differential responding to content and intonation components of a complex auditory stimulus by...

Analysis and Intervention in Developmental Disabilities, Vol. 6, pp. 109-125, 1986 0270-4684/86 $3.00+ .00 Printed in the USA. All rights reserved. Copyright © 1986 Pergamon Journals Ltd.

Differential Responding to Content and Intonation Components of

a Complex Auditory Stimulus by Nonverbal and Echolalic

Autistic Children

Laura Schreibman

University of California, San Diego

Barbara S. Kohlenberg

University of North Carolina, Greensboro

Karen R. Britten

University of California, Los Angeles

Anecdotal and scientific reports of the speech of autistic children suggest a possible differential responsiveness to limited components of speech stimuli. Previously nonverbal children who acquire speech seem to acquire content, but their speech is dysprosic, characterized by inappropriate intonation, pitch, rhythm, and modulation. In contrast, echolalic autistic children often display modulation and varied intonation but may demonstrate no comprehension o f the verbal stimulus being echoed. The present experiment was conducted to systematically determine whether echolalic and nonverbal autistic children respond overselectively to the intonation or to the content of a complex auditory stimulus dtffering along these two dimen- sions. Ten autistic children (five echolalic and five nonverbal) and six normal children, matched for MA, were trained to bar press at the sound of an auditory complex stimulus designated as S + and t¢ withhoM response to an auditory corn-

Requests for reprints should be addressed to Laura Schreibman, PhD, Department of Psychology, C-009, University of California, San Diego, La Jolla, CA 92093.

This research was supported by USPHS Research Grants MH39434 and MH28210 from the National Institute of Mental Health.

The authors are grateful to Debra Mills, Alison Stanley, and Karen Laski for their many contributions to this research.

109

110 L. Schreibman, B. S. Kohlenberg, and K. R. Britten

plex stimulus designated as S - . The auditory complex stimulus consisted o f two components, content and intonation. Test trials consisting o f various combinations o f the intonation and content components were then presented. Results indicated that whereas the normal children responded to both components or to content, the nonverbal children showed evidence o f selectively responding to content, and the echolalic children showed evidence o f selectively responding to intonation. These results are discussed in terms o f relating autistic stimulus overselectivity to deficits in speech production and to implications for designing language training programs for these children.

Stimulus overselectivity, or selective responding to only a limited portion of simultaneous multiple cues, is frequently noted among autistic children (cf. Lovaas, Schreibman, & Koegel, 1979; Schreibman & Koegel, 1982). This failure to respond to multiple cues has been demonstrated to occur when the cues are presented across sensory modalities (i.e., visual, auditory, and tac- tile) (e.g., Lovaas & Schreibman, 1971; Lovaas, Schreibman, Koegel, & Rehm, 1971), as well as when the cues are presented within the same sensory modality. Thus, evidence for stimulus overselectivity has accrued in the visual modality (e.g., Koegel & Schreibman, 1977; Koegel & Wilhelm, 1973; Schreib- man, 1975) and has been noted in the auditory modality (e.g., Reynolds, Newsom, & Lovaas, 1974; Schreibman, 1975).

For example, Koegel and Wilhelm (1973) presented a discrimination task within the visual modality to normal and autistic children. The children were required to discriminate between two visual complex stimuli, each consisting of two cues. Both groups of children acquired the discrimination, though in subsequent probe trials it became clear that, unlike the normal children, the autistic children had not learned the discrimination on the basis of both the available cues but rather had responded to only one of the two cues presented.

Reynolds, Newsom, and Lovaas (1974) trained both normal and autistic children to discriminate between two stimulus complexes in the auditory modality. One complex consisted of the simultaneous presentation of a con- tinuous high tone with periodic relay clicks; the other complex consisted of a low tone with periodic bursts of the sound of a motor. When the single components of these complexes were presented individually, it was found that whereas the normal children generally responded to both cues, the autistic children typically responded to only one.

The failure of autistic children to respond to multiple cues has been broadly implicated as contributing to the social, affective, and language development difficulties so often evident in these children (cf. Lovaas et al., 1971, 1979; Schreibman & Koegel, 1982). Although the area of stimulus overselectivity in the visual modality has been relatively well researched, and although this research has subsequently resulted in improved teaching methods (Koegel & Schreibman, 1977; Koegel, Schreibman, Britten, & Laitinen, 1979; Schover

Differential Responding to Content and Intonation 111

& Newsom, 1976; Schreibman, 1975; Schreibman & Charlop, 1981; Schreib- man, Charlop, & Koegel, 1982; Schreibman, Koegel, & Craig, 1977), stimulus overselectivity in the auditory modality has not been addressed as extensive- ly, the aforementioned studies by Reynolds et al. (1974) and Schreibman (1975) being exceptions. Both studies demonstrated that autistic children selectively responded to only one component of two-component auditory complex stimuli.

The potential implications of auditory overselectivity have been noted by researchers in the past. It has been suggested that impaired perception of sounds may partially account for the deficits shown by autistic children in auditory comprehension (Rutter, 1966, 1968). Furthermore, one would predict that speech development might be systematically affected by these deficits in auditory perception. Since speech consists of many acoustic components such as intonation, stress, rhythm, duration, and voice, as well as the distinctive phonetic features of verbal sounds that constitute actual words (Crystal, 1969), one could hypothesize that selectively responding to any one element, or limited elements, of speech would result in difficulties in learning to produce and to understand spoken words. Researchers have demonstrated that no single element of an acoustic stimulus will consistently result in accurate perception of phonemes or syllables (Jakobsen, Fant, & Halle, 1952; Liber- man, Cooper, Shankweiler, & Studdert-Kennedy, 1967). Rather, attention to multiple elements in an acoustic stimulus is necessary before accurate recognition occurs.

Autistic children have long been noted to manifest a variety of speech abnormalities. Most autistic children fall into one of two general categories: mute (nonverbal) or echolalic (e.g., Koegel, Rincover, & Egel, 1982; Rutter, 1968; Schreibman & Mills, 1983). Schuler (1976) has divided mutism into three categories: (a) total, (b) functional, and (c) semimute, with semimute defined as using a limited repertoire of words and word approximations in a functional manner to express immediate desires or dislikes. Echolalia is also subclassified into (a) immediate echolalia, which is the immediate repetition of the speaker's verbalizations, and (b) delayed echolalia, which is defined as the repetition of stored, usually echoic, utterances in new and typically inappropriate contexts (Fay & Schuler, 1982).

A hallmark feature of the speech of autistic children is dysprosody, or the faulty use of intonation, stress, and rhythm (Baltaxe, 1981, 1984; Baltaxe & Simmons, 1975). Anecdotal observations from our own laboratory and other treatment centers suggest that nonverbal children who eventually do acquire vocal speech (usually through training) frequently evidence a typical dysprosic pattern characterized by lack of inflection and intonation. Although speech content is accurate, these children frequently display little intonation-with intonation referring to the melody of speech over time. Conversely, echolalic


children frequently speak with varying, or even exaggerated, degrees of intonation, yet their speech is often contextually and semantically inaccurate.

Since speech consists of many elements, including the important components of intonation and content, selective responding to either component would lead to abnormal speech patterns such as those described above. In an attempt to relate autistic stimulus overselectivity to speech characteristics, this study was designed to systematically assess whether autistic nonverbal and echolalic children differentially respond to content versus intonation in speech stimuli (where content refers to the phonetic, rather than semantic, features of the verbal stimuli). As a basis for comparison, we were interested in assessing the response of normal children to the same stimuli.

METHOD

Participants

Ten male autistic children participated in this study. The chronological ages (CA) of Children 1, 2, 3, 4, and 5 were 4.9, 6.2, 11.2, 6.7, and 4.5 years, respectively (M= 6.7 years). The mental ages (MA) of these children as determined by standardized intelligence tests (e.g., Merrill Palmer, Stanford Binet, Alpern Boll, Slosson, Bayley Scales) were 2.7, 3.4, 2.1, 5.0, and 1.8 years, respectively (M= 3.0 years). These children emitted high rates of immediate and delayed echolalia as determined by teacher reports and direct observa- tion by the experimenters. The children all had shown patterns of echolalic speech prior to the age of 3 years and would reliably echo when presented with a stimulus to which they had no appropriate response (Carr, Schreib- man, & Lovaas, 1975). Children 1, 3, and 5 had extensive language training and could communicate basic desires. Children 2 and 4 had minimal language training, and their speech was almost entirely echolalic.

The CAs of Children 6, 7, 8, 9, and 10 were 4.8, 9.7, 12.0, 4.4, and 5.2 years, respectively (M--7.2 years). Their MAs were 4.1, 4.8, untestable, 3.9, and 2.7 years, respectively. These children were all initially mute, but had, through speech training, acquired limited receptive and expressive speech skills. The speech of these children could be described as monotonic, nonspon- taneous, and poorly articulated. Child 6 had extensive (2 ½ years) language training and, as a result, could produce, and respond to, fairly complex sentences. Child 9, after 1 year of language training, could communicate basic wants using one- or two-word utterances. Children 7, 8, and 10 had minimal language training and were essentially nonverbal.

These 10 children had been diagnosed as autistic by independent agencies using the diagnostic criteria of the National Society for Autistic Children (Rit-


vo & Freeman, 1978) and the criteria outlined in the American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders (3rd edi- tion, 1980). The children manifested severe deficits in social behavior, language, play, and learning. In addition, they engaged in self-stimulatory behavior (e.g., body rocking, arm flapping, spinning) and tantrums.

Six normal children, four boys and two girls, also participated in this study. Their mean CA was 4.0 years, with a range of 3.4 to 4.4 years.

None of the children had any type of organic hearing abnormality detected in the routine hearing evaluations conducted as part of their school or clinic assessments.

Apparatus

In order to ensure an unambiguous measure of the dependent variable, a lever press response was used. The apparatus was a 35.5cm x 24. lcm x 13.9 cm wooden box, painted black, with a 7.5 cm metal lever protruding f rom its front. The apparatus was placed before the child within easy reach. A microswitch beneath the lever was activated when the lever was depressed. Activation of the microswitch was used to define a lever press during this investigation.

Training Stimuli

All children were trained to lever press to the sound of an auditory complex stimulus designated as S + and to withhold response to an auditory complex stimulus designated as S - . Four different auditory stimuli were combined in pairs to serve as the S + and S - . Two sounds representing content ("min min" and "nur nur") and two intonations (monotone and varying) were combined so that each sound was paired with each modulation, totalling four possible stimulus pairs. The sounds were presented at 72 dB, measured by a General Radio Company sound pressure level meter. A single-subject design with across-subject control for assignment of S + and S - stimulus pairs was utilized. The assignment of complex stimulus pairs to the normal and autistic children is presented in Table 1.

The stimulus complexes were presented at 7 sec intervals, in a random order, on a prerecorded audio cassette tape. The randomized order was con- strained such that no S + or S - stimulus could be presented more than three times consecutively. The stimulus complexes were each recorded using the voice of a female adult. The first verbal unit of each complex was presented at an equivalent pitch level, with the second verbal unit being presented at either an identical pitch level or at a pitch level that was one musical fifth higher than the first verbal unit.


TABLE 1. The Assignment of Stimulus Complex Pairs to

Normal Children and to Nonverbal and Echolalic Autistic Children

Child Population S + S -

1 Autistic/echolalic min rain! nur nur 2 Autistic/echolalic rain min nur nur! 3 Autistic/echolalic nur nur! rain min 4 Autistic/echolalic nur nur min min! 5 Autistic/echolalic min min nur nur!

6 Autistic/nonverbal nur nur rain min! 7 Autistic/nonverbal nut nur! min min 8 Autistic/nonverbal min min! nur nur 9 Autistic/nonverbal min min nur nur!

10 Autistic/nonverbal nur nur rain min!

11 Normal min min! nur nur 12 Normal nur nur min min! 13 Normal nut nur min min! 14 Normal nur nur! min min 15 Normal min rain nur nur! 16 Normal min min! nur nut

Note. An exclamation point (!) represents varying intonation. No exclamation point represents monotone intonation.

Training Procedure

Each child was seated facing the exper imenter . The tape recorder was

located o f f to the side, placed such that the sound level meter would read 72

dB at the child 's ear level. The lever press appara tus was located on a small

table in be tween the exper imenter and the child. All sessions were conduc ted

in ei ther a small l abora to ry r o o m or in a quiet corner o f a c lassroom.

Tra in ing began by manua l ly placing the child's hand on the lever and then

immedia te ly p rov id ing re in fo rcement in the fo rm o f a small edible (e.g.,

raisin, peanut , candy). This manua l p rompt was delivered on five consecutive

occasions, af ter which the child 's responding was re inforced with food only

for unassisted lever presses. W h e n the child emi t ted at least two unassisted

lever presses within 1 minute , the t raining tape was act ivated. I f the child did

not emit unassisted lever presses, the manua l p r o m p t p rocedure was again uti l ized unti l unassisted lever presses occurred .

W h e n the t ra ining tape was act ivated, the child received five consecut ive

manua l p rompts to press the lever in the presence o f the S + and to wi thhold

responding in the presence o f S - . Responses (either response to S + or

wi thholding response to S - ) were re inforced on a cont inuous re inforcement schedule until the child responded correctly on five consecutive S + trials (lever


press) and five consecutive S - trials (withhold lever press), at which time the schedule of reinforcement was gradually lessened to a variable ratio three schedule. If the child responded incorrectly on five consecutive trials, two manual prompts were delivered. The child was trained until he or she responded correctly on 18 of 20 consecutive trials. To reduce the likelihood of unintended auditory stimuli confounding the procedure, the experimenter did not verbally interact with the child during the experimental sessions. One stimulus presentation occurred during each trial, and 40 trials were conducted during each training session.

Testing Procedure

After the child had acquired the discrimination between the given set of training stimuli, he or she was presented with the test tape which consisted of all four possible complex stimulus pairs (i.e., min min!, min min, nur nur!, nur nur, where the exclamation point indicates varying intonation). This procedure allowed assessment of the extent to which the child's responding was controlled by the content or the intonation component of the training stimulus complex. Approximately 60 training trials (approximately 30 of the original S + and 30 of the original S - ) were intermixed with approximately 40 probe test trials (approximately 20 trials each of the remaining combinations of content and intonation). The variations in the number of stimulus complex presentations were due to either accidental omissions or additions in the con- struction of the tape recording, or were due to the tape being stopped and started, resulting in the obliteration of a presentation of a stimulus complex. The variations in the number of presentations ranged from zero to two and are accounted for in the presentation of the results. Trial order was randomized with the constraint that no trial type could be presented more than twice consecutively. No reinforcement followed responses on probe trials. Reinforcement was available for correct responses on training trials. During testing, as during training, reinforcement was available on a variable ratio three schedule.

Modified Training Procedure

Two children, Child 5 and Child 10, failed to acquire the discrimination with the training procedure described. For these two children, the following procedures were utilized.

Child 5. Child 5 consistently failed to respond to the taped presentations of the stimulus complexes. In an attempt to increase responding to the taped presentations of S + and S - , the following procedure was employed.

Initially, the experimenter presented the child with 50 trials in which the


stimulus complexes were voiced by the experimenter without the tape recorder presentations. Following this, the experimenter voiced 39 trials with her hand over her mouth. Responses to S + and withholding responses to S - increased using this procedure. In an effort to transfer stimulus control f rom the experimenter 's voice to the tape, the experimenter 's voice and the taped complexes were presented simultaneously. The child did not respond, so voiced- only trials were reinstated, resulting in increased responding. Following this, a tape was constructed containing a series of simple commands to which the child successfully responded. Since the child was now responding to taped stimuli, we at tempted a return to the tape of experimental stimuli. The procedure was successful, and the child responded to the taped stimulus complexes. Training then continued in the standard manner as described above.

Child 10. This child also failed to respond to the recorded presentations of the stimulus complexes. A trainer, who was naive as to the purpose of the study, then voiced the stimuli concurrently with the tape, but the child still failed to respond. The tape was then discontinued, and voice-only trials were presented, with the volume of S + greatly elevated in order to make the stimuli highly discriminable. The loudness prompt was continued until the discrimination was acquired. Transfer to the tape was then attempted, using the methods described for Child 5. Transfer to the tape was never successful, and training was continued with the S + and S - presented by the experimenter, at equivalent levels. When the child reached criterion, he was presented with the test stimulus complexes, also voiced by the trainer.

Reliability

Reliability assessments were obtained on 33°7o of training and testing sessions for all children. During reliability sessions, a naive observer, viewing through a one-way mirror or f rom a corner of the room, scored the child's responses (microswitch activation) prior to the experimenter presenting any consequences to the child. Session reliability was calculated between the experimenter and observer by dividing the number of agreements by the number of agreements plus disagreements x 100. For all of the children except Child 10, reliability was 100070. For Child 10, reliability was 97°70 (range 94°70 to 10007o).

RESULTS

Training

All children acquired the training discrimination, although with considerable variability. Table 2 presents the number of trials to criterion for each child. Of the 10 autistic children, 8 reached criterion in fewer than 303 trials,

Differential Responding to Content and Intonation I17

TABLE 2. The Number of Training Trials to Criterion for

Normal Children and Nonverbal and Echolalic Autistic Children

Child Population Training Trials M

1 Autistic/echolalic 40 2 Autistic/echolalic 98 3 Autistic/echolalic 400 4 Autistic/echolalic 50 5 a Autistic/echolalic 280

6 Autistic/nonverbal 125 7 Autistic/nonverbal 31 8 Autistic/nonverbal 84 9 Autistic/nonverbal 302

l0 b Autistic/nonverbal 340

11 Normal 45 12 Normal 60 13 Normal 73 14 Normal 60 15 Noi'mal 27 16 Normal 130

173.6

176.4

65.8

aChild 5 required 280 standard trials to reach criterion, after the special training procedure had been implemented, bChild 10 required 340 voiced stimulus complex trials to acquire the discrimination.

with one child requi r ing 340 tr ials and a second requi r ing 400 tr ials . Al l o f the n o r m a l ch i ldren acqu i red the d i sc r imina t ion wi thin 130 tr ials . The mean number o f t r ia ls to c r i te r ion for the aut is t ic ch i ld ren was s imi lar regardless o f language characterist ics (i.e., 173.6 for echolalic children; 176.4 for nonverbal chi ldren) . In cont ras t , the n o r m a l ch i ld ren requi red , on the average, only 65.8 tr ials to learn the d i sc r imina t ion .

Testing

Figures 1 and 2 presen t the da t a for the 10 aut is t ic ch i ldren and the 6 nor - mal chi ldren , respect ively. Percent r e spond ing is presented on the o r d i n a t e for each chi ld. The four s t imulus complexes are presented a long the abscissa with the first two represent ing S + and S - , respectively, and the second two the i n tona t i on and con ten t p robe s t imuli , respect ively.

Dur ing test ing, the chi ldren genera l ly ma in t a ined a high level o f r e sponding to S + ; the only ones who responded on fewer than 90% of the S + presen- ta t ions were Ch i ld ren 1 and 5, whose scores were 86°7o and 76070, respect ively. In add i t ion , the chi ldren ei ther pa r t i a l ly (i .e. , less than 20070 responding) or comple te ly wi thhe ld r e spond ing to the S - t h r o u g h o u t the tes t ing tr ials (see Figures 1 and 2).

S1

$2

$3

$4

$5

100,

CHILD

REN

5O

r.~

MM

P N

N

NN

! M

M

MM

N

N!

NN

M

M!

NN

! MM

M

M!

NN

N

N

MM

! MM

N

N!

MM

N

N!

NN

M

M!

,z

~

100.

S

6 $7

S

8 $9

$1

0

NONV

ERRA

L 0

CHILD

REN

NN

M

M!

MM

N

N!

NN

! M

M

MM

! N

N

MM

! N

N

NN

! M

M

MM

N

N!

HN

MM

! N

N M

M!

MM

N

N!

• S

+

[]

S-

r'3

Into

natio

n 0

Con

tent

P

robe

P

robe

FIG

UR

E 1

. P

erce

nt r

esp

on

se t

o th

e fo

ur

stim

ulu

s co

mpl

exes

fo

r th

e ec

hola

lic

and

no

nv

erb

al

auti

stic

chi

ldre

n.


$11 $12 $13

L CHILDREN

50

0 ~- MM! NN NN! MM NN MM! MM NN! NN MM! MM NN!

NN! MM MM! NN MM NN! NN MM! MM! NN NN! MM

IBS ÷ I~ S- 1-11ntonati0n ~ Content Probe Probe

FIGURE 2. Percent response to the four stimulus complexes for the normal children.

Looking at the data for the probe trials, it can be seen that intonation controlled probe responding for four o f the five echolalic children (Children 1, 2, 3, and 5). In contrast , data for the probe trials for the nonverbal children (Children 6, 7, 8, 9, and 10) indicate that these children responded to the probe complex that matched the content of the S + on which they were trained more than they responded to the probe complex that matched the S + on intonation. This pattern o f response was particularly evident for Children 7, 9, and 10. Children 4, 6, and 8 had relatively lower scores on both probe types, sug- gesting that these children's responses had to some extent come under the control o f both components o f the training stimulus complex and hence showed minimal overselectivity (see Figure 1).

In Figure 2, it can be seen that three o f the six normal children demonstrated little evidence o f control by individual components . Interestingly, however, three o f these children (Children 12, 15, and 16) showed considerable control by the content componen t as did several o f the nonverbal autistic children. 1

'Statistical analyses were conducted to confirm the reliability of these findings. Briefly, an analysis of variance did not yield significant differences between groups (autistic vs. normal) on trials to criterion, and a two-way analysis of variance showed that all three groups of children main- tained the original discrimination and showed significantly more responses to the S + than to the S - stimuli. Additional ANOVAs showed that there were significant differences (a) between


It is also important to note that selective responding to intonation or content was not related to a particular type of S + stimulus. Thus, it occurred whether S + was monotonic or intoned.

DISCUSSION

Autistic nonverbal and echolalic children and normal children were trained to respond to an auditory complex stimulus. After the children acquired the discrimination, they were each tested with stimuli designed to identify the component of the training S + which had acquired the most control over responding. Four of the echolalic children gave evidence for selectively responding to the intonation component of the stimulus complex. All five of the nonverbal autistic children provided at least some evidence of selectively responding to the content component of the stimulus complex. The normal children showed minimal overselectivity in half of the cases, but did selectively respond to the content component of the stimulus complex in the rest of the cases. These results thus suggest that echolalic autistic children tend to overselect to intonation in a stimulus complex that consists of content and intonation, whereas nonverbal autistic children tend to overselect to content. Normal children, the results suggest, provide evidence of either not overselecting significantly or of overselecting to the content component of the stimulus complex.

These results are consistent with the overselectivity literature in two ways. First, these data provide additional support for previous research which has indicated that when autistic children are presented with a complex stimulus, they will generally respond to an overly restricted number of components of the stimulus (Lovaas et al., 1971). These data also support the finding of Reynolds et al. (1974) that when the stimulus complex presented to the autistic child is within the auditory modality, overselective responding occurs. Second, our results are consistent with Wilhelm and Lovaas's (1976) finding that the severity of overselectivity is linked to mental age, as the one echolalic child not overselecting to intonation was also the one with the highest mental age of the echolalic children (Child 4, MA = 5.0). Further, this child acquired the original training discrimination in 50 trials, a number on the low end of the range of autistic children acquiring the discrimination (range = 31 to 400). The two nonverbal children (Children 6 and 8) who least clearly overselected

the three groups in their responding to content and intonation probe trials and (b) between echolalic and nonverbal autistic children in that the echolalics responded primarily to intonaton and the nonverbals to content. Both nonverbal autistics and normals responded to content significantly more than to intonation. The details of the statistical analyses are available from the authors upon request.


to content also had acquired the discrimination on the low end of the range for the entire group of autistic children (125 and 84 trials, respectively).

In addition, it is interesting to note that the two children (Children 5 and 10) who required the special training procedures had among the lowest MAs of the participants. Perhaps they overselected to some feature of the training environment to the exclusion of the intended functional s t imulus - the audio tape. It is also interesting to note that although Child 10 was trained utilizing a loudness prompt, he ultimately came under the predominant control of the content component of the S + complex. While these findings are provocative, at this time we can only speculate on the controlling variables for the behaviors of these two children.

One of the most interesting findings of this research is the fact that the nonverbal autistic children and the normal children shared a tendency to respond to content over intonation. This is certainly understandable for the normal children, because content is the most relevant component of speech stimuli. That is, control by phonetic elements of speech independent of differences in pitch, etc., is certainly functional. We all must be able to discriminate the relevant element of content, because different people will undoubted- ly speak to us with different pitches, rhythms, rates, and intonation. It is therefore not surprising that those normal children who did show differential responding to the components responded to content. Ideally, however, they would respond to both content and intonation so that when they produced speech, both components would be correct (i.e., not monotonic as one often sees in previously nonverbal autistic individuals).

One might speculate that the nonverbal autistic children who responded more like the normal children would be among the most linguistically advanced. Looking at the children who participated in this study, this does not seem to be the case. For example, Child 6, who showed minimal differential responding to the content component, was the most linguistically advanced of all the autistic subjects. Also, Child 7 and Child 10, who showed clear control by content, were among the least linguistically skilled. (Child 10 was com- pletely nonverbal having no expressive nor receptive language.) In addition, looking at the echolalic autistic children, Children 1, 2, and 3, who clearly differentially responded to intonation, were the most linguistically advanced of this group.

Another possibility is that the pattern of responding may be a function of language training. During most language training programs (including ours), emphasis is placed on phonetic content over intonation, pitch, etc. This is because, as mentioned above, phonetic content is the most important feature for communication. Thus, reinforcement is associated with response to, and production of, correct phonetic content. While other prosodic features of speech are certainly included in speech training, it is likely that phonetic content is most closely tied to reinforcement. This might somewhat parallel


the early reinforcement history of normal children who are not expected to be skillful at production of language's prosodic features until after acquiring skill at phonetic content. Echolalic children, in contrast to normal and previously nonverbal autistic children, are generally known for accurate or exaggerated production of speech with minimal or no comprehension of content. These children, however, typically do not receive the same speech treatment in that emphasis is not placed on improving phonetic content per se. Rather, emphasis is placed on establishing functional referents for the speech. Perhaps this different training and reinforcement history is related to the different patterns of responding we see in the present study.

In order to tease out the effects of some of these variables (e.g., reinforcement history), it would be interesting to determine if differential responsiveness to speech components (e.g., phonetic content and intonation) is present before any speech-related reinforcement history. Ideally, we could conduct an assessment such as that used in the present study with autistic and normal children prior to any speech development. After the autistic children were distinguished as nonverbal or echolalic, one could look at the earlier response pattern. Perhaps children who early on respond differentially to intonation are highly likely to become echolalic while those whose early pattern indicates a responsiveness to content may be those likely to remain nonverbal (as op- posed to spontaneous development of echolalia). Unfortunately, at this time we are left to speculate on the causes of this differential pattern of responding between these two groups.

The intonational patterns of autistic children have previously been at- tributed to emotional problems assumed present in the children. Bettelheim (1967) identified intonational abnormalities as stemming from lack of feel- ing and difficulties in expressing anger. The monotonous speech often seen in mute children has been explained as a "failure to express personality" (Despert, 1956). A more plausible and useful paradigm might be to view intonational abnormalities in echolalic and nonverbal autistic children in light of the accumulated evidence for selective responding to complex stimulus situations.

Echolalia typically is an indicator of severe comprehension difficulites. Premack and Premack (1974) conducted research with a heterogenous population of children, including children labeled as emotionally disadvantaged, in- stitutionalized, and psychotic, and discovered that the more severe the echolalia present in the children, the more severe the comprehension problems. Carr et al. (1975) also provided evidence that supports the contention that echolalic autistic children echo words for which they have no response (i.e., that are incomprehensible to them). Several other researchers have also noted that echolalia is not indicative of receptive speech (Fay, 1967, 1969; Fay & Butler, 1968). One must be cautious, however, in discussing response to content in these investigations. Phonetic content, as assessed in the present study, and semantic content, as discussed in terms of comprehension, are quite dif-


ferent. It is possible, however, that the echolalic child's overselective response to intonation might be to exclusion, or detriment, of response to phonetic and /or semantic content in English.

Research has been conducted that has identified selective responding to auditory stimuli, although it has not been labeled as stimulus overselectivity. Frith (1969) conducted a memory study with 3- to 4-year-old autistic children in which she presented stressed and nonstressed words in sentences to the children and then looked at which words were more likely to be re- p e a t e d - t h e words that had been stressed or the words that were important due to their content. She noted that the autistic children tended to be influ- enced more by the physical qualities of the auditory stimuli than by their content. Similarly, Miller and Toca (1979) reported successfully teaching words to a 3-year-old autistic male who could "sing TV commercials" but had very little functional speech. They used a method called Melodic Intonation Ther- apy, which consisted of applying a specifically intoned melody to each new word that they taught him. They reported that he acquired language and that he continued to use the exact melody on which he had been trained for each word that he had acquired.

Relating the present results to prior findings, the observations that autistic children speak in either a highly intoned manner or in a monotonic manner, and that these two types of speech seem to be related to language comprehension, possibly can be viewed as resulting from overselective responding to complex auditory stimuli. However, one must be cautious in generalizing from these data. These results need to be replicated with additional subjects and with other, perhaps more functional, stimuli before we can be totally confi- dent in the phenomenon reported here. Additional research examining stimulus overselectivity in the auditory modality should be conducted in order to examine other speech components (such as stress, rhythm, and rate) to which autistic children may overselect and which subsequently could result in impaired speech expression and reception. Much of the research that has led to innovative approaches to modifying overselectivity in the visual modality, such as overtraining and thinning schedules of reinforcement (e.g., Koegel et al., 1979), and repeated exposure to multiple cues (Schreibman et al., 1982), could also be attempted with auditory overselectivity. Perhaps a more thorough examination of auditory overselectivity, and the subsequent development of treatment strategies aimed at its reduction, would result in a lessening of the pervasive difficulties that autistic individuals have in communicating effec- tively with their social environment.

REFERENCES

American Psychiatric Association (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC.

Baltaxe, C. A. M. (1981). Acoustic characteristics of prosody in autism. In P. Mittler (Ed.), Frontiers of knowledge in mental retardation (pp. 223-233). Baltimore: University Park Press.

124 L. Schreibman, B. S. Kohlenberg, and K. R. Bri t ten

Baltaxe, C. A. M. (1984). Use of contrastive stress in normal, aphasic, and autistic children. Journal o f Speech and Hearing Research 27, 97-105.

Baltaxe, C. A. M., & Simmons, J. Q. (1975). Language in childhood psychosis: A review. Journal o f Speech and Hearing Disorders, 40, 439-458.

Bettelheim, B. (1967). The empty fortress-infanti le autism and the birth o f self. New York: Free Press.

Carr, E. G., Schreibman, L., & Lovaas, O. I. (1975). control of echolalic speech in psychotic children. Journal o f Abnormal Child Psychology, 3, 331-351.

Crystal, D. (1969). Prosodic systems and intonation in English. London: Cambridge University Press.

Despert, J. L. (1956). Discussion--L. Kanner, irrelevant and metaphoric language in early infantile autism. American Journal o f Psychiatry, 103, 242-246.

Fay, W. H. (1967). Childhood echolalia: A group study of late abatement. Folia Phoniatrica, 19, 297-306.

Fay, W. H. (1969). On the basis of autistic echolalia. Journal o f Communication Disorders, 2, 38-47.

Fay, W. H., & Butler, B. V. (1968). Echo retention as an approach to semantic resolution. Journal o f Speech and Hearing Research, 11, 365-371.

Fay, W. H., & Schuler, L. L. (1982). Emerging language in autistic children. Baltimore: University Park Press.

Frith, U. (1969). Emphasis and meaning in recall in normal and autistic children. Language and Speech, 12, 29-38.

Jakobsen, R., Fant, C., & Halle, M. (1952). Preliminaries to speech analysis: The distinctive features and their correlates. Cambridge, MA: M.I.T. Press.

Koegel, R. L., Rincover, A., & Egel, A. L. (1982). Educating and understanding autistic children. San Diego, CA: College Pres's.

Koegel, R. L., & Schreibman, L. (1977). Teaching autistic children to respond to simultaneous multiple cues. Journal o f Experimental Child Psychology, 12, 299-311.

Koegel, R. L., Schreibman, L., Britten, K. R., & Laitinen, R. (1979). The effects of schedule of reinforcement on stimulus overselectivity in autistic children. Journal o f Autism and Developmental Disorders, 9, 383-397.

Koegel, R. L., & Wilhelm, H. (1973). Selective responding to the components of multiple visual cues by autistic children. Journal o f Experimental Child Psychology, 15, 442-453.

Liberman, A. M., Cooper, F. S., Shankweiler, D. P., & Studdert-Kennedy, M. (1967). Percep- tion of the speech code. Psychological Review, 74, 431-461.

Lovaas, O. I., & Schreibman, L. (1971). Stimulus overselectivity of autistic children in a two- stimulus situation. Behaviour Research and Therapy, 9, 305-310.

Lovaas, O. I., Schreibman, L., & Koegel, R. L. (1979). Stimulus overselectivity in autism: A review. Psychological Bulletin, 86, 1236-1254.

Lovaas, O. I., Schreibman, L., Koegel, R. L., & Rehm, R. (1971). Selective responding by autistic children to multiple sensory input. Journal o f Abnormal Psychology, 77, 211-222.

Miller, S. B., & Toca, J. M. (1979). Adapted melodic intonation therapy: A case study of an experimental language program for an autistic child. Journal o f Clinical Psychiatry, 40, 80-82.

Premack, D., & Premack, A. (1974). Teaching visual language to apes and language-deficient persons. In R. L. Scheifelbusch & L. L. Lloyd (Eds.), Languageperspectives-acquisition, retardation, and intervention (pp. 347-376). Baltimore: University Park Press.

Reynolds, B. S., Newsom, C. D., & Lovaas, O. I. (1974). Auditory overselectivity in autistic children. Journal o f Abnormal Child Psychology, 2, 253-263.

Ritvo, E. R., & Freeman, B. J. (1978). National Society for Autistic Children definition of the syndrome of autism. Journal o f Autism and Childhood Schizophrenia, 8, 162-167.

Rutter, M. (1966). Behavioral and cognitive characteristics of a series of psychotic children. In J. K. Wing (Ed.), Early childhood autism, (pp. 51-81). London: Pergamon Press.


Rutter, M. (1968). Concepts of autism: A review of research. Journal of Child Psychology and Psychiatry, 9, 1-25.

Schover, L. R., & Newsom, C. D. (1976). Overselectivity, developmental level, and overtraining in autistic and normal children. Journal of Abnormal Child Psychology, 4, 289-298.

Schreibman, L. (1975). Effects of within-stimulus and extra-stimulus prompting on discrimination learning in autistic children. Journal of Applied Behavior Analysis, 8, 91-112.

Schreibman, L., & Charlop, M. H. (1981). S + versus S - fading in prompting procedures with autistic children. Journal of Experimental Child Psychology, 31, 508-520.

Schreibman, L., Charlop, M. H., & Koegel, R. L. (1982). Teaching autistic children to use extra- stimulus prompts. Journal of Experimental Child Psychology, 32,475-491.

Schreibman, L., & Koegel, R. L. (1982). Multiple cue responding in autistic children. In P. Karoly & J. J. Steffen (Eds.), Advances in child behavior analysis and therapy (pp. 81-99). New York: John Wiley.

Schreibman, L., Koegel, R. L., & Craig, M. S. (1977). Reducing stimulus overselectivity in autistic children. Journal of Abnormal ChiM Psychology, 5, 425-436.

Schreibman, L., & Mills, J. I. (1983). Infantile autism. In T. H. Ollendick & M. Hersen (Eds.), Handbook of child psychopathology (pp. 123-149). New York: Plenum Press.

Schuler, A. L. (1976, November). Speech and language in autism, characteristics and treatment. Mini seminar at the annual meeting of the American Speech and Hearing Association, Houston.

Wilhelm, H., & Lovaas, O. I. (1976). Stimulus overselectivity: A common feature in autism and mental retardation. American Journal of Mental Deficiency, $1, 227-241.

differential responding to content and intonation components of a complex auditory stimulus by...

Documents