Journal of Abnormal Child Psychology, I1"ol.16, No. 1, 1988,pp. 45-56 Caregiver Interactions with Autistic Children Connie Kasari,~.2 Marian Sigman,1 Peter Mundy,1 and Nurit Yirmiya I Caregiver interactions with young autistic children were contrasted with those involving caregivers and developmentally matched mentally retarded and normal infants. Caregivers of autistic children were similar to other caregivers in their responsiveness to child nonverbal communication bids and in their engagement in mutually sustained play. Caregivers o f autistic children were similar to caregivers of mentally retarded children in their greater use of control strategies. However, these two groups of caregivers differed in the particular strategies they used to shape their children's behavior. Caregivers o f mentally retarded children pointed to objects while caregivers of autistic children spent more time physically holding their children on task. Individual differences within th e autistic sample indicated that caregivers regulated their children's behavior less and showed more mutual play and positive feedback to more communicatively able autistic children. These findings suggest that caregivers respond differentially to the specific deficiencies shown by their children. A u t i s t i c children show deficits in d e v e l o p m e n t a l tasks that are typically learned in interaction with others. T h e y rarely engage in social behaviors that d r a w a t t e n t i o n to oneself, such as showing o f f or displaying co y affect (Neu- Manuscript received in final fo[m May 21, 1987. Support for this research was provided by NIMH grant MH 33815 to the second author, and NIMH postdoctoral fellowship (MH 16381) to the first author. The second and fourth authors are also affiliated with the Department of Psychology, UCLA, and the third author with Olive View Medical Center, Sylmar. We thank Angie Pefia, Holly Flesh, Janet Miller, Anne Weickgenant, and Jon Schlichting for their assistance in data collection and preparation of this manuscript. Portions of this paper were presented at the International Conference on Infant Studies, Los Angeles, 1986. ~Department of Psychiatry, UCLA School of Medicine, Los Angeles, California 90024. 2Address all correspondence to Connie Kasari, Department of Psychiatry, UCLA School of Medicine, Los Angeles, California 90024. 45 00914)627/88/02004)045506.00/0 9 1988 Plenum Publishing Corporation 46 Kasari, Sigman, Mundy, and Yirmiya man & Hill, 1978; Spiker & Ricks, 1984), and rarely direct another's attention as a means of sharing interest in an object or event (Curcio, 1978; Loveland & Landry, 1986; Mundy, Sigman, Ungerer, & Sherman, 1986; Wetherby & Prutting, 1984). The inability of these children to share interpersonal involvement is considered to be pathognomonic of the autistic disorder (Kanner, 1943). Autistic children's disturbance in reciprocal interactions would seem to have a profound effect on others' attempts to engage them socially. Yet there are few studies that have directly investigated this effect. Studies that have examined autistic children's social interactions have focused on the interactive abilities of the child (Lord, 1984; McHale, 1983; Sigman, Mundy, Sherman, & Ungerer, 1986; Strain, Kerr, & Ragland, 1979) or the effect the child may have on siblings (McHale, Sloan, & Simeonsson, 1986). With the exception of the Sigman et al. paper, these studies have concentrated on the abilities of children over the age of 5. We know little about how adults, particularly caregivers, interact with these children in the preschool years. How adults interact with these children would seem a potentially important area of study for at least two reasons. First, autistic children display unique social behavior characteristics. In studies from our own laboratory, we have identified differences between the nonverbal communication behaviors of autistic children and those shown by controls. The autistic children were specifically deficient in their tendency to share or direct another person's focus of attention. They tended to show toys, point to objects or events, and follow another's point less often than normal children and mentally retarded children matched on mental age. This deficiency in initiating and responding to bids for joint attention appeared in interaction with both exPerimenter (Mundy et al., 1986) and caregiver (Sigman et al., 1986). Furthermore, these nonverbal joint attention skills were strongly associated with individual differences in language development. Since these child behaviors may be learned in interaction with others, differences in children's usage of nonverbal communication skills may result from differences in style of caregiver interaction. Second, according to the literature on normal child development, adults are viewed as guides in assisting t h e child's participation in activities (Vygotsky, 1978). The more difficulty the child has in achieving a goal, the more the caregiver is observed to provide a "scaffold" (i.e., support) that encourages the child toward success or completion of the task (Bruner & Sherwood, 1983). Therefore, caregivers may respond with different interaction strategies to children who exhibit specific social interaction deficits. Both of these alternatives suggest that style of caregiver interaction should be associated with differences in children's usage of nonverbal communication skills. To address this hypothesis, we contrasted the interactive behaviors of caregivers of autistic children with those shown by caregivers Caregiver Interactions 47 of mentally retarded and normal children of equivalent developmental level. We examined particular caregiver behaviors that paralleled the child socialcommunication behaviors that have been a focus of our program of research. In addition, we examined these caregiver interactive behaviors vis-~t-vis child nonverbal communication and language skills. METHOD Subjects Three groups of children and their caregivers participated in this study (see Table I). All children in the first group had received the diagnosis of infantile autism by clinicians of the UCLA Clinical Research Center. These diagnoses were made independent of the experimenters by two or more psychiatrists and psychologists who reached consensus using DSM-III criteria (American Psychiatric Association, 1980). The latter include the onset of symptoms before 30 months of age, pervasive lack of responsiveness to other people, gross deficits in language development, and bizarre responses to various aspects of the environment. Children with symptoms associated with known organic brain dysfunction were excluded from this study in order to create a more homogeneous sample. The second group was composed of children with mental retardation (MR). Half were children with Down syndrome and the other half were identified as suffering from retardation of unidentified origin. A third group comprised normally developing children. All groups were matched on mental age and maternal education. The autistic and mentally retarded groups were also matched on chronological age (see Table I). Procedure Caregiver-Child Interaction. Caregivers and children were videotaped for 12 minutes as they interacted with toys under a variety of semistructured conditions. Most children were observed interacting with their mothers; however, five children (one autistic, two MR, and two normal) were observed with their fathers. Interactions were carried out in the carpeted playroom of a university laboratory that included a sectional couch placed in a semicircle and a set of standard toys placed on the carpet in the center of the room. These toys included a wooden form board, a plastic shape-sorting puzzle, a doll, a doll bed and bottle, a car, blocks, a cup, and a rattle. The caregiver and the child were observed in the following situations: (1) free play with any of the toys, (2) play with the doll, bed, and bottle, (3) play with the shape-sorting puzzle, (4) a social game without the use of Kasari, Sigman, Mundy, and Yirmiya 48 Table I. Sample Characteristics Characteristic Chronological age (in months) Mental age a (in months) Maternal education (in years) Males M SD M SD M SD Autistic n = 18 Mentally retarded n = 18 Normal n = 18 53.28 11.84 25.72 9.14 13.80 1.99 14 50.22 17.93 26.00 9.74 13.70 2.93 9 22.22 6.09 25.50 8.38 14.10 2.72 14 ~As determined by a Cattell or Stanford-Binet score. toys, and (5) clean-up by putting all the toys away in a wicker basket. The free-play episode lasted 4 minutes, while the doll, puzzle, social game, and clean-up episodes were 2 minutes each and followed in order. Caregiver interactive behaviors were coded by independent observers from the videotaped records. Separate viewings of the videotape were used to rate the frequency and duration of caregiver behaviors. The caregiver behaviors were grouped into three categories: attention regulation, behavior regulation and responsiveness. Attention regulation included behaviors designed to elicit the child's attention to objects or events without the use of physical contact. These included the number of times the caregiver (1) showed objects, (2) pointed to objects and events, (3) modeled behavior or actions on objects, and (4) attempted to elicit eye contact with the child. Behavior regulation behaviors were used to request a specific action of the child or to control or manage the behavior of the child. These behaviors often included physical contact with the child and included the number of times that the caregiver (1) offered objects to the child, (2) physically prompted the child to perform an action, and the amount of time the caregiver (3) physically held the child on task, and (4) initiated a new activity with the child. Responsiveness included behaviors used to maintain the child's engagement with an activity. These behaviors included (1) the amount of mutual play between caregiver and child, defined as the total duration of time that the child and adult were mutually involved in the same activity, (2) the number of times the caregiver gave positive feedback, e.g., "That's right," patting child on back, and (3) the number of caregiver responses to child nonverbal communication bids scored when the caregiver responded to child points, shows, or offers. To examine interrater reliability, two raters independently coded 10 randomly selected caregiver-child sessions. Reliability was estimated with a generalizability study. A generalizability study yields intraclass coefficients (G-coefficients) that control for observer bias and have been recommended for use with observational studies utilizing data of a continuous nature (AI- Caregiver Interactions 49 gina, 1978; Berk, 1979; Mitchell, 1979). In this study, the G-coefficients represent the ratio of subject score variance over the sum of subject variance plus rater by error variance. The mean estimate of interrater reliability of the 11 caregiver behaviors was .94 (range .85-.99). Assessment of Nonverbal Communication Skills. Children were assessed using a form of the Early Social Communication Scales (ESCS; Seibert & Hogan, 1982) according to procedures described in detail elsewhere (Mundy et al., 1986). In this 25-minute procedure the experimenter and child were videotaped as they sat facing one another at a small table. A set of toys were in view but out of reach to the child. The experimenter presented the child with different toys, initiated social games and turn-taking activities, pointed to colorful posters that were hung on the walls of the room, and made simple requests of the child. Trained observers viewed the ESCS videotapes and rated the children's social communication behaviors according to three categories of child behavior: social interaction, indicating, and requesting. In the social interaction category, the focus was on behaviors that are used to engage in turn-taking with another-e.g., eye contact after tickle, and reach to adult in order to initiate a game. In the indicating category, the emphasis was on behaviors used to direct or share the attention of the experimenter to an object, such as pointing to objects and showing objects to experimenter. The focus in the requesting category was on behaviors used to elicit aid in obtaining objects or events, such as reaching to toys and responding to simple requests by the adult. The nonverbal communication data used in the present study consisted of the total number of behaviors exhibited by the children in each of the categories of social interaction, indicating, and requesting. Two observers independently rated the nonverbal communication behaviors of 19 randomly selected subjects in this study. Generalizability coefficients for the individual frequency scores within each category ranged from .67 to 1.0, mean = .84. Assessment of Child Language. The Reynell Developmental Language Scales (Reynell, 1977) were used to assess children's language abilities. These scales provide a standardized measure of both expressive and receptive language and are designed for use with children between the developmental ages of 1 and 7 years. Each child's performance on these scales was transformed into expressive and receptive language age equivalence scores. RESULTS Prefiminary Analyses Preliminary analyses were performed to determine whether caregiver-child interactions varied as a function of caregiver (mother vs. 50 Kasari, Sigman, Mundy, and Yirmiya Table II. Mean Frequency of Caregiver Attention Regulation Behavior with Autistic, Mentally Retarded, and Normal Children During Caregiver-Child Interaction Caregiver behavior Show Point Model Eye contact Mentally Autistic retarded Normal (n = 18) (n = 18) (n = 18) F(2, 51) 4.56 3.78 3.00 n.s. 11.17 18.28 11.06 4.96 15.56 14.50 12.33 n.s. 1.83 .94 .22 5.22 Significance of differences: Newman-Keuls comparisons A/MR A/N MR/N ~ " aMANOVA (Wilks's criterion), F(8, 96) = 2.51, p < .02. father) and child sex. No significant partner or sex differences were observed. Therefore, the data were pooled for subsequent analyses. Group Comparisons: Caregiver Interactive Behaviors To test for group effects on the caregiver behaviors, multivariate analysis of variance (MANOVA) was computed using the individual caregiver scores f r o m each category of caregiver behavior. Attention Regulation Behaviors. M A N O V A for the set of behaviors including caregiver shows objects, points to objects or events, models behaviors or actions on objects, and attempts to obtain eye contact with the child yielded a significant group effect (F(8, 96) = 2.51, p < .02). Data from the univariate F tests are presented in Table II. These data indicated that there were significant group effects for two caregiver behaviors: pointing and eliciting eye contact. Follow-up comparisons showed that caregivers of mentally retarded children pointed to objects or events significantly more often than caregivers in either of the other two groups. While the frequency of caregiver attempts to elicit eye contact was low for all three groups, caregivers of autistic children engaged in more of this behavior than caregivers of normal children but did not differ in this regard f r o m caregivers of mentally retarded children. The frequency of caregiver attempts to achieve eye contact with their children was highest during the social game for all three groups. In this situation we specifically asked caregivers to engage their children in a social game without the use of toys. Caregivers of autistic and mentally retarded children engaged in this behavior much more during the social game, although they did attempt to obtain eye contact with their children during the other play episodes as well. Caregivers of normal children, on the other hand, never engaged in this behavior except during the social game. Behavior Regulation Acts. To determine if caregivers differed in their use of control strategies, M A N O V A was computed for the frequency of Caregiver Interactions 51 Table II1. Mean Frequency (F) or Duration (D) of Caregiver Behavior Regulation Attempts with Autistic, Mentally Retarded, and Normal Children During Caregiver-Child Interaction Caregiver behavior Offer (F) Prompt (F) Hold on task (D) Initiate task (D) Mentally Autistic retarded Normal (n = 18) (n = 18) (n = 18) F(2, 51) 10.50 9.94 8.00 n.s. 6.22 4.89 1.44 4.00 2.39 .74 .20 10.79 3.98 2.81 2.28 3.85 Significance of differences: Newman-Keuls comparisons A/MR A/N MR/N' a a a a aMANOVA (Wilks's criterion), F(8, 96) = 3.16, p < .003. caregiver offers, the frequency of physical prompts, the amount of time the caregiver held the child on task, and the amount of time the caregiver initiated an activity with the child. This analysis revealed a significant group effect (F(8, 96) = 3.16, p < .003). As illustrated in Table III, caregivers of autistic children consistently exhibited more of the behavior regulation acts compared to caregivers in the other two groups. Follow-up tests for the individual variables yielded significant group differences for three out of the four caregiver behaviors. Caregivers of normal children engaged in m a n y fewer instances of physical prompting than caregivers of MR or autistic children, who did not differ significantly f r o m one another. Caregivers of autistic children spent a significantly greater amount of the session holding their children on task than did caregivers of either normal or MR children. Caregivers of autistic children also spent more time initiating an activity with their children than did caregivers of normal children, but not mentally retarded children. Subsequent analyses indicated that caregiver behavior regulation acts of initiating an activity and holding on task were significantly correlated with autistic children's indicating behaviors (see Table IV). In addition, our previous research found that autistic children display fewer indicating behaviors than do mentally retarded and normal children (Mundy et al., 1986). Since these caregiver behavior regulation acts are significantly correlated with child indicating behavior, and only in the autistic group, it may be that group differences in these caregiver behaviors were secondary to group differences in the children's use of indicating behaviors. In order to address this issue, we reanalyzed these data on caregiver behaviors by covarying out the differences between the groups in indicating behaviors. The results of this A N O C O V A yielded no significant caregiver differences between the groups on initiating an activity (F(2, 50) = .55, p < .58) but did yield significant caregiver differences on holding on task (F(2, 50) = 3.27, p < .05). Follow-up tests indicated that caregivers of autistic children held their children on task longer 52 Kasari, Sigman, Mundy, and Yirmiya even when the differences attributable to their children's indicating behaviors were covaried out. Responsiveness. To determine if caregivers varied in their responsiveness to their children's behavior, MANOVA was computed for the amount of time caregivers engaged in mutual play with their children, the frequency of positive feedback, and the frequency of responses to child nonverbal communication acts. To describe caregiver responsiveness to their children's nonverbal communication bids, the number of times the caregiver did not respond to the child's nonverbal communication acts of showing, giving, and pointing was computed. That is, no caregiver response was equal to the number of times the caregiver responded to the child's nonverbal communication acts subtracted from the number of nonverbal communication acts exhibited by the child. This method excluded the two autistic children who did not display nonverbal communication acts. MANOVA yielded a significant group effect F(6, 98) = 2.77, p < .02). Follow-up tests indicated only one significant group difference and this was in the frequency with which caregivers gave positive feedback (F(2, 51) = 5.57, p < .007). Caregivers of autistic children used significantly more positive feedback with their children than did caregivers of either mentally retarded or normal children (13.56, 9.22, 6.39 positive statements, respectively). Caregivers in all three groups engaged in mutual play with their children for over half of the 12-minute session (6.64, 7.88, 8.00 minutes for the autistic, MR, and normal groups, respectively). In addition, the group means for no caregiver response to child communication bids were all uniformly low (less than 2), indicating that caregivers rarely failed to respond to their children's nonverbal communication bids. Individual Differences: Caregiver and Child Characteristics These analyses were conducted to determine if caregiver behavior varied according to child nonverbal communication behavior and language abilities. Nonparametric analysis (Spearman rho correlation) was employed because of the marked skewness of the language and social-communication score distributions (Cohen & Cohen, 1983). The correlations between caregiver behavior and child abilities in the autistic sample are presented in Table IV. These data indicate that in the caregiver category of attention regulation few significant correlations were found between caregiver behaviors and child nonverbal communication and language abilities. Only one behavior, the amount of time the caregiver attempted to elicit eye contact with the child, was related to the children's nonverbal indicating skills, and this relationship was in the negative direction. That is, caregivers who actively attempt- Caregiver Interactions 53 Table IV. Correlations (rho) of Caregiver Interactive Behaviors Within Categories of Attention Regulation, Behavior Regulation, and Responsiveness and Autistic Children's Nonverbal Communication Behaviors and Language Abilities Child nonverbal communication and language skills Caregiver behaviors Attention regulation Show Point Model Eye contact Behavior regulation Offer Physical prompt Hold on task Initiate task Responsiveness Mutual play Positive feedback Responds to bids Nonverbal social Nonverbal indicating Nonverbal requesting Language Language comprehension expression - .004 -.12 .07 -.09 - .30 .18 - . 12 -.61 b - .33 .10 .06 -.28 - .04 .15 - .28 -.40 .002 .20 - .30 -.26 .19 .13 -.03 -.14 .28 -.24 -.55 a -.49 ~ .43 -.03 -.16 -.35 .32 -.37 -.28 -.43 .45 -.23 -.28 -.52 ~ .27 .20 .15 .50" .47 a .04 .43 .39 .00 .59 b .52 a .10 .50 a .43 -.20 ~p < .05. bp < .01. ed to elicit their children's attention to themselves had children who did less well on nonverbal indicating skills as measured by the ESCS. Caregiver behaviors in the attention regulation category were not associated with the chip dren's language abilities. In the behavior regulation category, the amount of time caregivers spent initiating an activity with the child was significantly negatively related to the children's nonverbal indicating behaviors and expressive language abilities. The amount of time caregivers held their children on task was also significantly negatively related to the children's nonverbal indicating skills. These behavior regulation correlations, as noted earlier, have implications for understanding group differences in caregiver behavior (see Behavior Regulation section). In the category of caregiver responsiveness, there were positive and consistent correlates between caregiver behaviors and child communication and language abilities. Individual differences in the amount of mutual play and positive feedback that caregivers engaged in were significantly correlated with individual differences in nonverbal communication skills and language acquisition among the autistic children. Both mutual play and positive feedback were positively correlated with children's nonverbal indicating skills and expressive language abilities. By way of comparison, we also looked at the pattern of correlations between caregiver and child behaviors in the mentally retarded and normal samples. In the mentally retarded sample, only two caregiver behaviors were associated with child behaviors. In the behavior regulation category, caregiver 54 Kasari, Sigman, Mundy, and Yirmiya initiating of a task was negatively associated with language comprehension (0 = - .60) and expression (O = - .63). In the responsiveness category, mutual play between caregiver and child was positively associated with nonverbal indicating (0 = .54), language comprehension (0 = .52), and language expression (0 = .68). Although fewer correlations overall were found in the mentally retarded sample, those that were significant were similar to what we observed in the autism sample. In contrast, the majority of correlations in the normal sample were in the area of language rather than nonverbal communication. Caregivers with normal children who were less advanced in expressive language skills attempted to elicit more eye contact with them (0 = -.46) and held them on task longer (0 = -.52). Caregivers initiated a task more if their children were less advanced in language comprehension (0 -- -.46). In the responsiveness category, mutual play was positively associated with the language variables (comprehension, 0 = .64, expression, 0 = .69). Caregivers with children who werelowin indicating behaviors used more positive comments (0 = -.49), the inverse of what we observed in the autism sample. DISCUSSION The results of this study identified relatively few differences in caregiver interactive behaviors of preschool-aged autistic, mentally retarded, and normal children. Caregivers of autistic children were just as responsive to their children as caregivers of mentally retarded and normal children, with all three groups spending over 50~ of the session in mutually sustained play. Moreover, caregivers used similar strategies to direct their children's attention to, and engagement with, objects. Caregivers of autistic children showed toys, modeled actions, and offered toys to their children as much as or more than caregivers of mentally retarded and normal children. Differences in caregiver interactions emerged mostly between the developmentally delayed and normal groups. In general, caregivers of autistic and mentally retarded children were similar in their use of control strategies. Both of these groups of caregivers physically prompted their children more and initiated activities for longer periods than did caregivers of normal children. However, the two groups of caregivers differed in terms of the particular strategies they used to shape their children's behavior. Caregivers of mentally retarded children pointed to objects while caregivers of autistic children spent greater amounts of time physically holding their children on task. These results are consistent with findings from other studies that have contrasted the interactions of caregivers of mentally retarded children and caregivers of normal children. Findings from these studies have shown that Caregiver Interactions 55 caregivers of handicapped children spend more time attempting to elicit behaviors from their children, initiate interactions more frequently, and are generally more directive in their play (Cunningham, Reuler, Blackwell, & Deck, 1981; Eheart, 1982; Jones, 1977). One explanation for these differences in caregiver behavior is that caregivers are compensating for child behavior that is insufficient or inappropriate by using particular interactive strategies. These strategies may serve to stimulate or delimit specific child responses to a level that is more consistent with the child's age and the adult's expectations (Bell, 1971). Caregivers of both groups of developmentally delayed children used more direct behavioral interventions with their children perhaps as a means of increasing their children's engagement in joint activities. Since caregivers in all three groups did not differ in terms of the amount of mutual play in which they were engaged, it would seem that the strategies employed by the caregivers of developmentally delayed children were successful. Our results also indicate that caregivers were sensitive to the particular characteristics of their children's social behaviors. In the autism sample, several caregiver behaviors were correlated with child joint attention (nonverbal indicating) and expressive language abilities. Specifically, we found that the less well their autistic children performed on indicating behaviors, the more caregivers tried to elicit their children's attention, hold their children on task, and initiate activities, and the less time caregivers spent in mutual play and positive feedback. In other words, caregivers regulated their children's behavior less and showed more mutual play and positive feedback to more communicatively able autistic children. The pattern and number of correlations was quite different in the other groups. Thus, the parents of autistic children seem to be responding differentially to the deficiency in joint attention behaviors shown by their children. The direction of associations cannot be inferred from these data. Caregivers are surely responding to characteristics of their children's behaviors. On the other hand, autistic children may be shaped by their rearing environments to some degree. Therefore, the possibility exists that parents who engaged in more mutual play and positive feedback encourage the nonverbal and verbal communicative skills of their autistic children. REFERENCES Algina, J. (1978). Comment on Bartko's "On various intraclass correlation reliability coefficients." Psychological Bulletin, 85, 133-138, American Psychiatric Association. (1980). Diagnostic and statistical manual o f mental disorders (3rd ed.). Washington, D,C.: Author. Bell, R. Q. (1971). Stimulus control of parent or caretaker behavior by offspring. Developmental Psychology, 4, 63-72. 56 Kasari, Sigman, Mundy, and Yirmiya Berk, R. A. (1979). Generalizability of behavioral observations: A clarification of interobserver agreement and interobserver reliability. American Journal of Mental Deficiency, 83, 460-472. Burner, J., & Sherwood, V. (1983). Thought, language, and interaction in infancy. In J. D. Call, F. Galenson, & R. L. Tyson (Eds.), Frontiers of infant psychiatry (pp. 38-55). New York: Basic Books. Cohen, J., & Cohen, P. (1983). Applied multiple regression~correlation analysis for the behavioral sciences. Hillsdale: New Jersey: Erlbaum. Cunningham, C., Reuler, E., Blackwell, J., & Deck, J. (1981). Behavioral and linguistic developments in the interactions of normal and retarded children with their mothers. Child Development, 52, 62-70. Curcio, F. (1978). Sensorimotor functioning and communication in mute autistic children. Journal of Autism and Childhood Schizophrenia, 2, 264-287. Eheart, B. K. (1982). Mother-child interaction with nonretarded and mentally retarded preschoolers. American Journal of Mental Deficiency, 87, 20-25. Jones, O. L. M. (1977). Mother-child communication with pre-linguistic Down's syndrome and normal infants. In H. Schaffer (Ed.), Studies in mother-infant interaction (pp. 379-401). New York: Academic Press. Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Child, 2, 217-250. Lord, C. (1984). Development of peer relations in children with autism. In F. Morrison, C. Lord, & D. Keating (Eds.), Applied developmentalpsychology (Vol. 1). New York: Academic Press. Loveland, K., & Landry, S. (1986). Joint attention and language in autism and developmental language delay. Journal of Autism and Developmental Disorders, 16, 335-349. McHale, S. (1983). Social interactions of autistic and non-handicapped children during free play. American Journal of Orthopsychiatry, 53, 81-91. McHale, S., Sloan, J., & Simeonsson, R. (1986). Sibling relationships of children with autistic, mentally retarded, and nonhandicapped brothers and sisters. JournalofAutism andDevelopmental Disorders, 16, 397-413. Mitchell, S. (1979). Interobserver agreement, reliability and generalizability of data collected in observational studies. Psychological Bulletin, 86, 376-390. Mundy, P., Sigman, M., Ungerer, J., & Sherman, T. (1986). Defining the social deficits of autism. The contribution of nonverbal communication measures. Journal of Child Psychology and Psychiatry, 27, 657-669. Neuman, C., & Hill, S. (1978). Self-recognition and stimulus preference in autistic children. Developmental Psychobiology, 11, 571-578. Reynell, J. (1977). ReynellDevelopmentalLanguage Scales (Revised). Windsor, England: NFER. Seibert, J., & Hogan, A. (1982). Procedures manualfor the Early Social-Communication Scales (ESCS). Miami: Mailman Center for Child Development, University of Miami. Sigman, M., Mundy, P., Sherman, T., & Ungerer, J. (1986). Social interactions of autistic, mentally retarded and normal children and their caregivers. Journal of Child Psychology and Psychiatry, 27, 647-655. Spiker, D., & Ricks, M. (1984). Visual self-recognition in autistic children: Developmental relationships. Child Development, 55, 214-225. Strain, P., Kerr, M., & Ragland, E. (1979). Effects of peer-mediated social initiations and prompting/reinforcement procedures in the social behavior of autistic children. Journal ofA utism and Developmental Disorders, 9, 41-54. Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Cambridge, Massachusetts: Harvard University Press. Wetherby, A., & Prutting, C. (1984). Profiles of communicative and cognitive-social abilities in autistic children. Journal of Speech and Hearing Research, 27, 364-377.