WHY
ENGLISH
SOUNDS
CHANGE
AIKEN
820.5
729
RONALD
A 1,020,944
PAAS
MADUR
S

***
TAMALLEETSİMULA: Saša
11817
SCIENTIA
ARTES
OF THE
VERITAS
UNIVERSITY OF MICHIGAN
LIBRARY
UMIBU:
PLUM
olet
**NOTETAA
Paraparauma jam salarygo* parket MsAP
mig at kontrar al die wat's allhane
TUEBOR
SIQUE AIS-PENINSULAM AMŒNAM”
FIRCUMSPICE
VAWAWAFANYA
TYA
BILUTE
RPADLARTELENTK
roy the Face Bangkok.
LASTA #Sugarciënekst:
WHY ENGLISH

V
SOUNDS CHANGE
By
JANET (RANKIN) AIKEN, PH.D.
LECTURER IN ENGLISH, COLUMBIA UNIVERSITY
друг
Ⓡ
THE RONALD PRESS COMPANY
NEW YORK
!
•
;
Copyright, 1929, by
THE RONALD PRESS COMPANY
All Rights Reserved
1
PREFATORY NOTE
The nature of this study has precluded extensive
use of authorities. A few standard works on his-
torical English have been quoted, in order to cite a
prevailing, or more or less accepted, statement of
linguistic fact or theory. The books so quoted are:
Jespersen's "Language" and "Philosophy of Gram-
mar," Krapp's "Pronunciation of Standard English
in America," Wright's "Old English Grammar,
Bloomfield's "Study of Language," and Wyld's
"Short History of English.'
19
Thanks are due to Professors William Witherle
Lawrence, Ashley H. Thorndike, A. V. Williams
Jackson, and Louis Herbert Gray, of Columbia
University, for reading and constructive criticism
of the manuscript. Helpful suggestions have been
made by Professor Harry Morgan Ayres. A greater
debt of gratitude is due Professor George Philip
Krapp, who not only has contributed largely to
the theories themselves, but has most patiently and
inspiringly supervised this book during the five years
of its working out.
JANET RANKIN AIKEN
West Redding, Connecticut,
September 3, 1929.
99
iii
CONTENTS
CONSONANTS AND VOWELS
ACCORD
Are They Generically Distinct?
Basis for Vowel-Consonant Correlation
Speech Sounds in Detail
General Conclusions
CHAPTER I
Definition of Accord
Approximation Accord
Ablation Accord
Mutation Accord
Recurrence Accord
CHAPTER II
Accord in Other Fields
Other Principles Operative
•
CHAPTER III
OTHER FACTORS IN SOUND CHANGE
The Modulation of Sounds
The Principle of Grading
Types of Grading in English
Tensity or Stress, not Quantity
Types of Stress Grading
The Principle of Conservation
Spelling Influence
The Intellectual Element
Summary .
•
•
•
3
6
10
18
22235
29
33
35
38
40
H*4ROBNON
41
43
45
49
50
55
57
60
62
PAGE
3
22
41
...
V
vi
CONTENTS
CHAPTER IV
ACCORD IN ENGLISH CONSONANTS
Few Instances Analyzed
Elimination of Gutturals
Other Languages Compared
A Theory of Linguistic Development
Palatalization of [k]
Satem and Centum
OE [g] Before Front Vowels
Medial [S] and [ž]
General Conclusions
CHAPTER V
ACCORD IN ENGLISH Vowels
Theories of Sound Change
Why Have Tense Vowels Become Raised?
Slack Vowels Not Raised
•
The Influence of [r]
Other Consonant Ablations.
Other Exceptions Due to Contiguous Consonant
Vowel Tensings
Slackenings in Vowels
Diphthongization
General Conclusions
•
•
CHAPTER VI
GENERAL CONSIDERATIONS
Newness of Accord
Immediate Causes of Sound Change
The Revival of Learning
Change Also Continuous
Time and Place Not Important
More Than One Principle of Change
Theory Re-stated
Comments on General System
•
·
•
•
•
***INNI??
76
80
82
85
90
93
96
99
100
102
104
106
109
111
113
116
117
119
122
123
125
PAGE
65
85
111
CONTENTS
vii
CHAPTER VII
ACCORD OF IDEAS
Need for a Simplified Terminology
Accord in Morphology
Accord in Syntax
Accord in Semantics
APPENDIX: A FEW RULES STATED
TABLES
•
•
127
131
135
138
I. Speech Sounds Classified According to Active Por-
tion of Tongue
II. Occurrence of Consonantal Sounds in English
III. Approximation Accord
IV. Ablation Accord
V. Factors in Sound Change
VI. Tense Vowel Changes
VII. Frequency of Consonant Occurrence
PAGE
127
141
9
12
26
31
282383
63
91
WHY ENGLISH
SOUNDS CHANGE
I
CHAPTER I
CONSONANTS AND VOWELS
Are They Generically Distinct?
To the remote inventors of the hieroglyphs, vowel
sounds probably mattered little. At least they were
not considered worth the trouble of recording. There
are no hieroglyphic vowels, nor does the vowel appear
in the early Semitic alphabets, which are supposedly
related in some way to the Egyptian. Sanskrit has
few symbols originally denoting vowels, and even
today we use an alphabet so meagerly furnished with
vowel symbols that each of the five we possess must
be made to carry an enormous orthographic load.
Today, however, the pendulum has swung, and the
vowel has become the pride and joy of linguists, the
consonant appearing comparatively uninteresting,
or at least comparatively unregarded.
The absence or paucity of vowel symbols in early
alphabets by no means indicates that vowel sounds
were absent or few in number. Contrary to the con-
clusions of the older school of linguists, who assigned
to Sanskrit, early Semitic, etc., only three vowel
sounds in all, it is now generally agreed that early
languages had their full proportion of vowels, prob-
ably as many as we have today. True, we cannot
reconstruct the systems, for lack of conclusive evi-
3
4
WHY ENGLISH SOUNDS CHANGE
dence; but we can be reasonably certain that a
linguistic enthusiast set down among the Akkadians
or the Sumerians would have found plenty of vowel
distinctions to analyze.
It is not difficult to understand how vowels might
easily be omitted in early alphabets. The consonant
pattern, generally speaking, is the word. G--rg-
W-sh-ngt-n w-s- gr--t m-n is simplicity itself to
read, while -eo-e -a--i---o- -a- a --ea- -a- could
hardly be read without further clue. The child
spelling by ear will seldom leave out a sounded
consonant, whereas vowels are frequently omitted,
e.g., fnish, dinr, apl, RNSR (answer; A. A. Milne,
Winnie-the-Pooh, page 46). The vowel, by the very
nature of its articulation, is a weak, "open" tone,
and acts largely as an accompaniment to the stronger
consonant.
At least so much must be said for the customary
opposition of vowels and consonants in current lin-
guistic writing, an opposition so hallowed by tradi-
tion as to be practiced automatically in every study
in this field. And yet a further question may be
asked. Granting that vowels are weaker than con-
sonants, that the distinction between them is fre-
quently useful and has been apparent in linguistic
practice since the very dawn of written language,
yet may vowels and consonants properly be treated
invariably as two distinct genera? Is the distinction
between them one of kind, or one of degree? And if
the latter, may not linguistics as a science have lost
CONSONANTS AND VOWELS
5
something by its invariable separation and opposi-
tion of the two?
Consonants and vowels are frequently differen-
tiated by the statement that vowels are the more
open, less obstructed impulses of the breath; and as
between [p], perhaps the most obviously obstructed
of English speech sounds, and [a], the most open
vowel, this distinction is easily apparent. Only one
sound, [h], is as open as [a]. Yet, oddly enough, [h] is
numbered among the consonants; and there we are,
with two excessively open sounds, one a vowel, the
other not a vowel.
Again, it is asserted that only a vowel can carry a
syllable; and yet phoneticians recognize syllabic n,
chicken, l, little, and r, dinner. Also we have a few
words such as sh, which contain no vowel. The
consonants n, 1, and r are frequently called semi-
vowels; but such a term in itself certainly does not
support the notion of any generic distinction between
the two groups.
Such another instance as the word tune, whose
pronunciation passes easily, almost imperceptibly,
from [tiun] to [tjun] and thence to [tsun], seems
to indicate some sort of affinity in the formation of
[1], [j], and [s], at least in these particular vocal
surroundings. Indeed, it seems impossible to find
an exact line of demarcation between vowels and
consonants such as those differentiating cats from
tables or red from B sharp. Like the colors, the
speech sounds appear to shade into one another in a
6
WHY ENGLISH SOUNDS CHANGE
continuous series from obstructed to open, from
strong to weak. And this being the case, it appears
feasible to treat them as one series.
The present study will first attempt an analysis
of English speech sounds in one group instead of
two, and on a single basis. Such an analysis, in
bringing out likenesses and correlations rather than
perpetuating differences, may perhaps cast light
upon some linguistic problems. No other radical
departure from traditional methods will need to be
made; indeed, few if any novel facts will have to be
sought. But perhaps simply the regrouping or re-
analyzing of the known facts of English linguistic
history may illumine some dark places.
Basis for Vowel-Consonant Correlation
The first problem obviously must consist in find-
ing a common basis for classifying English speech
sounds together as members of one system. To
accomplish this, the usual descriptive categories must
be modified, because these categories do not apply
equally to consonants and to vowels.
For example, consonants are ordinarily divided
into voiced (media) and breath (tenues). This dis-
tinction does not apply to vowels, which are all
voiced. The consonants are likewise distinguished
from one another by the character of the sound, as
continuants (opens, spirants), stops (surds, explo-
sives, plosives), nasals, fricatives, affricatives, sibi-
lants, aspirates, and a well-nigh endless list besides.
CONSONANTS AND VOWELS
7
These categories are meaningless for vowels, which
are all open sounds.
And on the other hand, there are vowel categories
seldom or never applied to consonants, as tense-
slack, rounded-unrounded, and high-mid-low. Upon
what basis then may all speech sounds profitably be
compared?
Unquestionably the tongue is an important factor
in sound modification, and it is the nature or place
of a given sound's formation or modification by the
tongue which will determine its place in the present
system. To accomplish this, it will be necessary to
take only a single step beyond the present practice,
since vowels are already classified minutely according
to the region of the tongue active in their formation,
into back-front-blade, and high-mid-low; and con-
sonants also are classified on the back-front basis,
though not hitherto as high, raid, or low.¹
If then we use the accepted data for speech sounds
as classified on the back-front system, and (for
vowels) on the high-mid-low system, about three-
fourths of the labor will be found already accom-
plished, for a reclassification of English speech
sounds, to show the part of the tongue active in
forming each sound, and the height of that active
It must not be supposed that the breath-voice, rounded-
unrounded, etc., categories will be disregarded in this study. They
are omitted from the table of sounds because it would be imprac-
ticable to show them there; but they will be utilized frequently in
making explanations of various linguistic facts. In general it may
be said that all consonants are tense in comparison with vowels,
and that most consonants (w an exception) are unrounded.
8
WHY ENGLISH SOUNDS CHANGE
part. Thus the task becomes by no means a diffi-
cult one.
Table I shows 42 English speech sounds graphi-
cally placed according to the portion of the tongue
which is active in the formation of each, and in such
a fashion as to correlate consonants and vowels.
This table will form a basis for subsequent discussion.
The sound positions which have been charted in
this table represent what may be taken as the cardi-
nal or normal, not the only, region of formation
for a given English sound. Many or most sounds
can be made or approximated with the tongue in
other positions, and few are definitely stable in their
region of origin. "There are as many varieties of
[h] as there are varieties of vowels before which it
may stand." (Krapp PSEA 15.) And the sound
[k] can be produced at any point of a fairly large
mouth area of contact between tongue and hard
palate. The vowels, especially the long ones, are
likewise unstable, shifting in endless variability.
These are extreme examples, but it is not too much
to say that no single sound in the speech of any
individual carries at all times precisely the same
value; and therefore the only question concerns the
degree of variability, not the fact of its presence.
Hence the chart of sounds in Table I must be taken
as an approximation, not a definitive plan.
Except where specifically stated otherwise, Amer-
ican English of a Middle Western type has been used
as standard. This is done for two reasons: first, the
in that
and
CONSONANTS AND VOWELS
6
TABLE I. SPEECH SOUNDS CLASSIFIED ACCORDING TO ACTIVE PORTION OF TONGUE
A. High
B. High-mid... [p] [b] [m]
[f] [v]
C. Mid.
D. Low-mid..
1. Lip
E. Low.
2. Point 3. Blade 4. Center
(Front)
[t] [d] [s] [z] | [s] [ž]
[b] [8] [n] [1]
[r] [1] [R]
[i] [y]
[I]
[j]
[k₂] [82]
[e] [A]
[9]
[ɛ] [ə]
[x]
5. Back
[u]
[U]
[o]
[W] [M]
[ɔ] [g.] [k] [n]
[a] [x] [3]
Note: The sound [h] is omitted since its position varies approximately from B3 to E5.
T

10
WHY ENGLISH SOUNDS CHANGE
author knows this type of speech best from use and
observation; and secondly, certain sound elements
([M], the inverted r, etc.) are preserved in this speech
although lost in Standard British English. Use of
this speech type of American English is a conven-
ience, and should not affect the validity of any con-
clusions which may be reached.
With these considerations in mind, let us examine
Table I in detail. The comments will relate largely
to height, since the factor of comparative frontness
is already well established.
Speech Sounds in Detail
B1. The lip consonants² [b], [p], [f], [v], [m], are
all classified as of high-mid height, although as a
matter of fact the height of the tongue is not the
significant factor in their production, and may vary
from mid (model, potter, father) to high-mid (pease,
feeble, veer). They may be described as of a neutral
high-mid tongue position, although on the whole
they are to be classed with the high consonants.
A2. The largest single group of sounds in Table I
is, as might be imagined, those consonants formed
by the active tip or point of the tongue. As a group,
these are the most stable of English speech sounds.
Only one, [1], varies in tongue position to any extent.
By the very character of their formation they neces-
sarily remain substantially the same in tongue posi-
2 The terms consonant and vowel will not of course be discarded
on account of their inclusion into a single system.
CONSONANTS AND VOWELS
II
tion, no matter what the contiguous sounds may be.
Such variation as may occur (e.g., between s in
stone and saw) does not materially affect the tongue
position. The production of [1] varies somewhat more
widely, e.g., in eel and awl.
Point consonants (A2 and B2) not only are the
largest single consonant group, but approximately
two of every three consonant sounds used in actual
speech belong to this group. This fact is set forth
in Table II, an analysis of some 5,000 consonant
speech sounds based upon four representative pho-
netic transcriptions giving sounds as used in ordinary
speech (Krapp PSEA 151-170). The transcriptions
were analyzed in two separate groups and the results
compared. As the proportions of the various sounds
were found substantially the same in both groups, it
is probable that the figures here given are fairly
representative of the frequency of consonant occur-
rence in general, although this could not be es'ab-
lished definitely without more extensive investigation
than is practicable here.
Table II will not be discussed in detail at this point.
The facts set forth will enter into subsequent discus-
sion, but here it is sufficient merely to note the large
preponderance of point consonants in English speech.
B2. The point consonant group includes, besides
the eight sounds placed as A2, two slightly lower
sounds, the semi-vowel [r], and its attenuated echo
[1]. These weak consonants are all that is left of
the OE strong trill [R]. It is supposed that [R] was
12
WHY ENGLISH SOUNDS CHANGE
TABLE II. OCCURRENCE of ConsonANTAL SOUNDS IN ENGLISH
Lip..
.[b] 140
[p] 163
[f]
141
[v] 157
[m] 242
Total lip........843
Point.....[d]
393
[t] 580
[s] 441
[z] 244
[b] 28
[૪] 237
[n] 496
[1] 353
[r] 223
[1] 319
Total point.....3314
Blade.....[S]
104
[ž] 29
Total blade......133
Front..
Back³
Front.....[il
Total front.
Back³.....[g]
45
79
[k]
207
[w]
139
[M] 36
[n] 99
Total back......560
SUMMARY AND PERCENTAGES
.. 45
Aspirate...[h] 118
Total aspirate....118
Grand Total....5013
Number Per cent of total
843
Lip...
16.8
Point.. .. 3314
66.1
Blade...... 133
2.7
45
0.9
560
11.2
Aspirate... 118
2.3
5013
100.0
3 The two sounds [g] and [k] are not divided into their so-called
guttural and palatal values, represented here as [g], [k], and [ga],
[kal, because this distinction was not made in the original transcrip-
tions. If this were done, the percentage of front sounds would be
considerably increased.
CONSONANTS AND VOWELS
13
a point trill, and it is so placed here; but the details
of its development to [r] are not entirely clear (see
pages 96-99).
The sound [r] is not stable even initially (compare
real and raw) and except at the beginning of a syllable
(rat, very) it has disappeared from the standard
speech of parts of England and of the United States.
In this respect [r] is like [h] and [w], both of which
have become sounds so weak as to be audible only
at the beginning of a syllable.
A3. Slightly farther back in the mouth than the
group of sounds just described are the pair [S] and
[ž]. Neither of these is stable (compare Shah and
she), but like all the other consonants, these are
sufficiently uniform to give the effect of the same
sound in various words. These two are perhaps
slightly lower than, e.g., [t] and [d], where the tip
of the tongue actually touches the roof of the mouth.
B3. The tense vowel [i], a "long" sound, is the
highest of the vowels, and is produced farthest front
in the mouth. Its rounded equivalent is [y], found
in OE and in modern French. The so-called "short"
value of [i], namely [1], is slack where [i] is tense, and
is slightly lower. A comparison of the vowel of deed
with that of geek will reveal the range of variability
of the pronunciation of [i], and [1] varies to much the
same extent (compare king and did).4
4 As "long" and "short" vowels are in fact entirely distinct
sounds, and are sufficiently differentiated by using separate symbols,
the phonetic "long sign" [:] will not be used in this study.
Į
14
WHY ENGLISH SOUNDS CHANGE
A4. The final group of high sounds includes [j]
(y in yet), [k2], and [g]. The first of these is a semi-
vowel, and is very near to [i] and [1], both in placing
and in pronunciation. Here we may see graphically
how the word tune passes from [tiun] to [tjun] and
thence to [tsun]. The three sounds [1], [j], and [s],
are easily transferable, being all continuant sounds
and all near one another in place of production. The
word duty may follow a similar sequence, [diuti],
[djuti], and [džuti].
The sounds [k] and [g] represent high or
"palatal" k or g. These values occur in conjunction
with high sounds. The k of keen, for example, would
be [k]; that of cart, [k,]; geese has [g], garden [g.].
These sounds are both slightly farther back than
[j]. High and low [k] and [g] will not always be differ-
entiated in this study, as the distinction is often
not essential to the argument.
C4. A group of five vowel sounds falls within
this area. First, there is the tense sound [e] (a in
mate) and its slack companion vowel [e]. A slack
vowel slightly back of [e] is [^] (u in up, nut, etc.).
Also in this group we have [9], the vowel heard
before [1] in the American "r-full" pronunciation of
bird, word, etc. This "inverted" vowel is discussed
at length in Chapter IV.5 And lastly, there is the
slack, “neutral” vowel [ə], very like [^] in pronunci-
5 Phoneticians also recognize [a:], the vowel of the "r-less" pro-
nunciation of hurt, bird, etc. This is not [A] lengthened but another,
tense, vowel. As the present study is based on American "r-full”
speech, it is not necessary to include [A:] in our table.
CONSONANTS AND VOWELS
15
ation but shorter and slacker. This is unquestionably
the most common vowel sound of English, since it
may appear in place of any other vowel in unstressed
position.
Why is it that we find so many vowels centering
in this area, just as we found so many consonants
(obstructed sounds) whose obstruction was caused
by the active tongue tip? It seems plain that these
vowels all involve a position of the tongue nearly
(in the case of [ə], quite) where it falls in a state of
rest. In other words, the vowels center about the
easiest position for emitting sound, the consonants
about the easiest mechanism for obstructing sound.
D4. Below these may be placed [æ], a sound of
very frequent occurrence in English, and a variant
with [a] in such words as grass, laugh; etc. In all the
words where [æ] alternates with [a], professional
speech experts (elocutionists) will advise pronouncing
[a]. Indeed, the sound [æ] is regarded by such
teachers as definitely distasteful and reprehensible
wherever avoidance of it is found possible.
So far as I know, no explanation of this avoidance
has been attempted; but is it not reasonable to sup-
pose that it is because of the facial expression in-
volved in pronouncing [æ]? The nose is slightly
elevated by the flattening of the lips, and the lip
corners are drawn down, giving an expression almost
precisely that which normally follows the apprehen-
sion of a bad odor or some other unpleasantness.
Compare with this the facial expression normal to
16
WHY ENGLISH SOUNDS CHANGE
X
[a], and I believe it can hardly be doubted that this
difference causes the avoidance of [æ].
B5. The highest sounds in the back series are
the tense vowel [u], and its slack companion [u].
If the vocal cavity be regarded as a pipe, with the
vocal chords at one end and the lips at the other,
then [a] would be called a "short-thick-pipe" sound,
since in its formation the lips are flattened back
against the teeth and the tongue held low, making
a sharp contrast to the production of the "long-thin-
pipe" sound [u], where the lips are protruded as if
for whistling, and the tongue held high to produce
an elongated, thin vocal cavity. All the rounded
vowels, as well as the consonant w, have this char-
acteristic, and hence are now and then called lip
sounds."
The consonants [w] and [M] show close affinity
with the vowel [u]. A simple fashion of teaching
foreign students the sound [w] is to substitute [u],
saying, e.g., u-ater for water. The breath equivalent
of [w], namely [M], is disappearing from standard
British English, as the breath equivalents of n, r,
and I have already done. Many Englishmen make
no distinction between when and wen, which and
witch, etc. This practice is frowned upon by lan-
guage teachers in this country.
Slightly lower than [w] is the tense rounded vowel
"The back vowels u, o, etc., are classed as the first or frontmost
series in Hebrew grammars, while w is called a lip consonant by
Wyld and others.
CONSONANTS AND VOWELS
17
///
[o]. In speech this vowel tends to diphthongize,
giving note [nout], etc. This tendency, which applies
to most of the tense vowels in British speech, is less
strong in American English.
D5. In the next mouth area fall three con-
sonantal and one (or two) vowel sounds. The symbol
[ɔ] in this study represents the tense vowel of law,
fault. The same symbol is often used to represent
the slack concomitant of [o], but there is an ambiguity
in the use of this symbol, inasmuch as the long or
tense [ɔ] is a sound different in quality from short or
slack [o]. Since, however, American speech tends to
use the tense sound in place of the slack in, e.g.,
God, frost, only the tense sound is represented in
our table.
To be exact, at least four values for k should be
recognized, instead of two. The pronunciation of k
differs perceptibly in key, coo, Coe, and caw. The
k-region (which is also the g-region) is perhaps larger
than that for any other sound except h. When like
a hammer the tongue strikes the hard palate at any
point within this area, a sound is produced which
we recognize as [k].
The final sound in this group does not vary over
a wide range. This is [n], a sound with the peculiarity
that it does not occur initially in English. In words
where n is followed by g or k, [n] is substituted for
[n] (anchor, anger).
E5. The low-back group contains as its only
ModE member the open vowel [a]. The open char-
that
Zei
18
WHY ENGLISH SOUNDS CHANGE
9
X
acter of [a] has long been exploited by physicians,
singers, and others. It is not a sound of particularly
frequent occurrence in ModE, unless we count its
presence in the diphthongs [ar] and [au].
Then there is the pair of guttural continuants
[x] and [3], which are absent from ModE but present
in OE and ModG. These are both definitely ob-
structed sounds, localized at the low and rear limits
of the English speech-producing area. They may
be called fringe sounds, falling as they do at the
extreme limits of this area.
Finally, it is impossible to classify [h], the most
unstable sound of English. This sound may be pro-
duced with the tongue in any position at all; its very
indefinite character makes it practically inaudible
except in initial position, although it may occur as a
slight aspiration in other positions.
General Conclusions
Taking the table of sound positions as a whole,
it is impossible not to be struck by the resemblance
of the plan to a rough diagram of the vocal cavity
with the teeth slightly apart and the tongue at rest.
This fact increases the conviction of its general cor-
rectness, since it would seem reasonable that sounds
should distribute themselves according to the com-
plete area of their modification.
It may likewise be noted that the vowels form,
as it were, a layer beneath an outer or upper layer
of consonants. This is equivalent to saying that the
CONSONANTS AND VOWELS
19
vowels are of a more open nature than the consonants,
and is again what we should expect to find in an
accurate representation.
In describing the sounds of this table, it has been
mentioned frequently that a given sound is unstable,
i.e., that it tends to accommodate itself to neighbor-
ing sounds. It is this phenomenon which forms the
subject of the present discussion, and to it I have
ventured to give the name accord.
Accord in speech sounds has long been given par-
tial recognition by linguists, under the various terms
combinative change, analogy, and assimilation.
That its full significance has not been perceived and
set forth is due in large measure to this confusion of
cumbrous or misleading terms, a confusion which has
effectually prevented clear thinking on the subject.
The term combinative change is too vague and un-
wieldy for scientific use. The term analogy is bad
on many counts (see page 130) but chiefly in its
implication of a conscious mental process favoring
change. And as for the term assimilation, perhaps
the most nearly exact of the three, let us examine in
more detail its qualifications for illuminating the
facts of sound accord.
Assimilation may mean making like or similar;
but is this what sound accommodations involve?
We have mentioned the word keen as employing [k 2]
because of the following high vowel [i]. But [k2]
and [i] are no more like or similar than [k,] and [i];
they merely accord much more nearly, as Table I
20
WHY ENGLISH SOUNDS CHANGE
graphically indicates. Evidently, assimilation should
not be used in this sense to describe this phenomenon.
Again, assimilation might conceivably refer to
the swallowing up of one sound by another. In ME
or early ModE night [nit], OE [x] has disappeared,
and this disappearance might be termed assimilation.
But the sound [x] has not been assimilated but aban-
doned, as when a river makes a cut-off to straighten
its course. The sound [x], between the high vowel [1]
and the high consonant [t], was in an awkward posi-
tion. It did not fit with either of these sounds. Its
range of variation, less than that of [k], was not
sufficient to enable it to harmonize with its sur-
roundings. Finally it was simply replaced by [j],
which then merged with the vowel to produce [i].
Can this loss of [x] be called an assimilation?
G
Enough has been said to show the unsatisfactory
character of the term assimilation as compared with
accord. In addition, the term accord has the advan-
tage of being capable of wide application to many
linguistic fields. Some of these wider applications
will be suggested in Chapters II and VII.
In addition to the lack of a suitable term, the
significance of sound accordings has been largely hid-
den by the lack of any basis for discussing speech
sounds as members of a single system. Table I
presents such a system graphically. It indicates
clearly why the tongue is at three different levels in
pronouncing the initial sounds of geese, goes, and gall.
It shows how, e.g., gleam [glim], from OE glam
CONSONANTS AND VOWELS
21
[glæm], is a change in the direction of sound accord.
Indeed, if I am not mistaken, this simple process of
correlating vowels and consonants will explain at
least a large proportion of the sound changes of
English phonological history. The detailed dis-
cussion of these will be reserved for later chapters.
Y
ܬܝܐܦܐ ܝܚ 3
1
CHAPTER II
ACCORD
Definition of Accord
In the foregoing chapter it was stated, what in-
deed must be obvious upon consideration to every
student of linguistics, that vowels and consonants are
not separate and distinct speech groups but are parts
of one continuous system. The universal separation
or opposition of them, practiced by scholars, may
possibly have operated to obscure linguistic problems.
Accordingly the writer has attempted to construct
a table which would show all English speech sounds
as members of one system. It was seen that this
classification could not be a hard-and-fast one; that
many or most speech sounds vary in their pronuncia-
tion, some slightly, others over a wide range. It was
found that this variation is determined, in part at
least, by the linguistic surroundings of a given sound,
that is, by the other sounds preceding and succeed-
ing it.
To this variation was given the name accord.
And it will now be necessary to complete the general
definition of this term, and to indicate a few of its
manifestations in the field of linguistics.
Accord in speech may be defined briefly as that
principle whereby a speaker tends to harmonize
22
ACCORD
23
speech sounds. In another aspect, it may be termed
the tendency of spoken language to become pro-
gressively more simplified, hence more efficient¹ in
its sound patterns. More specifically, accord is that
principle in language by whose operation sounds tend
to repeat or to approach in manner or place of pro-
duction, other sounds, whether immediately con-
tiguous in utterance or more remote..
It must not be supposed that these vocal accom-
modations which we are about to group under the
term accord take place quickly. The human tongue
is a most agile member, and when habituated to the
feat, can accomplish the most barbarous combina-
tions of sounds with no appreciable difficulty. Cus-
tom, conformity to social standards, the desire to be
understood, perpetuate the difficult sound combina-
tions, and it is ordinarily only over long periods of
time that any alteration is achieved. Linguistic
change is not a river rushing onward; rather it is a
glacier, whose motion is imperceptible to the casual
observer, but which nevertheless follows its own
definite laws of motion.
Nor must it be supposed that accord works in
the domain of conscious manipulation. The child
who discards useless motion in learning to walk does
not therefore understand and consciously use the
principles of physics. The adult who says eight may
realize that ModE has simplified OE eahta, but that
I Not always more "easy." The relation of accord to an "ease
theory" of speech development will be discussed on page 125.
24
WHY ENGLISH SOUNDS CHANGE
:
1
↓
realization is by no means a necessary concomitant
of the process of simplification. In one group of in-
stances, to be mentioned in this chapter, accord has
emerged from the realm of the unconscious into that
of conscious manipulation, but it has become thereby
a somewhat different thing, and manifests less com-
pletely the underlying principle.
Accord in its narrower sense might be stated
geometrically as a process of shortening the line be-
tween two points, or, physiologically, of reducing
tongue action. A simple instance may be given.
When a modern speaker says sea [si] instead of OE
să [sæ], he is pronouncing two sounds in each in-
stance, but nevertheless he is shortening the line of
sounds considerably. Let us analyze the change in
detail, on the basis of Table I.
In this table, the range of tongue-heights from
high to low has been lettered from A to E, and the
lip-back series has been numbered from 1 to 5, to
facilitate reference. Thus the consonant s, a high-
point sound, has the notation A2 in this system.
The low-mid-front sound [æ] is described on the same
plan as D4; while [i], a high-mid-blade sound, is
represented as B3.
The combination [sæ], OE sæ, will then be repre-
sented by the sequence A2, D4. The combination
sea, [si], will be in this notation A2, B3. Is it not
obvious that the sounds in the ModE word have
come to accord much more nearly than those of the
older word? Here as always with vowels, the process
懼​。
ACCORD
25
of shortening has been a gradual one, [æ] having
passed through the stages [e] and [e] before becom-
ing [i].
Approximation Accord
This change from să to sea is typical of what
may be called the first or approximation type of
sound accord. In such instances the line of sounds
shortens, making a more efficient vocal pattern.
This shortening may involve a change in the vowel,
the consonant, or both, as will be seen from the
further examples about to be given.
Not without misgivings, the writer has had re-
course to a graphic method to introduce the discus-
sion of the workings of accord. By this method, the
dots represent the various tongue positions, and a
line drawn from one to another shows the sequence
of these tongue positions in a given sound pattern.
A glance at Table III will show how this works out
for să and sea. The picture is merely another way
of saying that the sequence in să is A2, D4, and in
sea, A2, B3.
The defect in this plan of graphic representation
is that the line may be mistaken for a picture of an
actual motion of the tongue, an inference which
should not be drawn. In pronouncing sea, for exam-
ple, the point of the tongue does not move from
high-point to high-mid-blade. The line means
merely that these portions of the tongue are active
in sequence. If this fact is thoroughly understood,
26
WHY ENGLISH SOUNDS CHANGE
I
A
B
C
D
E
AB
C
D
E
A
B
C
D
E
A B C D w
E
TABLE III. APPROXIMATION ACCORD

1 2 3 4 5
OE sæ
1 2 3 4 5
OE bāt
1 2
3 4 5
OE top
1 2 3 4 5
OE sceap
A
B
C
D
Ladud
ABC
D
E
A
}
B
C
D
E
A B
B
C
D
E
1 2 3 4 5
1
ModE sea
2 3
3 4 5
ModE boat
1 2
3 4 5
Mode tooth
1 2 3 4 5
1
ModE sheep
ACCORD
27
the graphic method becomes a simple and effective
means of presenting mathematically the facts under
consideration.
Three further examples of this type of accord are
given in Table III. In bat-boat, the vowel has lifted
nearly to the level of the lip consonant which pre-
cedes it, and has also approached more nearly the level
of the succeeding high sound. In top tooth, the same
thing has occurred. The two high consonants have in-
duced a raising of the vowel to the highest vowel level.
At this point may be heard the objection that
the words given in Table III are not individual in-
stances, but are merely chosen examples of universal
sound laws of English; that, e.g., the change from
să to sea is only one of a large number of instances
where [æ] changed to [i]. This is of course true; but
it does not alter the further fact that this particular
change is an according of sounds to each other. At
present the task is merely to illustrate the various
types of accord. Later the large currents of sound
change will be considered.
In the final instance given in Table III, scēαpt
sheep, we have a change of a slightly different order.
Here the vowel has undergone its usual development,
being raised to a higher level. But in addition the
consonant [] has been substituted for the two for-
mer sounds [s] and [k]. The new sound is between
2
* It is, of course, possible that the change from [sk] to [S] may
have been complete at the beginning of the OE period; but the exact
chronology of the change is not essential to the argument here.
}
28
WHY ENGLISH SOUNDS CHANGE
the two older ones in region of production; both
former sounds are continuants; hence it is natural
that the one sound should replace the two. This
forms an instance of another type of approximation
accord, namely, substitution of sounds.
•
By the very nature of their formation consonants
cannot shade into one another as can vowels. A
singer may sing "ah - o - u" or "ee - ay - ah" but not
"p - t
- tk" without fresh breath impulses at each
change. Consonants, therefore, experience substi-
tutions where vowels approach near-by sounds.
merely by gradual approximation.
Another instance may be given. ModE yelp
• derives from OE gielp. The initial sound of the
older word undoubtedly was at one time [g],
although by the OE period it may have been [g 2],
or even [j]. It is certain that, at some time in the
history of this word, [g] ceased to be [g] and became
[j]. It was at this point that the substitution took
place.
Under this approximation type of accord should
also be grouped cases where the sound alters, not in
region of formation, but in some other respect. For
example, it is well known that d in the termination ed
is pronounced [d] after voiced sounds (buzzed, begged,
sorrowed, drifted,³ handed, hummed) and [t] after
breath sounds (dressed, picked, laughed). The d
becomes a breath sound where necessary to accord
3 The vowel following t constitutes voice
·
ACCORD
29
with a preceding breath sound. The same thing
occurs in the phrase have to, popularly pronounced
[hæftǝ]. The [v] of have has become [f], a breath
sound, to accord with the following [t]. Also round-
ing may be contagious, as in water [wɔtr], from OE ·
water, which should normally have given the flat
vowel [æ] in ModE, instead of the rounded vowel
[ɔ]. All these instances of accord may be regarded
as approximation types.
Arbican
Gernal
Ablation Accord
Far more numerous than examples of the approxi-
mation type of accord are those of the second or
ablation type. Under this heading fall all those
words where the process of according has involved
the loss of one or more sounds included in a former
pattern. The enormous majority of word changes
between OE and ModE involve such loss. More-
over, changes of this type are more spectacular,
showing in many instances a simplification quite
astonishingly radical.
The OE word eahta, already mentioned, is a good
example. In OE this word actually contained five
elements, ranging from high-point to low-back.
Today its elements number but two, neither falling
below mid-height.4 OE cneo and folc illustrate this
4 It is true that meticulous analysis of ModE eight would show
three elements, [eit], the first two forming a diphthong. The second
element of this diphthong, however, is in B3 position, or in the direct
}
30
WHY ENGLISH SOUNDS CHANGE
с
same simplifying tendency, and a final instance is
OE feoh, ModE fee, where two sounds now do the
work of four previous ones. Each of these examples,
shown graphically in Table IV, may profitably be
considered in some detail.
Let us first examine the instance folc-folk. Here
a short-cut, so to speak, is made to avoid following
a roundabout path. The l of folc is obviously an
unnecessary complication, coming as it does between
the two according sounds [o] and [k]. It is the accord
between these two, [o] and [k], which operates to
eliminate the extraneous [1].5 Note that we have
here two non-contiguous sounds illustrating the
principle of accord.
Words such as hafoc-hawk, hlaford-lord, illustrate
the same principle. In both these the lip consonant
is at a far remove, mechanically, from the two con-
tiguous low vowels, and finally becomes lost. The
instance eahta-eight is much the same; the low-back
consonant [x] is out of accord with its neighboring
line from 4 to A2. Such slight diphthongal elements will frequently
be disregarded in this study, which aims to confine itself to the
larger essentials of an extremely wide subject. Moreover, the first
element in eight is not universally pronounced as a diphthong. For
discussion of diphthongization, see Chapter V.
5 Wyld (SHE 159), following Luick, says incorrectly, "lis
retained initially and medially before vowels, and before point
(and blade-point) consonants; it is lost before back and lip con-
sonants."
Here an understanding of sound accord would have
avoided the error. This sound is lost before back and lip consonants
only when the preceding vowel is also low-back or mid-low-back;
that is, is lost only when low or low-mid sounds come both before
and after it. It is not always lost before lip and back consonants,
e.g., elm, elk, ilk, etc.
ACCORD
31
A
B
C
D
E
A
B
C
D
E
AB
B
C
D
E
A
B
C
D
E
TABLE IV. ABLATION ACCORD

1 2 3 4 5
OE eahta
1 2
2
3
3 4 5
W1
OE cneo
1 2 3 4 5
OE folc
1 2
3 4 5
OE feoh
A
B
C
D
E
A
B
C
D
Lad
E
A
B
C
D
E
A
ܐ
B
с
D
E
1 2 3 4 5
Mode eight
1 2 3 4 5
ModE knee
1 2 3 4 5
Mode folk
1 2 3 4 5
ModE.fee
32
WHY ENGLISH SOUNDS CHANGE
sounds, and is ablated accordingly. The second and
third vowel elements are likewise lost, although these
changes probably are due not to accord but to an
increase of stress on the first element of the word
(see page 51 below).
The examples cneo-knee and feoh-fee require fur-
ther elucidation. In one we have the ablation of an
initial sound, in the other of a final sound. It may
be asked, how can accord between sounds operate to
secure the loss of sounds at the beginning or end of
words, on the ablation or "short-cut" principle?
The answer to this question becomes simple if we
think of words, not as separate entities, but as com-
ponent eddies in a "flow of speech." To consider
feoh first, it is easy to perceive that only rarely
would this word end a sentence. And if medial in
a sentence, what kinds of sounds could possibly
follow? Only sounds higher and farther front—there
are no others in English speech. Thus, whatever
the position of feoh in the sentence, an ablating ten-
dency would generally be present, a tendency which
finally compassed the loss of the [x]."
This explanation indicates a solution for the word
cneo as well. The initial low consonant was followed,
and in practically every instance would be preceded,
by a sound definitely higher in region of production.
"It should be mentioned that the oblique cases of feoh already
in OE had ablated the [x]. In these instances, of course, the guttural
was always medial. The vowel loss and change in feoh is to be ex-
plained on the basis of stress and approximation, as with eahta above.
ACCORD
33
This condition finally brought about the disappear-
ance of the [k]."
It is interesting to note that as with the examples
in Table III, each of the words listed in Table IV
illustrates a general type of sound change. Thus,
OE h [x] always disappears in ModE, as it does in
eahta and feoh; OE cn when initial always becomes
simply [n], as in cnēo; and OE / between low-back
sounds always disappears, as in folc.
Occasionally a word combines the ablation and
approximation types of accord in the same change.
Take for example OE bohte, ModE bought. Here the
[x] has been eliminated as in eahta, but at the same
time we may observe a lowering in the vowel [o],
which after lengthening should normally give ModE
[u]. It can scarcely be doubted that this lowering of
[o] is an approximation to the low-back [x], which
before its ablation effected this lowering of the pre-
ceding vowel.
Mutation Accord
The third type of accord might well be deemed
incredible if brought forward here for the first time.
- How a sound in one syllable could be fancied to
operate to alter the vowel in another syllable, after-
ward itself disappearing entirely from the word,
seems on the face of it quite incapable of rational
7 It may be asked why then did cnèo not simplify earlier? Why
has it never simplified in German? These questions will be dis-
cussed later, in Chapter VII.
34
WHY ENGLISH SOUNDS CHANGE
demonstration. Yet this remarkable manifestation
of accord, under the name umlaut, is an accepted
fact of linguistic history, even though the rational
basis for the phenomenon has not been pointed out.
This rational basis is clearly found in our principle
of accord.
For umlaut works throughout in the direction of
according sounds. Wright (OEG 30, 32) defines both
palatal and guttural umlaut. The former he calls
"the modification (palatalization) of an accented
vowel through the influence of an i or j which
originally stood in the following syllable." Palatali-
zation, in the quaint vocabulary of linguistics, means
a raising or fronting or both. Wright, therefore,
here recognizes the nature of the change performed
by umlaut, though he does not relate it to any gen-
eral principle of sound change. That this principle
is however involved is easily apparent from a few
of his illustrations: Goth. andeis, OE ende; Goth.
badi, OE bedd; Goth. bugjan, OE bycgan; and so on
through a host of examples, all of the same general
nature.
Wright further defines guttural umlaut as "the
modification of an accented vowel (a, e, i) through
the influence of a primitive OE guttural vowel
(u, o, a) in the next syllable, whereby a guttural
glide was developed after the vowels a, e, i, which
then combined with them to form the diphthongs
* Something less than quaintness characterizes the application of
the descriptive term guttural to u, a high-back vowel!
ACCORD
35
་
·
ea, eo, io." Here again we may see clearly the oper-
ation of accord. Examples are perhaps unnecessary,
since guttural umlaut, according to Wright, is only
of a partial nature and occurs only in certain dialects
of OE. It seems to be a rule of linguistic history
that fronting changes among vowels are more fre-
quent and more thorough-going than retracting
changes; and perhaps the reason for this fact may
become apparent as we proceed.
Umlaut might reasonably be classed as one form
of approximation accord, since in all cases the first
vowel tends to approximate the second. But as the
sounds influencing each other are non-contiguous
throughout, and as umlaut itself is already such a
well-defined linguistic phenomenon, it seems better
to distinguish it completely from the usual type of
approximation accord. Other illustrations of the
mutation type, besides umlaut, might possibly be
found; but for the present we shall not search for
them, but shall pass on to the final division, the
recurrence type of linguistic accord. This final type
is well known by the terms rhyme, alliteration, and
assonance.
Recurrence Accord
All language appears to have an instinctive feeling
for rhyme, alliteration, assonance, even for repetition
pure and simple. In Ancient Hebrew, a tongue
devoid of these as formal metrical devices, we find
such fossilized rhyming phrases as thóhu wavóhu
36
WHY ENGLISH SOUNDS CHANGE
duinitive
bitta
(Gen. 1:2); and alliteration, assonance, and repetition
are likewise extremely frequent. The same thing is
true of Greek and Latin, among Indo-European
tongues; while Finnish and Bantu might be said to
be constructed largely on these principles." In
English we have pell-mell, helter-skelter, teeny-tiny,
dry-as-dust, by-and-by, and a host of others. OE
examples can scarcely be given, since this type of
unconscious repetition of sound plays small part in
highly formal literature such as that of the Anglo-
Saxon period.
Accord of this recurrence type, including rhyme,
assonance, and alliteration, in their spontaneous or
unstudied states, requires a more or less formal
definition to establish its character as accord. This
definition may be given as follows. In all these
phenomena, the tongue slips easily back into a cer-
tain position or pattern after a brief lapse of time,
and this return gives a sense of ease and familiarity
which in its turn produces æsthetic pleasure. Sound
patterns thus come to accord with one another, not
by the alteration or loss of their component parts,
but by their repetition. Indeed, repetition pure and
simple is the very first stage in this type of accord,
and this is shown in dozens of nursery words, papa,
mamma, by-by. Slightly more advanced is the stage
of rhyme, bow-wow, and of assonance, tick-tock,
9 But see page 73, note. Accord of this type is extremely primi-
tive, and is closely involved in the formation of inflectional systems.
Or else, having become conscious, it becomes the basis of a sophis-
ticated prosody.
ACCORD
37
7
fot
ধত
while alliteration by itself is apt to represent an even
more sophisticated stage, Simple Simon, Miss
Muffet.
In many languages the æsthetic pleasure of this
type of accord finally becomes self-conscious, and in
one or another of its manifestations is wedded to
that other even greater æsthetic pleasure of accents
-beats or strokes recurring regularly or irregularly.
Then we have poetry of the English, Old French,
or Anglo-Saxon types, depending on which aspect―
rhyme, assonance, or alliteration-is chosen. Fre-
quently, as in Hebrew, Greek, Latin, etc., the con-
nection is never made, and then we have a poetry
depending on accent or on quantity (another device
to the same purpose) with the ornaments of rhyme,
alliteration, assonance, and repetition entering only
more or less casually. 1°
IO
Obviously it would be difficult or impossible to
treat accord of this recurrence type within the narrow
limits of English linguistic history, and the present
writer will not attempt to do so, but will merely
suggest the fact that the same principle which governs
the Cockney when he says "'am and eggs," or,
emphatically "ham and h’eggs," and the child saying
"I hafta go," governs also the lofty sound patterns
of poetry. The "concord of sweet sounds" of the
IO
10 Note, however, that in tongues which do not use such “orna-
ments" for meter, the inflectional system usually makes up the lack
by providing plenty of sound accord. Thus, in such tongues, all
these four types of recurrence accord are already embedded deep in
the grammar.
38
WHY ENGLISH SOUNDS CHANGE
latter is in the former instances merely concord of
sounds; but it is concord in both.
Accord in Other Fields
It might be possible to pursue the search for
accord even farther; to pass from the mechanical
realm of sounds to the psychological one of ideas,
and to show that grammatical systems are based
upon definite principles, one of which is this har-
monizing tendency, this feeling for synthesis, which
we have named accord.
Accord in the field of ideas becomes a psycho-
logical principle, thus approaching nearer the plane
of conscious manipulation. Perhaps for this reason,
its manifestations are more familiar to the average
student, grouped as they mainly are under the two
heads, grammatical agreement and analogy.
It is easy to see that the feeling for synthesis of
expression lies at the very root of all inflectional
systems. Whether or not the inflectional elements
themselves were ever independent words, their use
and extension as such elements is due to accord of
ideas. Ordinarily this accord of ideas is expressed
through the use of identical or closely related sounds.
In Hebrew, for example, the termination ah regularly
or normally expresses the idea of the feminine gender,
whether in noun, adjective, or verb. In English the
related sounds [s] and [z] ([Iz]) express the idea of
plurality in nouns.
It is well to distinguish clearly the repetition or
ACCORD
39
echoing of sounds which is based on accord of ideas
from the repetition or approximation or ablation of
sounds which is sound accord. The child who says
by-by, ma-ma, is unconsciously utilizing this repeti-
tion type of sound accord. The child who says
sheeps, on the other hand, or heads, or mammas, is
using accord of ideas. The final s in these words has
no necessary companion or echo in the sentence. It
is not an approximation to any contiguous sound.
It is simply an unconscious echo of the general body
of s-plurals already established in the child mind.
According of ideas is also shown in grammatical
solecisms like It is him, and, All but he had fled. II A
37
It is shown in "popular etymologies" such as sparrow-
grass for asparagus. It is shown in most grammatical
"mistakes," but it must not be concluded that its
manifestations are confined to such mistakes, since
it is shown in some degree in every "correct" locution
as well. 12
The subject of accord of ideas will be sketched
in more detail in Chapter VII. Here it is necessary
I The double negative, among grammatical solecisms, I believe
to be largely a matter of sound accord, or at most a combination of
the two types. Jespersen, in his monograph on the subject, hints at
the according character of such negatives when he says that they
nearly always consist of some easily distinguishable sound element
repeated in speech. Thus the phenomenon becomes sound accord
of the fourth type, conditioned, however, upon the unconscious
desire on the part of the speaker to emphasize the negation.
12 So also with other linguistic forces. Stress is present with .
equal universality. As to ablaut, which may be a result of the
operation of stress, Wright (OEG 100) says “Every syllable of every
word of whatever part of speech contains some form of ablaut."

40
WHY ENGLISH SOUNDS CHANGE
only to mention this type of accord as a division of
our general subject.
Other Principles Operative
By this time the reader no doubt concludes that
accord is to be the famous lion's den into which all
went and from which none returned. To allay this
suspicion, as well as to complete our definitions, let
it be said that accord, potent and universal as it is,
is by no means the only principle involved in sound
change.
In all, four forces or principles regulating sound
change are recognized in this study. The first,
accord, may be likened to a centripetal force in
language, bringing sounds closer together through
ablation, approximation, and so forth. But this
principle can scarcely be made fully comprehensible
apart from the three others which remain to be de-
fined. It is frequently the interplay of these prin-
ciples which brings about phonological change.
Therefore, before considering in detail the appli-
cation of accord as formulated in this chapter, to the
large facts of the history of English, it will be neces-
sary to describe these other producers or modifiers
of sound change; and this description will occupy
the next chapter.
CHAPTER III
OTHER FACTORS IN SOUND CHANGE
The Modulation of Sounds
The principle of sound accord, as formulated in
the previous chapter, is concerned wholly with the
articulation. In other words, in order to explain
any given sound change by this principle, it must be
shown that articulations have thereby become more
completely harmonized in manner or place of pro-
duction. Such harmonization may have to do with
the character of articulations as high or low in the
mouth, as back or front, voiced or breath, rounded
or unrounded, tense or slack; in every case it is some
aspect of the articulations which becomes harmonized
by a given change.
But scarcely less important than the articulation
itself is its modulation, or modification, by other
factors which enter into all speech in a more or less
determinative degree. Thus a sound may be high
or low, not in region of utterance, but in pitch; and
the pitch itself may be rising, falling, or even. A
sound may be loud or soft, accented or unaccented,
prolonged or brief. Not only does all song or poetic
meter depend upon one or more of these types of
sound modulation, but in the everyday speech of
whatever tongue, each of them will be found to be
4I
42
WHY ENGLISH SOUNDS CHANGE
employed, either casually or as an integral factor
in the conveying of meaning.
Thus in Chinese, pitch is often the only mark to
distinguish between two separate meanings, and
according to tradition, sound length sometimes
played the same rôle in classic Greek. In English
all three elements, pitch, accent or stress, and quan-
tity, enter, though in widely varying degree, into oral
language as determinative of meaning.
The "rising inflection" of the question is an in-
stance in point; thus yes? is a different word from
yes, though the heard distinction is one of pitch only.
Accent is not infrequently used to distinguish mean-
ings from one another, e.g., project and project, differ
and defér, perfúme and pérfume. And occasionally
even quantity may be used to determine meaning;
compare ye-e-es (I suppose so; perhaps) with yes
(agreed).
Differences in sound modulation may assist in
producing large cumulative effects. The Shylock of
Shakespeare's time spoke the same words as the
Shylock of today; yet the effect was almost certainly
opposite. To give another instance, the writer saw
Shaw's "Pygmalion" as interpreted by the company
of Mrs. Patrick Campbell and, some years later, by
the Theatre Guild. The first production carried an
atmosphere of geniality, of spaciousness; the second,
of thin ill temper. Perhaps such differences are too
pervasive to be analyzed in detail; still they are
unmistakable, and are dependent largely upon the
OTHER FACTORS IN SOUND CHANGE
43
modulation of sounds, as well as gesture, expression,
and stage "business."
Sound modulation may be described somewhat
empirically as comprising all distinctions, other than
articulatory, made between sounds. Its main as-
pects are quantity, pitch, stress, and loudness. All
these aspects are present in some degree in any spoken
language, and one or more of them are likely to be
found basic factors in any given speech.
The standardized use of a certain form of sound
modulation tends to produce sound change, often as
radical as anything accord may effect. For example,
it is probable that the loss of English inflectional
terminations, certainly the most extensive single
change in the history of this language, was involved
with, if not entirely brought about by, an increased
stress falling upon the initial syllable of English
words. This change obviously has nothing to do
with according articulations of sounds. It is a stress
phenomenon, having to do with the modulation or
modification of sounds.
The Principle of Grading
A second principle of phonological change must
then be formulated, and from a superficial standpoint
this principle might well be termed stress, since stress
is its most common manifestation, particularly in
English. But this would be as misleading as to call
accord articulation. Stress is a cause or basis of the
changes we are about to consider, and not their
Y
44
WHY ENGLISH SOUNDS CHANGE
essence. Moreover, stress is only one aspect of the
modification of articulations, even though a thorough
study of language in general might well place it as
the most important.
As the articulation is the basis for the principle
of accord, so sound modulation is the basis for the
second principle of sound change, a principle for
which I venture to suggest the term grading. Grad-
ing may be defined as that principle whereby phono-
logical change is induced by the strengthening or
weakening of sounds in standardized or habitual
modulation of a given type.¹ It is a differentiating
principle, strengthening and emphasizing on the one
hand, suppressing and ablating on the other.
A word should be said regarding the choice of the
name grading. The term gradation is not new in
linguistics, having already been employed as synony-
mous with ablaut, to designate vowel permutation
in the derivatives of a root, e.g., sing, sang, sung, or
ride, rode, ridden. While the origin and cause of
ablaut are obscure, it is frequently held to be a
manifestation of the workings of stress, and hence a
phenomenon of sound modulation, of much the same
sort as those we are now discussing. To apply the
name gradation to all such phenomena, however,
might cause confusion and hence a related word has
been chosen, defined with a widened scope to include
¹ It need hardly be said that many sound modulations are
individual or eccentric, and it is only as modulation of a given sort
becomes generally adopted that it becomes effective as a cause of
sound change.
OTHER FACTORS IN SOUND CHANGE
45
consonants as well as vowels, and matters of pitch,
quantity, etc., as well as those of stress alone.
A simple illustration of grading is the formation
of doublets, too and to, off and of, Fr. me and moi, etc.
Here stress modulation has suppressed a word in one
set of surroundings and emphasized it in another,
with the result that two separate words have become
distinguished from each other. Additional illustra-
tions might be multiplied; but since this study is
concerned largely with accord in English phonology,
and since the general subject of grading may well
occupy a separate and extended analysis, we shall
not linger over the general principle, but pass on to
its specific applications in English.
In the previous chapter accord was compared to
a centripetal force, bringing sounds into closer har-
mony. By the same analogy the principle of grading
may stand for a centrifugal force, driving sounds
apart from each other. But this should not be
interpreted to mean that these principles oppose
each other, or work to opposite ends. One of the
commonest results of each is the ablation of sounds,
though it is usually possible to distinguish between
ablations of the two types. In general the principles
interplay and balance rather than oppose; and the
net result of each is usually a linguistic simplification.
Types of Grading in English
It will be admitted generally that while pitch is
present in English speech, and at times is even
46
WHY ENGLISH SOUNDS CHANGE
determinative of meaning, it has not been proved
to have any basic importance in producing English
linguistic change. We may therefore safely omit
pitch grading from further discussion, and loudness,
which is almost entirely unstandardized in English
and which in any event may be regarded as a factor
of pitch and stress, may likewise be disregarded.
The situation is somewhat different when we con-
sider quantity. It is the writer's belief that quantity
plays no more significant part than pitch, in English
speech; and yet the very terminology of linguistics
appears to oppose this view. We do not ordinarily
distinguish, it is true, long and short consonants; but
we do have what are termed long and short vowels,
and this distinction appears to be drawn as univer-
sally as that between consonants and vowels. What
justification may we find for seeking to eliminate
quantity from English modulational types?
Certainly our vowels do vary somewhat in time
of utterance. In the two phrases That's THE thing
and THAT'S the thing, the vowel of the varies in
numerous ways-in length, in stress, in tongue-
height, in tongue-tensity, etc. In the first illustration
the vowel is [il, a tense high-blade sound; in the
second it is [ə], a slack mid-front sound.
Here a
quantity distinction is only one of a number of
differences which are present, and it is perhaps the
least important of them.
Let us vary the illustration. Take these two
phrases: We meet John today and We met John today.
OTHER FACTORS IN SOUND CHANGE
47
The vowel of meet is termed long, of met short. But
is there any easily apparent (i.e., heard) difference of
time length between the [i] of meet and the [ɛ] of met?
Whether the verb or the object be stressed in both
phrases, the vowels appear to carry much the same
time value. Indeed, if John be emphatic in the
first sentence and met in the second, it is observable
that [e] will be a "longer" vowel than [i].
Linguists have long recognized that the ordinary
distinctions of kindergarten "phonics" between long
and short in vowels are invalid; that so-called
"short e" [e] is not "long e" [i] shortened; that
"long i" [ar] is a diphthong and "short i" [1] a high-
slack sound. But the temporal terminology still
lingers on, and is present, so far as I know, in all
phonological studies. Why has it persisted, if the
quantity distinction is indeed completely invalid?
It appears to be the common view that while
length is not a factor in vowel classifications in the
former naïve sense, it must nevertheless be present,
since there does exist a real distinction of some sort
between the two vowel groups called "long" and
"short." There is for example the [e:] of there
(British pronunciation) and the [e] of met; and these
are historically as well as experimentally distin-
2
2 Laboratory analysis has shown some extremely minute differ-
ences in number of vocal vibrations, e.g. between [i] and [ɛ] in speech
use; yet it is a question how much of the observed difference is
really stress, and also how much weight a minute difference of this
sort on either side should carry, since other effective modulational
distinctions are easily apparent to "the naked ear.”
1
48
WHY ENGLISH SOUNDS CHANGE
T
guishable. ME mete [mɛ:tǝ] (ModE meat) had a
different history from ME mette [metǝ] (ModE met).
If this difference is not one of quantity, what is it?
Wyld (SHE 45) suggests a correct answer, but
it is the merest suggestion, not recalled at any later
point in his treatise. In a general discussion of
tenseness in vowels he recognizes, though he does not
name, “two distinct series of vowels, one made with a
tense tongue, the other with a slack, but differing
in no other particular." In passing from one tense
to another tense, no slack stage is present. "If, for
instance, the tongue be slightly lowered from the High
Tense, we do not get a Slack vowel, but merely a
lowered Tense vowel, unless, of course, the tongue be
deliberately slackened, which is not at all necessary.'
It is astonishing that Wyld did not take the
obvious next step, that of identifying "long" with
tense, "short" with slack vowels. As a matter of
fact his own tabular representation shows all English
"long" vowels (except [a]) tense; all "short" vowels
slack; where he includes long-slack or short-tense,
the illustrations are taken from other tongues where
this quantity distinction is presumably preserved.
But nowhere does he appear to recognize the bearing
of his own statements.³
This identity of "long" with tense, "short" with
3 The phonological table of the New English Dictionary shows
the same reliance upon foreign words generally to furnish illustrations
of long-slack and short-tense, its editors seemingly blind to the
anomaly of requiring non-English illustrations for English pro-
nunciation!
OTHER FACTORS IN SOUND CHANGE
49
slack vowels, deserves more extended exemplification
and proof than is possible here. Its validity is not
necessarily wider than English.
That is, many
tongues undoubtedly employ real quantitative dis-
tinctions between vowels. Classic Greek and Latin
are traditionally held to have used such distinctions,
and this fact suggests the reason for the importation
of this nomenclature into English, so long held in
the Procrustean bed of Latin linguistic terminology.
Tensity or Stress, not Quantity
It would add to the scientific accuracy of English
phonology to discard altogether the modulational
terms long and short as applied to the two series
of vowel sounds, replacing them with some such
articulatory terms as tense and slack. In other
words, most of this matter of so-called quantitative
distinctions must be removed from the field of
modulation into the realm of articulation.
4
In the instances where what is regarded as quan-
tity in sound is indeed a modulational manifestation,
it is a manifestation of stress, not of quantity; or
of quantity only as a factor of stress. In the illus-
tration cited above, We meet JOHN today, and,
We MET John today, the tense vowel [i] of meet is
observably shorter than the slack vowel [e] of met.
4 The terms tense and slack are not to be insisted upon. If
Russell's terms mellow and metallic carry more significance to any
mind, they may be substituted. The important thing is to replace
a quantity (modulational) term by an articulatory term.
1
50
WHY ENGLISH SOUNDS CHANGE
But this length distinction arises out of the difference
in stress and is dependent upon it. In other words,
any stressed vowel in English will probably tend to
be longer in time of utterance than the same or
another vowel when unstressed. If the tense vowels
appear to occur in stressed position more frequently
than the slack, this also is explicable, as may appear
later on (Chapter V).5
We are left then with a tense-slack and a stressed-
unstressed distinction, as inheritors of what have
been called quantitative distinctions in English."
And so we finally arrive at the conclusion that every
important or considerable manifestation of the prin-
ciple of grading in English falls under the single
head of sound change resulting from the operation
of stress."
Types of Stress Grading
Sound gradings arising from the operation of
stress may be grouped under three heads: first,
5 Complete analysis of this matter would contemplate also
"checked" and "free" vowels, and would formulate the rules whereby
tense and slack tend to coincide with stressed and unstressed
respectively.
A former illustration, that of [e:] and [e], may require additional
comment. While for convenience the two sounds (compare [ɔ] slack
and tense, page 17) have been given the same symbol, they are
really distinct from each other, the difference lying precisely in the
matter of tongue-tensity. The same distinction may be drawn
between tense and slack [æ]. This tense-slack distinction is heard
more easily with high than with low vowels, a fact which may
account for the confusion.
7 Indications are not lacking that English is gradually altering
in the direction of a more resolved stress, and such indications, with
the possible causes for them, might well be made the subject of study.
OTHER FACTORS IN SOUND CHANGE
51
ablations due to unstress; second, obscurations or
suppressions from the same cause; and third,
strengthenings due to stress. While it will not be
possible to consider these types in detail here, a few
examples of each may be given.
8
Under the first heading will fall what is unques-
tionably the most extensive single linguistic change
in English, namely, the elimination of inflectional
terminations, in so far as this may be proved to be
a manifestation of stress. Closely allied with such
ablation is the loss of consonant elements from pre-
fixes, e.g., OE onweg, onbutan, ModE away, about;
or the loss of the prefix ge. It is noticeable that the
weak consonant sounds such as [n], [j], etc., seem
especially subject to such ablation, whereas the
stronger sibilant [s] has more than held its own in
suffix or prefix.
Vowels, being even weaker sounds than the
ablating consonants just mentioned, show an even
greater tendency toward disappearance in unstressed
position. It is quite possible that the loss of the
second element in OE diphthongs is an instance of
unstress ablation. Certainly this type of gradation
played a part in the simplification of OE hlaford,
hafoc, already referred to (page 30). And a host of
ModE words like separate [seprit], beloved [blavd] or
[blavid], show the same principle in force today.
8 Certainly stress would appear to be the main cause of these
ablations; but just as probably the intellectual element (see page 60)
was a factor.
52
WHY ENGLISH SOUNDS CHANGE
·
The second group of gradings based on stress
includes instances of the suppression or obscuration
of sounds. With regard to vowels, this tendency has
long been formulated into a rule, which may be
stated thus: In unstressed position, back vowels
become levelled as [ǝ]; front vowels as [ǝ] or [1].9
This rule is such a commonplace of linguistics that
it seems unnecessary to discuss it, except to point
out that in every instance the articulation becomes
or remains slack after the change. The rule thus
might almost be formulated: Unstress tends to
slacken a tense vowel.
Likewise with regard to consonants, unstress has
a definite effect, which may be summed up in the
statement, Unstress in consonants tends to voice,
stress to unvoice. The second half of this statement
anticipates the next section, dealing with the
strengthening of sounds by stress; but it is convenient
to consider the rule as a whole.
In the example have to [hæftǝ], given on page 29,
grading undoubtedly reinforces accord in producing
the breath sound [f], since the first syllable carries
considerable stress. Just the opposite is don't do it,
where the stress falls on the second syllable, with a
resulting merging of the [t] of don't with the voiced
[d] following.
9 The rule as here given differs slightly from, e.g., Wyld's state-
ment (SHE 155) which does not contemplate the reduction of front
vowels to anything but [1]. Yet we have statement [stetmənt],
eagle [igǝl], travel [trævəl] or [trævl), etc.
OTHER FACTORS IN SOUND CHANGE
53
It is somewhat difficult to exemplify this rule in
OE and ME, since several letters (ƒ, s, þ, etc.) carried
both the breath and voiced values, and it is difficult
to prove which were used in any given instance.
But in Mode the tendency is most clearly observed
in just such instances, because here the spelling has
less effect in determining the character of the sound.
Perhaps the best instance is ModE x, which is pro-
nounced both [ks] and [gz]. In execute, the first
syllable is unstressed, and we have the breath value
of x, namely, [ks]. In executive, the first syllable is
unstressed, and we have the voiced value [gz]. We
have exist, exhort, exonerate, with [gz], but exile,
extra, axiom, with [ks]. 1°
10
It is probable that this alternation of voiced and
breath consonants is far more prevalent than the
average individual, whose ideas of pronunciation are
involved inextricably with ideas of spelling, would
admit. Thus it appears doubtful that there is any
great distinction in popular speech between debtor
and deader, deafen and Devon. But the distinctions.
of spelling and meaning are so marked that this
identity would scarcely be admitted without labora-
tory proof.
10 Note that in all the above examples, x unstressed is followed
by a vowel (voiced sound). When the first syllable is unstressed
but a breath consonant follows the x, it is noteworthy that x has
its breath value [ks]. This fact is to be explained as an according
with the following consonant. Examples are extreme, extend, excess,
express, etc. So far as I know there are no instances in English of
unstressed x plus voiced consonant, but there is the German-
American dialect form exdreme [1gzdrim].
54
WHY ENGLISH SOUNDS CHANGE
This rule, which covers stress as well as unstress,
in consonants, takes us into the third division or
group of English gradings, a group including those
emphasizings or strengthenings caused by stress.
And with regard to stress in vowels, also, a rule may
be formulated. It may be stated thus: Stress in
vowels tends to tense, unstress to slack.
The second half of this rule has been exemplified
in a previous paragraph. It is exemplified also in all
those shortenings (i.e., slackenings) of vowels due to
the compoundings of word elements in English, as
wisdom, husband, bridal, stirrup, etc. The only
difference between these vowels and those of the
type of about [ǝbaut], lettuce [letis], is that in the
compounds vowels are not reduced invariably to the
neutral vowels [ə] and [1], but frequently show a
slackening less thorough-going.
The first part of the rule stated above, that stress
in vowels tends to tense, concerns us here particularly.
It has many exemplifications in English, but the
outstanding one, in my belief, is the regular length-
enings (i.e., tensings) of the vowel which took place
in the first (open) syllable of OE dissyllables.
If vowel slackening is admitted to be a phenom-
enon of stress grading, then this tensing clearly
belongs to the same group in general, as a reversed
counterpart of the first type. In other words, if
unstress be admitted to slacken a tense vowel, it
follows reasonably that stress will tense a slack one.
When OE mete [mete] (slack vowel) became ME
OTHER FACTORS IN SOUND CHANGE
55
mete [metǝ] (tense vowel) and finally ModE meat
[mit], the tensing undoubtedly took place because
of the accession of stress on the first syllable, the
very same factor which through unstress induced
or hastened the loss of the final syllable.
These tensings and slackenings will be considered
again in Chapter V, in their relation to the workings
of sound accord. They form the final example of
the operation of stress grading which will be given
here. The next task is to define the third general
force or principle affecting linguistic change, a prin-
ciple to which we may give the name conservation.
The Principle of Conservation
Thus far we have considered only forces making
for linguistic change, and the question may properly
be asked why, if these forces are so powerful, do they
not work more quickly and more completely, to
change the character of a language? Is there not a
force, corresponding somewhat to inertia in physics,
which to a large extent neutralizes and nullifies all
the tendencies toward linguistic change? It seems
apparent that such a force must exist, since in fact.
language changes with extreme deliberation and only
over long periods of time.
1
Does not the language itself, oral and written,
constitute such a force? It seems apparent that it
does; and so we shall recognize a linguistic principle
of conservation, defining this as the influence exerted
by the language itself, to impede and resist change.
56
WHY ENGLISH SOUNDS CHANGE
:
This influence is apparent in each of the two great
aspects of language, namely, speech and literature.
As with accord and grading, the operation of this
conservative principle is for the most part uncon-
scious or at least unreasoned. Take for example the
popular distinction between "easy" and "difficult”
in sounds. We call the Russian [sts] difficult or
even barbarous; the Russian applies the same term
to our [dž] and [b], and even our "easiest" sound [h].
What each means is that the sound does not conform
to his particular linguistic habit, i.e., to the body of
sounds actually in use in his linguistic group. Actual
physical ease or difficulty does not figure in the
situation. II

Conservation in speech sounds may well be named
linguistic habit, defining the term as indicating the
body of sounds in use by a group of speakers at a
given time. Linguistic habit is exemplified not only
by our judgments of easy and difficult in sounds,
but most widely in those changes incident to the
taking of words into one language from another.
Thus the dictionary itself is one mass of evidence for
the potency of such habit.
Change in sound values will come despite lin-
"If one be disposed to argue that English sounds are really
easier than, e.g., French, let him consider our word garage, pro-
nounced popularly garidge, where an extraneous sound has been
inserted into original [garaž], merely because English linguistic
habit does not favor [ž] initial or final in a word. In the same way
Russian will ordinarily replace [h] by xl, physically a much more
complicated or difficult sound.
[g]
OTHER FACTORS IN SOUND CHANGE
57
guistic habit. English loses its [x] and [3] French
its [b], German its [w]. English gains [S], and other
tongues likewise acquire new sounds. But such
changes take place only gradually and over long
periods of time. They occur in spite of linguistic
habit, not because of it.
In a primitive tongue, or one without a written
alphabet, this linguistic habit would constitute the
only check on linguistic change. But in most modern
languages, linguistic habit in sounds is powerfully
reinforced by another influence which in my belief
is scarcely less potent. This is the written word.
Spelling Influence
-
A great literature (Greek, Hebrew) is a force
toward stabilization of the vocabulary and grammar
of a language. It is safe to assert that English today
would be a perceptibly different tongue were it not
for the plays of Shakespeare and the King James
Version of the Bible. How far this stabilization
affects the values of sounds, however, it is difficult
to say. Probably when such a literature is orally
transmitted the influence upon sounds would not be
great. But where this literature, or indeed, any
message at all, is transmitted through the written
word the case is altered.
For most important, as regards sound stabiliza-
tion, is the influence of the written word, or spelling,
an influence operative almost universally in modern
languages by reason of the invention of printing and
58
WHY ENGLISH SOUNDS CHANGE
}
Į
the extension of literacy. This influence, as affecting
sound change, has been minimized by many linguists
who regard tense vowels almost to the exclusion of
other sounds. 12 For this reason, without attempting
to enter deeply into the subject, I shall state my
belief that the influence of spelling on sounds is
extremely potent, and give a few illustrations of what
I believe to be the fact.
For example, it is interesting to note that where
a single consonant represents both voiced and breath
sounds, these alternate easily. Thus we have north
[nɔr]] and northern [nɔrðårn], extra [ɛkstrə] and exist
[egzist], caps [kæps] and cabs [kæbz], etc., etc. But
where two separate symbols for voice and breath are
employed, alternation is consciously resisted. To say
debt [det], and debtor [dedor], would be regarded as
barbarous by the very people whose pronunciation
unconsciously approximates that very thing.
Spelling influence may turn back the tide of an
according change. How else shall we account for
[Indiǝn], [odiǝs], where [Indžiən], [odžiəs] (Wyld
SHE 160), were standard, recommended forms at the
beginning of the nineteenth century? There is evi-
dent a semi-conscious desire to keep as close as pos-
sible to the symbol, which desire will prevent sound
12 The alterations in the tense vowels came about so gradually
that this spelling influence was not felt. It is apparent in English
only when we compare the spelling values of our vowels with those
of French, German, etc. Alterations in consonants and slack vowels
involve substitution of sounds, a much more radical change, from
the spelling aspect.
OTHER FACTORS IN SOUND CHANGE
59
accordings or other linguistic change involving a
departure from spelling tradition, or even initiate
disaccordings to approach such tradition. Thus one
may hear religionists saying [kristiǝn] for [krist §än],
purists saying [Isju] for [Isu], and the man on the
street saying [ɔften] for [ɔfn]. ¹³
13
A most amusing illustration of the potency of
spelling is found in a little book by Robert Bridges.
His "Tract on the Present State of English Pronunci-
ation" is a product of the real agony this poet
experienced on seeing for the first time, set down in
uniform symbols, what English pronunciation really
looks like. In terror lest this monstrosity supplant
the spelling he loves, Mr. Bridges devises a new
"phonetic" alphabet, into which one "eye spelling"
after another finds its way, until he finishes with
some 70 symbols which are to him "readable," that
is, traditional.
As Mr. Bridges was so rudely forced to perceive,
traditional spelling does not express English pro-
nunciation. Reluctantly he admits that something
must be done, to ward off the approach of a scientific,
phonetic alphabet. So he makes a compromise,
keeping all the spelling illusions he possibly can,
throwing a few to the wolves of science. 14 The book
13 Not all "purisms" are disaccordings. The purist pronunciation
of rise (noun) as [rais] accords far better than the ordinary [raïz] in
the phrase where it is commonly employed, give rise to.
14 One feature of Mr. Bridge's alphabet deserves unstinted praise.
His new symbols are beautiful in themselves. If the phoneticians
would abandon their upside-down e's, backward c's, and ugly j's,
60
WHY ENGLISH SOUNDS CHANGE
may be recommended to all who would question this
matter of spelling influence as a most appealing, even
pitiful, human document, by a lover of English
spelling. Its flavor is intensified by the fact that
we all agree with the writer, at least in theory. We
all love the spelling we have acquired at so much cost.
Even the phoneticians have bowed to this idol.
In constructing their graphic systems, no attempt
ordinarily is made at any logical representation.
Convenience would dictate, for instance, some uni-
form mark to distinguish vowels from consonants,
breath sounds from voiced, tense from slack, and so
on. Sweet made some approach to a logical system
in his "History of English Sounds," but his system,
based on Ellis, rapidly became out of date, and
happily so. It is probable that the practical advan-
tages of a "spelling system" outweigh those of a
logical system, so potent is the compulsion of the
written symbol.
The Intellectual Element
Our fourth principle of sound change might be
omitted entirely without great damage to the argu-
ment. Conscious or directed thought plays so very
small a part in linguistic change as to be, broadly
speaking, negligible. When it does enter its effect
is largely permissive; that is, it may facilitate the
operation of one of the other principles, but seldom
for symbols which possessed both beauty and practicality, a really
phonetic spelling might come nearer adoption.
OTHER FACTORS IN SOUND CHANGE
61
$
if ever will it initiate sound change. Two examples,
which will now be given, are fairly typical.
The loss of inflectional elements (terminations
mainly) in English was unquestionably assisted by
the fact that these inflections were not necessary in
the conveying of meaning. Indeed, Jespersen seems
to regard this as the main cause of such loss, associa-
ting the simplifications with the growth of a fixed
sentence order. But this can scarcely be the case,
since a fixed sentence order is by no means peculiar
to those tongues which have lost inflectional elements.
The intellectual element here is to be regarded as
permissive, as facilitating the operation of unstress
grading.
Secondly, the double negative shows the intel-
lectual element operating in conjunction with recur-
rence accord. The former tends to emphasize the
negative, or at least to recur to it for more effective
transference of meaning. Accord reinforces this
tendency by adding its own tendency to repeat
sounds, and the result, in all the periods of English
development, has been the double, often the mul-
tiple, negative.
-
In this instance we see the intellect in a double
rôle, for having facilitated the development of the
double negative it began to apply the thumb-rule of
logic to it, calling I haven't got no cake a logical
affirmative, through the cancelling of the negatives
against each other. (One wonders, would three
negatives be allowable, as adding up to a final
62
WHY ENGLISH SOUNDS CHANGE
affirmative?) Again it appears probable that the
real argument against the double negative is the
unconscious tendency to eschew overemphatic modes
of speech (compare the academic dislike of love for
like, etc.) rather than any dependence upon abstract
reasoning.
It is difficult to find a term for this unconscious
rationalization which will not seem to lift it into a
plane of conscious tinkering with the language, and
as we have just seen, such tinkering is frequently
conscious so far as the precept is concerned. But
seldom if ever are the reasons adduced for the use
of certain locutions ("good English," "propriety in
speech," etc.), the real reasons. Thus the intellectual
element must be considered to lie wholly or at least
largely in the realm of unconscious mental processes.
Summary
We are now ready to summarize the effective
factors or principles of sound change, and this is done
in Table V. Succeeding chapters will attempt to
review the large or outstanding facts of English
sound change and to show the application of these
principles, particularly of accord, to them.
It should be emphasized that each of these prin-
ciples is universal in the sense that every radical
change probably involves all of them to at least a
slight extent. Thus when a change is referred to a
single principle, this is not intended to mean that
no other principle is involved. It is impossible as yet
OTHER FACTORS IN SOUND CHANGE
63
TABLE V. FACTORS IN SOUND CHANGE
I. Accord (basis, sound articulation)
1. Approximation
A. Shading of vowels
B. Substitution in consonants
2. Ablation
A. Vocalization of consonants
B. Loss of consonants
3. Mutation (umlaut)
4. Recurrence
A. Repetition
B. Rhyme
C. Assonance
D. Alliteration
II. Grading (basis, sound modulation)
1. Pitch (not important in English)
2. Quantity (not important in English)
3. Stress
A. Ablation through unstress
B. Obscuration through unstress
C. Strengthening through stress
III. Conservation
1. Linguistic habit (sounds)
2. Spelling influence (orthographic)
IV. The Intellectual Element (largely permissive)
64
WHY ENGLISH SOUNDS CHANGE
!
to trace accurately all the elements entering into all
changes, but it is safe to say that accord, grading,
conservation, even the unconscious operation of the
intelligence, are universal causes of sound change.
It is to be hoped that this analysis of sound change
may lead to deeper and more far-reaching analyses
of all the linguistic principles, so that phonology
will finally abandon its present status as a descriptive
science, and become an analytic one. Linguists have
too long been at pains to confine themselves to
descriptive research. The facts are available. It
remains only to correlate them, in order to make
linguistics rational instead of heterogeneous.
CHAPTER IV
ACCORD IN ENGLISH CONSONANTS
Few Instances Analyzed
Ideally, a demonstration of the principle of
accord, as formulated in Chapter II, should involve
the reconsideration of every type of sound change in
the history of the English language, to determine
the extent to which accord is an efficient factor.
Such an analysis is a task far beyond the aims of the
present study.
Moreover, in a majority of instances the according
character of a change is obvious as soon as the prin-
ciple is recognized. Take for example the progressive
simplifications of OE hn, hl, hr, and hw. Originally,
probably in primitive Germanic, these are believed
to have been true consonant combinations, [xn],
[x1], etc. Later the [x], a low-back consonant accord-
ing badly with the sounds following, was weakened
to [h], giving [hn], [hl], etc. Next the sounds were
simplified or accorded yet further by losing the first
element and becoming merely the breath semi-vowels
[n], [1], etc. Finally [n], [1], and [R] became voiced,
obviously to accord with the voiced character of the
vowel which would follow.¹
¹ An interesting phase of this development is the fact that
OE hw, ModE wh, apparently lagged behind the others of this group,
65
66
WHY ENGLISH SOUNDS CHANGE
Likewise accord is easily apparent in the progres-
sive changes which ushered the new sound [S] into
English phonology. In OE there was a consonant
combination sc, in which the c [k], originally [k],
had probably by now become [k]. This [k] still
accorded badly with the preceding [s], and finally
became ablated, at the same time causing a slight
retraction of the [s] to [S]. The stages were in all
probability [sk,], [sk₂], [sj], and [S].²
Since most accordings are of so obvious a nature,
it will be unnecessary, as well as impractical, to make
the present study an exhaustive catalog of English
sound changes. Therefore we shall consider in this
chapter only especially significant or outstanding
examples of consonant accord.
Elimination of Gutturals
Of these, one of the most interesting is the loss
of the OE gutturals h [x] and g [3]. This ablation
illustrates a wider aspect of accord than has yet
been mentioned, and illumines the development of
languages generally. The loss of these two OE
so that today we still have the breath sound [w], or, as it is com-
monly written, [M], in most speech areas of English. The explana-
tion of this lagging would seem to be that w, being a back sound,
accorded with [x] more nearly than did n, l, or r. Hence its simplifi-
cation was less early and less thorough-going.
That all these stages, or at any rate the first three, occurred
at some time before the ME period is evidenced by the fact that
words borrowed from ON (skip, skirt, sky, skull, etc.) do not exhibit
this according, although these were early importations into English.
Just why accordings should take place at one time rather than
another is too broad a question to be considered here. It will be
discussed in Chapter VI.
ACCORD IN ENGLISH CONSONANTS
67
sounds bears upon such apparently unrelated matters
as the loss of [w] in German, the elimination of gut-
turals in Hebrew, and the typical "neutral" vowels
in various tongues. It establishes accord as exem-
plifying what may be termed the centripetal ten-
dency in sound development, and on all these counts
deserves to be considered in some detail.
The fact of guttural elimination in English is
familiar to every student of historical linguistics.
This elimination must have been well under way by
the OE period, since OE initial h is already [h], not
[x] (OE hund, ModE hound; cf. ModG hund). OE
initial g too had begun to change in early Germanic.
Before front vowels it had become, at the beginning
of the OE period, [g], [gz], or perhaps even [j] (OE
geard, ModE yard); before back vowels this sound is
thought to have remained [3], but only until the late
period of OE, when it became [g] as at present (OE
god [30d]; late OE god [g,od]; ModE good [g,ud]).
Medially OE h [x] was very frequently found in
high-front surroundings. There were many words,
for example, in iht (niht, liht, gesiht, etc.). It is
evident that [x] in these surroundings would com-
pletely fail of accord, and hence it is not surprising
that here [x] early fronted to [x2], then to [j], and
then disappeared, leaving a lengthened vowel which
later diphthongized.
OE niht³ [nixt], ME night [nijt], then [nit], ModE night [nart].
3 Should it be questioned how such an awkward and barbarous
combination of sounds as [ixt] could occur, several replies may be
68
WHY ENGLISH SOUNDS CHANGE
Words containing [x] after front vowels experi-
enced the same ablation whether or not a t followed,
as did also [x] when final (OE neah, ModE nigh; OE
heah, ModE high). In the oblique cases of words
containing final [x] the [x] had already been ablated,
so that this sound in the nominative is itself merely
a survival. The ablation of a word's final sound has
already been explained, page 32. It would be gener-
ally true that a word following final [x] would begin
with a sound higher and more front, and such a
sound would act to produce an ablation according,
just as a t following within the word would do.
When [x] was combined with back vowels, its
ablation naturally occurred somewhat later, since in
this event there was less pressure toward an accord-
ing. When the ablation took place, it occasionally
involved a slight lowering of the vowel, in addition
to its tensing. And now and again the lip-rounding
present in most of the back vowels induced a sub-
stitution, giving the lip consonant [f]4 in place of
the former guttural [x].
made. First, [x] may have been raised to [x2] or even to [j] before
the vowel was raised to [i]. Second, as stated on page 23, the human
tongue is an agile member, capable of almost any distortion, and
with practice can bend itself into well-nigh impossible patterns and
even call them easy. For a general discussion of such "disaccord,"
see page 122.
4 This replacement of lip-rounding by [f] illustrates the affinity
between the lip and back sounds. It is not properly an according
but rather what may be termed an "infantile substitution." The
influence of accord is seen, however, in the fact that a high-front
consonant, nearer the consonant norm, replaces a former low-back
sound.
ACCORD IN ENGLISH CONSONANTS
69
Mode shoe [Su]
Mode bought [bɔt]
ModE rough [raf]
OE scoh (skox]
ME shoo [So]
OE bohte [boxte], [boux.tɛ] ME boughte [bɔtə]
OE ruh [Rux]
ME rough [ruj], [ruf]
The OE sound g [3], the voiced equivalent of [x],
disappeared under much the same conditions, except
that [3] occurs less frequently in high-front sur-
roundings. OE has few or no words in, e.g., igt; the
termination ig had probably already [32] or [j]. The
sound [3] occurred most often with back vowels or
medially between vowels, but in whatever surround-
ings, like [x], it underwent ablation.
ME owen [ɔwǝn]
OE agan [azan]
Mode own [on]
OE nægl [næ31] ME naile [næjlə] or [nælə] ModE nail [nel]
Finally, then, we have a language from which the
two sounds [x] and [3] have been eliminated, so com-
pletely that it requires some practice for an English-
speaking adult to pronounce properly German ich
or tage, Scottish loch or nicht.
It is easy to see why these two sounds should
disappear. In the first place, they accord more or
less badly in any sound pattern. Secondly, the
motion of the tongue in producing them is scarcely
an easy motion compared with that of the flexible
tip or point in pronouncing [d] or [s]. Unlike the
low-back vowel [a], which involves merely an opening
of the throat, they require an appreciable muscular
effort. And finally, they are on the extreme fringe
of the sound-producing area, and their elimination
would therefore tend to make the English speech
70
WHY ENGLISH SOUNDS CHANGE
1
mechanism more compact and hence more economical
and efficient.
Other Languages Compared
If accord is a universal tendency in language, we
should expect to find this phenomenon of guttural
ablation in other dialects as well as in English; and
in this we shall not be disappointed. It is illustrated
to a greater or less extent in the history of all the
Indo-European language families, as well as the
Semitic and many other tongues. It is in all respects
an according change, both in its narrower aspect as
accord within the individual word or sentence, and
in its wider application as a simplification of the
entire system of speech production. The process
might well be compared to centripetal action, in that
it narrows the area of sound production. It might
also be compared to the sloughing off of unnecessary
muscular actions in the perfecting of such activities
as learning to walk or to play the piano.
Ablations such as the one just analyzed, whereby
English has lost its gutturals, would be expected to
take place normally at the rear of the speech area
because, as has been pointed out, tongue action here
is more difficult. The rear mouth area is not so well
adapted to easy utterance as the tongue-tip region.
But neither are the lips so facile in action as the
tongue tip; [m], [b], [p], require appreciably more
muscular effort than [t], [d], [s]. If the experiment
be tried of reading rapidly with the mouth becoming
ACCORD IN ENGLISH CONSONANTS
71.
progressively more relaxed, it will be found that the
lip sounds will be entirely lost even before such back
sounds as [k] and [g] disappear.
It must not be supposed, therefore, that in every
instance the "fringe sounds" (see page 18) which are
eliminated will be gutturals. German, which in
several dialects has kept guttural sounds, has lost
the consonants [w] and [m], with their strong lip
action. Thus it has acquired a "center of gravity"
slightly farther back in the mouth than English,
which has lost its gutturals and retained its lip
action. French, on the other hand, has no gutturals,
but includes even more lip action (huile, oiseau) than
English, thus exhibiting a "center of gravity" yet
farther front.
These observations concerning modern French,
German, and English offer a fertile field for more
detailed study. It would, I believe, appear from
such analysis that German vowels, as well as con-
sonants, are characteristically or preponderantly
farther back than English vowels, and these latter
than French. For example, where [ə] is our "neutral"
vowel, in German the neutral vowel would be found
to be slightly farther back, and nearer to [o] or [ɔ],
and in French slightly farther front, and nearer to
[e]. Where fronted [k] becomes [s] or [S] in French,
and in English [t s], in German this sound does not
front, but remains [k].5
5 It may be asked, why do not all mouths have a normal, standard
center of gravity, if accord is a universal principle? Why, for
72:
WHY ENGLISH SOUNDS CHANGE
A Theory of Linguistic Development
In the widest aspect of all, these observations
point to a whole theory of linguistic development.
A primitive speech may be assumed to contain sounds
drawn from a wider speech area, and more awkwardly
put together, than will be the case after speech
development has taken place. Such accord as exists.
will be found mainly of the recurrence type. The
necessity, on the part of an undeveloped intelligence,
for being understood by other undeveloped intelli-
gences, will lead in general to elaborate speech
effort."
As the speech area is wider in the primitive than
in the mature stages of a given language, so also the
sound combinations will be more awkward and cum-
brous. The vocal aspect of language will exemplify
inefficiencies analogous to those of which Jespersen
wrote in his "Progress in Language." In addition
to the verbal circumlocutions and repetitions there
will be also maladjustments of sounds, sound patterns
which completely fail of accord. But gradually the
example, should German. "cohere" about a point farther back than
English? If accord were postulated as the only principle of linguistic
change, this question might constitute a real problem. But every
language is the resultant of many forces or principles, which must of
necessity lead to systems varying from one another.
• This does not necessarily mean that sounds are more in number
in a primitive speech, although this also may be the case; it means
that their range, i.e., the speech area, is apt to be wider. If a child
is asked to talk in a "new language" of his own, he will often "mouth"
his sounds, making various facial contortions and in other ways
widening the normal speech area, possibly even introducing inspir-
ational sounds as well as the normal expirational ones.
ACCORD IN ENGLISH CONSONANTS
73
influence of accord will be felt, against the weight of
linguistic habit, and finally the language will become
more compact and at least comparatively efficient.
Primitivity in speech, Professor Krapp has said,
is not a matter of chronology but of social experience.
Thus progress in speech adjustments depends upon
increased culture, wider national relationships, the
movements of trade, science, invention, and so forth.
As these develop and increase the intelligence of a
people, several results accrue. The mechanism of
the native speech becomes simplified and hence vastly
more effective, but this advance is partially neutral-
ized by the fact that the vocabulary increases enor-
mously, introducing innumerable new elements to be
assimilated into the existing body of speech. The
system of grammatical devices, only slightly affected
by vocabulary increase, becomes simpler and more
unified. A mature language thus presents what may
be called an enormous complexity of simplified ele-
ments, elements of sound, inflection, syntax, even of
word meanings.7
The course of this linguistic development may be
seen traced in the history of nearly any language
7 Superficially it might seem that languages such as Finnish and
Bantu, whose sounds accord to a large extent, contradict this
theory, since the degree of social experience which they reflect
could scarcely be called high, compared with the Indo-European
tongues. But the seeming accord of these comparatively primitive
tongues is not in fact simplification, but the manifestation of an
(extremely rigid system of inflection, which itself is symptomatic of
an undeveloped language. Recurrence accord (see page 36, note)
is differentiated from the other types in that it represents generally
an immature period of language structure.
San
Timo
Cp??, note & for
San Accordi
P
#7
14
!
po bester
74
WHY ENGLISH SOUNDS CHANGE
•
where sufficient records exist to make comparisons
possible. In this sense languages might accurately
be said all to be approaching the same end—that
synthesis which we have named accord.
Of course much remains to be done before such
a theory as the one just propounded could be called
established. Broad comparative studies should be
made, embracing languages of different stocks, to
study the direction of change. For example, a most
significant study could be made of two entirely
unrelated tongues such as English and Biblical
Hebrew, each the fruit of long centuries of social
development. It could be shown in detail how the
linguistic expedients adopted by each tended to
develop similarly, how the sound systems converged
into the same path, how the inflectional devices, the
syntax, even the semantics, have tended toward
syntheses of remarkably similar types.
All such syntheses are epitomized in the single
most significant consonant change of English, name-
ly, the ablation of the gutturals. We can allow our-
selves here only a hint of the wider ramifications of
this change, if we would avoid being led far beyond
the limits of this study.
Palatalization of [k]
The next consonant change to be considered,
only less important than guttural ablation, also has
ramifications far beyond English, connecting as it
does our tongue with all the so-called "satem"
ACCORD IN ENGLISH CONSONANTS
75
group of Indo-European dialects. This change is the
fronting of OE c.
Superficially it might seem possible to refer this
change to the same analysis as the one previously
described. It might seem reasonable that when the
gutturals [x] and [3] were lost, the next step would
be the elimination of [k] and [g], the pair of consonants
next most remote from the consonant norm. But
no such orderly process can have taken place. These
consonants were not eliminated save in front-vocal
surroundings, and even this partial loss did not fol-
low, but preceded, the loss of the gutturals. Hence
the fronting of OE [k] and [g] is to be regarded as an
instance of internal speech accord, rather than a
narrowing of the speech area as a whole.
The change from OE c to ModE ch is one of the
trite, almost threadbare, examples of "assimilative"
sound change. It is cited in every discussion of the
subject. Ordinarily it is cited uncritically and with
no detailed analysis. And yet this one change, with
its analogs in other languages, presents a mass of
problems, and corresponding opportunities for orig-
inal investigation. Why should OE [k] have become
[ts], whereas Sanskrit [k] became [s], and Latin [k],
French [s] or [s]? Why did OE [k] change, but not
OHG [k]? (At least a partial answer to this question
has been suggested in the previous discussion of Ger-
man as a "back" language.) Why has not [k]
changed in ModE keen, key, ken, etc.? It were
presumptuous to promise that all these questions
76
WHY ENGLISH SOUNDS CHANGE
may be answered offhand by reference to the prin-
ciple of sound accord; and yet perhaps some addi-
tional light may be thrown upon them.
It seems scarcely necessary to review the course
of the c-ch development. According to the current
statement, OE c [k], whenever it was followed by an
original (i.e., un-umlauted) front vowel, was [k₂], and
fronted to [ts]. If the vowel following the c had
been fronted by undergoing umlaut or other such
change, no fronting took place.
OE cinn [k,in]
OE ceap [k,eap]
OE cene [k,enɛ]
ME chin [tSin]
ME cheap [tsepl
ME kene [k,enə]
Mode chin [tSin]
ModE cheap [tſip]
ModE keen [k,in]
Ma
Satem and Centum
Changes regarded as similar to the c-ch develop-
ment are found to be widespread in language gener-
ally. All the branches of Indo-European are classified
as “satem” or “centum" tongues, according as IndE
[k] in words such as Sanskrit satem, hundred, has or
has not been replaced by [s], [ts], [s] or [ts]. In
the centum tongues the [k] has survived in all sur-
roundings; in the satem tongues it has been replaced
at least partially by point or blade consonants. Of
course not all the centum dialects have remained
centum. Thus French, from centum Latin, has had
three varieties of k-development, exemplified by the
three words corps [k], chanter [$], and civile [s].
Germanic is regarded as a centum family; yet in
T
ACCORD IN ENGLISH CONSONANTS
77
English [k] has been lost in instances such as those
just noted. In Modern German, on the other hand,
we have the curious spectacle of a language which has
experienced probably more consonant shifts than any
other IndE dialect, still clinging to the primitive
Germanic k (probably [k]) before front vowels
(kirche, church; kinn, chin). And this fact inten-
sifies our impression of German as a "back"
language.
8
But to return to the English c-ch sequence. Is
this change a simple matter of substitution accord,
analogous to the civilis-civile sequence? And if so,
why do we have [ts] and not [s], a simple sound?
The writer wishes to suggest that the determining
factor in changing OE c to ModE ch [ts] rather than
to [s] or [S] may reasonably have been a slight
i-glide (what the French call a sound mouillé) after
the [k]. The i-glide was present wherever OE c was
followed by tense or slack i or by diphthongs. It
was not present where c was followed by a single
vowel sound other than i, but appears frequently
to have been present with final c.9
8 Here is to be found a useful corrective to the attempts of some
linguists to determine the "primitive" or "mature" character of
a given language by some one or some few phonetic characteristics.
If the gutturals only be considered, German is far more primitive
than English; if the other consonants, it is far less so. As before
stated, primitivity must be determined by criteria other than these.
A given language shows its maturity by its general compactness
and simplification, not in any one sound loss or survival.
• In all discussions of final consonants it should be remembered
that in most or in many uses the "final" consonant was followed by
a vowel, the vowel of the inflectional termination.
#..
!
61
.7
78
WHY ENGLISH SOUNDS CHANGE
An i-glide in combination with any variety of
[k] will tend to change into [ts]. Take for example
our modern word cart. It could not by any possi-
bility pass directly into chart. But consider the word
in its Southern dialect form cyart [k,jat]. It will at
once be seen that this form could easily pass into
chart, and might indeed have done so except for the
restraining effect of the standard form cart. A little
experimentation will demonstrate that the same
thing holds for [k] plus front vowels. [kæt] could
not become [tsæt], nor [ket] [tset]; but [kjæt] and
[kjet] might conceivably be changed thus."
Let us consider very briefly a few exceptions to
the fronting and loss of OE c. First there are words
like OE cynn, ModE kin, where the rounded vowel
may offer some excuse for the exceptional develop-
ment (cf. cinn-chin). Then there is swylc-such, an
exceptional word on several counts; but the [ts] in
ModE is probably explainable by the fact that a
usual OE form was not swylc but swylce. But the
mass of exceptions are words like cæg-key, cene-keen,
centisc-Kentish. In all these the c is followed by a
single vowel, not a diphthong nor i. A cursory
examination seems to indicate that OE c combined
with a single front vowel (except i) regularly does
not front. This would seem evidence that the
10 Much the same reasoning should apply to languages where
[k] becomes [ts]; but in French the change seems simply a matter of
approximation accord (substitution). Note, however, the according
shown in the fact that French [s] is used with the highest, farthest
front vowels, while [§] is used with lower and less front sounds.
ACCORD IN ENGLISH CONSONANTS
79
fronting of OE c is indeed dependent on some factor
other than the mere frontness of the vowel. I
The great mass of regular developments from OE
c to ModE ch is exemplified by words like ceap-cheap,
ceorl-churl, cinn-chin, birce-birch, stenc(e)-stench. In
all of them there may have been present a slight
vowel glide after the [k], a sound which we may
represent by [j].
The change then becomes [ts] for former [kj],
and we have two substitutions. There is first the
substitution of [t] for [k], which latter had already
fronted to [k] to accord partially with its neighbor
sound. And next there was the substitution of [S]
for the i-glide, an alteration precisely analogous to
that involved in [tsun] for [tjun]. The popular
pronunciation [tun] could not become [tsun], any
more than [k,en] could become [tsin]. 12
To summarize, the presence of [ts] in these words
seems to be conditioned upon the former presence of
a slight vowel glide after the original c. The [S] fol-
lowing [t] is not a direct substitution, but appears
I I
Wright (OEG 153-155) enters into what seem quite unneces-
sary refinements of derivations, to separate "guttural" from
"palatal" [k]. The simple fact is that it is easiest to enunciate a
"palatal" [k] with a "palatal" vowel now, and that this condition
probably obtained in OE also.
12 This analysis, if correct, separates English from, e.g., French,
where direct substitution of [s] and [S] for former [k] has taken
place. It is regrettable that the breadth of our subject prevents the
elaboration or complete proof of many suggestions like this one, but
it seems a better alternative to give an unelaborated outline of
accord than to present a partial statement of the principle. Further
investigation must establish or discredit many of the detailed
suggestions.
80
WHY ENGLISH SOUNDS CHANGE
after [k] has been replaced by [t]. The entire process,
however, is an example of the operation of approxima-
tion accord as manifested in substitutions of sounds.
OE [g] Before Front Vowels
Since [g] is the voiced equivalent of [k], it might
be supposed that its history would be parallel with
that of [k]; but this was not the case; OE [g] failed
to become [dž], the voiced equivalent of [ts].
Late OE [g], as has been mentioned, had two
values, [g], before back vowels, and [g] before [i],
[I], [j], etc. This [g] became [j]. OE gear gives
ModE year, OE giest, ModE yeast, etc. This develop-
ment indicates that the i-glide, operative in words
with [k], was so strong in the g-words as to swallow
up the [g] sound entirely. This may be because [g], a
voiced sound, is closer in character to [j] than is [k].
In French also, Latin g developed differently from
Latin c.
Where c had three resultants, g had but
two, exemplified in the French words garrulité
(L. garrulitas) and genou (L. genu). In other words,
Latin g before back vowels remained; before front
vowels it became [z]. This fact will be discussed
further in the next section, which deals with [s]
and [ž] in English speech.
Medial [S] and []
The c-ch development helps to make easily
understandable the very large group of ModE words
where a former t, s, k, etc., has become [S] when
ACCORD IN ENGLISH CONSONANTS
81
·
followed by the i-glide already mentioned. A vol-
ume could profitably be written on the accordings
involved in the extension of the sound [S] in ModE,
its various sources and orthographic representations.
The "regular" development of this sound from OE
sc has been sketched at the beginning of this chapter.
But [S] in ModE is of more frequent occurrence in
numerous groups of words which have entered the
language since OE. In all these words, [S], a blade
sound just behind and below the breath point con-
sonants, is substituted for one of them (or for [k])
to accord more nearly with a following high vowel,
[u], [i], [1], or [j]. To use the termimology of Chapter
II, the first element in the sound sequences A2, B3,
or A2, A4, or A2, B5, has been altered to A3, a posi-
tion which makes for greater sound accord in every
instance. 13
The words sure, ocean, sensual, social, mansion,
anxious, mission, issue, and nation, (Krapp, PSEA
124 ff.) indicate the multiplicity of types under this
heading. It will not be necessary here to analyze
closely the accordings involved in each, as these are
easily apparent upon a little consideration.
While [ž] is the voiced equivalent of [S], it is a
much rarer sound in English speech, occurring almost
invariably in the combination [dž]. This sound [ž]
has not arisen from any OE source as has [S], and
13 It is not always only the consonant which is lost; sometimes
the vowel is lost (e.g., Christian) and sometimes both vowel and
consonant are replaced by [S] (ocean, cf. oceanic).
82
WHY ENGLISH SOUNDS CHANGE
its comparative rarity is therefore not a matter of
accord but of linguistic habit (page 56). The
English-speaking peoples are not unaccustomed to
the sound [ž] medially (treasure, usual, vision) but
its use is avoided in other positions, [dž] being sub-
stituted for it. Thus the recent borrowing garage
[garaž] becomes in popular speech [gæridž] or
[gǝrádž]. This substitution might properly be called
a disaccording due to linguistic habit; there is of
course nothing inherently more difficult or less
efficient about [garaž] than about [gæridž].
According is seen, however, in those words which
do contain [ž], in that the originals from which such
words (treasure, vision, gymnasium, etc.) were bor-
rowed had [z], not [ž]. The high-point consonant
has retracted, just as [s] and [t] were shown to have
done, to accord with a following high vowel. The
difference is that here, voice is present in both
original and derivative sounds.
General Conclusions
One does not finish discussing the operation of
accord in English consonantal changes; one merely
stops doing so. There is no end to the accordings,
individual and general, expressed in such instances
as moths [mɔöz] beside deaths [debs]; as anchor
[æŋkr] and anger [ængǝr], where the [n] becomes [n]
to accord with the back sounds [k] and [g]; as the
two popular pronunciations of fifth [fiþ] and [fift];
and there are innumerable other such examples.
ACCORD IN ENGLISH CONSONANTS
83
This chapter has reviewed briefly merely a few
of the consonant accordings of English phonological
history. We have seen the general tendency of
language to restrict its speech area, as exemplified
in guttural ablation; the tendency of [k] and [g],
both excessively unstable sounds, to pass their limits
of variation and become different sounds, when im-
pelled by the according tendency; the falling to-
gether of such near-by pairs as [s] and [8], [t] and
[j], and the consequent extension of the sound [S]
in ModE. Linguistic habit has been shown to
play a part in some consonant changes. Stress has
not been found important here, or at least, it has
not been found necessary to utilize the principle of
stress grading in making explanations. The nearly
uniform first-syllable stress from late OE well into
the ModE period, has made stress a constant factor
during these periods.
In conclusion, let us return to the glacier simile
employed in a previous chapter. The motion of a
glacier, according to recent research, is not smooth-
flowing like a river. Rather it is a somewhat spas-
modic alternation of pressure and release; that is,
pressure accumulates at a given point until it is
sufficient to overcome a specific obstacle, or the in-
ertia of the mass, or both; and then there is a burst.
of forward motion, if anything so deliberate may be
called a burst. Is there not here an analogy to the
process of sound change? Deliberate in the extreme,
nevertheless language does move toward greater
}
84
WHY ENGLISH SOUNDS CHANGE
accord, as the pressure of the unconscious urge
toward ease in speech becomes irresistible.
The glacier does not move all its mass equally;
motion is slower along the edges. So in language,
changes are seldom uniform or complete. Tenden-
cies which are strong enough to operate in one in-
stance fail to operate in another which seems nearly
related to the first. OE c becomes [ts]; g becomes,
not [dž], but [j]. ModE [t] plus [1] becomes [S];
ModE [d] plus [1] remains. 14
The scientist cannot yet predict precisely how
far a given impulse will carry a glacier; and it is the
same in language. No one, at least in the present
state of linguistic knowledge, could have predicted
beforehand that OE [n] would become ModE [n],
while [m] would remain [m]. No one could have
predicted that [S] would arise from OE [sk], while
[ž] would remain a sound foreign to the genius of
English speech. To vary the figure, sound change
may be compared in this respect to a rudderless boat.
When the engine moves, we know that the boat will
alter its place, and we know why this occurs, and the
general direction of the motion; but we cannot pre-
dict the precise point at which the boat will bring
up. That knowledge, with respect to sounds, may
be approximated by further investigation.
14 That the according tendency was and is present is shown by the
frequent pronunciations [odžas], [hidžas], [Indžan], for [odiǝs], [Indiǝn],
etc. Again the presence of voice in the [d] may account for the failure
of the tendency.
CHAPTER V
ACCORD IN ENGLISH VOWELS
Theories of Sound Change
Attempts to express the general nature of sound
change, or to state a reasonable cause or causes for
it, have been surprisingly few and feeble. For exam-
ple, it is sometimes intimated that sounds tend
toward articulations progressively farther front in
the mouth (palatalization). The analysis of English
consonants made in the previous chapter, an analy-
sis based on facts generally known and established,
by no means bears out this theory.
The consonants of English do exhibit a tendency
toward compactness leading to the elimination of
certain fringe sounds. Where substitutions take
place, moreover, they generally involve a closer
approach to the consonant norm, the point of the
tongue. But on the other hand, instances have been
adduced, such as [S] for [tj] or [sj], where a point
consonant has suffered retraction toward the rear of
the speech-producing area. And other changes, such
as the voicing of hn, hl, and hr, have nothing to do
with frontness.
It seems apparent, therefore, that sound change
is not primarily or consistently in the direction of
fronting. Let us then examine another attempted
85
86
WHY ENGLISH SOUNDS CHANGE
generalization, namely, that sounds tend toward
articulations progressively higher in the mouth (also
called palatalization).
This generalization has been a common one; in-
deed, so common that efforts have been made to
explain the alleged rule, or to point out its causes.
None of these explanations' has met with any general
acceptance, and as I can see no weight of conviction
in any of them, I shall notice them no more here
than to run over several by name.
There is, first, the occupation theory, by which
the spread of more sedentary, more indoor, less
vigorous occupations is alleged to make for a more
"mincing," less vigorous style of speech, in which the
mouth is not opened wide for a good volume of
sound, but is semi-closed, emitting therefore the
higher sounds.
Second, there is the civilization theory, whereby
the spread of culture and of cultivated, mannered
society may be supposed to have discouraged the
wide opening of the mouth (not a pretty sight at
best) for a "narrower," hence higher, type of speech.²
There is also the altitude theory, by which geo-
graphical altitude is credited in some more or less
occult fashion with a "raising" effect on sounds.
'The "imperfect transmission" theory is omitted here because
it is not an attempt at explaining the direction of sound change,
but merely at describing its character. The theory is noticed, e.g.,
on page 118.
Perhaps this statement is incompletely fair to this theory,
which contemplates also the nervous," "tensing" mental effect of
civilization.
ACCORD IN ENGLISH VOWELS
87
And finally, there is the weather theory, whereby the
artificial elimination of winter would tend to pro-
mote "freer" articulation.3
As just intimated, no one of these theories is
satisfactory to any large group of linguists. When
one of them is adduced, it is usually as a more or
less casual suggestion, rather than a convincing
analysis. Their tentative character has been so
obvious that linguists have appeared positively
relieved to forget them in descriptive research.
All these theories are conditioned upon the
notion of sound change as a series of sound risings,
and if this notion is without foundation in fact, they
become irrelevant. Has the "rising" concept of
sound change any justification?
Certainly this concept is not supported by the
facts of English consonant development. No single
one of the changes described in the previous chapter
is wholly a "rising" change, unless it be the substi-
tution of [j] for former [g]. Indeed, if we contemplate
merely the consonants, it is difficult to see how such
a theory could have arisen, so completely does it fail
to express the facts of their development.
But when we turn to the vowels, that is, to the
tense vowels, we can see at once how the "rising"
theory became current. It is quite true that every
3 Note that this and the first two theories tend to neutralize
each other. Obviously a man in a warm room can comfortably
open his mouth wider than a man in a snowstorm. The "wide"
sounds are the low ones. Therefore artificial summer should logi-
cally tend to lower, not to raise, sounds.
88
WHY ENGLISH SOUNDS CHANGE
regular English sound change where one tense vowel
has become another tense vowel, has involved a rise
in articulation.
OE [a], lãd, which has the notation E5 on our
diagram (page 9) became ME [5], lood, D5, and
ModE [o], load, C5. OE [æ], D4, became con-
secutively [e], C4, [e], C4, and [i], B3. It will be
found convenient to summarize such changes as
follows:
TABLE VI. TENSE VOWEL CHANGES
OE sound
[a] lād
[o] nōn
[e] rēc
[æ] sæ
ME sound
Low....
[] lood
[o] noon
[e] reke, reek
[ɛ] sea
Blade
High..... [i]
Mid.....
لا
[e]
Front
K [ε]
[æ]
ModE sound
[o] load
[u] noon
[i] reek
[e], [i] sea
Back
[u]
[o]
↑
↑
7[0]
[a]
This system of risings certainly constitutes the
most compact and striking body of sound changes in
English. This is the central feature of the story of
English phonological change, the citadel of its
kingdom. All or most else is subordinate in interest
to these surprisingly regular changes in the tense
vowels. It is these rises in articulation, and prac-
ACCORD IN ENGLISH VOWELS
89
tically these alone, which the phonologists have
contemplated, when attempting to interpret the gen-
eral nature of sound change. Considered by them-
selves, they present a picture of change as based on
rising articulation; and needless to add, it is this
partial picture which has occasioned the generaliza-
tion we are now considering.4
While the rising of the tense vowels constitutes
the most striking example of English sound change,
it cannot be regarded as typical of such change.
The consonants, as we have just seen, do not exem-
plify the same tendency in any regular fashion.
Neither do the slack vowels. Where these latter do
alter, it is not infrequently a lowering (OE pluma
[pluma], ModE plum [plam]). The tense vowels
themselves, in their so-called exceptional develop-
ments, frequently result in a lowering. All slacken-
ings (shortenings), as freond-friend, are of such a
nature, also thought from OE [boxte]. And finally,
how may we reconcile with this theory the regular
diphthongizations (OE is [is], ModE ice [ais]) which
have reduced the highest OE vowels to a combina-
tion of low and high?
The present study aims at the discarding of the
"rising" generalization, together with the varied
theories built upon it as attempted explanations. It
4 The inference is made even easier by the fact that similar or
analogous tense vowel risings are to be found in most or all IndE
dialects. The "iotacizing" tendency of Greek is an instance in
point, but illustrations are very numerous.
90
WHY ENGLISH SOUNDS CHANGE
aims at the establishment of another type of rational-
ization, based largely upon the principle of sound
accord, but contemplating also the influence of the
other linguistic principles which have been set forth.
Why Have Tense Vowels Become Raised?
Since it is an indubitable fact that English tense
vowels have actually become raised with much
regularity, any theory of sound change, to be accept-
able, must rationally explain this rising, as well as
the other changes in consonants and vowels. Hence
it is proper to ask at this point: does the principle of
accord offer a reasonable explanation of the phenom-
enon of vowel rising?
If the vowel risings are to be admitted as accord-
ings, we should find in sounds contiguous to these
vowels (probably following them; see page 141) in-
fluences exerted in the direction of higher articula-
tion. In other words, if it can be shown that the
long vowels were generally followed by higher sounds,
then accord might be postulated.
But is not precisely this the case? Let us recur
for a moment to the analysis of consonant sounds in
Chapter I. There, figures were given which show
that high-front consonants (lip, point, and blade)
comprise over 85 per cent, or 17 out of each 20, of
consonant sounds used in ordinary speech. An
additional analysis, covering approximately half of
the 5,000 consonants included in the first analysis,
indicates that an even larger proportion of high con-
ACCORD IN ENGLISH VOWELS
91
sonants, over 90 per cent, or 9 out of 10, is found in
post-vowel position.
TABLE VII. FREQUENCY OF CONSONANT OCCURRENCE
Point..
Blade..
Front.
Back.
Aspirate.
Total...
•
Lip consonants 297
910
53
27
191
•
Pre-vowel
No. Per Cent No. Per Cent No. Per Cent
64
19.3
59.0
3.4
1.7
12.4
4.2
Post-vowel
1542 100.0
128
687
16
73
14.1
76.0
1.8
8.1
Total
904 100.0
425
1597
69
27
264
64
17.4
65.3
2.8.
1.1
10.8
2.6
2446 100.0
What better basis could be procured for referring
tense vowel change to accord? The high consonant
which follows the vowel in practically every instance,
would by our theory induce a progressive rising of
the vowel. The tenth instance, such a word as
rēc-reek, may be explained as analogical in nature,
with the additional consideration that the low-back
consonants themselves are extremely unstable, and
may have been raised, or "palatalized" already,
through the influence of succeeding high sounds in
word or sentence.5
5 That [e] and [æ] should both have risen to [i] is perhaps explained
by rule 4, page 142.
92
WHY ENGLISH SOUNDS CHANGE
If further proof be required, one has only to run
over word lists such as those in Sweet's "History
of English Sounds," and determine in what propor-
tion of instances the rising of the vowel results in a
distinctly more efficient vocal pattern. In practically
every instance this will be found the case.
It is reasonable, then, to refer the phenomenon
of tense-vowel rising to the principle of sound accord,
operating to produce a closer approximation of the
vowel to contiguous sounds. And therefore accord
will constitute the present explanation of this body
of facts."
But immediately other problems present them-
selves. For example, the slack vowels were subject
to "raising" consonants as were the tense; why then
were they not raised? Likewise, not all tense vowels
rose regularly, even under apparently favorable con-
ditions; how can such exceptions be explained? And
finally, how account for the most puzzling change of
all, the diphthongization of OE [i] to ModE [ar], of
[u] to [av]? These questions, taken in the order
• In this matter of tense-vowel rising, we have an interesting
illustration of the interaction of accord and grading. As we have
seen (Chapter III) grading played a large part in the ablation of
inflectional terminations. As the inflectional termination was lost,
the final consonant of the stem came into the same syllable with
the preceding vowel, and so was freer to act upon it in the direction
of accord. This undoubtedly hastened the tense vowels' rising,
most of which occurred in the late ME and ModE periods. For
example, in words like OE na-ma, the vowel of the open syllable
was first tensed, giving ME na-me [na-mǝ]. Then as the vowel
termination became completely eliminated, giving [nam], the accord-
ing influence of the high consonant [m] was felt, and the vowel
gradually rose to [e], ModE name.
ACCORD IN ENGLISH VOWELS
93
just mentioned, will occupy the remainder of this
chapter.
Slack Vowels Not Raised
As with the tense vowels, the facts of the slack
vowels' development from OE to ModE are so
familiar as hardly to require restating. Broadly
speaking, the slack vowels have remained stable
during the entire course of their history in English.
The high blade vowel y [y], with its tense coun-
terpart, became unrounded to the nearest vowel
sounds, namely, [1] and [i]. Of the other OE slack
front vowels, [1] and [e] have remained unchanged,
while the change commonly attributed to OE [æ],
that of first retracting to ME [a] and then fronting
once more to ModE [æ], (OE at [æt]; ME at [at];
ModE at [æt]), appears dubious on the face of it,
depending as it largely does for evidence upon the
fact that the symbol æ disappeared from English
orthography in the early ME period. It seems a
much more probable hypothesis that the symbol
altered with no change in the sound, a fairly frequent
occurrence in linguistic history, than that the sound
itself should undergo such arbitrary and undirected
shifting.
The slack back vowels may have altered some-
what more than the front, from OE to ME. OE
slack [a] has practically disappeared from the lan-
guage, either tensing in ME, or falling in with the
94
WHY ENGLISH SOUNDS CHANGE
slack [æ] group." OE slack [o] (which was spelled o,
but may have been pronounced [o], [ɔ], or even [a],
in various dialects of OE) has had a dual develop-
ment, to ModE [o] or [a] (e.g., the two pronunciations
of ModE God, OE god). And finally, OE u [u], if
indeed it was so pronounced, frequently slackened
yet further to ModE [A] (OE wulf, ModE wolf
[wulf]; but OE hnutu, ModE nut [nat]).
From this brief summary it will be seen that it
is not necessary to admit that there has been any
change at all in the English slack vowels, between
OE and ModE, except the unrounding of y, and possi-
bly the fronting of a, where this has remained a slack
sound. All the others may credibly have been and
remained just what they now are.
If we cling to the former quantitative distinctions
of long and short as setting apart the vowel groups
from each other, we should be hard put to it to dis-
cover why mere vowel length should neutralize the
raising influence of the succeeding consonant and
leave the "short" vowels unchanged, while "long"
ones were raised. But the distinction of slack and
tense throws a different light upon the whole prob-
lem, and seems to bring it at least nearer to an
explanation.
First let it be said that the consonants (with the
7 The phrase Alma Mater illustrates the modern pronunciations
of the symbol a. As pronounced by different individuals, or at differ-
ent times by the same individual, it is either [alma mateı] or [ælmə
mete]. Where a is slack, it is either [æ] or [ə]; where tense, it is
[a] or [e].
ACCORD IN ENGLISH VOWELS
95
315
possible exception of h) are all tense, as compared
with the vowels. Thus they are much nearer in kind
to the tense vowels, and as rule 6 (page 142) states,
approximation accord, like physical cohesion, works
most effectively when the units are close together in
place or manner of production. (Otherwise accord
is apt to work by ablation.) The tense vowel, being
nearer in kind to the tense consonant, would nor-
mally tend toward approximation accord, to a
greater degree than the slack vowel.
It should also be remembered, in seeking an
explanation for the failure of the slack vowels to rise,
that the tongue, in pronouncing, e.g., slack [æ] fol-
lowed by tense [t], does not take two positions in
sequence. The vowel element may be described as
a state of the tongue, the consonant as an action.
This consonant action can be initiated from a slack
tongue state or from a tense one, but when the
tongue begins by being slack, the syllable merges
into one tongue action to greater degree than when
the tongue begins by being tense, as is the case when
a tense vowel is followed by a consonant. Thus a
syllable containing slack vowel plus consonant is in
more stable equilibrium, so to speak, than when the
vowel is tense, the difference being at bottom one of
muscular variation.
We see, then, that the failure of the slack vowels
to rise may be not incompatible with our theory of
accord. Let us proceed to the second question, that
of so-called exceptional or irregular vowel develop-
96
WHY ENGLISH SOUNDS CHANGE
ments. If the principle is to be found operative, we
should expect that these exceptions will likewise
prove the rule; that apparent irregularities are to be
explained along combinative lines. Can it be shown
that accord is the principle not only of the regular
but of the exceptional vowel changes?
Exceptions to the regular process of vowel devel-
opment are reducible in general to two types, and it
is as types rather than individual instances that we
shall consider them. One type has already been
mentioned, in the previous chapter. This is the very
large group where a consonant in vocalizing, or for
some other cause, has affected the vowel. The
second group comprises the multitudinous instances
of lengthening (tensing) and shortening (slackening)
of vowels. Let us consider these groups in their order.
The Influence of [r]
8
One consonant sound has occasioned so many
irregular or exceptional vowel changes that it deserves
detailed consideration. This is the consonant r,
which in modern speech has largely disappeared in
the very positions where formerly it had most effect
upon the vowel. 9
The usual description of the pronunciation of
8 Whether one prefers to call this r a vowel or a consonant is of
no moment in the present discussion.
• This ablation itself is an according like the elimination of l
in back surroundings, since it eliminates an "inverted" sound accord-
ing badly, not in place but in manner of production, with neigh-
boring sounds.
ACCORD IN ENGLISH VOWELS
97
OE r is that it was a sound like Modern Scottish r,
that is, a point trill, formed by allowing the point
of the tongue to tap or vibrate against the gums.
Without going into this question, let us note the
description by Krapp (PSEA 23) of another variety
of r, heard in the speech of Americans who "sound
their r's." This r, says Professor Krapp, is "pro-
duced by bending back the point of the tongue so
far that if it actually came in contact with the roof
of the mouth, it would strike about the middle of the
hard palate."
From this characteristic bending of the tongue,
this r is described as an inverted sound. It is also
necessarily a mid or low sound with reference to the
main body of the tongue, even though the tip of the
tongue may be high. If [a] be pronounced before
this variety of r it will be seen that no new position
of the tongue need be taken. The two sounds fit
together.
It is this variety of r which I believe to have been
present, not necessarily in the OE period, but at the
time when vowels were undergoing the modifications
caused by the presence of r. An r of this type alone
rationally explains the facts of the combinative
changes of vowels plus r. A back sound with much
tongue action alone could, in disappearing, have
caused the mutations which actually took place.
The vowel was unaffected by a final or intervocalic
r. Here the force of the consonant was somewhat
distributed and not confined entirely to a single
98
WHY ENGLISH SOUNDS CHANGE
syllable. So we have OE här, ModE hoar, eare-ear,
mōr-moor, ræran-rear, etc.
But when was followed by a consonant, crowd-
ing within the syllable permitted the maximum of
effect upon the vowel, and in this event we have
two possible lines of development. If the vowel was
low, it became [a] and remained so. OE deorc, ModE
dark; eart-art (vb.); earm-arm, deorling-darling,
steorra-star.10 That is to say, all low or low-mid
vowels tended to become levelled under [a], to accord
with a following plus consonant.
Where the vowel was high, a second variety of
levelling took place. This time the vowel [9] was
substituted for the former vowel. This new vowel
[9], then, becomes present in words originally of such
diverse vocalization as earth, bird, worth, herd, and
church. It is worth while pausing to examine [9] a
little more closely, and we may profitably quote
Professor Krapp's description of it (PSEA 35).
"The most characteristic quality of this vowel is due
to the fact that the point of the tongue is lifted up
and slightly inverted so that it is directed toward the
roof of the mouth. In other words, the tongue posi-
tion for is practically taken even while the vowel
is being pronounced. Could a better vowel be
devised, to precede this peculiar inverted consonant?
As r was ablated the tongue naturally tended to
""
10 Compare also clerk, Derby, hearth, etc., where even a former
[ɛ] has been lowered to accord with r in British English. With clerk
compare cleric and clerical, where no such lowering is found, because
the subsequent vowel divides the force of the r.
ACCORD IN ENGLISH VOWELS
99
smoothe from its inverted position, and we have the
final stage in the r-modified vowel, i.e., tense [^].
This is a mid vowel in which the point of the tongue
touches the lower teeth. It is the vowel pronounced
in bird, word, etc., in England and in parts of the
United States. But whether these words be pro-
nounced as in "r-less" British or "r-full" American
English, the vowels are conditioned upon the pro-
nunciation, lost or present, of an inverted r.
Other Consonant Ablations
As indicated in Chapter IV, vowels such as that
of thought [þɔt], OE pohte [boxtɛ], have actually low-
ered from their former values, through an according
with the low-back consonant [x]. This development
is found in a number of parallel words, sohte-sought,
brohte-brought, tæhte-taught, etc. It did not occur
in the combination of a high vowel with [x]. Indeed,
the vowel so affected seems to have been mainly o,
and so it may be questioned whether o in such words
may not have been already [ɔ] in OE.
The high-mid consonant w affected o in a differ-
ent fashion. The vowel of OE rowan, flowan,
growan, etc., instead of rising to [u] as is regular with
this tense vowel, remained [o]. The [w] disappeared,
leaving however a perceptible u-glide ([rou], etc.).
¹¹ In any event the change is a matter of accord, since normally
OE o should have risen to [u], and it was undoubtedly the low-back
[x] which prevented this development.
100
WHY ENGLISH SOUNDS CHANGE
The rounded character of [w] here exerted a stabil-
izing influence which prevented [o] from rising.
A few other accordings followed by consonant
ablation, where exceptional vowel development has
resulted, are OE bleow [bleow], ModE blue [blu];
strēaw-straw, græg-gray, etc. The accordings involved
are not difficult to trace. In a few instances, how-
ever, the vowel developed regularly despite an
ablating consonant; slōg-slew, trēo(w)-tree, sāwol-soul.
Probably the consonant sound in these words was
not marked.
Other Exceptions Due to Contiguous Consonant
All the foregoing instances involve an alteration
in the normal vowel development followed by the
disappearance of the consonant sound which caused
it. There are also many instances where a consonant
has affected vowel quality without itself becoming
ablated. These instances chiefly involve two con-
sonants, w and 1.12
Where OE w is followed by the unrounded vowels
[æ] or [a], these slack vowels, normally giving ModE
[æ], change instead to the lowest rounded vowel, [ɔ];
water-water, wascan-wash, etc. 13
Accord is patent
both in this "regular" exception and in the further
12 Note that it is the semi-vowels which cause most vowel change,
another illustration of rule 6, page 142.
13 A curious instance in warm, OE wearm. Here r and w are
mutually operative, and w prevails. Warm is not an exact rhyme
for arm or harm. Note also that slack vowels alone are involved
here. Tense vowels after initial w appear to develop regularly;
wā-woe, wōm-womb; but compare wic-week.
ACCORD IN ENGLISH VOWELS
IOI
exception to it, namely, words where the low-back
sounds [k] and [g] follow the vowel, and no rounding
takes place; wax, wagon.
There is also a small group of words (twā-two;
whā-who) where [w] appears to have "swallowed up"
the original vowel (possibly by stress grading) sub-
stituting for it the tense vowel nearest like itself,
i.e., [u]. This has not taken place in the compara-
tively unstressed word swā-so; here [a] has developed
regularly to [o]. In all these and a few other words
[sweoster-sister, sweord-sword) pre-vowel [w] has been
lost. Probably individual factors enter here; but it
is scarcely practicable to analyze closely such iso-
lated cases.
Somewhat less potent than the rounded consonant
[w] has been the lateral consonant [1]. Here again an
influence is exerted only when l is doubled or followed
by another consonant. Under these circumstances,
a low vowel preceding / tends to become the slightly
raised and rounded vowel [o]; weall-wall, sealt-salt.
Where I ends a syllable the vowel frequently is
not affected (Alfred; but cf. altar). Here the influence
of l is so distributed as not to be uniformly effective.
Ablated I (walk, talk, etc.) appears sometimes to have
had the same rounding effect on the vowel as has
which has remained in pronunciation. ¹4
So much, then, for the exceptions in vowel devel-
14 It seems almost necessary to apologize for reviewing examples
of sound change so well known to linguists. The writer's only
excuse for repeating them is to make them understood better as
connected phenomena, related to each other through general laws.
.
102
WHY ENGLISH SOUNDS CHANGE
opment caused by the influence of contiguous con-
sonants, whether or not these were later ablated.
The exceptional development frequently takes place
as a preliminary to such ablation, and this likewise
points to accord as a cause, since the very lack of
accord which would lead to the final loss of the con-
sonant might be expected first to bring about a par-
tial accommodation of the vowel. We have seen that
the exceptions to the rule of vowel change are not
exceptional, if the nature of vowel change is correctly
stated as accord rather than merely a raising of the
vowel. Each change makes for greater harmony or
efficiency of sounds.
The other most striking group of "exceptions" to
the rule of tense vowel change is to be found in the
lengthenings (tensings) and shortenings (slackenings)
which are so very frequent in the transitions from
OE to ModE. It is the writer's belief that in this
department of English phonological history we have
phenomena of stress grading, not of accord. Here
is to be seen clearly the same cause which may have
effected the loss of English inflectional terminations,
namely, the rise somewhere during the OE period of
a strong first-syllable stress, which by the early ME
period had intensified, operating in practically every
word.
Vowel Tensings
Tensings most often occurred in the vowel of an
open (initial) syllable of a two-syllable word. Thus
OE nama [nama] became ME name [name] (tense
ACCORD IN ENGLISH VOWELS
103
vowel), ModE name [nem]; OE mete, ME mete,
ModE meat. The tensing took place because the
progressive accession of stress on the first syllable
tended to make tense the stressed vowel, and finally
did so. Stress invariably tends toward tenseness,
lack of stress toward slackness (page 54).
+
Accord is seen, however, in the minor fact that
the tense vowel substituted for the slack would nor-
mally be slightly higher than its predecessor, and in
the more important fact that, once tensed, these
vowels rose into closer accord with their succeeding
consonants, just as vowels originally tense did.
There is need for a thorough study of the influence
of the consonant in closed syllables, as these became
closed in ME through the loss of inflectional termi-
nations (page 92, note). For example, we have just
seen that the slack vowel in the open syllable tends
to become tense. Alterations such as this obviously
must occur while the termination was still effectively
present, i.e., in late OE and early ME. Then accord,
made more potent by the closing of the syllable,
operates to raise the vowel, as in the word meat, just
instanced. But in closed syllables there is also
undoubtedly a slackening influence upon the vowel,
since the consonant itself absorbs much of the tongue
action of the syllable; and this fact probably led to
slackenings such as breath, bread, to be discussed in
the next section. 15
15 Such slackenings may be even more prevalent in other lan-
guages. Compare French biftek for our beefsteak, etc.
104
WHY ENGLISH SOUNDS CHANGE
Most linguists recognize, besides the tensing of
the open vowel in dissyllables, certain "lengthening
combinations" of consonants, which combinations,
they assert, gave rise to the long vowels in words like
child, comb, find, etc. Numerous exceptions make
this analysis uncertain, and while it is impracticable
to enter into the subject here, it should be said that
the entire subject of lengthening and shortening com-
binations needs a thorough reconsideration. Thus
many "shortening combinations" are practically
identical with lengthening ones. 16
Slackenings in Vowels
We have seen that vowel tensings, or lengthen-
ings, to use the ordinary term, fall into two groups:
tensings in the open syllable, and so-called tensings
by consonant combinations. The first is involved
with the accession of first-syllable stress, while the
second is of dubious validity.
Vowel slackenings may also be treated under two
heads. First, there is the stress (or rather, unstress)
phenomenon of slackening in the elements of word
compounds. As the elements of a compound were
welded into one whole, generally stress became fixed
upon the first syllable. In this event the vowels of
subsequent syllables tended to slacken; if the stress
16 Wyld (SHE 76, 114) gives to OE cild, and its plural, originally
a short vowel. He then asserts that this vowel was lengthened in
both by the combination ld; but he also makes ld plus consonant a
shortening combination and says that the ME plural childre had a
short vowel through its influence. This is simply confusion.
ACCORD IN ENGLISH VOWELS
105
remained diffuse, both vowels might slacken. Thus
we have wisdom from wis dom, hussy from hūs wif,
Kingston from cyninges tūn, etc. The same tendency
is well illustrated in double pronunciations, e.g.,
I [ar] and [ə], the [ði] and [ðə].
Another group of vowel slackenings, fully as im-
portant, is exemplified by ModE words such as
breath, dead, good, friend, all of which show by their
spelling as well as by other linguistic evidence that a
tense vowel must have persisted well into the ModE
period. Indeed, most such words still retain a tense
vowel in one or more regional dialects of English.
It is possible that these slackenings may be the
result of a tendency of a consonant or consonant
combination to divert stress and therefore tensity
from a preceding vowel, since in all instances they
occur in a closed syllable. In this event they fall into
line with what has been said in the previous section
about the influence of a consonant in a closed syllable.
They may possibly likewise be connected with an ob-
servable tendency in ModE to a more diffused stress,
since in late OE or early ME, strong initial stress
worked in an opposite direction. But much remains
to be done for the full elucidation of these slackenings.
And now, what do we find? Except for a few
individual instances such as briar, shall, merry, the
explanation of which will not be attempted here, the
discussion so far has taken cognizance of all the
important types of vowel change since OE, always
excepting the thorniest question of all, the problem
106
WHY ENGLISH SOUNDS CHANGE
of diphthongization. Thus far the case for accord
seems fairly clear in both regular and exceptional
changes. The rising of the tense vowels exemplifies
it, and except for so-called "quantitative" change,
all the exceptional types fall under the head of con-
sonant influence, a matter of accord in itself.
Diphthongization
The writer will not attempt to describe the
sources of the OE diphthongs, to rationalize these, or
even to analyze their disappearance. Falling diph-
thongs, almost without exception, their history in
general is that of their first (or stressed) element,
and at that this study is content to leave them. The
consideration of diphthongization will contemplate
only those diphthongs present in ModE, and of
these, will omit the rising diphthong [Iu], which has
been mentioned before.
It is a curious and significant fact that ModE
diphthongs fall naturally into two groups. Where a
ME or Mode vowel is diphthongized by the entrance
of a sound previous to the main or true vowel, as fine
[fain] from [fin], this extraneous or added sound is low
and back, usually [a], less frequently [ɔ]. Where diph-
thongization, on the other hand, works by the en-
trance of a sound subsequent to the true vowel, it is
generally a higher vowel sound, as note [nout] from
[not], eight [eit] from [et], play [pler] from [ple].17
17 There are similar situations in other IndE dialects. There is
classical Greek, where all diphthongs end in iota or upsilon, German
ACCORD IN ENGLISH VOWELS
107
}
The second type of diphthong is very easy to
explain by our principle of accord. If the high con-
sonants subsequent to the first vowels in such words
or phrases as note, eight, play ball, play the game, etc.,
are operating to raise the vowel, what more likely
than that the first indication of a rising should be a
vowel glide in the direction of a rise? The sequence
in the word note might well become [not], [nóut],
[nóút], [noút], [nut].
But the other and very common type where a
low-back sound appears preceding the vowel, is
another matter. One of the most invariable rules in
English phonology is that OE i [i] becomes ModE
[a1], and OE u [u], ModE [au] (mīn-mine, hūs-house).
How may this apparent anomaly be reconciled with
our system of accordings? While it is true that a
lesser proportion of pre-vowel than of post-vowel con-
sonants are high (about 82 as compared with 92 per
cent) still the rules of accord certainly do not explain
the introduction of a low-back vowel glide just before
the very highest vowel sounds.
One thing at least is clear; the explanation must
be sought outside the field of accord. Tenseness
might conceivably enter as a factor, since undoubt-
edly the higher vowels are progressively tenser than
the low. The body of the tongue, which has become
where ei is a diphthong but ie a single sound, etc. I do not say that
these two English types apply to all languages, however. English
linguistic habit appears to be conditioned upon an enormous pre-
ponderance of point consonants. So far as I know other tongues
have not been studied from this point of view.
108
WHY ENGLISH SOUNDS CHANGE
E
raised, may unconsciously be seeking lower, more
normal levels.
An explanation which appears more probable, and
which yet is advanced only tentatively, is as follows:
The sound [a] represents as nearly as any sound, pure
voice unmodified and unobstructed. The sounds [i]
and [u] represent such voice as little as any vowel
sound may, being very close in quality and region to
the consonantal [j] and [w]. But consonants alone
cannot "carry" a syllable. They may be likened to
the bony structure, which to function must be held
together by the muscles or tendons of speech, i.e.,
the vowels.
Hence as the high vowels become higher, accord-
ing more and more closely with the high consonants,
there comes a stage at which their vocal quality
becomes attenuated so much as to endanger the
carrying or voicing of the word or syllable, and it
is at this point that pure voice, in the shape of a lower
vowel-glide, enters. 18
This glide may become so strong as to "swallow"
the vowel, as in Southern dialect [tam] for standard
[taim], or Cockney [ladı] for former Cockney [laıdı].
In this event the whole raising process must begin
anew.
18 If this introduction of voice was gradual, as the current state-
ment holds, this fact does not affect the argument. All approxi-
mation accord is made up of gradual approaches, and the sequence
in the instance fine may well have been [fin], [fəin], [fain], with many
intermediate steps as well. Of course the present statement requires
much elaboration to be complete.
ACCORD IN ENGLISH VOWELS
109
The process may be compared to the stretching
of a rubber band. The vowels are, must be, based in
voice. While there may be a straining away from
this base, a long-maintained equilibrium of height,
still when a change takes place, it is likely to initiate
a return to pure voice.
This points to the really basic difference between
consonants and vowels, a difference of course well
known to phoneticians, but obscured by a mass of
non-essential and petty distinctions. The basic dis-
tinction must remain that of voice, the essential in
the vowel, the non-essential in the consonant. All
other distinctions may profitably be swept away,
but this one must remain.
General Conclusions
It is immediately apparent from the foregoing
analysis of English vowel change, that principles
other than accord operate with far greater frequency
than is the case with English consonants. For ex-
ample, stress grading enters here to a considerable
extent.
And yet the working of accord is also patent.
Not only the main, regular English vowel changes
are explicable in the light of this principle, but the
so-called irregular developments reveal themselves
for the most part as consistent manifestations of
accord.
Of course in this brief analysis the surface of the
subject has scarcely been scratched. The great
110
WHY ENGLISH SOUNDS CHANGE
system of umlaut, itself definitely according in char-
acter, has been completely disregarded, as has all
consideration of the influence of dialects and dialect
variation. Studies along these and other lines are
needed, in order to increase order in English phonol-
ogy by referring its facts to understood principles.
CHAPTER VI
GENERAL CONSIDERATIONS
Newness of Accord
It is perhaps idle to put the question whether the
principle set forth in the foregoing chapters is indeed
a novel idea in linguistics. For generations linguists
have recognized accord in a partial and unorganized
fashion, under the term assimilation. But no such
recognition, so far as the writer knows, has done
more than point to a few vocal accommodations as
phenomena more or less isolated in character.
For example, Jespersen comes as near accord as
any one, when he says (L 264), "In the great majority
of linguistic changes we have to consider the ease or
difficulty, not of the isolated sound, but of the sound
in that particular conjunction with other sounds in
which it occurs in words." But Jespersen does not
in fact consider this in any large proportion of his
instances. He does not even take the necessary first
step of constructing a system upon which all speech
sounds may profitably be compared.
Bloomfield also makes an approach to a statement
of accord. He says (SL 218), "We may suppose that
every sound change is assimilative in nature, chang-
ing some discordant element in the habits of pro-
nunciation into an articulation harmonious with the
III
I12
WHY ENGLISH SOUNDS CHANGE
•
*
total speech activities of the time." At first glance
this statement appears to define accord very well.
But then one asks "What are 'total speech activi-
ties?'" The phrase betrays the essential vagueness
of Bloomfield's concept. Moreover, he himself
immediately qualifies the entire statement by saying
that in so speaking he is "not naming its cause, but
merely describing, in part, the general nature of
sound change."
""
Indeed, Bloomfield specifically denies that the
fact of assimilation can possibly throw any light
upon the causes of sound change, because, he says
(SL 210), "the results do not indicate why the
change took place when and where it did." The
combination [kn], as in OE cneo, says Bloomfield,
had been pronounced so from time immemorial.
Why should it have been in England only (cf. ModG
knabe, etc.) and during a specific century (the
seventeenth, he infers) that this [kn] was simplified
to [n]?
Even if it were the fact that we cannot know why
a given sound should change at one time rather than
another, still this is no argument against assuming
accord as a final cause, if not an efficient or imme-
diate cause, of the phenomenon. The causes of
earthquakes are fairly well known; yet it may be
impossible to account scientifically for the time and
place of any particular shrinkage of the earth's crust.
We know why a dam bursts; but who can predict
the exact time and place of its bursting? Even so
GENERAL CONSIDERATIONS
113
in language, accord is no less a cause because it has
not yet been determined what is the immediate
or moving force behind a specific according. We
may say accurately that whenever these moving
forces which make in general for change become
actively operative, then change will take place,
caused ultimately in large measure by the according
tendency.
What then are these immediate or moving forces?
What, in language, corresponds to the rising water
which bursts a dam, to the accumulation of pressure
which impels a glacier?
Immediate Causes of Sound Change
Language being fundamentally a social institu-
tion, it is reasonable to suppose that the immediate
causes facilitating linguistic change will be the same
as those favoring social change of other varieties.
These causes have been traced in detail by historian
and sociologist. Conquest, trade, invention, educa-
tion, economic pressure, the mingling of diverse
racial stocks, are a few of the forces which are behind
the slow movement toward unification of society.
If this hypothesis is correct, it is by no means a
coincidence that the period of most rapid linguistic
flux in England is likewise the period of most violent
social upheaval. The eleventh and twelfth centuries
were times of conquest and consequent social
reorganization such as England has never seen since,
and they also witnessed more sound changes than
114
WHY ENGLISH SOUNDS CHANGE
any equal period of time since the first coming of the
Anglo-Saxon tribes to England.¹
During this transition period Saxon, Norman, and
Dane not only fought but lived together, the three
cultures mingling on a basis not too unequal to admit
of permeation. English had the determining advan-
tages of numbers and tradition, French the weight
of social and political domination, and Norse the
facts of its family likeness to English and its posses-
sion of a number of simplified, convenient pronominal
and other forms.²
One product of the middle period of English
literature illustrates so well the results of this battle
of tongues, that a detailed study of this aspect of it
might well be made. The poem "Sir Gawain and
the Green Knight" is believed to have originated in
an aristocratic (French) level of society, in the west
midland of England, a borderland between Scandi-
navian and English. The vocabulary seems almost
equally apportioned among the three tongues. The
meter derives from French, the word order and
syntax generally from English. Norse influence is
C
¹ To perceive this fact clearly, one must take a broad view of
sound change, not confining it to changes in the tense vowels, as
seems the usual procedure, but including also the vocalization of
consonants, levelling of inflectional terminations, etc. The tense
vowels did not rise to any extent until after the root syllable had
become closed by the loss of the termination.
2 Danish was the speech of many closely knit communities in
England, and within these communities it has been very persistent.
A Danish acquaintance told the writer that before she learned
English she could understand much of the speech of the inhabitants
of certain English west coast towns.
GENERAL CONSIDERATIONS
115
less strong, but is seen in vocabulary and in some
syntactical and inflectional forms.
Old English, which at one time seemed almost
submerged in this welter of tongues, finally emerged
in the ME period a language simplified and har-
monized, like a boulder which has been smoothed by
wave action. If we compare any late ME poem
(the Canterbury Tales, Piers Plowman) with a
representative poem in OE, this simplification is most
evident. The English of the later date is relatively
unencumbered with inflections; it does not proceed
by indirection, leaping, halting, veering in its ad-
vance, but comparatively at least, each sentence
progresses logically and steadily to its goal.³ The
simplification of consonants is practically complete,
and the rise of the tense vowels is under way.
This simplification, it must be remembered, is qual-
itative, not quantitative. Quantitatively the English
language expanded enormously during this period, in
vocabulary and in linguistic processes-idioms, syn-
tactical devices, etc. Numerically, speech sounds
may have increased also, though the elimination of
the gutturals and voiced semi-vowels makes this
fact uncertain. Simplification is shown in the more
efficient adjustment of these multiplied factors to one
another; in a word, by increased linguistic accord.
3 It should not be inferred that simplification originates from
any one of the three linguistic ancestors of ME, that is, from OE,
French, or Danish. ME has accorded far more than any one of
them. The accord results from the mixing, not from any single
ingredient.
116
WHY ENGLISH SOUNDS CHANGE
The Revival of Learning
One other period of English linguistic history is
comparable with the period of rapid change just
mentioned. This latter embraces the sixteenth and
seventeenth centuries, which constitute the transi-
tion from Middle to Modern English. Here we find
a period of flux second only to that of the eleventh
and twelfth centuries. Not only does the vocabulary
of English increase enormously, but the processes of
according and grading are unusually active.
But do these facts bear out the theory that lin-
guistic change is conditioned upon social upheaval?
This later period presents no such picture of military
conquest and mingling of races as does the former.
Civil strife existed to a considerable extent, but this
was distinctly subordinate to other social processes.
The question is answered easily by the reminder
that war and military conquest, so long assumed to
constitute history, are in reality only a part of the
forces making for social reorganization. War of itself
will not operate toward linguistic alteration, as wit-
ness the Anglo-Saxon conquest of the Britons, the
European conquest of the American Indian, etc. It
happened that the Norman Conquest, unlike these
others, involved a real mingling of cultures, which
latter was the important factor in inducing social
as well as linguistic change. Conquest by itself
would have effected little or nothing in these
directions.
GENERAL CONSIDERATIONS
117
The factors which in the Conquest made for lin-
guistic accord were present in high degree in the
second period, though to be sure with a difference.
Where previously cultures had mingled in physical
impact and continuous contact, now the mingling
was less physical, more intellectual and spiritual.
This latter is a period of mental and commercial
exploration such as has never been surpassed in the
history of the English peoples. We see a more
peaceful penetration of outside cultures, together
with the development of trade and industry, the
reorganization of religion and government, and in
the intellectual realm a depth of interest in classic and
foreign models combined with an immense surge of
creative expression. These were some of the many
factors which, I believe, caused the linguistic flux.
With the two great periods of linguistic change
closely paralleling the great periods of English social
reorganization, the hypothesis seems at least reason-
able that such upheaval is indeed the genesis of
sound accord. If this hypothesis can be established
through further study and comparison with other
tongues, then we have advanced a step nearer an
understanding of the nature and cause of sound
change.
Change Also Continuous
The foregoing analysis must not be interpreted
as meaning that linguistic change takes place only in
periods of upheaval. To a certain extent, sound
118
WHY ENGLISH SOUNDS CHANGE
change is continuous. Take for example Iceland,
which for the past six or seven centuries has been
a fairly homogeneous, isolated community. The
vocabulary, the grammar of Icelandic have altered
so little in this period that an English student of Old
Icelandic could read a modern Icelandic newspaper
with no great difficulty. Yet the sounds have altered
so much that this same student could not understand
the spoken language without some practice.
////
The reason for this undercurrent of continuous
change appears to lie in the fact that social change
is in some degree continuous, and in the further fact
that language, like other social habits, undergoes a
continuous process of transmission from one genera-
tion to another. Children are apt to learn im-
perfectly the most awkward sound combinations,
and so, very gradually, the sound pattern itself be-
comes simplified. An analysis of the patterns of
"baby talk" would show clearly the operation of this
simplifying tendency, which finally may affect even
adult speech.4
If "baby talk" were merely a fortuitous substi-
tution of sounds undirected by the according ten-

4 A somewhat analogous instance is the learning of English by
foreigners. The changes of the Period of Conquest were doubtless
facilitated by the fact that large groups learned English as adults.
But here we must reckon also with the operation of linguistic habit.
For example, the modern German's substitution of [v] for [w] is of
course due merely to the absence of [w] in his native speech. Accord
probably induces the choice of [v] (a lip sound nearer the consonant
norm than w) instead of some other sound. And so with the results
of other "foreign accents."
GENERAL CONSIDERATIONS
119
dency, no norm of baby talk could exist, and no
effect on adult speech would be possible. Thus the
very existence of this linguistic type is a possible
proof of our principle.
To summarize the discussion so far, we may con-
clude that the final cause of sound according is
the "unstable equilibrium" of awkward sound com-
binations. The immediate causes are: first, the
continuous shifting of the personnel of the language-
using group; and second, the forces of social progress
and reorganization, namely, war, education, inven-
tion, trade, race amalgamation, and so forth. These
latter may be called blind forces in that they do not
determine the direction of change. This direction is
determined in English almost wholly by the interplay
of the two principles of accord and grading.
Time and Place Not Important
Is it necessary to a comprehension of linguistic
change as conditioned upon these principles, to
account for the time and place of each specific mani-
festation of accord? Compare accord, if you please,
to a most fixed human law, that of gravitation. Com-
pare specific changes to the rolling of pebbles down-
ward following the tilting of a plate. Let accord
stand for the law, a period of social reorganization
for the tiltings, and individual accordings for the
motions of the pebbles. We may say that the rounder
pebbles (i.e., unstable combinations) will move first;
and that the more the plate is tilted the more pebbles
120
WHY ENGLISH SOUNDS CHANGE
will move.
But does it affect our acceptance of
gravitation as a final cause, tilting as an immediate
cause, if we admit that we do not know which pebble
will move first, or the precise instant of its moving?
Just so, we do not always know why one according
antedates another in English sound development;
we could not yet predict the precise time of its
occurrence; but none the less we may know definitely
its immediate and final causes.
Gravitation may be reduced to predictable results
under laboratory conditions, it is true; and the devel-
opment of language, covering as it does centuries in
time and continents in area, may yet be made sub-
ject to analysis of this sort. But even should this be
accomplished, it will in all probability merely verify
our knowledge of the fundamental causes of sound
change.
It may be well to render more specific the state-
ments just made, by analyzing in detail one par-
ticular change. Let us select for this purpose the
change from OE sc [sk] to ModE sh [S].
Any OE sound combination involving the back
consonant [k] would be more or less awkward and
hence susceptible to change. It is significant that
OE [k] has survived initially only with back or low
vowels (coal, calf), with the back consonant [w]
(quick, queen), and with the semi-vowels [r] and [1],
which as we have already seen (pages 98, 101) have
an affinity for low-back vowels (claw, crab). Initial
[k] has disappeared before a high vowel or vowel
GENERAL CONSIDERATIONS
121
glide (chin), before the point consonant [n] (know),
and after the point consonant [s] (short). And these
instances cover the use of this sound initially in
English.
Why OE sc should have simplified some centuries
earlier than OE cn it is perhaps not possible, certainly
not essential, to know. Suffice it to say that by
the end of the OE period this [sk] had become [S].
The change marks a considerable advance in vocal
efficiency. We now have one sound in place of two,
and that one far better accommodated to speech
sounds in general.
In the early ME period there entered the lan-
guage a considerable group of ON words in [sk], and
these have remained unchanged in consonant pat-
tern (skin, sky, skirt, etc.). The question may now
be asked, why did OE sc change, but not ON sc?
This question seems capable of a definite answer.
It is reasonable to suppose that ON sc was prevented
from altering by the very fact that OE sc had
changed. It would not have been difficult for any
hearer to grasp the fact that the initial sounds of,
e.g., short and sky were entirely distinct, and being
distinct, they naturally tended to be kept so. The
speakers had no consciousness of any traditional con-
nection between the [S] of short and the [sk] of sky.
On the contrary, they had the very best aural proof
of difference.5
5 To be sure, there are instances where sounds originally distinct
have fallen together, i.e., where the according tendency has been
122
WHY ENGLISH SOUNDS CHANGE
Finally, the question may be asked, is not the net
result of these phonological developments a compli-
cation rather than a simplification, since English is
left with two phonograms for one? Where OE had
only [sk], we now have both [sk] from ON and [S]
from OE. The answer seems to be that a simplifica-
tion has taken place, even though it has been incom-
plete. In so far as [sk] has become [S], the language
has simplified, even though all [sk]'s have not passed
into [S]'s.
More Than One Principle of Change
Indeed, the aim of this study has not been accom-
plished if the reader is left with any idea that all
changes must be complete and in a single direction.
Were this the case, languages everywhere would by
now be mechanisms of super-human efficiency, in-
stead of the makeshifts we find them. Sound change
is frequently incomplete. More than this, we now
and then encounter a change which can only be
called disaccord, where the principle we have been
describing is positively reversed!
An excellent example of such disaccord is the
diphthongization of the high vowels (page 108), where
the low-back sound [a] enters between two high
sounds, and even tends to swallow up the succeeding
vowel. On the face of it, nothing could more com-
able to overcome this tendency to keep separate sounds distinct from
one another. But such instances do not make invalid the present
analysis of ME [S] and [sk].
GENERAL CONSIDERATIONS
123
pletely deny the principle we have tried to prove
operative.
In like manner the law of gravitation is seem-
ingly set aside, by a balloon, a feather, a kite, by
muscular action, and by the slower process of physical
growth. Every physicist knows that these excep-
tions are apparent and not real, and can analyze the
opposing factors which temporarily neutralize the
law. So in linguistics we may admit the principle
of accord in spite of disaccording instances. Such
instances may be set down as due to any of a number
of linguistic forces.
Theory Re-stated
To return, then, to a final statement of the gen-
eral theory of language first set forth in Chapter IV.
It seems a tenable assumption that a language in its
earliest stages is "something rich and strange"-rich
in inflections, in tongue contortion, in vocal action;
and strange in all these. Sounds are apt to be drawn
from a wide speech range, and to be most awkwardly
put together into syllables.
As social progress takes place, the principle of
accord, which may be called the unconscious urge
toward ease in speech, operates.
operates. Continuously
through the influence of new generations of speakers,
and often radically through social amalgamation and
general upheaval, changes occur, changes which are
furthered, modified, obstructed, or even reversed by
countless influences affecting now a single individual,
124
WHY ENGLISH SOUNDS CHANGE
now the entire language group. The changes may
conform to any one of a number of linguistic prin-
ciples, among which grading and accord are easily
pre-eminent in the history of English. Grading tends
to separate and distinguish sounds, accord to har-
monize and simplify them. In the long run, when
we evaluate the alterations, we see that the accord-
ings are probably the more numerous and radical.
Frequently we can rationalize yet further, and
see why a particular situation should arise in a given
language. It is easy, for example, to conclude that
the English linguistic "center of gravity" (page 71)
is between those of French and German because the
language itself is between the two, formed in large
degree from their union. We cannot trace in detail
the processes which made German a "back" language
and French a "front" language; but this we can
assert: in so far as linguistic pressure (through con-
quest, trade, etc.) of a front phonological character
is brought to bear upon, e.g., German, that tongue
will tend to be modified in a fronting direction. And
in any event its modifications, such as they are, will
be explainable by our basic principles. This may
not be all we could desire in the way of scientific in-
sight, but it is something.
With all this interplay of forces and principles
it would be strange indeed if languages did not con-
tinuously vary widely from one another in their
❝ Of course grading may and does also lead to simplifications, by
its tendency to obscure and eliminate unstressed elements.
GENERAL CONSIDERATIONS
125
phonological systems. We need no explanation of
such variation; it is natural and even inevitable.
Theoretically one may construct an ideal system of
according sounds, a harmonious divine whole toward
which human sound creations might be said to move.
But practically, such a system is far from realization.
Comments on General System
A few comments on the general theory just set
forth may be in place. In the first place, it is observ-
able that the total number of speech sounds is not
important to our system. This number, indeed, is
likely to remain much the same; that is, coincident.
with the ablation of disaccording elements may occur
the introduction of new sounds nearer the norm.
But while this is true, it should be remembered that
these new sounds are not necessarily or ordinarily
substitutions for the old, and that the number of
sounds in itself has no special significance to our
theory.
Secondly, it should be said that this is not an
'ease theory" of speech development, as the term is
commonly used. The proponents of "ease" as an
explanation of sound change contemplate at least to
some extent the ease or difficulty of certain isolated
sounds, a matter quite unessential to the argument
here.
While it is true that sounds do vary from one
another in the amount of what we may call mouth
distortion necessary to their production, still this
126
WHY ENGLISH SOUNDS CHANGE
!
variation, difficult or impossible to measure, is not
a factor in sound change, since habitual use will
make a familiar "difficult" sound easier than an
unfamiliar "easy" one.
It is only with respect to sound combinations that
the present analysis could be called an ease theory;
it might better be termed an efficiency theory. But
the phrase, theory of accord, best expresses the
matter of this analysis, since it deals with sounds only
in relation to each other, not with a sound, per se.
The writer has been at pains to emphasize that
we have as yet little more than a theory, an outline
which must be filled in by both broader and narrower
studies by many investigators before it can be re-
garded as established. The fact of sound accord as
a main principle of linguistic change I regard as
demonstrated; its full statement must involve a
great deal of further research.
CHAPTER VII
ACCORD OF IDEAS
Need for a Simplified Terminology
It will be observed that here and there through-
out this study, statements which may apply to de-
partments of linguistics other than phonology have
been made. Especially was this true in setting forth
a general theory of language development. Will the
term accord then be found useful in a wider, general
sense, to illumine phenomena of morphology, of
syntax, even of semantics?
On the surface, there seems grave objection to
adding another to the already extended array of lin-
guistic terms. Indeed, to the casual observer the
terminology of linguistics might well appear in a
hopeless state of mist and muddle. On the one hand,
we have a single term such as spirant, which in
Webster's Dictionary is given no fewer than four
incompatible definitions, each backed by authorita-
tive usage. On the other hand, we have a single
construction such as the good in the good are happy,
for which Jespersen (PG 106 note) lists nine terms in
common use. In addition there are such words as
velar, palatalization, dark and bright vowels, etc.,
which are either positively misleading, or cursed
with a vagueness of intent which cannot but pre-
A
127
128
WHY ENGLISH SOUNDS CHANGE
clude scientific exactness. A new term, to contribute
toward the resolving of this tangle, must have strong
qualities of definiteness and simplicity.
The great need of linguistics has been for a new
and better terminology, and in phonology at least,
this need is being met. Easy, definite terms, such
as high, slack, back, voiced, etc., are replacing the
former romantic or cumbrous ones, and have gained
wide currency among those scholars who do not
regard clarity as incompatible with erudition. The
change is comparable to that from Roman to Arabic
notation in mathematics. It becomes suddenly easy
to perceive principles and relationships which were
hidden under the old system.
Throughout the present study I have been en-
deavoring to establish accord as such an illuminating
term, which by its very use must operate to clarify
the phenomena it describes, a thing the older term,
assimilation, failed to do. I now wish to suggest as
a possibility that this same term, accord, may well
be found useful in other departments of linguistics.
To prove this and to indicate a few of the resulting
clarifications, will be the aim of this chapter. It will
be impossible to give more than the barest hints at
a synthesis of this vast field, more than an outline
to be filled in by further detailed investigation.
It was said (page 22) that accord in speech is that
principle whereby a speaker unconsciously tends to
harmonize speech sounds. With the single verbal
substitution of ideas for sounds, this definition will
A
ACCORD OF IDEAS
129
serve very well here. Accord in "grammar" is that
principle whereby a speaker tends to harmonize
ideas. It is the tendency of spoken or written lan-
guage to become progressively more efficient, more
synthesized, in expression.
Several examples of this type of accord were
given in Chapter II, pages 38 and 39. A simple one
is the child's formation of a plural sheeps from sheep.
With no conscious mental effort, the plural is con-
formed or accorded to the vast body of s-plurals.
I
Sound accord will accompany accord of ideas.
If the word to be pluralized ends with a voiced sound
or a sibilant, the child will make the according sound
adjustments, saying, e.g., pigs [pigz] or dishes [di siz],
without in the least realizing that sound accord, as
well as accord of ideas, is present. Of course the
two factors are entirely distinct. The adding of a
sibilant to pig is idea accord; while the voiced
character of that sibilant is sound accord.
At this point the comment may be heard: what
you have been describing is entirely familiar to
linguists under the term analogy. The child adds
[s] to sheep by analogy between sheeps and the vast
body of s-plurals. There is no reason for calling this
process accord of ideas, with another term already
established in usage.
• The illustration sheeps should not be interpreted to mean that
such accord of ideas is apt to produce a result grammatically incor-
rect. In the great majority of words, of course, a correct result will
be obtained by the according process, e.g., dogs, cats, houses, etc.
130
WHY ENGLISH SOUNDS CHANGE
If it is true that the term analogy accurately and
completely describes such phenomena, there is indeed
no place for another term. But does it?
In its general use outside linguistics, analogy con-
notes a more or less conscious connection between
two things (processes, sequences, etc.) of not too
unequal rank or value. The term is largely used
with reference to reasoning of a distinctly high order.
Thus one would scarcely say "Jean bobbed her hair
by analogy between her own and the vast array of
bobbed heads in Chicago." The comparison is too
unequal. Neither would one say, "A baby cries at
birth by analogy between itself and other newborn
babies." The process is too unconscious. One
might, however, say, "There is a real analogy be-
tween the process of invention and that of physical
growth," or "No analogy exists between the vague
hypotheses of astrology and the facts of astronomy."
If this summary is correct, then the current state-
ment that the child adds s to sheep by analogy
between sheeps and the vast body of s-plurals, is bad
English, a definite misuse of words. The process is
entirely unconscious, the comparison wholly unequal.
The situation is somewhat improved, it may be ad-
mitted, if we rephrase the statement thus: sheeps is a
formation analogous to that of the normal s-plural.
Here the conscious reasoning is not so directly
referred to the child who says sheeps, nor the com-
parison phrased to appear so unequal. But obviously
we cannot get along with the adjective analogous,
ACCORD OF IDEAS
131
and so omit the misleading noun analogy from
linguistic terminology.
There are other reasons for objecting to analogy
as a descriptive term for these instances. It is too
pompous, too inflated, too vague, for this purpose.
Just as front and high have replaced the older palatal,
so will scientific exactness be served by substituting
an easy, accurate, definite term for the cumbrous
analogy.
Accord is in itself a desirable term in this con-
nection. It has the necessary qualities of ease and
definiteness, and in addition it bridges the gap be-
tween phonology and grammar, establishing in itself
a synthesis between them.2
No further apology, then, will be made for the
adoption of the term accord, or accord of ideas, in
place of analogy. Let us proceed to a consideration
of a few of its workings in the various departments
of linguistics.
Accord in Morphology
This heading might almost be phrased "Mor-
phology is Accord," so completely do morphological
systems depend upon according of ideas. It seems
reasonable to assume, with many linguists, that all
or most inflectional elements were at one stage of
2 Were it not for this fact, the word agreement would do as well
as accord; but agreement is also used in the specialized sense of
formal grammatical agreement, e.g., between a noun and its verb,
adjective and noun, etc.
132
WHY ENGLISH SOUNDS CHANGE
J
the language separate words, even though one may
balk at the task of ferreting out just what those
words were. In the nature of language, once such
an element had merged its identity with the govern-
ing words, it would in all probability disappear,
leaving no independent survivor in speech. In a few
instances, e.g., in the Semitic tongues, we are reason-
ably sure of the progenitors of many of the inflectional
elements, but even here the attempt to complete
such information raises more difficulties than it
resolves.
Contracted forms such as I've, he'll, shan't, show
stages in the paths taken by syntactical units in the
process of becoming morphological elements. Such
elements-ve for have, ll for will, nt for not—if found
regularly in a given relation to other stronger sen-
tence words, might easily, especially in a language
without spelling, become merged with these words,
either as suffix or prefix, depending on the sentence
order and sentence stress characteristic of the
language.
In every instance the merging of linguistic ele-
ments represents or reflects a union or according of
ideas. This according process may be extremely
active. It is probable that in early stages languages
use far more and finer morphological groupings, just
as some primitive tongues are said to have separate
"genders" for long things, round things, hard things,
and so forth. Possibly the strong and weak cate-
gories in Germanic, the Latin and Greek declensions
.
ACCORD OF IDEAS
133
and conjugations, are survivals of some such group-
ings.
At present we have only a few traces of such
former categories, but these are still numerous
enough to keep most of our systems from complete
uniformity. Thus we have the various "irregular”
verbs, plurals like oxen, children, etc., and unaccord-
ing systems of pronouns, where the recently altered
it alone follows the usual (noun) plan of inflection,
with objective like the nominative, and possessive
formed by the addition of s.
Minor according groups may even grow up within
the limits of a larger system according generally with-
in itself. An example of such development is the
English plurals of fish and game words. Nouns
indicating varieties of fish or game, quail, deer,
salmon, bear, pike, trout, etc., have come to exhibit
a plural identical with the singular, especially when
used in the dialect of outdoor sport. This is not a
traditional condition; e.g., OE fisc formed its plural
regularly, fiscas. These words have accorded to a
few such words with no separate plural form.
In general, however, inflectional systems of
accordings are remarkably uniform and universal
within a given language, exhibiting a strong ten-
dency toward obliterating any exceptions which may
happen to exist. Whether these accordings are by
terminations incorporated into the individual word,
as illustrated by the third person singular of verbs
(goes, comes) or in word phrases, as the possessive
134
WHY ENGLISH SOUNDS CHANGE
pronoun plus a noun (my book, his table), they are
none the less manifestations of the according prin-
ciple. Chinese with its separate grammatical ele-
ments, Hebrew with its combination of independent
and compounded accordings, indeed, any language
at all, exhibits similar syntheses.
Perhaps it is somewhat easier to perceive accord
of ideas in morphological "mistakes" than in "cor-
rect" locutions, since in the former instances the
according principle works in opposition to tradition
rather than in consonance with it, and hence is
easier to isolate as a factor. Take for example the
words nearer and drownded, both traditionally mis-
takes, and both illustrating the same type of error,
that of a double according, where a second inflec-
tional element has been added through an imperfect
realization of the first one as actually added and not
a part of the word. This imperfect realization has
also tended to establish the form with a single in-
flection element, near, drownd, as the primary word.
In one of the two instances the tendency has been
successful. Accord is here seen to have overridden
tradition and to have established a new form as
standard. 3
To re-state briefly the function of accord in
3 What part the repetition type of sound accord has played in
this development it is impossible to say with any certainty; but I
believe it to be very small, else why not higherer, sinneded, etc.?
On the other hand, there is sounded, by tradition properly souned.
Certainly there is a tendency to add to a final voiced point continu-
ant [n], the corresponding voiced stop [d], to round off the close of a
word.
ACCORD OF IDEAS
135
morphology, we may say that accord of ideas is the
principle active in producing ever more completely
synthesized inflectional systems. Occasionally minor
according groups, separate from the main body, are
found; but in the long run these also tend to become
absorbed. When a conflict of linguistic ideas occurs,
accord may lead to locutions unaccepted by the
standard speech; and in course of time many or most
of these are received into the language.
Worn-down tongues such as English, Danish,
Hebrew, Aramaic, etc., best illustrate this morpho-
logical according. Declensions and conjugations have
merged and so become fewer in number. Former
morphological distinctions have been obliterated. In
general the language has approached nearer and
nearer the Esperanto ideal. 4
Accord in Syntax
In syntax, we have a type of accord appreciably
more nearly in the realm of conscious mental activity
than morphological accord. Grammatical "correct-
ness" is not by any means always instinctively
attained, even by native speakers of a language. It
is one of the favorite pastimes of grammarians to
4 The part played by a fixed word order in facilitating the loss of
inflectional distinctions has been mentioned by linguists. Its effect
is by no means negligible, but may have been overrated. Highly
inflectional languages frequently possess a fairly fixed word order;
and on the other hand, the word order of ModE is if anything less
rigid than that of German, a much more highly inflectional tongue.
A stock example in English is the line "The plowman homeward
plods his weary way," which may be transposed in surprising variety.
!
136
WHY ENGLISH SOUNDS CHANGE
compile lengthy lists of every sort of grammatical
solecism, committed by literary artists of the highest
repute. Such lists should be well adapted to comfort
and encourage lesser wielders of the pen, by demon-
strating the unattainability of syntactical perfection.
Where agreement is followed out according to the
usual standards of grammatical correctness, the
according is simple and obvious. I go, he goes, she
herself, need no special analysis. Where a "mistake"
has arisen, it will be apparent upon examination that
a mental conflict exists, a conflict between an idea
of form and an idea of content. Let us analyze a
simple instance.
In grammatical practice it is ordinarily found
that the subjective case precedes, the objective case
follows, the verb; e.g., John hit him, we sent to them,
etc. And as English word order almost always places
subject and verb at the beginning of a sentence, the
rule in practice, that is, in the semi-conscious practice
of the ordinary speaker, might be stated thus:
Before the verb, use the subjective; after the verb,
use the objective.
But this levelling according tendency does not
operate so uniformly as the speaker semi-conscious
of his language might desire. For one thing, we are
told that the copula is not a true verb but an expres-
sion of identity, and hence must not take the ob-
jective case. That is, the idea of identity expressed
in It is I, should rule out the idea of standard form
expressed in It is me. But can it in fact do so? In
ACCORD OF IDEAS
137
this as in most cases, accord of form is strong enough
to outweigh the idea of identity, and we have it is me
triumphantly riding to general acceptance over the
dead bodies of innumerable purists.
Another less obvious instance of the working of
the same rule may be given. The poet wrote:
The boy stood on the burning deck
Whence all but he had fled.
“All but him,” immediately says the grammarian.
But here is clearly a preposition, and prepositions
govern the objective case. The phrase as written
seems preferable, showing as it does a sound regard
for the tacitly adopted rule of subjective before verb,
objective afterward. If the rhyme had required:
The boy stood on the burning deck
Whence all had fled but him
}
no doubt it would naturally have been written that
way.
"Between you and I," almost invariably a pre-
verb phrase, illustrates precisely this same tendency
toward larger syntactical accord; and many other
examples might be cited.
These instances all indicate clearly the leveling
nature of syntactical accord, which works toward
grammatical unity even beyond that permitted by
grammatical rules. One larger aspect of the rule
set forth in the foregoing paragraphs is worthy of
138
WHY ENGLISH SOUNDS CHANGE
note. The rule is one of sentence order. The older
rules call for accord in the smaller grammatical units
of prepositional phrase, or copula plus pronoun. The
present practice therefore illustrates a tendency
toward a larger synthesis which takes in the sentence
as a whole, instead of in its minor parts; and this
tendency is perhaps characteristic of progress in
grammatical accord.
Any particular rule of syntactical accord, working
toward general acceptance in any particular language,
is by no means universally applicable to language as
a whole. We might postulate a syntactical norm or
"center of gravity" in each language, a norm differ-
ing in each just as phonological "centers of gravity"
have been found to vary. In other words, the appli-
cation of the according principle will in all probability
be found peculiar to one language; the principle itself
is universal.
-
Accord in Semantics
Fascinating as it would be to delve deeper into
the workings of syntactical accord, it would not be
profitable in the present study, since most such
accordings are either "regular" or of the type just
analyzed, that is, proceeding from a conflict of
grammatical ideas.
But when we turn to the field of semantics, or
the meanings of words, we rise at once to the plane
of pure ideas relatively unhampered by form or
system. Wherein does accord play a part in this
ACCORD OF IDEAS
139
immense conglomeration? And what may be said
of accord in its still wider reaches, of the endless
shades and tones of blended meanings of words,
which, woven with highest artistry, we call literary
style?
The English instructor who told his fair pupil,
"Horses sweat; men perspire; but ladies merely
glow!" had a clear perception of the fitness of words
to one another. The lady who boasted to a friend,
"You just can't realize the enormity of Mr. Smith's Cohen
business!" likewise had a clear perception of what
she wanted to express. We smile because her verbal
pistol went off in the wrong direction. We have one
connotation for the word enormity; she had another.
All malapropisms are of this character; a word
properly according with one set of meanings is
wrenched into an incongruous environment, and the
result is verbal humor. But the consideration of such
misplaced words merely illumines the truth that our
entire verbal system is a vast interplay of according
ideas. The extent to which such accord is carried
out efficiently constitutes clarity of style; the extent
to which accord is manifested brilliantly constitutes
style itself.
Oliver Wendell Holmes speaks somewhere of
words which have been betrothed from the cradle,
but never mated until joined by the master hand.
The highest literary art blends sound and sense so
delicately as to afford exquisite enjoyment. Accord
here has become to a great extent a self-conscious
140
WHY ENGLISH SOUNDS CHANGE
I
art, but it is not the less accord; and the highest art,
for example that of Shakespeare, takes account of
every variety of accord, from the harmony of sounds
to the most lofty blending of ideas.
The fields of semantics and of literary style have
never been reduced to order by research. Students
of the former have been somewhat like children pick-
ing up pebbles on the seashore; each word is exam-
ined in turn and abandoned. Studies in semantics
have thus almost without exception been supplements
to dictionaries; and significant studies in style have
been few.
Probably the highest literary expression will al-
ways elude complete analysis; but countless more
pedestrian subjects are in need of light. For example,
many studies of recurrence accord as a factor in
style might well be undertaken. A study of obso-
lescence is needed, to cast light on the causes why
words disappear from the language. Also verbal
humor might well be studied along the lines just
suggested.
Indeed, the charting of accord in sounds appears
a comparatively simple matter beside its analysis in
these largely unexplored fields.
APPENDIX
A FEW RULES STATED
At the risk of some repetition, it appears wise to
assemble here certain rules of phonology, most of
which have been formulated and discussed here or
there in the body of this study, others of which have
merely been assumed to be true, and will now be
stated for the first time. These rules are brought
together for convenience of reference, with no great
pretense at completeness or logical arrangement.
Several have to do with stress (grading), others with
accord, others with neither. In general it may be
said that all of them present what the writer believes
to be facts important to the present argument, but
facts largely undiscovered or disregarded in current
linguistic writing.
1. In general, sounds tend to influence preceding
more than succeeding sounds. This is a matter of ob-
served fact which the writer will not attempt to
explain. Its bearing upon the discussion of accord
is obvious, since it means that we generally look to
a succeeding sound for the explanation of a change in
a preceding one. Perhaps the most radical example
of this rule is in umlaut (page 33); but it is exempli-
fied in practically all accordings.
2. For the terms long and short as used of vowels,
141
142
APPENDIX
tense and slack should be substituted. For discussion,
see pages 46 to 50.
3. Vowel sounds change more readily than con-
sonants, tense vowels than slack.
4. Sound change in vowels normally occurs with
reference to tongue height rather than tongue region.
That is, the back vowels ordinarily stay in their own
series without becoming front, and vice versa; mōr
does not become meer, nor mate, moat. The motor
picture of the active part of the tongue in making a
given sound appears to be clear, but the exact height
of that active part is not so clear. A violin player
may finger a string slightly too far up or down, but
he is not likely to pass from one string to another
unknowingly. Just so, the tongue appears capable
of more accurate action in regard to region than to
height.
5. Accord in vowels works generally by approxi-
mation. This is due to the unstable character of
the vowel.
6. Accord in consonants works in general by sub-
stitution when the according elements are near each
other, by ablation when they are far apart in the
mouth. Tune [tjun] passes easily by substitution
to [tsun], [sjur] to [Sur]; but in niht, cealc, noth-
ing but an ablation is likely to straighten the line
of sounds.
7. An increase of stress on the initial syllable of
words facilitated the leveling and final dropping of
English inflectional terminations (page 51).
A FEW RULES STATED
143
8. Stress in consonants tends to unvoice, unstress
to voice (pages 52, 53).
9. Stress in vowels tends to tense, unstress to slack
(pages 50-54).
10. Unstressed back vowels tend to become leveled
as [ə], unstressed front vowels as [ə] or [1] (page 52).
11. Sound change is more likely to take place under
stressed conditions. This is illustrated by the frequent
alteration of initial consonants before, or by excep-
tion to, medial or final ones. It is also shown to some
degree by the fact that tense (usually stressed)
vowels became raised, while slack vowels remained
at the same level.
12. Linguistic habit consists of that body of sounds
to which a people is accustomed (pages 55 to 57).
13. Accordings not involving spelling take place
easily; others with greater difficulty (pages 57 to 60).
Ablation accord, 29 ff., 63
Ablaut, 44
Accord,
defined, 19, 22 ff., 39, 40
in morphology, 38, 131 ff.
in semantics, 138 ff.
in syntax, 39, 135 ff.
in tense vowels, 90 ff.
of ideas, 38 ff., 129 ff.
types of, 25 ff., 63, 65 ff.
Analogy, objections to term, 129
ff.
Approximation accord, 25 ff., 63
in consonants, 65 ff.
umlaut as, 35
Assimilation, objections to term,
19, 20
Bantu, 36, 73
BLOOMFIELD, quoted, 111, 112
Breath-voiced,
INDEX
of consonants, 6, 7, 28–29, 65
rule, 52 ff.
BRIDGES, quoted, 59 ff.
Causes of linguistic change, 113 ff.
"Center of gravity" in languages,
71, 124
Chinese, 134
63
Consonants,
accord in, 65 ff.
categories applied to, 6
Consonants-Continued
classified, 9
frequency of, 12, 91
influence on vowels, 96 ff.
opposed to vowels, 4-6
Danish, 114, 135
Diphthongization, 106 ff.
Disaccord, 68, 122
Ease theory, 23 note, 56, 125, 126
ELLIS, 60
Final sounds, change in, 32, 68
Finnish, 36, 73
French, 13, 45, 56, 57, 71, 75, 76,
78, 79, 80, 103 note, 114, 115,
124
Fringe sounds, 18, 69, 71
German, 57, 67, 71, 77, 106 note,
124
Grading,
defined, 43 ff.
types of, 45 ff., 63
Greek, 36, 37, 49, 89, 106 note
Gutturals, elimination of, 66 ff.
Conservation, principle of, 55 ff., Icelandic, 118
Hebrew, 35, 37, 38, 74, 134, 135
Intellectual element, the, 60 ff.
JESPERSEN, 39 note, 61, 72, 111,
127
145
146
INDEX
A
KRAPP, 8, 11, 73, 81, 97, 98
Latin, 36, 37, 49, 75
Linguistic habit, 56-57, 63, 82
LUICK, 30 note
MILNE, 4
Modulation, sound, types of, 41
ff., 63
Mutation accord, 33–35, 63
Pitch, 41-43, 45-46, 63
Point consonants, 9-13
Quantity, 42, 46 ff., 63
Recurrence accord, 35 ff., 63
defined, 36
Rounded-unrounded,
of consonants, 7 note, 29
of vowels, 7 note, 99 ff.
RUSSELL, 49 note
Russian, 56
Sanskrit, 3, 75, 76
Spelling influence, 57 ff., 63
Stress, 44 ff., types of s. grading
63
SWEET, 60, 92
Tense slack,
of consonants, 7 note
of vowels, 7 note, 13 note, 47 ff.
rule for vowels, 52 ff.
slack vowel changes, 93 ff.
slackenings in vowels, 104 ff.
tensings in vowels, 102 ff.
tense vowel changes, 88 ff.
Theories of sound change, 85 ff.
Transmission theory of language,
86 note, 118 ff.
Vowels,
accord in, 85 ff.
categories applied to, 7
classified, 9
in early alphabets, 3, 4
opposed to consonants, 4-6
WRIGHT, 34, 35, 39 note, 79 note
WYLD, 30 note, 48, 58, 104
!!
*

r
.
3
::
:

Cat Martin ( 1 2 WERE NE SE
----
----
1
UNIVERSITY OF MICHIGAN
3 9015 00577 0394
DO NOT REMOVE
OR
MUTILATE CARD
Tanan kegaga
M
thông đ đến 3 ngư