The loci of stuttering: Grammar or prosody?

The loci of stuttering: Grammar or prosody?

JOURNAL OF COMMUNICATION DISORDERS THE LOCI OF STUTTERING: 12 (I 979). 283-290 283 GRAMMAR OR PROSODY? MARCEL E. WINGATE Depurtment ofSpeech. ...

551KB Sizes 0 Downloads 41 Views

JOURNAL OF COMMUNICATION

DISORDERS

THE LOCI OF STUTTERING:

12 (I 979). 283-290

283

GRAMMAR

OR PROSODY?

MARCEL E. WINGATE

Depurtment ofSpeech. Washington State Universit.v, Pullman.

Washington 99164

Much of the research dealing with linguistic dimensions in stuttering has emphasized the various aspects of grammar, particularly as these aspects contribute to the meaning of utterances. However, data from several sources indicate that meaning and grammatical features are not likely to be the important language variables related to stuttering, but that stuttering occurrence is linked more closely to the prosodic dimension. The published literature in this area reports work done with English-speaking subjects; comparable material from other languages would be of considerable value to our understanding of linguistic determinants in stuttering.

Over a period of 10 years, beginning in 193.5, a series of articles authored principally by Brown (Johnson and Brown, 1935, 1938, 1939; Brown, 1937, 1938a, 1938b, 1938~; Brown and Moren, 1942; Brown, 1945) reported the discovery of certain features of the English language which were evidently related to the occurrence of stuttering. These findings were corroborated by the results of similar studies undertaken by Hahn (1942a, 1942b) and by the results of related work done by Eisenson and Horowitz (I 945). For some time these findings evidently did not arouse a great deal of interest, since the topic was not pursued further at that time. Then in the 1960’s, evidently as a reflection of the expanding interest in the field of linguistics, there developed a rather vigourous resurgence of attention to this area of stuttering research. Much of the renewed research took Brown’s work as its point of departure and extended this area of investigation into linguistic influences on stuttering occurrence. Throughout the course of the recent revival of interest in this area of study, inquiry focused primarily on one or more of the four word-related features of the language which Brown identified and discussed at considerable length in his series of publications. These four features-grammatical class, initial sound, sentence position, and word length-were repeatedly found to be associated with the occurrence of stuttering. The results of much research agree in revealing a higher incidence of stuttering on: 1) “content” words (nouns, verbs, adjectives, adverbs) as compared to “function” words, (articles, prepositions, conjunctions,

Address correspondence to: Marcel E. Wingate, Ph.D., 201 Daggy Hall, Department Washington State University, Pullman, Washington 99164.

B Elsevier North Holland, Inc., 1979

of Speech,

0021-9924/79/04283-08$01.75

284

M.E. WINGATE

pronouns); 2) words beginning with consonants’ (as compared to vowels); 3) words occurring very early in a sentence; and 4) words that are five letters or longer in length. The continued preoccupation with the four features mentioned above was conditioned not only by their discovery by Brown but also by the fact that he considered them to be the essential linguistic features relevant to stuttering. He devoted the great bulk of his analysis to a consideration of these four features and to the development of a rationale that would account for their influence. Brown’s (1945) conviction regarding the significance of these features is reflected in his statement that ‘*. Ithese features] . . appear to be by far the most important determinants of the loci of stutterings, and may be the only important ones. ” While Brown considered some alternative explanation of his findings, he maintained a clear preference for a psychological interpretation-specifically, in terms of the psychology of stuttering incorporated in the “evaluation” theory of Wendell Johnson. Basically. the frame of reference for his interpretation of these findings involved the conception that the stutterer’s speaking is undertaken in an atmosphere of concern about both his adequacy in conveying his message and his success in doing so. Supposedly, the stutterer makes evaluations of what it is in the speech sequence that is most essential to convey meaning and he is sensitized (though perhaps not consciously) to these features. Since, according to this viewpoint, these acts of evaluation are the precipitants of stuttering, instances of stuttering are therefore most likely to occur in association with those features of the language which are thought to contribute most to the meaning of an utterance, or which otherwise might assume a psychological prominence in oral-verbal communication. For each of the four features identified earlier (grammatical function, word length, sentence position, and initial phoneme) Brown provided a rationale to account for their importance in contributing to **. the importance or conspicuousness of the various words in the speech sequence : .” (Brown, 1945). Much of his rationale centered around the variable of “meaning. ” Thus, he pointed out that more stuttering was associated with ‘*. words upon which sentence meaning depended most ” (B rown. 1038c).’ Similarly, Brown (1938~) accounted for the inllucncc of the type of initial phoneme by the “. inference that the

‘Both Brown (Johnson and Brown.

1935) and Hahn (1942b) reported a few consonants which also

were associated with a low incidencn of stuttering. However.

recent research has tended to simply

make the distinction between consonants and vowels in their analyses. ‘The feature of grammatical class lends itself very readily to a Justification of this sort. Since “content” words refer to obJects. events. actions and qualities, they can be said to carry more informatmn than “function” words. The point has been elaborated in subsequent research. However. persuasive arguments can be mounted to contradict the claim that content words regularly carry mwr information than function words.

LOCI OF STUlTERING:

GRAMMAR

OR PROSODY?

285

greater difficulty of consonants as compared with vowels is in part due to their greater relative importance for clarity and distinctness, and hence meaning .” He explained the relative importance of word position (Brown, 1943, 1945) as reflecting a ‘*gradient of meaning” in which early words in a sequence are appraised as more important to meaning than words occurring later in the sequence. In regard to the variable of word length, Brown (1942, 1945) contended that longer words are evaluated by the speaker as more important due to their greater prominence in the speech sequence. Brown devoted relatively little attention to the finding that the preponderance of stutterings involved the initial sounds of words. Again, his interpretation of the significance of this feature was presented in terms of the conspicuousness, and therefore importance, of the beginning of a word. He did not develop further this rationale of the “. . . psychological primacy of the initial part of the word for the stutterer . . ” (Brown, 1938b) and the reader is left to accept the explanation as it stands. It is surprising that the matter of word-initial sound was accorded so little attention. In the first place, even if considered as a separate feature it was found to be associated much more frequently with stuttering than any of the other features; Brown (1938~) reported that slightly over 92% of stutterings were associated with initial sounds of words. More importantly, initial sound seems clearly to be more basic than any of the other features. That is, it is evidently independent of a word’s grammatical class, length, position in a sentence, or the type of sound with which a word begins. At the same time, the feature of initial sound almost completely overlaps the other four features. Complete overlap is prevented only by the matter of word length: if all words in the language were one syllable in length then the incidence of stuttering associated with word-initial sounds would undoubtedly have approached unity. This point will be developed presently. Brown (1938b) also gave only secondary consideration to his findings regarding the association of “word accent” with the occurrence of stuttering. It is curious that he did not devote more attention to this matter, particularly in view of findings that indicate the substantial effects of this variable. Brown’s data regarding the influence of word accent are not very clear since he reported them somewhat differently than the results relating to the four features he considered to be important. Brown reported the data regarding word accent in terms of the number of subjects who evidenced the effect, rather than in terms of the frequency with which this effect occurred. However, from his statements it is quite clear that all of his subjects stuttered substantially more on accented syllables. It is also evident that stuttering occurred most often on an accented syllable regardless of the position of the syllable in a word. Brown dealt with the features of word-initial position and accented syllable as though they were independent variables. At the same time, he explained their

M.E.

286

WINCIATE

influence as emanating from the same source; i.e., his preferred interpretation was that both initial position and accent were features of “prominence” or “conspicuousness” and therefore evaluated by the speaker as being of more importance to the message. The subsequent research not only directed its attention toward the four wordrelated features cited by Brown, but was also greatly influenced by his explanation of these phenomena. Thus, all of the later research concerned itself with one or more of the features that Brown emphasized and a substantial proportion of that work was addressed to some aspect of the presumed relationship between “meaning” and stuttering (Conway and Quarrington, 1963; Hannah and Gardner, 1968; Kroll and Hood. 1976; Lanyon and Duprez, 1970; Peterson, 1969; Quarrington, 1965; Schlesinger et al, 1965; Soderberg, 1967, 1969, 1971; Taylor, 1966).3 The findings of some later work has provided grounds for questioning whether “meaning,” in the sense intended by Brown, is the important aspect that elicits stuttering (Lanyon. 1969; Lanyon and Duprez, 1970; Peterson et al, 1969). Further, some studies have found reason to question the significance of the “grammatical class” feature. For instance, Soderberg (1967) suggested that the apparent effect of the grammatical class of words is more a matter of uncertainty in word selection. Taylor ( 1966) believed her data yielded evidence that the grammatical class of words masks the more fundamental distinction which can be drawn in terms of phoneme type (i.e., consonant vs vowel). Quarrington et al ( 1962) found a different “gradient” of stuttering among content words than the one reported by Brown. Actually. doubts about the relevance of grammatical class per SP could well have been inferred from certain of Brown’s early findings. For instance, Brown (1938a) reported that there was consiaerably more stuttering in the reading of contextual material than on the same words read from a list. This finding should have been sufficient to suggest that the apparent relation of stuttering to word type reflects something beyond the fact that a word belongs to a certain grammatical category. Certain results reported later by Eisenson and Horowitz (1945) were even more pertinent. These authors found that context produced a differential effect on the relation between stuttering and content or function words: an increase in contextual organization of reading material was associated with increased stuttering on content words but a decrease in stuttering on function words.

3Thr

majority

as the

“Semantic

techniquub. which

There

of these

arc‘ many

has been most

the prebent efficiently

studies

have

Differential”

discussion

uidely for

by approaching

invoked

and objections employed

purposes the major

relatively

measures

of

to be raised m studies

of contrast. point

more

sophisticated

“information ahout

these

of stuttering. the intent directly.

of this

measures

load”

based

tcchniquea,

While

such

presentation

of

“meaning.”

such

on forward-guessing particularly

criticisms

the

latter.

are relevant

can be achieved

to

more

LOCI OF STU’ITERING:

GRAMMAR

OR PROSODY?

281

These findings, from both the initial and the more recent studies, indicate that it is not the grammatical status of the words themselves that is related differentially to stuttering occurrence, but that the important variable is one that reflects certain properties of words as conditioned by the connected verbal sequence which they constitute. Such properties are well represented in prosodic variables, dimensions of the speech process which should immediately come to mind when considering connected speech. There is some evidence in the literature which suggests that prosodic variables are importantly involved in stuttering. In a study reported several years ago (Wingate, 1966) I found that the adaptation effect could be significantly altered by creating even rather limited prosodic changes in the reading passage. Somewhat later (Wingate, 1967) I found the performance of a group of stutterers to be significantly inferior to a matched group of nonstutterers in the solution of “Slurvianisms, ” a task which evidently relies heavily on prosodic manipulation fcr successful achievement. More recently (Wingate, 1976) I have presented an analysis of many sources of evidence which indicates that stuttering is essentially a prosodic defect. So far no research has made the attempt to investigate a specific prosodic variable. However, data from several sources have particular bearing on this matter. The reports of three studies have called particular attention to a preponderance of stuttering in word-initial position. Johnson and Brown (1935) reported that 92% of stutterings were associated with the initial sound of a word; Hahn (1942b) reported 98%; and Taylor (1966) reported 97%. These statistical statements are corroborated by notations made in less formal sources (e.g., Bluemel, 1913; Emerick, 1963; Johnson, 1955) that stuttering occurs essentially on initial sounds. In effect, this observation is evidently so common that it is taken for granted, and its significance thereby ignored. The potential import of initial position becomes even more compelling if it is considered in terms of s~lluhleinitial position. Brown (1938~) reported that over 99.5% of all stutterings occurred “. . on the initial sound of a syllable either at the beginning of a word or within it. ” This value, which approaches unity, is of particular significance when considering that it expresses a function occurring in a disorder that otherwise is claimed to show so much variability. The next step is to consider the nature of the syllables whose initial sound is stuttered. The reports of both Brown (1938b) and Hahn (1942b) clearly related the occurrence of stuttering to the initial sound of the stressed4 syllable. That is, in most instances when stuttering did not occur in word-initial position, the initial syllable in that word was not the stressed syllable; stuttering occurred on the syllable in that word which bore the stress. If we now take note of the fact that the 4Brown and Hahn used the word “accent”

to identify this feature

288

M.E. WINGATE

great bulk of English words carry the stress on the initial syllable (Bolinger, 1968; Delattre, 1963), the repeated finding that a preponderance of stuttering (in English-speaking stutterers) involves the word-initial position becomes an expression of a more fundamental principle-i.e., that stuttering is some function of linguistic stress. There is other evidence to suggest that stuttering occurrence is associated with linguistic stress. Earlier in this paper reference was made to results which had revealed that the occurrence of stuttering on “content” as compared to “function” words was affected differently by contextual factors. That is, content words were found to be stuttered more frequently when spoken in a meaningful context than when spoken in a sequence having low propositional value. The inverse effect was observed for function words. These findings suggest that content and function words were affected in opposite fashion through some factor that comes into play when words are spoken as parts of a syntactic arrangement. It seems highly probable that the factor responsible for this effect is the linguistic feature of stress-in view of what is known about stress and its occurrence in connected English speech. Sources in the field of linguistics have identified a relationship between stress and words of the content-function dichotomy. For present purposes a statement by Smith (1959, p. 69) provides a succinct description of this relationship: it is essential to realize that in English only one primary stress can occur in a stretch of utterance bounded by a terminal juncture Also it is necessary to bear in mind that the various classes of words in the language can appear only with certain stresses and not with others. For instance, except in “compounds,” nouns, adjectives, adverbs, and main verbs never occur with less than secondary stress; auxiliary verbs, prepositions and conjunctions never appear with more than tertiary. Articles almost always occur with weak stress-never more than tertiary-and personal pronouns regularly with no more than tertiary.’ though they quite frequently are “nominalized” by virtue of being said with secondary. Of course, any word can be said with primary stress when a higher degree of pitch is shifted to it for reasons of emphasis, in which case the word which would normally carry the primary stress in the sequence is reduced to secondary. These are$ict.s of the structure of the language, internalized by all native speakers outside of awareness and binding on writer and reader alike. Some linguists use only a 3-level system of stress description, but the distinction is not crucial to the present point. The essential matter is that, whether favoring a 4-ievel or 3-level system, there is regular agreement that in connected spoken English the content words-those most frequently associated with stuttering-are the words most likely to receive heavy stress, and the function words-those least frequently associated with stuttering-are most likely to receive weak stress (be unstressed). Note, in Smith’s description, that there is some latitude in respect to the words on which some particular level of stress might possibly occur. Thus there is some

LOCI OF STUTTERING:

GRAMMAR

OR PROSODY?

289

variability of stress placement among words within each category, and to a much lesser extent, across the two categories. However, on the whole the distribution of stress placement is pretty well determined relative to word type. This pattern is highly consistent with what is found in respect to stuttering on content and function words, including the fact that different studies report a different “gradient” of stuttering among words of the content category. It should be noted that the relative latitude in stress placement may be occasioned not only by the particular context of an utterance but also by individual differences in speaking-or reading aloud. This might account substantially for the frequent finding that not all stutterers have difficulty on the same words, even when they all read the same material. At the same time, a common pattern is evident among subjects as a group. A few years ago (Wingate, 1972) I made a graphic level-recorder tracing of an “exemplary” recording of a spoken text which showed that intensity peaks (accepted as indicators of stress) corresponded with group data on stuttering occurrence in the same text. Almost all of the evidence relating stress and stuttering loci has been obtained from readings of prepared material. However, the present author’s casual observation of many stutterers has suggested that the same relationship obtains in the spontaneous speech of stutterers. Hejna (1972) has reported work which corroborates these observations. To date, investigation relating stuttering to linguistic variables has been reported for English only. It would be of considerable value to stuttering theory to have comparable data in other languages.

References Bleumel, C. S. (1913). Stammering and Cognate Defects of Speech. New York: G. E. Stechert. Bolinger. D. (1968). Aspects of Language. New York: Harcourt Brace. Brown, S. F. (1938a). A further study of stuttering in relation to various speech sounds. Q, J. Speech 24: 390-397. Brown, S. F. (1938b). Stuttering with relation to word accent and word position. J. Abn. Sot. Psychol. 33: 112-120. Brown, S. F. (1937). The influence of grammatical function on the incidence of stuttering. J. Speech Dis. 33: 112-120. Brown, S. F. (1945). The loci of stutterings in the speech sequence. J. Speech Dis. 10: 181-192. Brown, S. F. (1938~). The theoretical importance of certain factors influencing the incidence of stuttering. J. Speech Dis. 3: 22%230. Brown, S. F., and Moren, Adelaide. (1942). The frequency of stuttering in relation to word length during oral reading. J. Speech Dis. 7: 1X3-159. Conway, J., and Quarrington, B. (1963). Positional effects in the stuttering of contextually organized verbal material. J. Abn. Sm. Psychol. 61: 29%303. Delattre, P. (1963). Comparing the prosodic features of English, German, Spanish, and French. IRAL 1: 193-210. Eisenson, J., and Horowitz, Esther. (1945). The influence of propositionahty on stuttering. J. Speech Dis. 10: 19>197.

290

M.E.

Emerick.

L. L. ( 1963). A clinical

WINGATE

on the “tinal”

observation

J. Sprrch

stuttering.

Huu-ing

Dis. 28:

lYblY5. Hahn,

(1942a).A

E. F. factors

Hannah,

Elaine,

P.,

and

J. Spmh

nonfluency. Hejna.

study

in oral reading.

R. F. (1972).

of the relationship

J. Spwch

Dk.

Gardner,

Joanne

Hecrring Rex

The relationship

between

G.

stuttering

occurrence

(196X).

A note

on

syntactic

relationships

in

I I: 83.%860.

between

accent

or stress

and stuttering

during

spontaneous

Asha 14: 479 (abstr.).

speech.

Johnson. W. (1955). A study of the onset and development of stuttering. Sturrrving in Children und Arlult~. Minneapolis: University of Minnesota Johnson,

and grammatical

I.

7: 143-15

W., and Brown,

S. F. (lY3Y).

Stuttering

in relation

to various

In Johnson, Press.

speech

sounds:

W. (ed.)

a correction.

u. J. Sprech 25: 2@22. Kroll.

R. M.. and Hood, adaptation

S. B. (1976).

effect in stutterers

Lanyon.

R. I. (1969).

Lanyon,

R. J..

Speech:

and

relation

Duprez,

Peterson,

H.,

Reick.

of task presentation

meaning

J. Cvmmun.

speakers.

of nonfluency

D. A. (1970).

PsychoI. 76: 93-97. Peterson. H. A. ( 1969). Affective J. Speech Hruring

The influence

and normal

to information

Nonfluency,

of words

Mary

Beth,

and

Hoff,

B. Conway,

J.. and Siegel,

strength.

as rated by stuttering

and superior

Rita

K. (1969).

N. (1962).

G. A. (1967).

Soderberg,

G. A. (I97 I). Relationship

J. Ahn. readers.

value

as a function

and sentence

of

position

of

study

R. Stuttering,

of some

properties

of

information

load,

and

Linguistic

English prosody. Stuck> in Linguisfic’s of adaptation trends in the oral reading

factors

inferior

Dis. 2: Y% 108. J. Sprech Heurirrg Rrs.

in stuttering.

of word

14(3]4). of stutterers,

information

and word length

IO:801 -810.

to stuttering

disfluen-

J. Commurr. L)is. 4: Yl4.

I. K. ( 1966).

Taylor,

164: 451-452. length.

and nonstuttering

A test of satiation

An experimental

Rcs. 5: 387-394. B., and Melkman,

J Commun.

speakers.

Soderberg. ties.

word

J. Speech Heurirzg Dis. 30: 32 -36.

Smith. H. L. (1959). Toward redefining Soderberg, G. A. (IY6Y). A comparison speakers

and

70: 22 I-224.

stuttered words. J. Sprech Heuring Schlesinger, I. M.. Forte. M., Fried, response

load on the

Res. 12: 337 -343.

J. Ahn. Sot. P.ywhol.

Quarrington.

SGnce

value.

information

adaptation in stuttering. J. Sprrch Haring Rc?;. 12: I IO- 117. Quarrington. B. (1965). Stuttering as a function of the information words.

and information Dis. Y: Y> I IO.

The properties

of stuttered

words.

J. Vcrh. Lrcmw~g

Verb. BehaL,. 5: I 12-

I 18.

M. E. (1966). M. E. (1976).

Wingate. Wingate,

M. E. (1976). Stuttering: Theory und Trrutment. New York: Irvington-Wiley, (Chap. 9). M. E. (I 972). Symposium on linguistic-motor determinants of stuttering. American Speech

and Hearing

Prosody Slurvian

in stuttering adaptation. J. Speech Hawing Res. 9: 550-556. skill of stutterers. J. Speech HewinK Rex. 10: 844-848.

Wingate, Wingate,

Association

Convention.

San Francisco.

No. 20.