sound change and language change: a sociolinguistic interpretation

S o u n d C h a n g e a n d L a n g u a g e C h a n g e : A

S o c i o l i n g u i s t i e I n t e r p r e t a t i o n

Fred C. C. Peng

h2ternational Christian University

A BS TRA CT

Recently, an article by Lawrence Johnson stated that in language change "The time span considered can be across several centuries or as few as two demographic generation" (1976:165). tie thus concludes that "Specifying the terms 'fast' and 'slow', we have given some support to the claim that change begins slowly and accelerates in succeeding generations, and we have given evidence that change advances more rapidly in urban than in rural communities" (1976:171). In this article, I examine the validity of these claims. An alternative theory is then proposed, whereby[language change is viewed as a result of the accumula- tion of Changes in language behavior among living humans, i.e., a change in the norm of individual speech activities within a language community. In light of this view, language change can not only be observed and captured, while in progress, but also taken as a manifestation of the process of human change fia general within each generation. Concrete evidence is prpvided from Japanese, in the realms of sound, vocabulary, and grammar, in order to support my theorical construct.

THE PROBLEM

In my recent article, enti t led "-A New Explanation of Language Change: The Soeiolinguistic Approach," I pointed out that all previous studies of language change share without except ion one common feature that stands in sharp contrast with my view. This common feature, which has become some sort of myth in the literature, is that changes occur across the boundaries o f two or more generations. I added that even Labov's observation o f language change in Martha's Vineyard, involving a claim that sound change

- 9 4 -

LANGUAGE SCIENCES, VOLUME 1, NUMBER 1 (1978-79)

may be captured while in progress, takes in three generations (Peng 1976:70).

This common feature was repeated once more by Johnson recently (1976) who claims that "The time span considered can be across several centuries or as few as two demographic generations" (1976:165). He thus concludes that "Specifying the terms 'fast ' and 'slow', we have given some support to the claim that change begins slowly and accelerates in succeeding generations, and we have given evidence that change advances more rapidly in urban than in rural communities" (1976:171).

In view of this (unfortunate) development, several questions need to be raised here, so as to eradicate, once and for all, the myth that seems to persist in the literature. For the sake o f convenience, these questions are asked below in the order to which I shall address myself.

QI. Is it linguistically plaus~le to construct a theory of sound change that is based on the assumption that sound change takes place across the boundaries of two or more generations?

Q2. Is it true that change begins slowly and accelerates in succeeding generations?

Q3. Is it fah to state that whatever changes that have taken place or are being observed in one sector of language can be regarded as indicative or a model of changes in general, Le., language change as a whole?

Q4. Is it theoretically sound to generalize from type of changes in one language the same type of changes in other languages?

Q5 Can linguists, historical lingutstis in particular, do themselves justice by ignoring non-linguistic change, when they deal with linguistic change?

All these questions are interrelated and must be answered accord- ingly. But I shall delay answering them and take upon myself the task of doing so only after I have presented enough materials to set the stage for a fruitful discussion that will follow. The needed materials will be provided in two parts; namely, (1) Previous Study on Sound Change within One Generation and (2) Data Base for the Present Study.

- 9 5 -

SOUND CHANGE AND LANGUAGE CttANGE

PREVIOUS STUDY ON SOUND CHANGE WITHIN ONE GENERATION.

To begin with, let me quickly review what I said in the article mentioned above concerning sound change within one generation. First, I took Nomoto's 1950 study and 1971 study and came up with the result that each individual seems to continue developing his or her speech beyond 13 years of age, at an ever decreasing rate, until the age of 35 or thereabouts. I added that "Such is the case in most of the phonetic parameters" (1976:82). I thus proceeded to ask a question. If changes can be directly observed taking place within one generation, what are the mechanisms of sound change as may be discerned from the study? Five.mechanisms were then singled out: Age factor, Educational background, Phonetic parameter (i.e., the choice of speech sound), Oscillation (for what Weinreich called retrograde), and Life expectancy. Each mechanism was elaborated on the basis of supporting data (1976:83-90).

Second, I took Jespersen's metaphor and compared it with my alternative, schematic representation of language change, which may be recapitulated as follows (1976:91).

Bkth Death Parent

Transmission

Child

Transmission

Granclchld

Figure 1: Illustration of Jespersen's Metaphor

- 9 6 -


Birth Death

marriage 35

Parent

Child

Transmission

Grandchild

marliage 35

Transmission

Figure 2: Alternative Schematic Representation o f

Language Change

This alternative theory suggests that the child can learn Iris language perfectly; that in spite of his perfect learning, the language in question still changes, because the model the child learns his language from had changed considerably before the child was born; and that the child's model had, in turn, learned from quite a different model, just as the child will serve as quite a different model to his own child. In this way, I concluded that a sound change, be it abrupt or not phonetically, can only be gradual in terms of behavior within each individual, with smooth (i.e., perfect) transmission from generation to generation (1976:92).

DATA BASE FOR THE PRESENT STUDY.

Having reviewed briefly my previous study, I am now in the position to construct my data base which will again be based on Nomoto'sstudies in Tsuruoka.

- 9 7 -


Beside segmental sounds, Nomoto and his associates also looked into grammar and vocabulary and came up with two kinds of changes in language behavior which they called area changes and individual changes. Area changes are obtained by comparing the sample from 1950, judged against the point of reference at that time, and the other sample from 1971, again judged against the point of reference at that time. By contrast, individual changes are determined in two steps. First, in 1971, as many members as possible from the 1950 sample were located and their language behavior was re-examined. Second, their records from 1950 were dug out for the purpose of comparison. In this context, I shall again follow Nomoto's suggestion and refer to the study that pertains to area changes as area study and the study that pertains to individual changes as panel study.

It should be mentioned in passing that there exists no other large scale study of language change based on the investigation of the same people over a long interval. By contrast, the two Japanese studies were conducted at an interval of twenty years or so, using the same questionnaire, in the same geographic region, and with, in many cases, the same subjects. This being the case, linguistic factors as well as non-linguistic factors, for individuals and a whole geographic area, can now be examined diachronically in the realms of grammar and vocabulary with certainty for the first time in the history of linguistics.

D a t a o n G r a m m a r f r o m t h e A r e a Study

Eleven items were investigated in the 1971 study of which nine had also been used in the 1950 study. To illustrate the diachronic aspect of the two studies, I shall refer to eight items (cf. Table 5). But only two of them will be used in this section to account for the grammatical change in different age groups; namely, the imperative of okiru ' to get up' and the negative of sunt 'to do'.

The imperative of okint has two variants, viz., okiro and okir~ The former is the standard form whereas the latter, the dialectal form. There are other variants that may be considered the equivalents; namely, okinasai(yo). These are included in "Others" in Table I below. Likewise, the negative of surtt has two variants, viz., sinaide

- 9 8 -

LANGUAGE SCIENCES, VOLUME I, NUMBER 1 (1978-79)

and sanegude, the former being the standard form and the latter, the dialectal form. "Others" include sende, seihen, etc. Table 1 summa- rizes the result of okinl and Table 2, those of sunt.

Observe now that these two verbs have entirely different distribu- tions; okiru has the dialectal variant as the dominant form, whereas suru has the reverse, with the standard variant being the cominant form, in both studies. Observe next that while the frequency of the standard form of sullt is the converse of the age factor, no such relationship exists in the dialectal form of okint. But observe further that when education is taken into account, the frequency of the dialectal form of okiru is conversely proportionate to the level of education and yet the frequency of the standard form of sttrtt is properly proportionate to the level of education, in both studies.

All this suggests that the same group of people may have different behaviors with regard to different verb inflections. Moreover, if these behaviors are taken over time (i.e., ff we compare the figures within and those without the parentheses), it is of interest to note that some age groups have changed more than some others, e.g., the 35-44 bracket and the 55-69 bracket. Furthermore, one would expect that people with higher education tend to change to the standard form. But such is not the case in either verb; as a matter of fact, a good number of them have changed to the dialectal form, especially for the verb okiru, indicating that the same non-linguistic factor may affect different forms differently.

Data o n V o c a b u l a r y f r o m t h e A r e a Study

In the case of vocabulary, seven items were investigated in both studies, although there were ten in the 1950 study and eleven in the 1971 study. Of the lexical items used in both studies, I shall refer to five (of. Table 7) but present only the result of two in this section; namely, ntsuban ' to house-sit' and moo 'any longer'.

The first item has three variants, rusuban, rttsui, and yosuri. The f'trst one is the standard form, whereas the other two are considered dialectal variants. Some replied rttstt sunt or its equivalents all of which are included in "Others." The second item also has three variants, moo, ato, and ado, with moo being the"standard form and the rest, dialectal. "Others" include tot temo and totemo. The results

- 9 9 -

S O U N D C H A N G E A N D L A N G U A G E C H A N G E

0

o

.~

.o

Z

o

o

0

I I I I I I

t t ~ O O

oO kO

o q . q ~

eq ,--~ q

oo '~D Cq

S'

e.)

O

~

O~

C6

- 1 0 0 -

LANGUAGE SCIENCES, VOLUME I) NUMBER 1 (1978-79)

0

Z

<

o o.

~J

~0

0 c~

0 [~

O0 ,-~

~- (~ 0 (~4 C4

"N a ~

~ ~ 0

~ 0 ~ ~

I l l l l l

O0 r r

O ' , O O O %

~ O q ~

O%

0

O%

o Z

- 1 0 1 -

SOUND CHANGE A N D LANGUAGE CHANGE

o

Z

o

c~

c~

os

c~

0 oo

C~ it%

~ o .

s

oo c'~ c-I

~i. os

I'~ oO

C'41

* q . ~ Q

t '~ c-I rr

. . . . . . ~

q q ~ q ~ ~ N N

~ 0 ~ ~

I I I I I I ~ 0 ~ ~

t"- 0 0 r.-.

r ,.-q 11-~ r

0

0

r , .~ ~

0 E

r

- 1 0 2 -

L A N G U A G E SCIENCES, VOLUME 1, NUMBER 1 ( 1 9 7 8 - 7 9 )

@

v

, ~ r - -

0

oo oO

~ ' ~ ~

I I I I I I

~ c ~ o

~. o0. q

~ 0 ~ "

r t~X tr~ O0 ~ C-I

r ~ - ,--q o O ,--q ~/')

@ CTx

" 0

t ~ ~D

0

r

Ox �9

~ ~ Z

- 1 0 3 -

SOUND CtlANGE SND LANGUAGE CHANGE

of these two items are summarized in Table 3 and Table 4. In Table 4, ato and ado are put together in one column.

As in grammar, the two lexieal items presented here also display interesting behaviors. Note first that while the first item leans heavily towards the standard form, the second item has the standard and dialectal forms more or less equally divided for the 1971 study. This phenomenon suggests that the same group of people may, too, have different behaviors with regard to different lexical items. Note next that in the 1950 study m o o has much higher frequency in all age groups than its dialectal counterpart. Consequently, if these behaviors are taken over time (i.e., if we compare the figures within and those without the parenthese) it is of significance to note further that rusuban has gained while m o o has lost; that is, more people have changed to the standard form in the case of n t s u b a n but about an equal number of people have changed to the dialectal variants in the case of moo. But, unlike grammar, people with higher education have changed to the standard form for n t suban but to the dialectal form for moo; in this respect, people with lower education have changed to the standard form much more than people with higher education, in the case of rusuban (i.e., from 48.0 to 76.7 compared with 68.0 to 78.4) whereas just about an equal number (percent) of people in each level of education have changed to the dialect form, in the case of moo.

Degrees of Change in Grammar.

In view of the variations in language behavior that have been presented, it can be anticipated that other items investigated may have variations as well. To show this, I have arranged the degrees of change in both the panel study and the area study with regard to the eight items shared in them. Each degree of change is the difference taken between the two studies on each item. The results are mapped into a ranking profile, so that not only the relative variations of all items can be compared but the two studies on the same items may be compared as well. Tables 5 and 6 summarized these results.

- 1 0 4 -


Table 5 Degrees of Change in the Panel Study

(Grammar)

1950 1971 Difference Rank

- s i n a i d e 67.3% 58.9% - 8.4 6 m i n i 59.8 47.6 -12.2 3 t s u y o k a t t a 59.8 49.6 - 10.2 4 - t a c h i 53.3 36.9 -16.4 1 - k a r a 50.5 38.3 -12.2 3 - k e r e d o m o 29.0 25.2 - 3.8 7 o k i r o 19,5 10.2 - 9.3 5 - n a r a ( b a ) 32.7 16.8 -15.9 2

SOURCE: Nomoto et al. (1974:252), adapted from Table IX-14.

Table 6

Degrees of Change in the Area Study (Grammar)


- s i n a i d e 66.6% 70.0% + 3.4 5 m i n i 57.5 46.4 - 1 1 . 1 3

t s u y o k a t t a 56.2 42.9 -13.3 2 - t a c h i 52.3 38.1 -14.2 1 - k a r a 42.8 33.5 - 9.3 4 - k e r e d o m o 29.1 29.1 0 7 o k i r o 19.9 17.5 - 2.4 6 - n a r a ( b a ) 28.8 15.5 -13.3 2

SOURCE: Nomoto et al. (1974:137--43), adapted from Tables VI-9 through VI-18.

- 1 0 5 -

SOUND CIIANGE AND LANGUAGE CttANGE

Notice that in each table, the suffix, tachi, ranks highest whereas the negative of sum (eft sinaide) and the imperative of okim (cf. okiro) are the reverse of each other in rank between the two tables. In general, however, the two ranking profiles look very much alike, which is to say that the longitidinal study, represented by Table 5, and the cross-sectional study, represented by Table 6, do not differ very much from each other, suggesting some kind of stability in the direction of change. The minor differences are due to the fact that the area study was based on a random sample but the panel study was not. The same is true of sound change, as was akeady reported in Pcng (1976:82-4).

Degrees of Change in Vocabulary

In the same fashion, I can show the variations in language behavior with respect to vocabulary. In keeping with the degrees of change in grammar, I have arranged the degrees of change in both the panel study and the area study with regard to the five lexical items shared in them. Each degree of change is again the difference taken between the two studies on each item. The results here are also mapped into a ranking profde for the comparison of the relative variations among the lexical items and between the two studies. Table 7 and 8 summarize these results.

T a b l e 7

Degrees of Change in the Panel Study (Vocabulary)


itsumo 84.1% 84.1% 0 rusuban 76.6 89.7 +13.1 hazukasii 92.5 90.6 - 1.9 odoroita 58.9 60.8 + 1.9 moo 73.8 46.7 -27.1

4 2 3 3 I

SOURCE: Nomoto et al. (1974), adapted from Table IX-17.,

- - 1 0 6 - -


JTable 8 Degrees of Change in the Area Study

(Vocabulary)


itsumo 88.2% 93.7% + 5.5 5 rusuban 53.9 80.7 +26.8 1 hazukasii 86.8 92.1 + 5.3 4 odoroita 61.2 78.7 +17.5 3 moo 70.4 44.9 -25 .5 2

SOURCE: Nomoto et al. (1974), adapted from Table VI-23 through VI-28.

Here, notice that except for the reverse in rank of msuban and moo between the two studies, the two ranking profdes look very much alike, the minor differences being due to the sampling in the two studies (i.e., random in the area study versus non-random in the panel study). The important thing to observe is, of course, the fact that the same group of people, or even the same individuals, for that matter, have changed differently with regard to different lexical items. I shall have more to say in this connection later.

DISCUSSION

With the materials presented above in mind, let me now return to the questions originally asked. Recall that the first two questions are specifically concerned with sound change, whereas the remainder are more general. I shall, therefore, take up the f'trst two questions in the next section and treat the others in the following sections.

S o u n d C h a n g e

First, I should mention that it is rather unfortunate that Johnson repeats the misconception that sound change must take place across generation boundaries. The reason is that this view, shared by most

- 1 0 7 -

SOUND CIIANGE AND LANGUAGE CttANGE

historical linguists, raises a serious theoretical problem; namely, that somehow language changes, be they sound changes or lexical changes or what not, pop up all of a sudden when one passes his language to the next generation, while during his entire life span nothing changes in his own linguistic system (Peng 1976:70). I have also pointed out that if language change need not take place across generation boundaries, the only alternative is that language change is a continuum of constant changes and, therefore, need not wait for a new generation in order to take effect or to be observed (Peng 1976: 70).

In light of this t'mding, coupled with the empirical evidence presented above that sound change takes place not only within each individual but at an ever decreasing rate, that is, taken cross-sectionally, a person may change his linguistic system within his life span but gradually reduces his rate of change until the age of 35, even though changes may continue to take place after the age of 35 (but at a much reduced rate), it is hard to believe that sound change must take place across generation boundaries. Of course, what happens at each boundary is not the source of change (or imperfect learning or errors) but a part of the continuum which serves as transmission from generation to generation. It must follow, then, that any theoretical construct of sound change that assumes changes across boundaries of two or more generations cannot be considered plausible.

Second, given the above finding that sound change takes place within each individual at an ever decreasing rate, I must now ask whether it is true that change begins slowly and accelerates in succeeding generations. Although the data presented by Johnson may seem suggestive of this tendency, a closer look at his data indicates otherwise (especially when they are compared with ours), simply because ours can account for changes within one generation, whereas Johnson's (which include several sources) contain materials from at least three generations, each having a different age bracket and being younger than the preceding generation. For instance, he uses Labov's materials from Martha's Vineyard (aw) that cover three generations; namely, Oldest Generation, Middle Generation, and Youngest Genera- tion. But note that the three generations correspond to age level 61 to 90, age level 31 to 60, and age level under 30, respectively, and

- 1 0 8 -

LANGUAGE SCIENCES, VOLUME l, NUMBER I (197'_.~-79)

that there is no information about the changes of the younger age groups when they reach the older bracket; that is, nothing is known about the changes of the people in the 31 to 60 group, when they reach the next level, or of the people under 30, when they become 60. Thus, when the numerical values (Johnson 1976:168) of 0.06, 0.37, and 0.88 are compared, the differences do not represent the acceleration of change rate in three succeeding generations; rather, they indicate three static manifestations of one continuous change taken cross-sectionally. In other words, the difference, for example, between 0.06 and 0.37 does not constitute a speed-up, nor is it illustrative of acceleration in the rate of change in two successive generations; it merely suggests the possibility that people of tile Middle Generation will not continue centralizing ([au] ~ [eu]) at tile same rate when they reach the next age level (61-90) and nothing else. In order to get the dynamics of change, however, what Johnson should have done would be something like this: Wait for the people of the younger generation to reach tile next age level (i.e., 30 years) and, then, compare their centralization with that of the older generation at the same age level. For instance, he should have got tile numeric value of the Youngest Generation (under 30) when they reach the next age level (31-60), and compare it with the numeric value of the Middle Generation when they are still at the level of 31-60, and do likewise for the Middle Generation and tile Oldest Generation. But nothing of this sort has been done. Consequently, he has no data whatsoever to support the claim of acceleration in the rate of change.

By contrast, our data from the area study show exactly this kind of dynamics pertaining to change. That is, the results of all age groups investigated in 1950 were compared, 21 years later, with those of similar age groups investigated in 1971. Thus, we have information not only of two comparable age groups, say, 35-44, one taken from the 1950 study and the other from the 1971 study, for comparison pertaining to change, but also of different age groups taken cross-sectionally for comparison pertaining to the rate of change. The data from the area study are, then, backed up almost one to one by our data from the panel study. Thus, in the case of sound change, we can comfortably conclude that all age groups have

- 1 0 9 -

SOUND CItANGE AND LANGUAGE CI1ANGE

changed but that the rate o f change goes down, as the age goes up, within each generation. This finding is based on the result o f our data in Figure 3 and Figure 4 which summarize the f'mdings o f the 1950 and the 1971 studies.

% 100.(

90

80

70

60

50

40

30

96.5

8.7

Average 83.2

68.1 66.5 ~ - ~

61.0,:'/ "

Average 57.1

82.9

"•'----•79.5

~ 6 2 . 9 N61.0

\ %. . . . . . . . .

\

X

\ \

\N, 43.2

",,..%

"-.36.1

i i | i I

15 20 25 35 45 55 Age I I i i I i 19 24 34 44 54 69

1971 Study . . . . 1950 Study

Male 81.3 Male 53.9 Female 84.5 Female

Figure 3: Percentage of Standard Responses with Respect to Phonetic Quail W Plotted for Different Age Groups.

- 1 1 0 -

LANGUAGE SCIENCES, VOLI.~IE I, NUMBER 1 (1978-79)

% 90.0

80

70

60

20

I0

66.5..-- . . . . .

61.0~ ~

- 6~..1 . ,...,.., ~ 6 3 . 9

"" "-.,,61.0

- - 1971 Study

. . . . . 1950 Study

. . . . . . . D~ference between the Two Studies

21.3

"" ,~17.0

"'- e2.9

15 20 25 35 Age 1 I I I

19 24 34 44

Figure 4: Differentiation in the Percentage of Phonetic Quality for the Same Age Groups in Each Study

As may be seen in Figure 3, in 1950 the age group which conformed most closely to the standard dialect was the 2 5 - 3 4 year old group, whereas in 1971 the younger tile subjects were, the more they conformed. (It is to be understood that file comparisons apply only within the range of 15 to 69 years o f age, s~lce subjects in both surveys fell only within this age span.)

- 1 1 1 -

SOUND CHANGE AND LANGUAGE CHANGE

The individual differences between the two surveys can be more readily compared if the two curves of Figure 3 are aligned by moving the 1950 curve 21 years to the right, since the interval betwen the two studies is 21 years. That is, the 15-19 year age group of 1950 is now directly under the 35---44 year age group of 1971, since the subject who was 15 years old in 1950 was 36 years old in 1971. The differences of the two curves thus readjusted are indicated in Figure 4. It is interesting to note that the resultant differentiation between the two studies is a straight line, and that all age groups have been affected to some extent by the trend to conform to the standard dialect.

From the aforementioned it must follow that there is a certain degree of incongruency in Johnson's data. For instance, how can he be sure that the first generation (Oldest Generation) did not have a faster rate of change when they were younger and that the third generation (Youngest Generation) will not slow down when they grow older? In fact, his data support precisely what we have found (cf. Figure 4), if his three generations are regarded as cross-sectional, which is to say that the rate of change will be reduced in all cases, e.g., Martha's Vineyard, as one goes from the Youngest Generation (0.88) through the Middle Generation (0.37) to the Oldest Genera- tion (0.06). The fact that Johnson has no data for each individual within one generation (which, by contrast, we have in the panel study) regarding his or her changes suggests that he cannot be sure of the rate of change being faster in each succeeding generation, that is, accelerating in succeeding generations. To demonstrate this fact, let me resort to my schematic representation of language change which was stated earlier as an alternative to Jespersen's Metaphor.

To this schematic representation, however, I must add in each generation the rate of sound change taken from Figure 4 which depicts sound change within one's own life time (notably from 15 to 44 years of age) with a plotted extension (dotted line) beyond 44. I have also circled three places which correspond to Labor's three age levels utilized in Johnson's data. The result of these modifica- tions in the schematic representation is recapitulated as follows:

- - 1 1 2 -

Bkth


0.06

. . . . Q Death $ V r r f f r f ' .

Age 10 20 30 40 50 60 70

Birth

3 7 . . . .

f r f f f

80

Oldest G.

~ m

f ! I

Age 10 20 30 40 50 60

Figure 5:

Birth

"<k 70

Middle G.

Ir ir I f Ir f It

Age 10 20 30 40 50

Youngest G.

Schematic Representation of Sound Change and Its Rate

Observe now that this schematic representation shows that what Johnson has done is pick the three age brackets, one from each generation, with differing numeric values of vowel centralization (each of which falls in line with, and can be explained by, the rate of sound change therein). From my point of view, then, that tile Oldest Generation has tile lowest numeric value is not because, as Johnson has claimed, change begins slowly at first, but because the age bracket (61-90)picked has according to the.schematic representation already slowed down the rate of change; and likewise, that the

- 1 1 3 -

SOUND CttANGE AND LANGUAGE CHANGE

Middle Generation and the Youngest Generation have successively increased their numeric values may also be explained by the fact that in the schematic representation they are younger in age and, therefore, stand higher in the rate of change and, consequently, is not at all because change begins slowly and accelerates in succeeding generations, as claimed by Johnson who also lends support to the claim of Wang and Cheng (in their discussion of lexical diffusion) and of Bailey that sound change follows an S-curve (Johnson 1976: 168). (By an S-curve is meant that sound change begins slowly and then increases rapidly [in some cases leaving residue] .) In light of my explanation above, it should be clear that none of the assertions made by Johnson and others is true.

S o u n d Change versus Language Change

Having discussed the f'trst two questions, I am now in the position to address myself to the third question raised at the beginning of this article. The key issue here is the problem of generalization in theorizing, a point that will be taken up again for discussion in the next section. It should be noted, however, that this kind of problem has existed in linguistics (historical linguistics in particular) at least since the time of Neogrammarians (and perhaps since the beginning of modem linguistics [ 1786]), simply because historical linguistics has always been concerned mainly with sound change.

Given this overall background, it is only natural that linguists like Johnson would want to theorize a statement as generally as possible. But generalization of this kind can often get out of hand. This is precisely what has happened to Johnson's Index, when he says that "It is also suggested that the Index can be applied to syntactic and lexical changes as well" (1976:165), although in his conclusion he tones down considerably by saying that "As we compile more data to measure particular changes, we can expand our inquiry to a sector of language, such as phonology or morphology, and f'mally, via the uniformitarian principle (Labor 1972c), gain more insight into historical changes" (1976:171).

One reason for this criticism is that, although it has been known that sound change may lead to grammatical change (and lexical change) and vice versa, this is true only when the end points have

- 1 1 4 -


been achieved. Even so, a sound change does not necessarily affect all grammatical properties that subsume the sound involved, nor does it influence in any significant way the lexical status of the items containing the sound in question. At times, moreover, a lexical change is due not to any sound change but to human errors. In short, we know very little, in fact, too little about the relationship between sound change and other changes to warrant any bold attempt or unfounded statement for generalization regarding language change. A good example is English/~] the development of which as a phoneme must have been due to a pressure for greater regulariza'tion in the English phonemic system (Peng 1976:70). But note that this particular sound change as an end point has affected such lexical items as sing ( < AS. shzgan) and khTg ( < AS. cyng, a contr, of cynhlg) to produce a contrast with shz and khl, respectively, as an end point. And yet in the case of shlk ( < AS. s#zcan), although the grammatical property has changed, the end point of the phoneme in question has never materialized. Moreover, think comes from thenchen which was confused with thhzchen ' to seem' (cf. AS. thencan), a result of change not due so much to sound change as to human error and other changes. To complicate the issue, the factor of borrowing may be added, as in file case of apron [by faulty separation of a napron] (cf. ME. napron; OFr. l~peron). It seems, then, clear tlaat sound change and language change are not synonymous and that until linguists have come to know more about the relationship between sound change and other changes, care must be taken at all times to avoid making shallow generalizations regarding language change.

Another criticism in this connection is the very fact that Johnson focuses on particular ongoing sound changes, not to mention his faulty focus stated earlier, which, in the absence of any data on ongoing syntactic and lexical changes, automatically prevents him from making any sensible generalization of his Index from sound change to other changes. To uphold my criticism, let me point out that when dealing with sound change in progress, the number of sounds involved are relatively small (by virtue of the fact that the total number of phonemes in the phonemic inventory of any language is small to begin with) and are easily

- 1 1 5 -

SOUND CttANGE AND LANGUAGE CtlANGE

identifiable, because they occur not only in lexical items that may undergo ongoing changes but also in lexical items that are stable, even though they do not occur alone, whereas in the case of lexical change in progress, the number of lexical items in the vocabulary of any language are usually enormous to begin with and, therefore, those undergoing changes are more often than not larger than the number of sounds involved in ongoing changes and, worse still, much more difficult to identify. To complicate the matter further, I could mention that unlike sounds, which could not occur in isolation, lexical items do occur in isolation, e.g., imperative forms and names of merchandise. For this reason, when an expected form fails to appear, and a substitute is used, in an investigation of a sample such as those examined by Nomoto and his associates, unless altemants are known ahead of time to be in existence in the community under investigation, the same method is employed, and the same items are tested over time (as in the case of Tsuruoka with an interval of 21 years), it is very difficult to pin down tile lexical items that are supposedly undergoing changes. Furthermore, altemants at times bear no resemblances to the items being replaced, e.g., rttsttban and yosuri in Table 3 or shmide and sanegude in Table 2, for that matter, so much so that unless care is taken to garner geographic and social distribu- tions of target items in advance, it is simply impossible to focus on particular ongoing lexical changes. I suspect that partly because of these reasons Nomoto and his associates chose only ten and eleven items in their two investigations on vocabulary in the area study. Consequently it is rather unrealistic, to say the least, for Jotmson to try to generalize his Index, which is constructed on the basis of sound change (albeit erroneously), for syntactic and lexical changes.

N o n - u n i f o r m i t y of Changes

As were already indicated earlier, my data on both grammar and vocabulary show that the same group of people may have different behaviors with regard to different verb inflections (cf. Table 5) and lexical items (cf. Table 7). I must now add that in the case of my previous study on sound change, I have already

- I 1 6 -

LANGUAGE SCIENCES, VOLUME l, NUMBER I (1978-79)

pointed out that "the choice of language sounds may also play an important role as a mechanism in sound change." That is, certain sounds are more susceptible to changes than certain others (Peng 1976:84). In the same vein, I have also said the following:

" . . . it is clear that in Japanese, vowels are probably less susceptible to change than consonants. On the other hand, if one undertakes an investigation of this kind on English, he may well find that vowels are more suscept~le to change than consonants, llistoricaUy, this conjecture seems plausible, as may be evidenced by the English Great Vowel Shift. It remains to be seen, however, how other phonetic parameters undergo change in Japanese (1976:90).

Given these views, which are supported by factual data from Japanese (in my previous and present studies), language change in progress cannot be lumped together in one package as if all sectors (or elements of one sector) progress in the same direction or at the same pace. This is because, as I have said already before, even the same group of people, or the same individuals for. that matter, will change differently with regard to different phonemes, lexical items, and grammatical properties, a point that is strictly in line with the Malinowskian definition that language is only the general norm of human speech (or sign, in case of non-verbal) activities (Peng 1976:93). From this point of view, then, the norm will change when individual (speech/sign) activities change, a fact that reflects the dynamics of language change. By non-uniformity of changes is, thus, meant the individual activities, taken over time, that are reflected in the choice of phonemes, lexical items, and grammatical properties. In other words, these individual activities, taken over time, show a certain patterning in one sector (or with regard to certain elements within one sector) but this kind of patterning, as was already pointed out with respect to my answer to Q3, may or may not be observed in another sector (or among other elements of the same sector).

In concrete terms, then, let me cite English as an example which shows marked ongoing changes in vowels, rather than in consonants. This is, of course, historically true as well. Therefore,

- 1 1 7 -


it is probably for tiffs reason that so many recent studies on sound change in progress based on English have concentrated on vowels. Johnson has certainly made good use of this tendency. Itowever, another language like Japanese does not necessarily follow suit; my own study mentioned earlier in this section clearly suggests that in Japanese consonants are much more susceptible to change, although both vowels and consonants are included in the statistics used in the study. Thus, to answer Q4, whatever there is to discover regarding sound change in progress based, say, on English vowels, cannot and rriust not be generalized for the vowel system of another language, because of the principle of non-uniformity of changes stated above, even though the technique employed in the study of one language can be adapted for the study of another in terms of change. The reason for this cautious statement is because linquists have had the tendency of jumping the gun to look for some kind of universality, when more work on individual languages is badly needed. I hope Iffstorical linguists have learned the lesson from the past, never to repeat the same mistake in the current explora- tion of language change.

U n e v e n n e s s o f the Variables Affecting Changes

Recall now that in the previous sections ihave statedthat the level of education affects the choice of neither the verbal forms nor the vocabulary in a predictable way. For instance, one might expect to see more educated people changing to standard forms across the hoard, with less educated people preferring dialectal forms. But this is contrary to hard fact; the truth of the matter is that the same level of education affects different forms or sounds differently. Figure 6 is illustrative of this tendency. Notice that the scores of S (Sound) in 1971 are higher than those in 1950, whereas in G (Grammar) the situation is reverse.

The same is true of age. That is, one might also expect to see younger people changing to standard forms across the board, with older people preferring dialectal forms. But, again, this expectation is unwarranted. The reason is that, while in Table 1 it is true that

- 1 1 8 -

%

80-

70'

60-

50"

40"

30-

20"

10-


1971 i950

S (Sound)

G (Grammar)

A (Accent)

/ /

/ /

o"

I I I Ed. Low Middle High

Figure 6: Scores according to Education in the Panel Study

younger people do change more to the standard form, in Table 2 it is older people who change more to the standard form; moreover, in Table 3 there is no denying that the standard form (msuban) receives a higher rate of change from younger people than from older people, in Table 4 more people change to the dialectal form (ado) regardless of age. Figure 7 reflects this tendency. Notice, also, that the positions of S and G in both studies are the reverse of each other across all age groups.

- 1 1 9 -

%

80-

70

60"

50"

40-

30-

20-

10

SOUND CIlANGE AND LANGUAGE CItANGE

S (Sound)

G (Grammar)

1971 1950

. . . . . - 0

A (Accent)

Age

Figure 7:

" ' O

I I I 54 69

Scores according to Age in the Panel S t u d y

All this seems to suggest that whatever (social) variables there may be that affect changes, they affect them quite unevenly; that is, one set of variables affects a certain type of change in a degree greater than that in which another set o f variables affects a certain other type of change.

More and more linguists are becoming aware that synchronically language is not a self-contained entity. But to what extent it is not self-contained, when taken over time, is not only an intriguing but also an important question that awaits an answer from historical linguists. As a first step in this direction, I have mentioned five

- 1 2 0 -


mechanisms of sound change in Peng (1976:83) which were reiterated briefly at the beginning. From the point of view of these mechanisms, it-seems reasonable to say that language is not self-contained at least to the extent that, when taken over time, historical linguists can no longer do themselves justice by ignoring non-linguistic change while dealing with linguistic change.

C O N C L U S I O N

I have in the preceding sections been concerned with four things; namely, the reality of sound change, the human behavior of other linguistic changes, the relationship of sound change and other linguistic changes, and the rate of change. Let me now summarize by proposing three points, so as to bring the whole presentation to a close.

Firstly, I believe I have provided enough evidence upon which to draw a conclusion that although linguists have been aware that when we speak of change it is people who change, and language change is simply a manifestation (or symptom) of human change, not enough research is being done in the area, or attention paid to the probe, of what I have called th e dynamics of change. This kind of study requires both cross-sectional and longitudinal investigations of fairly large samples, in the same areas, with the same method, and at an interval of hopefully 20 years. Since a research of this nature is often painstaking and costly, historical linguists should turn to linguistic geographers for assistance in the provision of advice and materials. With the exception of linguists like Kurylowicz who once renounced for internal reconstruction all support from linguistic geography and other social sciences (1964), historical linguists must actively seek more cooperation from linguistic geographers as well as from all other social scientists. It is through this kind of cross-fertilization that language scientists can hope to achieve the goal of dealing with the dynamics of change, among other things.

Secondly, I have also presented sufficient evidence to support my claim that if it is people who change, the change'itself must take place within each individual to begin with, whose rate of change is

- 1 2 1 -

SOUND CHANGE AND LANGUAGE CHANGE

affected by his or her own physical condition (age or maturation) as well as by the environment. Thus, as each individual increases his or her age, the rate o f change decreases. Nobody knows, however, what will happen, if life expentency is extended beyond 100 years o f age, to the rate of change. According to William J. O'Neill, people in the future may live for 200 years (The Japan Times, June 2, 1977). Here is the excerpt o f what he reports:

Some scientists think the time is coming when people can expect to live for 200 years or more.

As 1977 began, there were about 13,000 people in the United States who were born in 1876 or earlier - a small number in a population that exceeds 216 million. Yet in some parts of the world, living more than a century is not unusual

Gerontologist Alexander Leaf a few years ago visited Abkhazia in the Soviet Union, Vilcabamba in Ecuador, and Hunza in Kashmir, and found that in all three places 100 years was regarded as a normal lifespan.

In Abkhazia, he met a woman who was 130. Dr. Leaf noted: "Many elderly couples had been married 70, 80, or even 100 years." He reported that most of the centernarians worked and led active lives.

Since the structure of the molecule DNA f'trst was descried by scientists in 1953, the poss~ility has grown that aging may be delayed and life drammaticaUy prolonged.

Of course, life expectency alone is not the influencing factor o f human change; the envkonment counts heavily in this regard, the foremost influencing factor in the environment being human inter- action. Note here that although the Japanese now live longer (perhaps longest? ), 57% of Japanese population crowded on only 2% of land, according to the lastest report prepared by the Prime Minister's Office (The Japan Times, June 27, 1977). In this respect, then, Johnson is probably right in saying that "change proceeds more rapidly in urban than in rural areas" (1976:165). I have reached a similar (albeit more substantial and elaborated) conclusion in my "Urbanization and Language Sciences: The Japanese Case" (Peng 1978).

Thirdly, and Fmally, I should mention that if human change is the key to language change (and sound change in particular), more

- 1 2 2 -


rigorous research is needed in such realms of specialization as phonetics, neurolinguistics, sociolinguistics, and pedolinguistics to help determine the change and development in the total behaviors of humans as organisms.

R E F E R E N C E S

Johnson, Lawrence 1976 "A Rate of Change Index for Language," Language h2

Society 5.165-72. Kurytowicz, Jerzy

1964 "On the Methods of Internal Reconstruction," in Pro- ceedings of the Ninth blternational Congress of L#zguists, pp.9-31, H. G. Lunt (ed.), The Itague: Mouton.

Nomoto, Kikuo, et al. 1974 Chiiki Shakai no Gengo Seikatsu (Language Behavior of a

Speech Community), Report 52, Tokyo: National Language Research Institute.

Peng, Fred C. C. , 1976 "A New Explanation of Language Change: The Socio-

linguistic Approach," Forum Ling~tisticum 1.67-94. 1978 "Urbanization and Language Sciences: The Japanese

Case," in Lang~tage in Context, pp. 1-26, Fred C. C. Peng (ed.), Hiroshima: Bunka Hyoron Publishing Company.

N . B . This article is a revised version of the paper presented at The Third International Conference on ltistorical Linguistics held in Hamburg during the summer of 1977. I am grateful to the Ministry of Education for making the trip possible.

- 1 2 3 -

sound change and language change: a sociolinguistic interpretation

Documents