You are on page 1of 5

JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR 6, 5 6 0 - - 5 6 4

(1967)

Hesitations in Children's Speech During Explanation and Description ~


HARRY LEVIN, IRENE SILVERMAN, AND BOYCE L. FORD Cornell University, Ithaca, New York Twenty-four children, six each from the kindergarten, second, fourth, and sixth grades, were shown three simple physical demonstrations. They described and explained what they saw. For children of all ages, explanation compared with description was characterized by more words, pauses, hesitations, longer pauses, and a slower rate of speaking. This study concerns the relationships between thought and speech. W e use the word, "speech," deliberately, as distinguished from "language." The variables with which we are concerned are a subcategory of paralinguistics (Traeger, 1958), the hesitations and ttuencies in speech. As such, they describe the form of speech and are applicable to language regardless of the content or the grammatical forms of the language. Broadly, we are predicting that the process of explanation is mirrored in hesitant speech whereas description is more fluent. Goldman-Eisler (1962), has demonstrated clear relationships between automatic speech and speech which requires new cognitive constructions. For example, she presented adult Ss with N e w Yorker cartoons which they first h a d to describe and then interpret. Hesitations increased under the second condition. Similarly, we have found that the patterns of hesitations could be interpreted in terms of those which were responsive to stress and those which reflect instructions for high intellectual quality (Levin and Silverman, 1965). T h e present study is analogous to Gold1 This study was supported by Grant MH-07226 from the National Institute of Mental Health. 560 man-Eisler's, b u t explores the p h e n o m e n a with children varying in age from 5 to 12 years. The children are shown three physical demonstrations and are asked, for each, to describe and then to explain w h a t they have seen. METHOD Subjects. The Ss for this study were 24 children attehding the Cayuga Heights Elementary School, Ithaca, N.Y. This school serves an upper middleclass neighborhood where, in general, the children may be characterized as having advanced verbal skills. Three boys and three girls were randomly selected from each of four grades: kindergarten, second, fourth and sixth. The children were assigned without bias to one of three Es. ~ The Tasks. Three physical demonstrations were carried out by E for each child, who was systematically queried about each. The interaction between S and E was tape-recorded. Each child was seen in a room apart from his classroom, during a single session. The following three demonstrations were presented to each child and always in this order: 1. Two balloons of different colors and sizes were inflated with gas, the larger one with helium, the small one with air. The E holds both in one hand, then releases them so that the helium-filled one rises and the other falls. 2. Two discs of identical size, shape, and color We wish to thank the teachers and principal of Cayuga Heights School for their cooperation and Anita Ianeo and Gary Shaw for their parts in gathering the data.

HESITATIONS IN CHILDREN'S SPEECH are made of different materials, one of lead, the other of wood. They are placed in front of S who is told to lift the heavier and then the lighter one. 3. Two colorless liquids are presented to S in separate beakers. They are mixed together. The result is a magenta-colored liquid. It will be noted that the three demonstrations were chosen for their counter-intuitive natures: the larger balloon rises, two objects look alike but "feel" different, colorless liquids that result in a vivid color. We expected the surprise value to fascinate the children and keep them involved with the task. The Queries. The E fastened a lavalier microphone around the child's neck, meanwhile explain-' ing that he was going to show him some things, that they would talk about the things he saw and that the microphone and tape-recorder were used because we wanted to have a record of what we talked about. To put the child at ease, E first talked with him about what he was doing in his classroom. The E placed the materials f6r each demonstration in front of the child and asked the following questions in this order: 1. Tell me what you see here. 2. Now I am going to do something. Watch and see what happens. (With D1 and D3) What happened? (With D2) Pick up this one, then this one. How did they feel? 3. Why do you think that happened? (D2: VV'hy did they feel that way?). 4. I'd like you to tell me, as though you're telling a story, from the time I brought these over, what they looked like, what I did, what happened, and why you thought it happened. It was often necessary for E to probe further when S did not respond, especially after the request for explanation. The E used sueh questions as, What colors are they? Do they look the same or different? Why did one go up and one fall down? Scoring the Hesitation Phenomena. Each session was transcribed verbatim and all hesitations, including repetitions, sighs, giggles, deep breaths, and tongue clicks were included. Pronunciation as we heard it rather than conventional spelling ("wanna," not "want to") was typed. The following variables were entered on the typescript by an E who listened to the tape. These variables have previously been described in detail (Levin and Silverman, 1965): 1. Vocal segregates: "uh, urn, er." g. Grammatical correction: "The balloons gowent up." 3. Word order correction: "The red is water-The water is red." 4. Lexical correction: "The circles-the discs are the same." 5. Accentual correction: "When did you do that? When did ydu do that?" 6. Phonological correction: "That's

561

the idear-the idea." 7. Sentence ineompletion: "The two balloons-The red one went up." 8. Repetitions, including phrase, words, syllable, etc.: "Then-then-you put them together." 9. Slips of the tongue. 10. Omission of words or parts of words. "Then you started to put-away." 11. Parenthetie remarks: for example, I mean, etc. 12. DraMs: nonphonemie lengthening of a phoneme. la. Zero segregates (unfilled pauses) were reeorded visually by playing the tapes through a rectifier into a Brush Reeorder. Pauses of 80 msee or longer were noted on the transeriptions. Reliability. Two seorers, one (IS) highly experienced in seoring hesitation variables, eoded several tapes independently. A total of 147 items was scored on which the two agreed on 142 items or 97%. Thereafter, during independent scoring of the tapes, problematie instances were discussed by the two coders, Fluency Variables. The following were eounted for each part of eaeh demonstration: number of words, syllables and time, in seeonds. Description and Explanation. The questions were designed to elicit either description or explanation. The typescript for each demonstration was divided into sections in this manner: I. Presentation of materials for demonstration and request to S for description of materials. II. The demonstration itself and description of what happened. III. S's response to why? IIIs. Spontaneous explanation by S. IV. Summary of total demonstration. RESULTS AND DISCUSSION T h e basic variables d e s c r i b e d a b o v e were c o m b i n e d into the following indices w h i c h form the variables used h e n c e f o r t h in the analysis: 1. T o t a l n u m b e r of words ( W ) . 2. T o t a l n u m b e r of hesitations ( H ) . This m e a s u r e is the sum of all hesitations scored, excluding the n u m b e r of pauses. No single h e s i t a t i o n variable, other t h a n the f r e q u e n c y of pausing, o c c u r r e d o f t e n e n o u g h in o u r data to b e a n a l y z e d separately. W e , therefore, s u m m e d all hesitations into a single index. 3. T o t a l n u m b e r of pauses ( P ) . This m e a s u r e i n c l u d e s all pauses l o n g e r t h a n 80 msee. 4. Rate. Ratio b e t w e e n n u m b e r of words a n d seconds speaking. 5. T o t a l a m o u n t of time p a u s i n g ( T P ) . 6. Ratio of n u m b e r of hesitations to

562

LEVIN, SILVEBMAN,

AND FOBD

number of words ( H / W ) . As various investigators have pointed out (GoldmanEisler, 1962; Levin and Silverman, 1965), the frequency of hesitations in speech is related to the S's productivity. This measure controls for productivity. For consistency with other studies, we have chosen number of words as the denominator, although we could have used syllables or morphemes. 7. Ratio of number of pauses to number of words ( P / W ) . 8. Average
TABLE 1

M E A N SCORESFOR EACH PART Part Measure Words Hesitations Pauses Rate Total Pausing Hes./Words Pause/Words Ave. Pause I 26.03 ~. 9 2 4.51 ~. 64 3.08 .10 .15 .54 II 15.21 1.46 ~. ~4 3 . ~4 1.26 .08 .13 .40 III 55.96 8.74 1~. 81 ~. 3 4 11.24 .15 . ~1 .91 IV 84.65 9.31 ~0.29 ~. 41 11.51 .1~ . ~4 .60

length of pauses (AP), i.e. total time pausing and pauses. The means of each ables for the four parts are ble 1.

ratio between frequency of of these varishown in Ta-

Intercorrelations Among Variables. Table 2 presents the intercorrelations among the eight variables. Each score was based
TABLE

on the total session, on all three demonstrations. Obviously, the variables are, on the whole, strongly related to each other. Children who talked a lot also hesitated and paused frequently. They spent more time, overall, pausing, but productivity has no relationship to the average length of pauses. This general effect of productivity replicates a finding of GoldmanEisler's (1962) and confirms our earlier speculation that hesitations and pauses can be described as occupying a single dimension (Levin and Silverman, 1965). Further, it can be seen that our attempt to describe hesitations and pauses independent of words is effective (W X H / W : r--- .11). The relationships with rate (words/sec) present a clear picture. The overall number of words spoken does not influence rate. On the other hand, all variables having to do with hesitations and pauses are inversely related to rate. That is, the time spent in hesitations and pauses adds to the time spent in speaking but not to the number of words spoken, so that we may think of hesitations displacing words and diminishing the rate. Even when corrected for productivity, it should be noted that frequency of hesitations and frequency of pauses are almost interchangeable variables (r --- .79). Finally, the average length of pause, with the exception of the two relationships
2

INTERCORRELATIONS: TOTAL SCORES Measure Words Hesitations Pauses Rate Total Pausing Hes./Words Pause/Words Ave. Pause 1 -2 .75 -3 .8~ .81 -4 - . 07 -. 37 -. 43 -5 .53 .55 .66 -.68 -6 .17 .67 .54 -.66 .48 -7 .11 .41 .61 --.77 .55 .79 -8 - . 08 - 07 - . 0~ -.57 ,69 . 17 . ~5

HESITATIONS IN CHILDREN'S SPEECH

56~

noted above, tends to be independent of the other variables. Tests of Hypotheses. Our hypotheses state that speech under instruction to explain will be more hesitant than when Ss are asked to describe. The prime comparisons are between the responses to Question II, What happened?, and Question III, Why? The first question was principally a warm-up to which the answers were generally laconic. Question IV, the summary, really is a memory exercise, containing both explanation and description. To test these hypotheses t tests between Parts II and III were calculated for each of the eight indices. The differences between these two parts is statistically significant. When we compare description with explanation, Ss in the latter instances speak more, with more hesitations and pauses, and the average length of the pauses is longer. Also, the children spoke more slowly when explaining the physical phenomena than when describing them. These findings match closely those reported by Goldman-Eisler (1962) for adults. Yet, there are several aspects of our procedure which might have influenced these results and should be considered. The request for explanation followed, in almost every ease, the child's description, so that the greater hesitations may be due to the sequence rather than to the task of explanation. Two observations suggest that the sequence effect does not adequately interpret the findings. If there were a "rundown" with time, it would be accentuated across the three demonstrations. The differences between explanation and description, however, hold for all of the three demonstrations. A second line of evidence comes from twelve instances of spontaneous explanations. These explanations also followed Part II but were not elicited by instructions to explain. Spontaneous explanations

are brief but more fluent than responses to Why? Therefore, since they share a sequence with solicited explanations but are not responsive to the probe, they suggest that the hesitancy of explanation compared to description is due specifically to E's request that the child explain the physical phenomena. The Es, at their discretion, asked questions of the children when they were especially hesitant or were not addressing themselves. In Part lII, 226 probes were recorded compared with 92 in Part II. The frequency of probes was correlated with the various indices. There are n o relationships in Part II. In Part III, the number of probes are not related to either the child's number of words or number of hesitations. They are, however, related to the other six variables. These findings suggest that E asked leading questions when the child spoke slowly or paused a great deal. Nevertheless, these probes did not have the effect of increasing the child's fluency compared to his performance during description. There are no variations in our data attributable to the different ages of the Ss. It seemed to us, from a perusal of the means of the eight variables, that the kindergarten children were often lower than the other three groups. However, the large variability within the five-year-old group precluded significant differences which we might attribute to age. With only six children in each age group, these results can only be suggestive. Several findings merit comparison with our previous study (Levin and Silverman, 1965). In that study, the Ss were all sixthgrade children whose task was to tell stories in response to assigned story themes. The rate of talking under those more spontaneous circumstances were 2.10 words/sec, compared to the overall rate of 2.66 words/sec, or 2.88 for the most similar

,~4

LEVIN, SILVERR'IAN~ AND FORD

present group, the sixth-graders. We may conclude, therefore, that spontaneous speech is no more rapid than speech under the test-like conditions of the present study. We inferred earlier that length of the unfilled pauses was responsive to whether the S was under stress or was performing cognitive operations prior to talking. The finding that pause length increases under instructions to explain bolsters our former assumption. To understand the differences in fluency when a child is asked to explain or to describe an event, it is useful to contrast the task requirements of the two processes. Take, as an example, the demonstration of the two balloons. At the end of the demonstration, the child is asked to tell what he saw. One balloon is nestling against the ceiling, the other is on the floor. The props for his storv are directly available to him. He must rely on short-term memory, hut this requires storage and retrieval over, at the most, a half-minute interval. His job, then, is to encode a brief series of salient events with the objects in full view. It seems to us that his choices consist of appropriate lexieal items and the past tense for his utterance. We have no reason to believe that the words to encode the event are not immediately available to all these children, especially since they were asked in the first query to describe (attach words to) the simple apparatus. Almost every child, it seemed to us, understood that a different kind of discourse was required by the "Why?" question. Many of the answers were, of course, incorrect. Some younger children simply

stated a phenotypic attribute: one balloon is red and the other is blue. Even this simple explanation requires a choice from a large array of .characteristics that he might have used. Also, we are not convinced that the child who answers in this way simply does not have an idea and uses "description-explanations" to satisfy the E's demands. Therefore, if the child does not know the answer yet is forced to respond, we expect his speech to be hesitant. We may contrast this lack of fluency with the spontaneous explanations offered to at least one demonstration by 12 Ss. Spontaneous explanations were briefer, but on almost all variables were more fluent: fewer hesitations and pauses, briefer pauses and faster rate. These results confirm our expectations that explanations which the child has available are less hesitant than those which must be constructed. In other words, the hesitations in speech inversely mirror the automatieity of the cognitive process. To encode an event whose features are available to the child is automatic. To search one's memory, to accept or reject an idea that comes to mind, to put ideas together-in short, to t h i n k is not automatic and results, as we have seen, in slow, pause-filled, hesitant speech.
~EFEll.ENCES GOLDMAN-EISLER,

F. Speech and thought. Dis-

covert.i, April, 1962.

LEVlN, H., AND SmVER_XIAN, I. Hesitation phenomena in children's speech. Language and Speech, 1965, 8, 67-85. TRAECE~, (;. L. Paralanguage: a first approximation. Studies in Linguistics, 19.58, 13, 1. (Received November 3, 1965)

You might also like