German
classroom, or of the motives of romanticism in that of
second-year French is a deplorable farce.

The texts, then, as the pupil grows familiar with the

language and at the same time progresses towards matu-
rity, should be selected more and more for their inner
content. From the simplest elementary selections we may
proceed to easy short stories, then to more serious histor-
ic, descriptive, and narrative prose and to drama and

poetry. Toward the end of the course summaries of the

literary, cultural, and political history, preparing for —
possible college courses in these subjects, should be —
read.

References. The English reader will find details

9.
about the methods of language-teaching in the two follow-
ing books and in the bibliographies which they con-
tain:

Otto Jespersen, ITow to Teach a Foreign Language^

London and New York (Macmillan) 1904 and 1908.
Bloom field, Studj of LkDguuge 20
306 THE TEACHING OF LANGUAGES

Leopold Bahlsen, The Teaching of Modern Languages,

Boston (Ginn) 1905.
The latter book contains a brief review of the history
of language-teaching in Europe, which shows plainly that
our language-teaching differs from that of the European
countries not as a mere difference in choice of methods
(e. g. that they use the 'direct' and we some other 'meth-

od'), but that most of our practice is half a century

or so behind that of the European schools, which has
kept better pace with scientific insight into language.
 CHAPTER X.

THE STUDY OF LANGUAGE.

1. The
origin of linguistic science. Linguistics (Ger-
man Sprachivissenschaft , French linguistiqiie) took its
beginniDg, historically, in the study of writings which
were preserved for their religious or esthetic value. As
these texts antiquated^ interpretation of their language
became necessary and led finally to a grammatical codifi-
cation of their forms. In this way the study of philology
German Fhilologie, Yrench philologie), — that is, of nation-
al cultural tradition, —
came to include a linguistic
discipline whose aim was the practical one of making
intelligible and preserving certain writings. Thus origina-
ted the treatises of the Indian grammarians (chief among
them Panini, fourth century B. C), the Ancient Greek
grammar (especially Dionysios Thrax, second century B C,
and Apollonios Dyskolos, second century of our era), the
Latin grammars (Donatus, fourth century, Priscian, sixth
century), the Hebrew grammar, and so on.
The linguistic study at this stage was properly a means
to an end, a prodrome to philology. Nevertheless, there
were always scholars, who, be it from a genuine but mis-
guided interest in language or from sheer pedantry, con-
fined themselves to this grammatical study. Thus there
developed a pseudo-linguistics, which occupied itself with
grammatical dissection of texts, with haphazard etymolo-
303 THE STQDY OF LANGUAGE

gies,and witli vague theorizing as to origins.^) A further

impulse to this grammatical study was felt when the pop-
ular language deviated from that of the texts to the
point where the latter became unintelligible, or when
people of alien speech adopted the culture and with it
the philologic studies of a more advanced nation. Both of
these conditions were given in medieval Europe, where
classical Latin had become unintelligible to the people of
Romance ton<]jue and was foreign to the northern nations.
At first the teaching of Latin (and^ when it was revived,
that of Greek) was conducted on a sensible basis: the
language was spoken, written, and read until the student
had firm command of it and easy access to the classical
literature. Later, however, pedantry prevailed: in spite of
such great educators as (in the sixteenth century) Ascham
and and Comenius,
(in the seventeenth century) Ratichius
theoretical grammar came more and more to be looked
upon as a means of learning the ancient languages. This
went so far that, for example, up to very recent times
English schoolboys had to memorize the entire contents
of a Latin grammar before they were allowed any real
contact with the language. It was only a slight allevia-
tion of this barbarity when the rules of grammar were
at least illustrated by disconnected sentences. This latter
method prevailed when, early in the nineteenth century,
modern languages came to be studied in Europe for practi-

1) Owingto this occupation the term 'philology' has come

to be misused in English first as meaning linguistics and then
even in reference to misplaced and piddling grammatical study.
The best usage, however, — that, for instance, of the greatest
of English-speaking linguistic scholars, the American William
Dwight Whitney, —does not sanction this; philology is the
study of national cultural values, especially as preserved in the
writings of a people, linguistics the study of man's function of
language
 THE ORIGIN OF LINGUISTIC SCIENCE 309

cal purposes; accordingly, the grammatical facts of these

were codified in imitation of Latin and Greek and became
the basis of instruction. This sentence-method, used in
the books of Ahn, Ollendorff, and many successors, is
stiU, in various modifications, supreme in American schools
In this way pseudo-linguistics, supported by a false
pedagogic idea, held the field until the nineteenth cen-
tury. There had, to be sure, been attempts in the preced-
ing centuries to attain a genuine understanding of lan-
guage, but these were frustrated chiefly by the aprioristic,
purely logical —unhistorical and unpsychologic —
manner
of consideration and also, in spite of the comparison of
Arabic, Hebrew, and the writers' own modern languages,
by the confinement of the study to a narrow and acciden-
tal group of idioms. The work of such men as Schottelius,
de Brosses, Fulda, and even, early in the nineteenth cen-
tury, Bernhardi, remained, therefore, without direct re-
sults.

It was the opening to Europe of India and the widen-

ing of cultural and scientific interests which we call the
romantic movement, that led to a more fruitful study of
laDjjuaore.
The romantic interest in things ancient and distant
made European thinkers ready to receive the Indian cul-
turewhich such men as William Jones and Colebrooke
brought from the East. This culture included, in the manner
above described, grammatical treatises dealing with San-
skrit, the sacred and literary language of India, — trea-
tises which European scholarship found a linguistic
in
achievement beyond any it had known. For, while the
Sanskrit grammar had not attained to the idea of a science
of language and served in India the same purpose of
mis-instruction of the young that Latin grammar had
fulfilled in Europe, its original task of preserving through
310 THE STUDY OF LAl^v>UAGB

millenia the norm of usage was satisfied by a

classical
highly exact description of Sanskrit pronunciation and
word-formation. The former of these, especially, was a
revelation to European students, who had never given
attention to the articulations of speech. Medical investiga-
tors, meanwhile, —
owing, again, to the romantic impetus,
— came to study the physiology of language, until we
find the two tendencies, represented, for example, by the
physiologist Briicke and the linguist and philologian
Scherer, culminating in the modern discipline of phonet-
ics. In respect to word-formation also, the transparency

of the Sanskrit lansuaoje and the excellent treatment it

had received from the Hindu grammarians, afforded a
new insight into the development of linguistic forms.
Modern linguistics more than any other phase of our
cultural life, is a heritage from India.
The romantic impulse led to a widening of the group
of languages studied, which, with the insight afforded by
Sanskrit, resulted in the recognition that a number of
languages of Europe and Asia are related. This recogni-
tion,made by William Jones and Friedrich Schlegel, was
shaped by Franz Bopp (1791 —
1867) into a scientific
investigation, which showed definitively that these lan-
guages are divergent forms of an earlier uniform parent
language. This investigation, brought into fuller and more
accurate form and subjected to more careful method by the
work of such men as August Friedrich Pott (1802 1887) —
—
and August Schleicher (1823 1868), has grown into the
study of Indo-European linguistics, which to this day has
remained the central and best-known field of linguistic
science.
The progress of Indo-European gave new
linguistics
interest to the study of Latin, Greek, Sanskrit, and the
modern European languages. The first three were directly
 THE ORIGIN OF LINGUISTIC SCIENCE 311

involved in the origin of Indo-European study, and the

groups of modern languages soon received their individual
treatment. This has been fullest in the Germanic and the
Romance languages. Jacob Grimm (17S5 1863) laid —
the foundation of the former in his monumental Deutsche
Grammatik ('German', —
we should say tod ly 'Germanic'
— Grammar), the first great scientific linguistic work of
the world, and perhaps even today the greatest. On Grimm's
model C. F. Diez (1794—1876) founded the study of the
Romance languages in his Grammatik der romanischen
Sprachen. The scientific study of the Celtic group received its
basis in the Gramumtica Celtica of J. K. Zeuss (1806—1 856)
and that of the Slavic languages in the VerglcicJtende Gram-
mnfik der slaivisclicn Sprachen of Franz von Miklosich
(1813—1891).
The new interest in linguistics did not, of course, con-
fine itself to theIndo-European languages: it led also to
the study of language in general. This study received its
foundation at the hands of the Prussian statesman and
scholar Wilhelm von Humboldt (1767 —
1835), especially
in the first volume of his work on Kavi, the literary lan-
guage of Java, entitled Uher die Verschicdeuhcit des meiisch-
lirhcn Sprachhaues und ihren Fin/hi/^ auf die gcistige Ent-
icickchinfj dcs McnscJiengcscJde Ms ('On the Variety of the
Structure of Language and its upon the Mental
Influence
Development of the Human R;:ce'). Humboldt's work has
been followed in two directions. The study of the lan-
guages of the world has resulted in a series of disciplines
paiallel to Indo-European lingui.«:tics, each studying a set
of related idioms. The chief families today so recognized
are the Semitic, tlie Hamitic (these two are thought to
be in turn descended from a common earlier speech), the
Uralic (Hungarian, Finnish, and other languages), the
Altaic (Turkish, Tartar, etc.; these two groups also are
312 THE STUDY OF LANGUAGE

thought by many to be related), the Caucasinn (in the

Caucasus; the most important language is Georgian), the
Malayo-Polynesian, the Indo-Chinese (Tibetan, Burmese,
iSiamese, Chinese), the Dravidian, the Bantu, and the vari-
ous American families, such as the Athapascan and the
Algonquian. These have progressed in various degrees
toward a scientific comprehension like that which we have
of the Indo-European languages. Thestndy of the American
languages, though supported in praiseworthy manner by
our government, is hampered by many external conditions,
includinor the lack of investiojators with linsjuistic and
especially phonetic training.
The other direction in which Humboldt may be said
to have led the way, — although here the older gramma-
rians have been not without influeuce, — is the study

of the conditions and laws of language: its psychic and

social character and its historical development. This study
was furthered by the growth of psychologic insight and
of the historical point of view and method, — both of
which are from the beginninor related to the linf]fuistic
studies by the common origin in the romantic movement.
Especially active in the psychologic interpretation of lan-
—
guage was H. Steinthal (1823 1899); the American schol-
ar W. D. Whitney (1827 —
1894) applied to the historic
phase a remarkable clearness and truth of comprehension,
to be appreciated in a field from which mystic vngueness
and haphazard theory have been slow to recede. Both of
these men have been followed by numerous investigators
who have contributed to our understanding of the men-
tal processes of speech and of its change and develop-
ment in time; the great advance of psychology in recent
decades and the rise of social and ethnologic studies have
been, of course, of the highest benefit to this phase of
the science of language
 HOW TO STTjDY linguistics 313

2. How to study linguistics, a) The student who

wishes to devote all or any considerable part of his time
to the study of language should begin with that language
whose facts are immediately accessible to him, — of course,
his own. He
sbould diligently watch his articulations,
practise their phonetic notation, and observe individual
and local variations from his own usage. This observation
must be accompanied by an elementary study of phonet-
ics, for which one of the following books, in the beginn-

ing preferably the last named (which contains a brief

phonetic text in three varieties of English, including
American), should be used:
Henry Sweet, A Primer of FhoneticSj third edition, Ox-
ford 1906.
Otto Jespersen, LcJirhuch der Phonetikj second edition,
Leipzig and Berlin 1913.
Paul Passy, Petite phonctigue compare'e, second edition,
Leipzig and Berlin 1913.
A fuller and by far the best treatise on phonetics, which
the student should later use, is:
Eduard Sievers, Grund:iige der Phonetih, fifth edition,
Leipzig 1901.
The learner should then go on to the morphology and
syntax and finally the phraseologic and stylistic features
of the language he hears and speaks every day. There
are, unfortunately,few descriptions of modem English
which can be consulted in this connection. The southern
British usage is- given in:
Henry Sweet, A Primer of Spolicn English, fourth edi-
tion, Oxford 190G.
The northern British usage, more conservative and more
like the American, is given in:
R. J. Lloyd, Northern English: Phonetics, Grammar^
Texts, Leipzig 1899.
314 THE STUDY OF LANGUAGE

b) The approacli to the historic development of lan-

guage should then be made through the medium of Eng-
lish. For this the aids are copious. One may first read:

Otto Jespersen, Groivth and Structure of the English

Language, Leipzig 1905.
J. B. Greenough and G. L. Kittredge^ Words and their
Ways in English Speech, New York
(Macmillan) 1901,
and then the various 'readers' and 'primers', true' models
of their kind, of the late Henry Sweet, published by the
Oxford University Press, viz., for Old English:
Henry Sweet, Anglo-Saxon Primer, eighth edition;
Henry Sweet, Anglo-Saxon Beadcr, ei<^bth edition;
Henry Sweet, A
Second Anglo-Saxon Header: Archaic
and Dialectal,
and for Middle Enojlish:
Henry Sweet, First Middle English Primer, second
edition;
Henry Sweet, Second Middle English Primer: Extracts
from Chaucer, second edition.
These should be supplemented by the historical accounts
of the development of English in:
Heny Sweet, Primer of Historical English Grammar,
which is a condensed version of:
Henry Sweet, Short Historical English Grammar, which,
in turn, is a separate publication of part of the historical
material of:
Henry Sweet, Neiv English Grammar (two volumes).
Especially important is the last of these, which con-
and fairly complete account of the phonet-
tains a readable
ic, morphologic, and syntactic development of English

from Old English to the present time.^) This historical

1) The general linguistic and grammatical disquisitions at

the beginning of the book arc not, however, to be recommended
 now TO STUDY LINGUISTICS 315

study of Eiiglisli should be rapid and extensive rather

than intensive, — unless, indeed, one intends to take
English for one's special field, — for it is more important
at this stage to get a general idea of linguistic develop-
ment than to learn the particular historic facts of English.
Simultaneously with the preceding study the general
c)
facts and principles of linguistics should be the subject of a
course of somewhat more intensive reading. If one has
not studied psychology, some modern text of it should
be read. The beginning of linguistics is best made with
one of Whitney's books:
Whitney, Language and the Study of Language,
AY. D.
New York (Scribner) 18G7 (and successive reprints),
W. D. Whitney, Tlie Life and Growth of Language,
New Y^'ork (Appleton) 1875 (and successive reprints).
These books, though today incomplete, are fundamental
works of our science and are, moreover, written in a
style of remarkable clearness and dignity. After this,
one should read, for the principles and methods of modern
hnguistics:
B. Delbriick, Einleitung in das Studlum der indogerma-
nischen SpracJicn, fifth edition, Leipzig 1908,
H. Paul, Prinzipicn der Sprachgeschidite^ fourth edition,
Halle 1909. (An English adaptation of the second, 188G,
edition is Strong, Logeman, and Wheeler, Inlroduciion to

the Study of the History of Language, London 1891),

H. Oertel, Lectures on the Study of Language^ New York
(Scribner) 190^.
The semantic phase of linguistic development is clever-
ly and interestingly, though, unfortunately, from the
standpoint of 'popular' psychology, discussed in
M. Ureal, Essai deScm((nti(jac, fourth edition, Paris 1908.
(An Ejjglish translation of the third, 1897, edition by
Mrs IL Cust appeared in London in 1900).
 L

316 THE STUDY OF LANGUAGE

Later it is advisable, because tbe books so far named

are for the most part not fully modern as to psychologic
interpretation, to study carefully the great linguistic work
of the philosopher and psychologist Wundt:
W. Wundt, VolJcerpsychologie, 1. und 2. Band, Die SprachCy
third edition, Leipzig 191
It is convenient to supplement this with the Indo-
European linguist Delbriick's critique and valuation, which
appeared in answer to the first edition (1900) of Wundt's
book:
B. Delbriick, Grundfragen der Sprachforschungy StraB-
burg 1901,
and with Wundt's answering statement, important as
to the relation of psychology, descriptive linguistics, and
historical linguistics:
W. Wundt, SpracJigeschichte und SpracJipst/chologie,
Leipzig 1901.
A highly suggestive book on the history of language is

Otto Jespersen, Progress in Language ^ with Special

JReference to English^ London 1894 (and reprints).

d) The general aspects of language cannot be under-

stood without at least some acquaintance with divergent
forms of speech. The best aid for this is the clear little

description of eight languages of widely different types

(with an illustrative text of each),
F, N. Finck, Die HaupUypen des Sprachbaues, Leipzig
1910.
A very useful of the languages of the earth,
list

arranged in families, —
though perhaps too optimistic
in the assumption of relationships, —
is another booklet

by the same author:

F. N. Finck, Die Sprachstdmme des ErdJireises, Leipzig
1909
 •HOW TO STUDY LINGUISTICS 317

Brief summaries, valuable for reference, of the gramma-

tical facts of a large part of the languages of the world
are given in
F. M tiller, Gnmdriss der Sprachivissenschaft^ four vol-
umes, Vienna 1876, £f.

The more general, in part the practical aspects of lin-

guistics, are treated in the lively, if not always fully
modern book,
G. von der Gabelentz, Die Sprachwissenschafty second
edition, Leipzig 1901.
The relation of linguistics to ethnology, strangely neg-
lected in all these books, is briefly discussed by Pro-
fessor Boas in the Introduction of the Fortieth Bulletin
of the Bureau of Ethnology of the Smithsonian Institu-
tion, namely:
F. Boas, Ilandhooh of the American Indian Languages^
Parti, Washington 1911.
A good introduction to ethnology, containing an ex-
cellent chapter on language, is;
R. R. Marrett, Anthropology j New York (Holt) and
London [1011].
e) Meanwhile the student will have chosen some lan-
guage or group of languages as his special field of study,
— as, for example, English, German, French, Latin, Greek,

or Sanskrit, or, of groups, the Germanic, the Romance,

or the Slavic languages. If, as is usually with us the
cnse, some Indo-European language or group is chosen,
the study should be accompanied by that of the Indo-
European family in general. There are two excellent
Vjrief compendia of what is known about this group; the

first fuller and more exact, the second better suited to

continuous reading:
K. Brugmann, Ktirze vergleirhcnde Grammatilc der indo'
ger muni ache n Sprachcn, Struliburg 1004.
318 THE STUDY OF LANGUAGE

A. Meillet, Introduction a VEtude Comparative des

Langues Indo-Europeennes^ third edition, Paris 1913.

These books contain ample bibliography, not only of

Indo-European publications, but also of those on the
various groups constituting the family. A fuller account,
with complete bibliography, is

K. Brugmann and B. Delbriick, Gnmdrifi der ver-

gleichenden Grammatih der indogermanischen Sprachen,
first edition, five volumes, StraBburg, 1886 — 1900, second
edition, first three volumes, ibid. —
1897 1911. (For those
parts which have not yet appeared in the second edition
the first must be used; the first two volumes of this have
appeared in an Eoglish translation in four volumes:
J. Wright, R. S. Conway, and W. A. Rouse, Elements of
the Comparative Grammar of the Indo- Germanic Languages,
New York, Westermann, 1888—1895).
If the student's chosen language belongs to any one
of the large branches of Indo-European, he should make
also a study of this branch and of the other languages
in it;— thus, if he is specializing in English, he should
not neglect Germanic linguistics and the study of Frisian,
German (High and Low, including Dutch), Scandinavian,
and Gothic. The nucleus of one's work should be, how-
ever, the intensive study of some one language or group,
based, if possible, on the present speech as heard, as well
as on texts, — for comparative purposes of course on
the oldest, — and on the standard books and articles
about the subject. In this work the student will learn
to understand also the general principles more thorough-
ly than is possible at second hand. In time he will find
gaps in our knowledge or errors in our interpretation
which he will be able to fill out or to correct, if he is
willing to devote himself to a strict adherence to historic
 RELATION OF LINGUISTICS TO OTHER SCIENCES 319

fact in all its detailsand to an inflexible discrimioation

between mere surmise and scientific certainty.
3. Kelation of linguistics to other sciences, a) To
philology. Linguistics, we have seen, took its origin in
philology, —in the study of national culture. The re-
lations between the two sciences are still manifold. The
most original of these relations, the practical one, is ob-
vious: for philologic study thorough knowledge of the
language of the community and of its writings is a nec-
essary instrument. If the community has a long cultural
history, as in the case of France, Germany, or England,
this knowledge must extend to the various historic forms
of the language and will naturally shape itself into a
study of the linguistic history of the nation. The philol-
ogist must not, however, mistaking the means for the
end, confine himself to this linguistic study: if he wishes
to remain philologist, his aim must be the understanding
of the more conscious cultural activities of the nation;
ifhe wishes to go over to linguistics, it will be his duty to
study also the elements and principles of this science and,
to some extent, the linguistics of other nations. The few
who have been successful!}"
scholars active in both philol-
ogy and linguistics made a study of both sciences, —
a twofold task exceeding the abilities of most men; there
has been on the other hand some confusion, beyond that
in name, of the two sciences, usually in the shape of
philologists who neglected the genuine values of their
own science for amateurish but pedantic i)seudo-lin-
guistics.
Aside from the practical relation of linguistics to philol-
ogy, there is an intrinsic connection between them, which,
however, has been overestimated lather than neglected
This connection inheres in the fact that language is the
most elementary cultural activity and bears traces, al-
320 THE STUDY OF LANGUAGE

ways, of tlie more deliberate cultural achievements of a

nation, both in the clearness and flexibility of its syn-
tactic and stylistic forms and in the vocabulary. The
cultural features of the latter are revealed, of course, by
etymology, in the study of semantic changes and word-
borrowings.
b) To literary history and criticism (German Literatur-
tvissenschaft). The science of literary history, recently
also named (in my opinion, misnamed) 'comparative liter-
ature',has like linguistics, grown out of philology, with
the aim of studying not the cultural achievements of this
or that nation, but the development of literature (story-
telling, poetry,drama, and so on) among nations, groups
of nations (such as western Europe), and among mankind
universally. As the instrument of literature is language,
the student of literature needs a general, if elementary
knowledge of the nature and development of language;
as, on the other hand, the use of language in literature

is a powerful factor in the history of the former, the lin-

guist must often consult the student of letters.^)

To history. Since language changes, in time, its his-

c)

tory is part of that of the speaking community. This is

true most evidently of the external history of language, —
of its differentiation into dialects, its uniformization by
a standard form of speech, its spread over tributary
peoples, or of a nation's adoption of an alien language

1) Utterly unscientificthe notion that linguistics is in some

is
way an illegitimate rival of the study of literature, and that
any and all linguistic students ought properly to transfer their
activity to the latter field. This notion is an ofi^shoot of the
idea that only professional study of literature enables one to
love or understand it. As a matter of fact, linguistic scholars,
owing to their contact with texts of various languages, are often
fair connoisseurs of literature
 RELATION OF LINGUISTICS TO OTHER SCIENCES 321

under conquest. It is no less true, however, of the internal

history, where every change of language is of course real-
ly a historic event.
Nevertheless, historians have not, as a rule, included
linguistic history in their studies or treatises. The ex-
ceptions are twofold. Wherever there has been a con-
scious linguistic activity, especially in the formation or
spread of a standard or literary language, history has
taken notice. Thus, of the history of the English lan-
guage, the rise of London English during the Chaucerian
period, of the history of German, the origin of Standard
German in the imperial offices and in Luther's Bible-
translation, have alone been included in the histories of
these nations. While this exclusion of most linguistic
history has been tacitly made, it can be justified, if one
limits history to those events in which deliberate individ-
ual action has demonstrably or presumably played a
part. The second phase of linguistic development which
has been included in history is the testimony of vocabu-
lary to material surroundings and events; thus the stratum
of Romance loan-words introduced into English after
the Norman conquest finds mention in histories of the
English people. Li so far as the data so furnished by
language come directly from historic periods, they are
useful in cultural history (German Kulfurgeschi elite)] in
so far as they are derived, by the comparative method,
from prehistoric times, they are of moment, though —
the methods of interpretation are not yet certain, in —
pre-history (German Urgesehiehte).
To etJinology. Language is the most purely com-
d)
munal of human activities, —
the one least amenable to
modification by individuals and least obscured by the
secondary rationalizing explanations familiar to ethnol-
ogists. The unconscious communal grouping of ideas
Bloomfleld, Study of Language 21
322 THE STl DY OF LANGUAGE

(formation of categories) takes place nowhere so freely

as in lan<^ung*e. This is true not only of the grammatical
groupings and those implied in the vocabulary, for —
every word involves a classification of experiences (p. 63),
— but even of the sonnd-system, which represents a
communal selection of a limited number of places of artic-
ulation and manners of articulation from among a pos-'
sible infinite variety (p. 53).
Thus the language of any single community at a given
time is an important part of the ethnologic data concern-
ing it. This does not mean, however, that linguistics is
part of ethnology, for it is only the descriptive data
which the two sciences have in common. The linguist
can collect these data and proceed to the interpretation
of their origin and relation to other forms of speech, or
at least to their insertion into a general scheme of lin-
guistic development and distribution; their comparison
with the other ethnic data, such as those of religion,
myth, and custom, with view to a characterization of the
community, must be left to the ethnologist.
e) To psychology. The relation of linguistics to psy-
chology on the one hand, implied in the basic position
is,

of the latter among the mental sciences. These sciences,

studying the various activities of man, demand in differ-
ing degrees but none the less universally, a constant
psychologic interpretation. Perhaps this is but negatively
true: perhaps the student of a mental science could and
ideally should refrain from any running psychologic inter-
pretation; in practice, how^ever, such interpretation is un-
avoidable. In describing an analogic or semantic change,
for instance, linguists most usually outline the conditions
of mental predisposition which brought it about. If they
do not do this in terms of scientific psychology, they
will resort to rationalizing 'popular psychology', — to
 RELATION OF LINGUISTICS TO OTHER SCIENCES 323

such explanations as that the new form was desired for

crreater 'clearness' or 'convenience'. As lancruao-e is in
its forms the least deliberate of human activities, the one
in which rationalizing explanations are most grossly out
of place, linguistics is, of all the mental sciences, most
in need of guidance at every step by the best psychologic
insight available.
On the other hand, psychology makes a wide use of
the results of linguistics. Modern psychology recognizes
two sources of information. The one is introspective
.

analysis under the control of mechanical (experimental)

devices which record the physical correlates of the men-
tal process. The information so obtained applies to the
activity of the individual human mind. This activity is

always conditioned, however, in varying degrees, on past

experiences which in themselves are products of mental
action of other individuals. Thus, when one speaks a
sentence, the form it takes is duo to the utterances which
the speaker, since infancy, has heard from the other
members of his community. It is due, in other words,
to a series of connected mental processes extending in-
definitely back into time and occurring in an indefinite
number of individuals. Such mental processes, then, as
those involved in the utterance of speech cannot find
their explanation in the individual, — he receives his
speech-habits from others, — but must be traced for ex-
planation from individual to individual ad infinitum.
They are products of the mental action not of a single
person, but of a community of individuals. These prod-
ucts, —not only hmguage but also myth, art, and
custom, —
are the data uliich mnke possible the second
phase of psychology, social psychohjgy, (German VoJhcr-
psycholofjte). As language, moreover, is less subject than
these other activities to individual deliberate actions
21*
324 THE STUDY OF LANGUAGE

whicli interfere with the commnnal nexus, it is the most

important domain in the study of social psychology.
f) To philosophy I shall not presume to enter here
upon the epistemologic problems in which linguistic con-
siderations must play a part. Far more of our experience
than one generally assumes is shaped by the linguistic
habits in which we live. The apparatus of logic, more
especially, depends upon the language we speak: the
logical forms, in other words, must develop historically
with the langunge. Not only our more abstract concepts,
but also those of qualities and actions are due to lin-
guistic forms, or rather, are the subjective phase of lin-
guistic forms, which have been evolved in the course of
time. Much of our philosophy, in consequence, moves
captive in the plane of its authors' language, which it

should, for freedom, transcend, — as it can only through

the study of language.
To come to a simpler matter, the development of lan-
guage occupies a peculiar and interesting position in the
universal growth of things which philosophy essays to
study. Faster than biologic evolution, so fast indeed that
a change like the one from tbe Indo-European parent lan-
guage to modern English takes place, as it were, under
our very eyes, yet incomparably slower and. more unifi-
able, to our comprehension, than the historic change in
other human activities, linguistic development may rep-
resent to us a type of progress intermediate between
these
The unfolding of the unconscious into consciousness
takes place nowhere to our direct knowledge so clearly
as in the activity of speech. In the spread of single
languages over whole continents, and in the more con-
scious shaping of these languages, lies the beginning of
a growing rationalization of speech. If movements such
 RELATION OP LINGUISTICS TO OTHER SCIENCES 325

as that for an artificial world-language, which would

seem to be here in the current of natural progress, have
met with failure, this is because they have been but
superficially rational and for the most part mere distor-
tions of the languages we have unconsciously developed.
Linguistic science has not come to a point where the
artificial creation or preservation of a la?.guage is pos-
sible or even conceivable. Nevertheless, such misplaced
attempts throw light upon the growing consciousness in
the domain of communicative activity. One need think
only of international signals, numerals, the division of
time, the metric system, and the like, to see the increas-
ing amenability of this domain to purposeful modification.
It is in this development, — in such phases of it as the
teaching of reading and writing and of standard languages
and foreiom languao-es in schools, in the treatment of the
deaf and dumb, in stenography, in the preparation of
international means of communication, —
that linguistic
science finds more and more its active part in human
progress. In short, linguistic science is a step in the self-
realization of maxL
 INDICES.
The numbers refer to pages.
Words in brackets are to be taken as cross-references.

1. AUTHORS, etc.

Aasen 290 von der Gabelentz 317

Ahn 309 Goethe 289
Apollonios Dyskolos 307 Greenouffh 314
Ascham 308 Grimm 208, 311.

Bahlsen 306 Van Helmont 237

Bernhardi 309 Herder 14
Bible, Kincr James translation Herodotus 13
289 ; Luther's translation 290, Homer 292
321 Horace 181
Boas 317 von Humboldt 311, 312.
Hopp 310
breal 172, 316 International Phonetic Associa
de Drosses 309 tion 23.
Briicke 310
Jespersen 305, 313, 314, 316
•Brugmann 317, 318.
Jones 309, 310.
Tarroll 237
Karadjic 290
Chaucer 59, 195, 289, 321
Kittredge 314
Cicero 289
Koran 292.
Colebrooke 309
Comenius 308. Lloyd 313
Luther 296, 821.
Dante 289
Delbriick 315, 316, 318 Marrett 317
Diez 311 Meillet 318
Dionysioa Thrax 307 von Aliklobicb 311
Donatus 307. Muller 317.

K.lkins 19 Oertel 315

Kpicureana 14. Olleudorf 309.

Finck 316 Panini 307

Fulda 309. Pa.sBy 99,318
328 INDICES

Paul 316 Steinthal 312

Pott 310 Sterne 249
Priscian 307. Stoics 14
Sweet 313, 314.
Ratichius 308
Eousseau 14. Vedas 292
Scherer 310 Verner 278.
von Schlegel 310
"Whitney V, 308, 312, 816
Schleicher 310
Wordsworth 247
Schottelius 309
Shakspere 195, 249, 289
Wundt VI, 316.

Sievers 313 Zeufi 311.

2. LANGUAGES.
English, mentioned on almost every page, is not here inclndcd;
see Table of Contents and of. also West-Germanic, Germanic,
and Indo-European.
Albanese: relationship 270, 272 Avestan: relationship 272, his-
(Indo-European) tory 272, 276, 277, 278, study
Algonquian lauguages (Mes- 292 (Iranian)
quaki) 171, 312 Aztec =Nahwatl.
Altaic languages (Tartar, Tur-
kish) 311 Baltic languages (Lettish, Lithu-
American Indians gesture-lan-
: anian, Prussian): qualities
guage 4, 6; languages of with object 106, relationship
(Algonquian, Athapascan, Ca- 270, 272, f., decrease 265 (In-
ribbean,Chinook j argon,Green- do-European)
landish, Lule, Nahv,atl, Tsim- Bantu languages (Kafir, Subiya)
shian): objectivity 63, f., gen- 312 genders 109, 143, number
;

ders 109, diversity 262, de- and person as gender 143,

crease 262, 264, loan-words in congruence 153, 182
English 282, study 19, 312; Basque: numbers 90
picture-writing 7 Bohemian =
Cechish
Arabic: sounds 24, 33, 64, pos- Bulgarian: relationship 270,
sessor with object 107, 149, history 225, 278 (Slavic)
loan-words through A. into Burmese: relationship 312 (Indo-
English 282, literary language Chinese)
290, study 293, 309, (Semitic) Bushman (Kham) 90 sounds 27.
:

Armenian: sounds 36, 40, rela-

tionship 270, 272, history 276, Canarese: sounds 54 (Dravidian)
277 (Indo-European) Caribbean languages (Arowak)
Arowak numbers 89 (Caribbean)
: 89: loan-words in English
Aryan = Indo-Iranian 282, 283
Athapascan languages 312: dis- Caucasian languages (Georgian)
1

tribution 266, f. j
312
 INDICES 329
Cechisli: Bonnds 26, 29, word- 244, 248, 255—8, 274, 287,
stress 49, 101, relationship spread 262, 264, standard lan-
270, history 215 (Slavic) guage 264, 289, f., relation-
Celtic languages (Irish) emotion-
: ship 266, loan-words in English
al relations 171, relation- 212, 225, 281—4, 287 study
ship 270, 272, decrease 265, VI, 300, 305, 317, 319 (Ro-
study 311 (Indo-European) mance)
Chinese: sounds 24, 51, 54, 55, Frisian: history 267, 274, f., re-
writing 22, words 85, f., 98, lationship 266—9, study 318
word-form 93, 101, derivation (West-Germanic).
152, 168, homonymy 207, com-
pounds 97, 161, 189, parts of Georgian: sounds 40, sentence
speech 126, f., 128 —not as 110, 173, f., relationship 312.
in English 112, f., 126, sen- (Caucasian)
tence-stress 53, sentence-pitch German sounds 19, 24, 28—40, 61
:

177, word-order 113, 115—7, —3, 55, 195, 210, 219, writing
119, 188, congruence 130, f., 22, words 49, 75, 81, f., 86—
tense 68, f., 144, number 108, 9, 162, 164, derivation 106,
142, interrogation 92, dialects 207, genders 109, 129, f., 142,
22, relationship 312, literary f, 151, inflection 87, 93, 129,
f., 143, f., 147, f., 153, f., 166,
language 22, 292, study 19,
292 (Indo-Chinese) 180, 184, 186, sentence 48, 98,
Chinook jargon 262 173, 191, 193, f., history 208,
Cistercian monks' gesture-lan- f., 213, f., 216, f., 230, 232—

guage 6. 5, 242, 244, f., 249, f., 274, f.,

277, f, 281—6, loans from
Danish: sounds 29, f., 33, 40, Latin 216, f., 281, 283, loan-
genders 109, relationship 268, words in English 70, 281, re-
influence of Latin 282 (Scan- lationship 264— 7,269,8tandard
dinavian) language 265, 289, f, 321,
Dayak: 157 (Malayan) study 301, 303—5, 317, 319
Dravidian languages (Canarese) (West-Germanic)
312: influence upon Indie 220 Germanic lauL-uages (Gothic,
(of. 192) Scandinavian, West-Germanic)
Dutch 237: sounds 28, relation- 269; history 201, 206, 208,
ship 266, f., 269, history 230, 211, 214—6, 218, 221. 229 f,
standard language 265, study 234, 272, 276—9, 285, rela-
318 (West Germanic). tionship 269, f., study 311,
317, f. (Indo-European)
Finnish: cases 107, f., 144, re- Gothic: history 272, 275—9, re-
lationship 311 (IJralic) lationship 269, study VI, 218.
French: sounds 27—40, 44, f, (Germanic)
writing 22, 300, words 87—90, Greek: sounds 82, 47, 61, 152,
163, liaison 99—102,257, gen- writing 20, derivation 152,
ders 109, sentence 4H, 53, 99, compoHition 164, inflection 92,
171, 173, 175, 257, f, history lO'J, 116, f, 142, 145—8, 156
90, 214, 226, 232, 235, 241, f., — 7, lGi,8enlLuce 116, f., 184,
 —

330 INDICES

history 116, f., 217, f., 230, 128, f., 131, 151, 257, sentence
243, 265 272, 276—9, loan-
f., 175, 256, f., history 272, 276
words from G. 237, 282—4, —8 Celtic)
relationship 261, f 270, lite-
, Italian: sounds 29, 31, f. , 45, 53,
rary languages 263, 289, f., 92, derivation 105, 165, verb
study 292, 307— 10,. 317 (In- 107, pronoun 88, genders 132,
do-European) history 214, f, 225, 227, 274,
Greenlandish: sounds 33, 54, in- relationship 266, literary lan-
. flection 107, 110, f., 135, 149, guage 289 (Romance) •

f., 174, objectivity 104, f„ languages (Latin) 270, 272,

Italic
sentence 110, f., 179, 190, f. 285 (Indo-European).
Hamitic languages 311 Japanese 20, 48, 70, 88, 167
Hebrew: sounds 24, loan-words Javanese, see Kavi.
through H. into English 282,
literary language 290, study Kafir153—5, 182 (Bantu)
292, 307, 309 (Semitic) Kavi 311 (Malayan)
Hindustani 262 (Indie) Kham 90 (Bushman).
Hottentot 27 Latin (for modern development
Hungarian 31, 311 (Uralic). see Romance): writing 20, f.,
Icelandic: word-stress 49, 101 inflection 135, 154 7, tenses
relationship 268, history 275, 68, 144, voices 115, 145, 173,
277, f. (Scandinavian) genders 109, cases 92, 108,
India, languages of: sounds 28, 115, f., 144, 167, 185-7, pro-
30, 54, 256, writing 20, sen- nouns 88, f,, 118, 176, adjec-
tence 192; see Dravidian and tive 106, sentence 68, f, 98,
Indie f., 107, 111, f., 118, 148, 162,
Indians see Americans Indians, 168, f., 172, 176, 179, 192,
Indie languages (Hindustani, 194, 253, congruence 181,
Sanskrit) 219, f., 270 (Indo- word-order 171, 186, f., history
Iranian) 212—5, 217, f., 225, 230, 232,
Indo-Chinese languages (Chinese) —
241 4, 248, 255, f., 272, 274,
312 276—8, 283, 287, loans from
Indo-European languages (Al- other Italic languages 285,
banese, Armenian, Baltic, Cel- loans to En^>lish 106, 212, 281,
tic, Germanic, Greek, Indo- 284, 287, to German 281, 283,
Iranian, Italic, Slavic, Tocba- influence or other languages
ric) 269—73, 276— 80, sentence 282, relationship 270, spread
172, Primitive I.-E. 106, 201, 257, 262—4, 266, f., 289, lite-
225, 229, 234, 254, 256, f., rary language 289, f., study
study 310—2, 317, f. 292. 307—10, 317 (Italic)
Indo-Iranian languages (Indie, Lettish 24, 270 (Baltic)
Iranian) 270, 277, f. (Indo-Eu- Lithuanian: sounds 31, 47, 51,
ropean) derivation 106, relationship
Iranian languages ( A vestan) 270, 270, history 272, f., 277, f
272 (Indo-Iranian). (Baltic)
Irish: sound-variation 102, f., Lule 150, 174.
 INDICES 331
Malay 85, f., 86, 88, 93, 98, 156, 215, influence of other lan-
f., 261,(Malayan)
f. guages 282, relationship 270,
Malayanlanguaj^e8(Dayak,Kavi, spread 262, literary language
Malay) 132, 157. Malajo- Po- 290 (Slavic).
lynesian)
Malayo-Polynesian languages Sanskrit: sounds 26, 51, 64,
(Malayan, Polynesian) 312 sandhi 102, 128, inflection 155,
Mesquaki 171 (Algonqu an). numbers 142, cases 144, 167,
184, voicesl45,f., conjugations
Xahwatl : inflection 135, 146, 145, f., reduplication 156, f.,
149, 157, comj^ounds 160, 164, compounds 106, 160, f., 164,
f., 254, sentence 98, 167, 169,
sentence 192, 194, relation-
179, 253 ship 230, history 218, 220, 230,
Norlh Germanic Scandinavian = 272, 276—9, literary language
Norwegian: sounds 31, 37, f., 290, study 292, 307, 309, f.,
51, 53, 92, f., 152, 164,
100, 317 (Indie)
177, pronouns 89, genders 109, Scandinavian languages (Danish,
derivation 152, composition Icelandic, Norwegian, Swedish)
164, sentence 173, 177, rela- 143, 173, 268, f., relationship
tionship 261, 268, influence of 269, history 230, 275, loans
Latin 282, literary languages to English 285, study 318
290 (Scandinavian). (Germanic)
Semitic languages (Arabic, He-
Oscan 270 (Italic). brew) 33, 107, 133, 311
Servian 270, 290 (Slavic)
Polish: sounds 31, f., 49, 101, Siamese 312 (Indo-Chinese)
pronouns 88, history 215,- re- Slavic languages fBulgarian,
lationship 270 (Slavic) Cech, Polish, Russian, Ser-
Polynesian languages 54, f. (Ma- vian): sounds 29, f., 32, 40, f.,
layo-Polynesian) 44, f., 47, 53, f., pronoun 89,
Portuguese 264, 266, 282 (Ro- genders 142, manner 145, de-
mance). rivation 106, sentence 92,
history 215, 218, 225, 227, 265,
Romance languages (French, 272, f., influence of either lan-
Ital an, Portuguese, Rouma- guages 282, relationship 270,
nian, Spanish;: sounds 40, 44, study 220, 311, 317 (Indo-
f, 47, 53, f., derivation 106, European)
248, genders 143, relation- Spanish: sounds 28, 31, f, his-
ship 262, f., 266, f., history tory 248, 274, loans to Eng-
265, f., 274, spread 264, study lish 282, f, relationship 266,
220, 311, 317, 321 (See Latin) spread 262, 264, st.indard
Roumanian 26<) (Romance) language 264 (Homance)
Russian: sounds 31, 86, 38, 40, Subiya 143, 182 (Bantu)
f., 45, 48,61, 65, 152, pro-
f., Swedish: sounds 30, 37, 61, 63,
noun 88, adjective 111,
66. 100. f, 162, relationship 268,
185, manner 145. sen ten ee 1)2, influence of Latin 282 (Scan-
111, 17^-4, history 212, f.. dinavian).
 9 3

332 INDICES

Tartar 311 (Altaic) Uralic languages (Finnish, Hun-

Tibetan 312 (Indo-Chinese) garian) 107, 144, 311 (Ural-
Tocharic 272 (Indo-European) Altaic).
Tsimshian 151, 157, 253
Turkish: sounds 36, inflection West - Germanic languages
95, f., sentence 192, relation- (Dutch, English, Frisian, Ger-
ship 311 (Altaic). man) 200, 266, f., relation-
ship 268, f., history 229, 274,
Umbrian 270 (Italic)
f.,279, 281, 285 (Germanic)
Ural-Altaic languages (Altaic, Wolof 174.
Uralic) 311

8. SUBJECTS.
Ablaut 153, 229 82, 120, 133, f., 139—41, 197,
abnormal sibilants 31 219, 221—51
absolutive 178, f., 254 attribute, attribution 61, 110, f.,
abstract words 65, f. 122, 149, f.
action-words 65, f. attributive languages, see objec-
actor and action categories 67, tive
f., 112, 115, 121, 148, 172-5 automatic sound -variation 23,
adaptation 225, f. 54, f., 151, 155, f., 220, f., 250.
adjective 122, f,
adverb 123, f. Back vowels 34
affix 153—6 bilabials 28
alphabet 20 blade 30, f.
alveolars 28 breath 9, 24, 26
analogic change 59, f, 196, 221 breathed, see unvoiced.
—37
analysis of experience 59 — 63, Cartilage glottis 26
85—90, 142, 237, f. case 107, f., 122, 143, f., 183, f
anaphoric words 89 categories 67 —
animals 56, f. cerebrals 30
aphasia 67 change 16, f, 195—258
apperception 57, f., 60, f. child 10—3, 223
article 117, f., 175, f. choke 40; cf glottal stop
articulation 19—55, 195, f., 299, f. close syllable stress 47
arytenoids 24, 26 command 76, 121
aspect, see maimer comparative method 200, f., 274
aspirate initial 26, 33, 40; a. — 80
stop 40, 53, f. compound syllable pitch 51, f;
assimilation 59, f., 196, f., 219, c. 8- stress 47, 52
221—51, 283, f. compound words 96 — 8, 104, lOG,
assimilation of articulations 214 140, 159—66, 235, f., 254
—6 concept 58, 63, 65, f., 85—7
association 57, f., 66, f., 69, f., condensation 241 —
 6 6

INDICES 333
corgrnence 127—81, 180—2 Geminate, see doubled
conjunction 124, 193, f. gender 109, 129, f.,-142, f, 182
—
consonants 28 33, 163 gestures 4 — 7, 14, f.
contamination 224, f. glides 40, f.

coronals 28—30 glottal stop 24, f., 33, 40

cross-reference 178 80. — 24—6, 40
glottis
government 182 —
Deaf-mutes 5 grammar 289, 302—4, 307—9
declarative, see statement Grimm's law 208
definite and indefinite categories group stress 48 — 50
117, f., 175, f.
gums 28—30
deictic words 64, f., 88, f,
Haplology 217
dentals 28—31
high vowels 34
derivation 141, 150, £ homonymy 125, 157, f., 181, 186,
dialects 260, f.
206, f.
diminutive 105, f.
hypotaxis 191, 193, f.
diphthong 43
discursive relations 60 — 2, 110
Imperative 92, 147, f.
—4, 1G9— 70 inclusive 88
dissimiliation 216—8, 283
infinitive 118, f., 121
dominant element of experience
infix 155, f.
7, 58, 63, 83, f., 238—51; d. e.
inflection 140—50
in sentence 50, f., 113, f., 170,
intensity, see stress
f., 175.
interdental 28
dorsals 28, 30—5
doubled 45, 52, t
interjection 73 7, 124 —
interrogative, see question
duration 52, f.
invention of words 12, 236, t

Emotional relations in sentence Kernel 153, f.

49, f., 113, 170, f.; of. domi-
nant element Labials 28
enclitics 49, 100 labialized 41
ethnology 256, 317, 321, f. labiodentals 28
etymology 244, f., 281, f. languages 261, f.
evolution of language 252 —
laryngeals 24, f., 83
exclamation 70, 73—7, 91, f., 121 larynx 24, 33
exclusive 88 laterals 29, f., 43
exocentric compounds 161 lenes 39, 63, f.
explosives, see stops length, see duration
expressive movements 1 10. — liaison, see sandhi
lips 28, 31, 34, 41
Formational elements 62, f, 79, literary languages 290, 292
f, 93-6, 103—9, 221—37 loan-words 70, lOG, l.i'i, 280—6
fortes 39, 53, f. local relations 107, f, 116, f
fricatives, see spiranti logical, see discursive
&ODt vowels 84. loose, see wide
 9 6 6 6 7

334 INDICES

loudness, see stress onomatopoeia 81

low vowels '34 open syllable stress 47
lungs 9, 24. oral articulation 27 38 —
origin of language 13 —
Manner of action 144, f. orthography, see writing
manner of articulation 54, 208 outcry 9, 73.
material relations in sentence
114—9, 171—4 Palatals 31, f.

meaning, see association, seman- palatalization 214, f.

tic change palatalized articulation 31, 41
metaphor 247 — palate 27—31
metathesis 216, f., 283 parallel forms 141
mid vowels 34 parataxis 191, 193, f.
mixed vowels 34 participle, see verljal adjective
mixture of articulations 41 parts of speech 112, 120 —
mixture of dialects 284, f. person 121, 148, f.
mode 146—8 philology 807, f., 319, f.
morphologic categories 68 — 70, phonetic alphabet '23, 300
103, 141—50 phonetic change, ph. law, see
morphologic classes 108 — 10,120 sound-change
—40, 221—37 phonetic -semantic parallelism
morphologic sound-variation 151 136, f.

—8 phonetic -semantic word -classes

morphology 110, 120—66, 221 131—5 '

—37 phonetics 19, 299, 313

murmur 25, f.
picture-writing 7, 20
musical sound 27. pitch 25, 51, f, 151, f., 177, f.
place of articulation 54, 208
Names 247, f. plain stops, see pure
narrow vowels 84 post-dentals 28—30
nasals "27, f., 43 predicate, predication 61, 64,
nasalized 27, f., 38 110, f, 12 L
natural syllables 42 — prefix 155 —
noise 27 preposition 117, 124, 143
nominal languages, sep objective primitive creations 235
non-syllabics 42, f. proclitics 49, 100
noun 111, f., 121, f., 136—8 pronoun 64, f, 87—9, 123, 143
number 121, f., 141, f., 148, f. pronunciation, see articulation
numerals 89, f. proportional analogy 226, f.
numeratives 130, f. psychology VI, 14, 222—4
nursery words 11, f. pun 99, f.
pure 40 initial
Object affected 115, f, 121, 127, pure stop 40, 53, f.
143, 148, f., 176, 187
objective languages 64, 104, f., Quality-words 65, f.
107, 111 quantity, see duration
object- words 63, f. question 52, 71, 91, £
 5 6 4 7

INDICES 335

Reduplication 156, f. •
syllable-pitch 51, f.
relation 66, 105, 107, f. syllable-stress 43 —
relative pronoun 193, f, syntactic categories 68, f, 112,
root 154 115, 117, f., 121, 174—6
rounded 31, 34, 41. syntax 119, 167—94.

Sandhi 101, f. Teeth 29—32

semantic change 7, 16, 78, 237 tense 62, 68—70, 121, 144, 183
—51 tense vowels, see narrow
semantic parallelism 139 tone-color 27—29, f.
semi-vowels 42 tongue 27—38
sentence 48—53, 60—3, 76, 110 total experience 56 — 63
—9, 167—94 transitive verbs 176; t. words 127
sentence-equivalents 170 translation 300, f.

sentence-pitch 51, f., 92, 176—8 trills 29, 33, 43

sentence-word 64, 111, f., 253 triphthong 43.
serial relation 113, 124
set phrase 116, f., 122, 124, 188, Umlaut 152, 230, f
f, 248, f., 286, f. unrounded 34
sibilants 30, f unvoiced 25
soft palate 32; cf. velum uvula 27, 33
song 10 uvulars 32, f.
sonority 42 —
sound 8, f., 14 Velars 32
sound-change 16, 77, 202—21 velum 26, f.
sound-symbolism 79, f., 93, 235, f. verb 111, f., 121, 138, f.
sound- variation, see automatic, verbal adjectire 122, 191, f.
morphologic verbal noun 122
spelling, see writing Verner's law 216, 229
spirants 27, f. vocal chords 9, f., 24—6, 33
standi ard languages 288 90. —
vocative 92, 144
statement 71, 91, f. voice 9, 25
stops 27, f., 39, f. voiced 25
stress 25, 42—53, 152, 177, f. voices of verb 145, f.
stress-group 48 50 — vowels 27, 33—8, 152, f.
stress-syllable 44—8, 52, f.

subject 61, 110, f., 115, 121 ^Vliisper 26

subordinate clause 124, 190 — wide vowels 34
substitution of" sounds 219, f. word 48—51, 62—70, 82—90,
suction-sounds 27 93—110, 120-66, 221—51
suffix 154 word-order 92, 114—8, 127, f.,
supplction 158, C 1S5-8, 193, f.
syllabaries 20 word-pitch 51
yllabics 42—7 —
word-stress 48 50
syllalde 42—53 writing 7, f., 19—24, 287, I
yllable-boundary 48 —
isH
43

-p
o
•p I

o
•p 4
o
-g

STUUItS
PONTIFICAL INSTITUTE OK MEDIAEVAL
59 QUEEN'S PARKCRESCENT
T0R0NT0~5, CANADA
?ai73-