JOURNAL

OF COMMUNICATION

DISORDERS

9 (1976),

121-128

PERCEPTIONS OF AUDITORY COMPONENTS OF STUTTERED SPEECH MOYA

L. ANDREWS

AND RAYMOND

G. SMITH

Speech and Hearing Center, Indiana UniversiQ, Bloomington, Indiana 47401

Two student groups, one of speech pathology majors and one a class of mixed background fulfilling a humanities requirement, listened to the same set of sentences spoken first in nonstuttered and then in stuttered form. Subjects rated each replication along a set of 114 semantic differential-type adjectives but on IO-point unipolar scales. These scales had previously been factor analyzed and represented 31 positive and 29 negative perceptual factors. The results indicate that all college students react favorably to stuttered speech. However, six factor differences significant at the 0.05 level or above emerged between the two groups when evaluating the stuttered version. These results suggest that pathology students do not perceive stuttered speech as do nonpathology students; they are more favorably disposed on every dimemsion.

Introduction Responses to stuttered speech exist only in the mind of the observer. The present study explores the responses of the listener. It addresses the question of the nature of the changes in message perception brought about by the change from nonstuttered to stuttered speech to determine which responses are modified when all other aspects of the message are held constant. It further asks whether such differences are a function of sensitization by speech pathology majors to stuttered speech or are in fact common responses of all college students. Despite the obvious importance of the area, prior investigation into listener perceptions of stuttered speech is limited. In their research review of stuttering, Beech and Fransella (1968) report no studies dealing with listener perceptions. Reported studies are mostly tangential to the focus of the present study. For example, Giolas and Williams (1957) found that children of kindergarten and second grade level could discriminate between fluent and disfluent speech but preferred storytellers who were fluent. Miller and Hewgill (1964) found differences in amounts of nonfluency to be related to the degree of confidence with which the audience viewed the speaker. As amount of disfluency increased, ratings of the speaker’s persuasiveness tended to be less favorable. Glennie (1966) concluded that stuttering was not a factor in oral argumentation. She found no difference in audience acceptance of oral arguments when those arguments were spoken by the same stutterer speaking disfluently or fluently. Wingate and Hamre (1967), building upon the earlier work of Bloodstein (1953, compared the perceptions of stutterers and nonstutterers with film segments of persons purportedly listening to stuttering speakers and found no differences. Their finding confirmed the results of the earlier Bloodstein study. 0 American

Elsevier Publishing

Company,

Inc.,

1976

121

122

MOYA

L. ANDREWS

and RAYMOND

G. SMITH

Fransella (1%5) compared ratings of stutterers and nonstutterers for the concept Stutterers on a semantic differential developed by Smith (1962). A factor analysis of these ratings showed the groups to be similar in response patterns, both placing the concept at the “painful,” “severe,” “difficult,” and “tense,” poles of the Pleasantness dimension. The response of the listener is undoubtedly influenced both by expectation and by training. Sander (1965) found that a mother’s estimate of the number of disfluencies in a sample of her child’s speech was affected by the type of listening instructions given. A listener’s tendencies to respond in a habitual manner to persons who stutter may be affected by such factors as previous exposure, professional training, or course work in the area of stuttering. It is widely accepted that group norms and mores affect perceptions of the group members. Materials

and Subjects

In the course of a prior investigation Andrews (197 1) made a series of recordings for each of approximately 70 stutterers repeatedly reading the same excerpt from “ Androcles and the Lion.” These stutterers had all been classified by a panel of three experienced speech clinicians as either 2s or 3s on a 3-point scale labeled mild, moderate, and severe. To provide stimulus materials for the present study these recordings were screened for stuttered and nonstuttered renditions of identical sentences by the same speakers. Selection criteria included clarity of articulation, freedom from background noise, general audibility, and pronounced auditory symptoms of stuttering, primarily repetition and prolongation. From three to five complete sentences in each form were obtained from each of five males and one female. It was felt that sentences supplied by genuine stutterers should yield far more valid results than sentences supplied by normal speakers who “fake” stuttering. These paired sentences were separately grouped to constitute the nonstuttered and stuttered stimulus forms, respectively. The nonstuttered form required a delivery time of about 1 min; the stuttered form was approximately twice as long. Speaker order within forms was randomized, but was the same for both forms. In order to obtain an unbiased reference base for the critical comparison, the nonstuttered form was administered first to each group of subjects. The perceptual scale used as the measuring instrument was developed in the following manner. Approximately 500 terms thought to describe listeners’ perceptions of messages were rated by about 800 subjects along a lo-point scale in response to the question, “How important is this quality (e.g., clearness) to any These ratings were factor analyzed,l and of the scales, 114 were message?” ‘Factor analytic programs used included Univesity University of Miami Program MFACTOR.

of California Biomedical

Program BMD03M

and

PERCEPTIONS

OF AUDITORY

COMPONENTS

OF STUTTERED

SPEECH

123

sufficiently loaded on a principal factor (< 0.60) and minimally contaminated with any other factor (< 0.30) to be acceptable for a measuring instrument of this type. The factor analyses showed the scales to represent 31 positive and 29 negative factors with from one to four usable scales making up each factor. The test-retest reliability of the 60 factors was 0.88. Scale ratings within factors were averaged to constitute factor scores. For example, the factor labeled clarity is indexed as the mean scale score for the scales clear, coherent, and consistent. In order to avoid the introduction of the questionable semantic assumption of bipolarity, positive and negative scales were rated and factor analyzed separately. In the interest of increasing sensitivity, 10 rating steps (&9) were provided for each adjective. At this point an operational assumption was introduced. It was assumed that a judgment along any adjectival scale consists of two elements: First, a weight or importance judgment that answers the question, “How important do I feel that the (e.g., clearness) is to any message?” and second, “How quality of (clear) is this particular message. Y’~ For this score the subject was asked to rate the set of stimulus sentences as a whole. Each set of 114 scales was therefore administered three times to each subject: first to secure an estimate of the weight or importance of the trait in general, second to obtain an estimate of the amount of the trait ascribed to the set of sentences in nonstuttered form, and finally, to obtain ratings of the stimulus set in stuttered form. Scale scores were computed as geometric means of paired importance ratings and message scale ratings; factor ratings were simple averages of their component scale scores. Two groups of subjects were used, the first consisting of 21 juniors and seniors majoring in speech pathology from classes in Practicum and Voice Disorders. These were termed the speech pathology (P) group. The second, a group of 17 education and humanities students, juniors and seniors taking a general speech communication course, were designated the humanities (H) group.

Statistical

Analyses

Due to the fact that each factor is theoretically orthogonal and independent, the multiple t-test is the appropriate statistical tool. Four sets, for a total of 240 tests, were run between the 60 factors, two sets of correlated t-tests within groups

%is

is actually a restatement of Anderson’s information integration model where the response R = is the weight or importance of the information item and s is its scale value. For a complete statement of the theory see Norman H. Anderson, Injknarion integration theory; n brief survey. Technical Report CHIP 24. San Diego: Center for Human Information Processing, University of California, April, 1972. WiSi; w

MOYA

124

L. ANDREWS

and RAYMOND

G. SMITH

between the message and its stuttered replicationa and two sets of uncorrelated ts between the P- and H-group scores for each message form.4 Alpha was preset at 0.05. Out of 60 independent tests, exactly 3 might be expected to reach significance by chance alone. A finding of more than three would be indicative of real differences between scores. Results Table 1 shows factors and levels of significance for all differences meeting criterion for the humanities students. When nonstuttered and stuttered versions are compared for this group, 10 significant differences appear. These subjects rate the stuttered version as less dislikable, less unpersuasive,5 less indirect, less unclear, less facetious, less ignorant, less dull, less imaginative, less worthless, and less run-of-the-mill. Note that all significant differences appear on negative factors. Despite the fact that orthographic opposites were included for some of the positive scales, no significant differences emerged.

Humanistic Factor

TABLE 1 Group Factor Comparisons for Nonstuttered Xl

K2

vs. Stuttered Sentence Sets

t

P

2.51 3.06 2.17 3.00 2.75 2.35 3.16 3.43 2.99 3.39

0.05 0.01 0.05 0.01 0.05 0.05 0.01 0.01 0.01 0.01

Positive (none) Negative Dislikableness Unpersuasiveness Indirectness Unclarity Facetiousness Ignorance Dullness Unimaginativeness Worthlessness Run-of-the-milltress

%re

4.35 6.66 5.08 5.37 4.05 5.26 7.03 6.21 5.99 6.66

3.63 5.68 4.14 3.75 2.97 4.10 5.85 4.56 4.52 4.40

program

used was the Wrubel Computing

Center, Indiana University,

MULTIT.

“The program

used was the Wrubel Computing

Center, Indiana University,

PRDOBS.

%eparation of positive and negative scale terms, both for instrument development and application, rather than the more usual bipolar adjectival format was adopted in order to avoid the assumption of semantic to orthographic isomorphism. For example, for some subjects the term incomperenr anchored the negative pole of the wise dimension. We therefore used only the unipolar form. In describing results this forces the use of the double negative, however, with a certain amount of awkwardness. For example, with a negative unipolar factor labeled incompetence, one is forced to state that one message is more or less imcompetent than another; he could say that one was more or less competent only if he were able to demonstrate that the two terms were in fact end anchors for a competent-incompetent dimension. Our factor analyses demonstrated clearly that this was not the case.

PERCEPTIONS

Pathology

OF AUDITORY

COMPONENTS

Group Factor Comparisons

Factor

_

OF STUTTERED

SPEECH

125

TABLE 2 for Nonstuttered vs. Stuttered Sentence Sets x2

r

P

2.45 3.31 2.85 3.31 2.15 2.53

0.05 0.05 0.01 0.01 0.05 0.05

3.41 3.75 2.66 2.72 2.09 2.49 2.23 4.65 2.12 4.34 2.89 3.23 2.43 3.35 4.05 3.84 4.77 3.38 4.48

0.01 0.01 0.01 0.01 0.05 0.05 0.05 0.01 0.05 0.01 0.01 0.01 0.05 0.01 0.01 0.01 0.01 0.01 0.01

Positive Likability colorfulness Clarity Honesty Activity Alertness

4.06 2.84 5.59 4.47 2.06 3.92

4.67 4.19 4.63 5.70 2.89 4.57 Negative

Dislikableness Unpersuasiveness Colorlessness Noisiness Bias Indirectness Unclarity Unknowledgableness Facetiousness Ignorance Untraining Boastfulness Unscholarliness Passivity Dullness Conservativeness Unimaginativeness Worthlessness Run-of-the-milhress

3.96 5.70 5.54 3.07 3.37 3.41 4.34 4.47 2.98 4.83 4.53 1.95 4.35 4.25 6.63 4.16 5.82 4.59 6.60

3.23 4.59 3.95 2.40 2.42 2.60 3.42 3.10 2.53 3.50 5.29 1.13 3.25 3.45 5.21 3.05 4.24 2.96 3.94

When two versions are compared for the pathology students, however, a much more elaborate pattern of differences emerges. Table 2 again shows all nine of the above noted significant differences at probabilities of 0.01 except for the factors labeled indirectness, unclarity, and facetiousness, which appear between 0.01 and 0.05. In addition, however, this group found the stuttered version to be significantly less dislikable, less colorless, less noisy, less biased, less unknowledgable, less boastful, less unscholarly, less passive, less conservative, but more untrained. For positive factors Table 2 shows six differences. The P-group rated the stuttered version as more likable, more colorful, more honest, more active, more alert, but less clear. Table 3 shows far fewer differences when the P- and H-groups are compared for the nonstuttered version. The t-tests between the two groups on the 60 factors for these sentence sets show only three significant, any or all of which could be significant by chance alone. The H-group apparently found the nonstuttered version to be more believable, more disordered, and more indirect than did the P-group.

126

Factor Comparisons Factor

MOYA

L. ANDREWS

between

Humanities Xl

and RAYMOND

TABLE 3 and Pathology x2

G. SMITH

Groups for Nonstuttered t

P

2.21

0.04

2.84 2.31

0.01 0.03

Sentence

Sets

Positive Believability

2.92

4.48 Negative

Disorderliness Indirectness

4.37 3.41

6.12 5.01

When the two groups are compared for the stuttered version, however, six significant differences emerge as shown in Table 4. The P-group viewed it as much more likable, more colorful, more thoughtful, more scholarly, more alert, but less worthless.

Discussion Although the biases noted in the above results may only reflect current awareness of problems of all minority groups, the speech pathology students clearly demonstrate more bias. For the humanities students differences emerged in favor of the stuttered version on 9 of 60 factors tested. The speech pathology students who, either because of training or predisposition might logically be expected to be more sensitive to speech defects, manifested a much stronger bias, with significant differences on 25 out of the 60 factors tested. If this finding truly represents a lack of objectivity on the part of pathology students generally, it has clear implications for both professional training and clinical practice. It raises the important question whether these results arise from a lowered threshold of sensitivity to speech problems or from an exaggerated tendency to perceive such problems as less severe than do students in other disciplines.

Factor Comparisons _____ Factor

between

Humanities Xl

TABLE 4 and Pathology x2

Groups for Stuttered Sentence t

P

3.73 2.07 2.56 2.09 2.12

0.001 0.05 0.02 0.05 0.05

2.06

0.05

Positive Likability Colorfulness Thoughtfulness Scholarliness Alertness

4.63 4.19 4.42 3.96 4.57

2.86 2.93 3.11 2.81 3.36 Negative

Worthlessness

2.96

4.52

Sets

PERCEPTIONS

OF AUDITORY

COMPONENTS

OF STUTTERED

SPEECH

127

The data analyses presented here yield a partial answer to this question. The tests within each group (correlated ts) speak to the sensitivity hypothesis and clearly demonstrate that pathology students are more keenly aware of speech problems. The fact, however, that six significant differences emerged for the between-groups comparisons of the stuttered version suggests tenability for the bias hypothesis. Zealous students in speech pathology courses may identify so closely with instructor or textbook reports of the difficulties experienced by stutterers as to cause their perceptions to be unrealistic, even given a general cultural bias. Either an inability to accept normative reactions to stuttered speech or an unwillingness to do so could well jeopardize honest therapeutic relationships. The present study shows that biases do exist over a wide spectrum of perceptual factors. Personnel involved in the supervision of professional training as well as students undergoing such training need to be alerted to such bias and its implications for affecting professional behaviors. It is difficult to see how a stutterer could be taught to view his handicap objectively if his therapist, himself, is nonobjective. As a concluding caveat, this study shares with numerous similar studies some obvious weaknesses, including the hazard of extrapolating from small samples to larger populations and the absence of complete randomization in sample selection. It should be noted, however, that the within-subjects comparison reduces such error substantially. Also, a sufficient number of tests were run to establish the probability of bias. Even if one is reluctant to accept the conclusion that the t statistic is robust when applied to data of this type, the internal validity of the present study is sufficient to demonstrate the existence of bias. Furthermore, a substantial number of factor differences were noted lying between the 0.05 and 0.10 levels of significance. An increase of sample size would answer the question of significance for these marginal factors. Even if parametric tests are ruled out, simple observation of factor means makes the bias obvious. Of the 57 factors out of 60 that changed as a result of the stuttered version for the H-group, 32, a chance number, changed in the more favorable direction; of the 59 of 60 factors that changed for the P-group, however, 43 changed in the more favorable direction. The probability of a difference this large between proportions occurring by chance alone is less than 0.001. The conclusion seems inescapable that these pathology students were more biased in favor of stuttered speech than were the humanities group. This finding appears to be worthy of additional attention by concerned speech pathologists. References Andrews, M.L., The effect of certain types of auditory stimulation on overt stuttering behavior. Unpublished doctoral dissertation. Columbia Univ. 1971. Beech, H. R., and Fransella, F. Research and experiment in sfurrering. Oxford: Pergamon Press 1968.

128

MOYA

Bloodstein,

0.) and Bloodstein,

L. ANDREWS

and RAYMOND

A. Interpretations

G. SMITH

of facial reactions to stuttering. J. Speech Heuring

Dis., 1955, 20, 148-155.

Giolas, T. G., and Williams, D. E. Children’s reactions to nonfluencies in adult speech. J. Speech Hearing Res., 1958, 1, 86-93. Fransella, F. An experimental evaluation of the speech correction semantic differential. Speech Monographs, 1965, 32,448-45 1. Glennie, J. S. Stutteringas afactor in oral persuasion. Unpublished master’s thesis. Indiana Univ. 1966. Miller, G. R., and Hewgill, M, A. The effect of variations in nonfluency on audience ratings of source credibility. Quarferfy J. Speech, 1964, 50, 3644. Sander, E. K. Comments on investigating listener reaction to speech disfluency. J. Speech Hearing Es., 1965, 30, 159-165. Smith, R. G. A semantic differential for speech correction concepts. Speech Monographs, 1962,29, 32-31.

Wingate,

M. E., and Hamre, C. E. Stutterer’s projection of 1istenerreaction.J. 10, 339-343.

1967,

Speech Hearing

Res.,

Perceptions of auditory components of stuttered speech.

Two student groups, one of speech pathology majors and one a class of mixed background fulfilling a humanities requirement, listened to the same set o...
489KB Sizes 0 Downloads 0 Views