Journal of Speech and Hearing Research, Volume 37, 913-923, August 1994

Influence of CommunicativeCompetence on Peer Preferences ina Preschool Classroom

Bethany L. GertnerSpeech Services

Associated Children's TherapiesChattanooga, TN

Mabel L. RiceChild Language Program

University of Kansas, Lawrence

Pamela A. HadleyDepartment of Speech and

Hearing SciencesArizona State University, Tempe

Recent research suggests that children's linguistic competence may play a central role inestablishing social acceptance. That possibility was evaluated by examining children's peerrelationships in a preschool classroom attended by children with varying degrees of communi-cation ability. Three groups of children were compared: children with normally developinglanguage skills (ND), children with speech and/or language impairments (S/LI), and childrenlearning English as a second language (ESL). Two sociometric tasks were used to measurepeer popularity: positive nominations and negative nominations. Children in the ND groupreceived more positive nominations than the children in either the ESL or S/LI groups. When thechildren's positive and negative nominations were combined to classify them as Liked, Disliked,Low Impact, or Mixed, the ND children predominated in the Uked cell, whereas the other twogroups of children fell into the Disliked or Low Impact cells. In addition, the PPVT-R, a receptivemeasure of single-word vocabulary, was found to be the best predictor of peer popularity. Thefindings are discussed in terms of a social consequences account of language limitations.

KEY WORDS: specific language Impairment (SLI), peer relations, English as secondlanguage (ESL), preschool Intervention

The purpose of the present study was to explore the relationship between children'sability to use language skillfully and their acceptance among peers. Howes (1983)suggested that complex social interaction skills are learned best in stable dyads andthat social skills develop within a friendship relationship. If this is the case, thensecuring and maintaining mutual friendships early in a child's life is important. Duringtheir preschool years, children begin to differentiate between friends and playmates(Howes, 1988). Preschool friends, as compared to playmates, become more respon-sive in conversation and begin to exchange more positive and less negativebehaviors during interactions. However, the friendship will not mature if one child isnot accepted by another. There are various reasons why children are not accepted bytheir peers. One reason may be that they are unable to use language effectively.Preschoolers use their communicative competence to make friends. Thus, if childrenexhibit poor communicative competence, they will often be denied access to theirpeer group (Howes, 1988). It has also been found that these children tend to becomeless positive in their affect with peers over time (Howes, 1988).

There is growing recognition that young children with language impairment are atspecial risk for failure to develop social interactions with their peers (cf. Brinton &Fujiki, 1993; Craig, 1993; Gallagher, 1991, 1993). The close association betweenchildren's linguistic competence and their patterns of peer interaction is demonstratedin a series of studies carried out in preschool classrooms (Hadley & Rice, 1991; Rice,1993; Rice, Sell, & Hadley, 1991). Rice and her colleagues demonstrated that thechildren who were sought out as preferred conversational partners in an integratedpreschool setting were generally those with normally developing language skills

C 1994, American Speech-Language-Hearing Association

M

913 0022-4685/94/3704.0913

914 Joumal of Speech and Hearing Research

rather than children with speech and language impairmentsor those learning English as a second language. In addition,the children with limited language abilities were more likely toinitiate verbal interactions with adults than with normallydeveloping peers. Rice (1993) argues that children withlimited language abilities encounter a number of socialconsequences. One consequence is that they are not fullyincorporated into peer interactions, a conclusion supportedby converging evidence from a number of studies (e.g., Craig& Washington, 1993). Extending this perspective, it is rea-sonable to predict that adjustments in conversational inter-actions could also be associated with social status in the peergroup. Children who are less able to engage peers inconversational interactions are less well-equipped with thecrucial skills necessary to transform social relationships intofriendships.

A connection between children's discourse abilities andtheir peer popularity has been established for normallydeveloping children (Black & Hazen, 1990; Hazen & Black,1989; Place & Becker, 1991). This connection has yet to bedemonstrated for children whose only developmental deficitis speech and/or language impairment(s). That is the topic ofthe study reported here. In this introduction, a selectivereview of the normative literature is provided, followed by asummary of related findings reported for children with cogni-tive deficits. The latter sample is of interest because theseyoungsters also demonstrate communication deficits. At thesame time, the multiplicity of their limitations does not allowfor a clear connection between social status and communi-cation limitations. What is needed is evidence from childrenwhose cognitive and social functioning is relatively intact butwhose communication skills are not. Finally, there is a shortsection reporting on children's attitudes toward their peerswho have speech problems. Although this information is notdirect evidence for low peer status, it does support the ideathat children's peers are sensitive to communication skillsand tend to regard deviations from normative expectations ina negative manner. In the three sections to follow, in theinterest of brevity a few studies are discussed that arerepresentative of larger bodies of work. Readers interested ina more comprehensive review of these topics can consultthese references for related studies.

Social Status and Discourse Skills in NormallyDeveloping Children

Hazen and Black (1989) investigated the relationshipbetween social status and discourse skills in normally devel-oping preschool children. Each child in the classroom re-ceived a Liked or Disliked score based on the combination ofpositive and negative nominations he or she received. Re-sults indicated that socially accepted children had betterskills than less accepted children for initiating, maintaining,and reinitiating coherent discourse across interaction con-texts. In a similar study, Black and Hazen (1990) studied theresponsiveness of liked and disliked preschoolers to familiarand unfamiliar peers. The results of this study indicated thatdisliked children made more irrelevant comments, were lessresponsive, and were less likely to clearly direct their initia-

tions to both familiar and unfamiliar peers as compared totheir liked counterparts.

Similarly, Place and Becker (1991) conducted a studyusing a rating scale methodology that examined the impact ofpragmatic skills on likability. Pragmatic skills refer to theappropriate use of language in social contexts. In this study,third and fourth grade girls were asked to listen to four taperecordings of a 10-year-old girl simulating an interactionepisode with a librarian. In each recorded scenario, the girlappropriately or inappropriately used four different pragmaticskills. The girl was judged as more likeable and describedmore positively when she displayed pragmatic competencethan when she displayed inappropriate pragmatic behaviorssuch as requesting inappropriately, interrupting, or failing tomaintain the logic of the conversation.

Communicative Competence and PopularityStatus in Children with Cognitive Deficits

Studies assessing communicative competence and popu-larity status have also been conducted with children withvarious handicapping conditions (Guralnick, 1981; Hemphill& Siperstein, 1990). These studies support the notion thatcommunication handicaps parallel limited socioverbal inter-actions and judgments of social immaturity.

For example, Guralnick (1981) studied the peer relations inmainstreamed playgroups of preschoolers with and withoutcognitive deficits. Results of this study indicated that thechildren with cognitive deficits communicated less effectivelyand were not as well accepted as their peers without suchdeficits. Not surprisingly then, it was found that the childrenwith cognitive limitations and the children without such limi-tations tended to form polarized social networks. Guralnick(1981) concluded that "handicapped children were perceivedas being of lower status and are treated accordingly" (p.287). Exactly why the children with cognitive deficits haddifficulty with social skills is unclear. The social interactiondifferences may be related to these children's cognitive orsocial skills deficits, physical appearance, and/or their par-ticular limitations with interactive speech and language skills.

Hemphill and Siperstein (1990) investigated the relation-ship between the conversational competence of children withmild cognitive limitations and their acceptance by regulareducation students. The regularly educated elementaryschool students were asked to watch a videotape in whichchildren with mild cognitive deficits displayed either skill orlack of skill at conversational management. The studentsresponded more favorably as measured by a descriptivequestionnaire and two bipolar scales when the children withcognitive deficits displayed skilled conversational manage-ment.

Attitudes Toward Children and Adults withSpeech and Language Impairment

Crowe Hall (1991) researched the social implications ofhaving even a mild articulation disorder in the elementaryyears. Her study involved using a semantic differential toassess the attitudes of fourth and sixth graders toward their

37 913-923 August 1994

Gertner et al.: Peer Preferences 915

videotaped fifth grade peers with (a) normal articulation, (b)articulation errors on /r/, and (c) articulation errors on /s/ and/z/. The results revealed that the normal speakers receivedmore positive attitude judgments than did the speakers witharticulatory errors. Thus even relatively mild articulationerrors can affect the popularity and peer acceptance ofelementary-age children.

The previous studies, then, document that discourse abil-ities are associated with peer status for young normallydeveloping children and, plausibly, for children with cognitivedeficits as well. Furthermore, relatively mild speech problemsinfluence the peer judgments of elementary-age children.Although these findings are consistent with the predictionthat children with speech and language impairments are atrisk for failure to develop peer friendships, they do not directlyestablish that connection. In this study, the relationshipbetween linguistic competence and social status was exam-ined for three groups of children enrolled in a mainstreamedpreschool classroom: children who are developing languagenormally (henceforth referred to as ND), children with speechand/or language impairments (S/LI) and those learning En-glish as a second language (ESL). Of particular interest werethe following questions: Are there differences among thethree groups of children in peer status? and Do measures oflanguage ability predict peer status, after controlling forchronological age and intelligence? It was predicted thatchildren with better speech and language skills (the NDgroup) would enjoy higher peer status than the children withlimited speech and language skills (the S/LI and ESLgroups), and that language measures would predict peerstatus.

Method

Subjects

The subjects were 31 children, 19 males and 12 females,who were enrolled in the Language Acquisition Preschool(LAP) located at the University of Kansas. Sixteen childrenwere enrolled in the morning class and 15 in the afternoonclass. None of the children had physical or visual handicaps.All children had normal hearing abilities as determined byregular hearing screenings conducted by the certified staffaudiologist. Hearing was screened at 20 dB HL (ANSI, 1970)at 250, 500, 1000, 2000 and 4000 Hz in both the right and leftears. All but two of the children demonstrated average orabove average intelligence as determined by the ProcessingComposite of the Kaufman Assessment Battery for Children(KABC) (Kaufman & Kaufman, 1983). The two exceptionswere sisters whom the clinical staff regarded as demonstrat-ing S/LI status. One sister scored 78 on the KaufmanProcessing Composite and the other scored 58 on the sametest. Because this study examined friendship patterns inpreschool classrooms, these children were included as sub-jects, even though their Kaufman scores were lower thanthose of the other children with speech and/or languagedisorders.

At the time of initial enrollment, children were placed in oneof three language status groups. Placement was determined

according to each child's performance on a battery of testsand descriptive language measures1: (a) children developinglanguage normally (ND), (b) children with speech and/orlanguage disorders (S/LI) (cf. Rice & Wilcox, in press), and(c) children learning English as a second language (ESL).The language measures in the test battery included thePeabody Picture Vocabulary Test-Revised (PPVT-R; Dunn &Dunn, 1981), the Reynell Developmental Language Scales-Revised (Reynell & Gruber, 1990), the Goldman-Fristoe Testof Articulation (GFTA; Goldman & Fristoe, 1986), and aspontaneous language sample used to obtain a mean lengthof utterance (MLU) and determine mastery of age-appropri-ate grammatical morphemes.

Children classified as normal-language models were re-quired to score within normal limits on all standardizedmeasures, possess an MLU within the predicted range forchronological age (Miller, 1981), and use age-appropriategrammatical morphemes (de Villiers & de Villiers, 1973).Upon initial enrollment in LAP, children in the S/LI group wererequired to meet two or more of the following criteria: (a)score one or more standard deviations below the mean onthe PPVT-R, (b) score one or more standard deviationsbelow the mean on the receptive and/or expressive portionsof the Reynell; (c) possess an MLU one or more standarddeviations below the mean for chronological age (Miller,1981), (d) lack mastery of at least two age-appropriategrammatical morphemes (de Villiers & de Villiers, 1973), and(e) score below the 16th percentile on the GFTA. In addition,children scoring below the 16th percentile on the GFTA withsignificantly limited intelligibility in conversational speechwere also enrolled, regardless of their language abilities.Only one child in this study, S/LI-12, was enrolled based onthis criterion.

Standardized measures of speech and language develop-ment obtained within 1 month of the sociometric data collec-tion are reported in Table 1. It is important to note thatbetween initial enrollment and the time of the sociometricdata collection, two children in the S/LI group had madeconsiderable gains in their speech and language abilities.The intelligibility of the child previously mentioned, S/LI-12,currently scored at the 15th percentile on the GFTA. Asecond child, S/LI-8, was now scoring within normal limits onall measures of language development. However, a numberof speech sound substitutions remained. He scored at the 4thpercentile on the GFTA, although these errors did not ad-versely affect his overall intelligibility. Despite the currentcommunicative status of these children, they were retained inthe S/LI group because of their history of communicativedifficulties. Whereas friendship patterns may be establishedwithin a period of a few months, it is possible that, onceestablished, these patterns may be relatively resistant tochange (cf. Denham & Holt, 1993). Thus, it is likely that thecommunicative histories of these children could influencetheir peer status.

'These classifications are for the purpose of LAP clinical documentationrecords only. The children were never grouped in this way in the classroom.There was no pull-out treatment, and no time when the children were groupedby language ability for classroom activities.

���--��--

916 Journal of Speech and Hearing Research 37 913-923 August 1994

TABLE 1. Subject descriptors.

ChildID Sex Age IQ1 PPVTb Rey-Rec c Rey-Expd GFTA ·

ND-1 F 65 115 122 111 118 86ND-2 M 63 96 106 116 99 70ND-3 M 64 105 123 111 93 59ND-4 F 61 108 116 99 92 88ND-5 F 64 112 106 106 121 99ND-6 F 47 92 93 83 83 67ND-7 M 68 101 94 94 94 78ND-8 F 43 102 123 123 109 67ND-9 M 58 94 96 106 108 99S/LI-1 M 58 104 81 64 64 7S/LI-2 M 54 89 104 78 76 12S/LI-3 F 66 58 57 65 65 1S/LI-4 F 47 78 83 63 63 9S/LI-5 M 50 101 94 63 93 23S/LI-6 M 46 86 98 71 75 16S/LI-7 M 70 92 122 125 64 1S/LI-8 M 60 120 119 108 101 4S/LI-9 M 60 90 97 64 64 4S/LI-10 M 57 93 99 99 76 9S/LI-11 F 47 95 101 73 91 11S/LI-12 M 57 136 120 103 99 15

ESL-1 M 64 122 104 96 93ESL-2 M 62 109 89 82 95ESL-3 F 58 96 81 63 64ESL-4 M 49 114 63 63 63ESL-5 F 51 115 98 82 88ESL-6 M 55 106 61 69 64ESL-7 M 57 110 72 84 85ESL-8 F 45 131 68 63 63ESL-9 F 45 102 77 78 102ESL-10 M 49 106 55 63 63

alQ as measured by the Kaufman Assessment Battery for Children (Kaufman & Kaufman, 1983).bPeabody Picture Vocabulary Test-Revised (Dunn & Dunn, 1981). Receptive portion of the ReynellDevelopmental Language Scales-US Edition (Reynell & Gruber, 1990). dExpressive portion of theReynell Developmental Language Scales-US Edition (Reynell & Gruber, 1990). eGoldman-FristoeTest of Articulation (Goldman & Fristoe, 1986).

The ESL children were generally drawn from families whohad recently moved to the U.S. for the purpose of highereducation. Typically, one or both parents were enrolled inadvanced graduate study. The children were selected to beas young as possible at the time of initial enrollment with noprevious exposure to the English language. There was oneexception to these enrollment criteria. A hearing child of deafparents, ESL-3, was enrolled in the preschool because of herunique oral language learning needs. Her first language wasAmerican Sign Language (ASL). The other children's nativelanguages included Chinese, Spanish, Farsi, Korean, andPortuguese. Normal first language acquisition was confirmedby parent report. Singletons or firstborn ESL children werepreferred so that school-age siblings did not affect languagelearning. All ESL children tested within the normal range ofintelligence on the KABC. Testing was conducted in thechildren's native language. There was considerable variabil-ity in English skills across the 10 ESL children, as a conse-quence of variable amount of time enrolled in the preschool.Five of the children could be described as demonstratingrelatively average receptive and expressive language abili-ties at the time of the study, that is, standard scores werewithin 1 1/2 standard deviations of the mean on both the

receptive and expressive portions of the Reynell (Receptiverange = 78 to 96; Expressive range = 85 to 102). In contrast,the other five ESL children had receptive and expressivelanguage skills two or more standard deviations below themean (Receptive range = 63 to 69; Expressive range = 63to 64). Importantly, a comparison of the children's raw scoreson these measures revealed the same division.

Of the 31 children, 9 were in the ND group, 10 were ESLchildren, and 12 were in the S/LI group. The children rangedin age from 43 to 70 months with mean ages (and standarddeviations) for each group as follows: ND, M = 59.2 (8.53);S/LI, M = 56.0 (7.60); ESL, M = 53.5 (6.74). There were nostatistically significant differences between the groups as afunction of age [F(2, 28) = 1.33, p > .05]. Both morning andafternoon classes contained approximately an equal numberfrom each of the three groups.

Procedures

Positive and negative nomination measures, followingBlack and Hazen (1990) and Hazen and Black (1989) wereadministered to both the morning and afternoon classes.

Gertner et al.: Peer Preferences 917

Each child (i.e., rater) was taken individually into a smallroom near the LAP classroom. To familiarize children withthe procedures of the test, they first participated in anorientation activity to acclimate them to the requirements ofeach sociometric task. The orientation procedures simulatedthe actual sociometric task. This familiarization process wasintended to ensure the validity of the child's responses.Several pictures of various types of food were laid out on thetable in the testing room. The child was asked to point to thekind of food he or she liked to eat. Subsequently, the selectedpicture was removed from the selection choices. This proce-dure was repeated two more times. Thus, children chosethree types of food they liked to eat. Next, without returningthe "liked" food pictures to the array, the child was asked topoint to three kinds of food he or she didn't like to eat.Similarly, the selected pictures were removed from theselection choices and the question repeated two more times.

After the orientation activity, the examiner displayed pho-tographs of all the children in the rater's classroom. The raterwas asked to name each of his/her classmates after theexaminer pointed to a child's picture. This refreshed therater's memory and ensured that he/she attended to all of thephotographs giving each classmate an equal chance of beingnominated. The aforementioned procedure was includedbecause, during the piloting phase of this study, it was foundthat not all of the children, particularly the children with S/LI,knew the names of all their classmates.

The examiner began the procedure by instructing the raterto "Point to the picture of who you like to play with in dramaticplay." "Dramatic play" is an activity in the LAP classroomwhere the children can participate in role play and imaginaryplay. During this activity, children are free to choose their playpartners and their conversational partners. This specificreferent was included to introduce the same social context toall child raters in order to reduce the possibility that variabilityin ratings would be attributable to variability in imaginedsocial contexts. After the child pointed to the picture, theexaminer removed that picture from the array. Then, theexaminer repeated the positive question two more times byasking the rater, "Who else do you like to play with indramatic play?" Thus, each rater selected a total of threeliked classmates.

Without returning the positively nominated pictures to thearray, the examiner continued the procedure by requesting,"Now, point to who you do not like to play with." After thechild pointed to the picture, the examiner removed thatpicture from the selection spread. The examiner repeated thenegative question two more times by asking the rater, "Whoelse don't you like to play with?" At the end of the procedure,a grand total of six pictures had been removed from thearray. Consequently, it was impossible for a child to benominated both positively and negatively. The child's positiveand negative responses were documented on a summarysheet.

ResultsThe first set of analyses addressed the question of whether

or not there was a difference between the three groups of

TABLE 2. Group means for the soclometric measures: Positivenomination, negative nomination.

Soclometricmeasure ND S/LI ESL

Positive nominationM 4.22ab 2.33a 2.60 bSD 1.86 1.50 1.08

Negative nominationM 1.78 3.58 3.40SD 1.20 2.11 2.88

Note. Like letters denote significant differences, p < .05.

children on any of the sociometric measures. The sociomet-ric tasks yielded two peer nomination measures based on thetotal number of nominations received. That is, each nomina-tion was scored with a numerical rating of "1." The nomina-tions were not weighted according to the order in which thechildren were chosen. The total number of positive nomina-tions and the total number of negative nominations wereeach summed yielding two summary measures, POSNOMand NEGNOM, respectively. Group means were computedby summing the total number of nominations each childreceived and dividing by the number of children in eachlanguage group. For example, the POSNOM mean for thechildren in the ND group was 4.22. This indicated that, onaverage, each ND child was positively nominated by approx-imately 4 of his or her 14 or 15 classmates.

Two one-way univariate analyses of variance (ANOVA)were conducted to test for differences between the threegroup means (ND, S/LI, ESL) on each sociometric measure.The group means and standard deviations for each measureare reported in Table 2. The POSNOM means for thechildren in the ND, S/LI, and ESL groups were 4.22, 2.33,and 2.60, respectively. The NEGNOM means for the afore-mentioned groups of children were 1.78, 3.58, and 3.40.Thus, a mean score of 3.40 indicates that each child in theESL group was selected for negative nominations an aver-age of 3.40 times. The ANOVAs revealed that group differ-ences were statistically significant for the positive nominationmeasure only [F(2,28) = 4.55, p < .05]. Post-hoc Schefftpairwise comparisons revealed that the ND children receivedmore positive nominations than either the children in the ESLor S/LI groups. Group differences for the negative nominationwere not significant, although differences in the mean scoreswere in the predicted direction.

To further examine the relationship between languageabilities and peer preferences, the two children with morelimited intellectual abilities (i.e., S/LI-3 and S/LI-4) and thetwo S/LI children with age-appropriate language abilities (i.e.,S/LI-8 and S/LI-12) were removed and the analysis wasrepeated. It should be noted that although the nominationsthese children received were removed from the data, thenominations they provided for their peers could not bedropped. When these children were eliminated, the meanpositive nomination measure for the S/LI group was reduced(M = 2.00, SD = 1.07) and the mean negative nominationwas increased (M = 4.13, SD = 2.42). Although this strategyprovides stronger evidence for the relationship betweencommunication abilities and peer status, it comes at the

918 Journal of Speech and Hearing Research37 913-923 August 1994

Sociometric Status Groups

r--------------J. Liked

I D I

I ® I

IIa T

A A V U -

[]

~- -O l |- b

0 ND The number nrade ead

ESL ividual ditd (d. Table 1)

I mr-

7------------- IIMixedIIIIIIII

WIUsrW

1Z

0rcnZfa~

oI u

VW

(b~~~~~~~~~~~~~~

O~~~~~~~~~~~~

4n 1ke--~~~~~~--t--~~~~~~~~~~~-~~~~---J~~~~~~

I-1.25 &.

-2.5 T1Low mpact

L - - - ;.-- -4 - - -41 - 4-. -I . . I

-2.4 -1.8 -1.2 -06 0 0.6 1.2 1.8 2.4

Negative Nomination Z-Score

FIGURE 1. Soclomric status groups.

expense of decreasing our S/1-1 sample by 33%. Thus, wepreferred to retain these children in the remaining analyses,acknowledging that a history of communication limitations aswell as cognitive limitations may also influence current levelsof peer status in the classroom.

Following the strategy of Hazen and Black (1989), it isinformative to reconfigure the nomination measures accordingto a 2 x 2 classification system wherein the children are placedin one of four social status groups (see Figure 1). To do this, thechildren's raw scores were converted to z-scores for eachmeasure, following the procedures of Black and Hazen (1 990).The four status groups can be regarded, conceptually, aschildren whose status is Liked, Disliked, of Low Impact, or of aMixed profile. The Liked group is composed of those childrenwho received a POSNOM z-score greater than zero and aNEGNOM z-score less than zero. Conversely, the children

placed in the Disliked group received a POSNOM z-score lessthan zero and a NEGNOM z-score greater than zero. Thechildren found in the Low Impact group are those whosePOSNOM and NEGNOM z-scores were both less than zero.Hence, Low Impact children can be thought of as those who arenot conspicuously liked, but are not obviously disliked either.Accordingly then, these children are often characteristic of amore "neglected" or "overlooked" group. And lastly, thechildren whose POSNOM and NEGNOM z-scores were bothgreater than zero were assigned to the Mixed status group.Mixed children are those who are more controversial innature, meaning they may receive a high number of bothpositive and negative nominations. Thus, these children tendto acquire polarized degrees of social stature. In other words,they are usually either clearly liked by some peers orblatantly disliked by others.

Gertner et al.: Peer Preferences 919

The distribution among the four social status groups re-vealed that the ND group had the highest representation inthe Liked Cell. In fact, five of the six ND children received apositive nomination score that was significantly above theclassroom averages (i.e., POSNOM Z-score > 1.00). More-over, those children who were not in the Liked cell all fellwithin the average range of positive nominations. That is,none of the children in the ND group had POSNOM Z-scoresbelow -0.59. In contrast, only one child in the S/LI group wasclearly placed in the Liked cell. This child, S/LI-8, had thestrongest expressive language scores in the S/LI languagegroup (refer to Table 1). Ten of the 12 children in the S/LIgroup and 8 of the 10 children in the ESL group received anaverage number of positive nominations (i.e., POSNOMZ-score between -1.00 and 1.00). Thus, the majority of thechildren in these groups were distributed in a band along thehorizontal axis with relatively neutral negative nominationZ-scores as well. Only five children fell below the averagerange of peer popularity. Two S/LI children, S/LI-1 andS/LI-11, were clearly Disliked, receiving significantly fewerpositive nominations and more negative nominations thantheir peers. The third child, S/LI-3, was nominated positivelyby only one peer; however, no one identified her as a childpeers disliked playing with. The two ESL children whoreceived significantly fewer positive nominations, ESL-4 andESL-6, both were among the ESL subgroup with limitedEnglish proficiency, who scored two or more standard devi-ations below the mean on all English standardized tests.

The second research question addressed the notion thatthere are, perhaps, factors in addition to a child's languagecompetency that contribute to social status. It was postulatedthat his or her age or IQ could also play a role. Within thissample, the children's ages ranged from 43 to 70 months.Within such a preschool group, it might be expected thatolder children would have an advantage for social status. Ina similar fashion, a relatively wide range of intelligence levelswas evident in the sample, with scores ranging from 58 to136. Although there is not a clear prediction to be drawn fromthe normative literature, it is plausible that children withhigher IQs may have a social advantage. To investigate therelationship between the children's sociometric data and theirchronological age, IQ, and language abilities, the LAP data-base containing the age and all the test scores for each childwas used. The language tests of interest were the PPVT-R,the receptive (REY-REC) and expressive (REY-EXP) lan-guage portions of the Reynell Developmental LanguageScales-Revised, the GFTA and the processing composite(KABC-PRC) of the Kaufman Assessment Battery for Chil-dren, which is regarded by the test developers to be thepreferred single measure of intelligence for this test(Kaufman & Kaufman, 1983). This battery of tests yieldsstandardized measures of receptive and expressive lan-guage abilities, articulation skills, and intelligence.

To investigate the relationship between the aforemen-tioned variables and peer preference, the correlations be-tween the POSNOM and NEGNOM scores with age, IQ, andthe individual speech and language scores were examined(see Table 3). Results of the correlational analyses revealedthat POSNOM was moderately positively correlated with age,PPVT-R, REY-REC, REY-EXP, and GFTA (range r = .44 to

TABLE 3. Correlations between: Soclometric measures, age,test scores.

POSNOM NEGNOM

Age and IQAge .47** .00KABC-PRC .24 -. 05Standardized speech and

language measuresPPVT-R .49** -. 17REY-REC .55*** -. 29REY-EXP .48** -. 22GFTAa .44* -44*

Note. aGFTA scores for the ESL children were not available. Thus,correlations with GFTA scores were computed without the ESLchildren (n = 21). Significant levels for one-tailed correlations: ***p <.001, **p < .01, *p < .05.

r = .55, p < .05). NEGNOM was moderately negativelycorrelated with GFTA (r = -. 44, p < .05). It is important tonote that GFTA scores were not available for the ESLchildren. Therefore, the GFTA correlation was computed withthe ND and S/LI groups only. In summary, the children'spositive nomination total was moderately correlated withchronological age, measures of receptive and expressivelanguage skills, and articulation ability. The negative nomi-nation total was correlated with children's articulation abilityonly.

A moderate negative intercorrelation was also apparentbetween POSNOM and NEGNOM (r = -.50, p < .01). Thissuggests that these two measures tapped into the sameconstruct as they allowed the children to reflect on thecumulative effect of both the positive and negative experi-ences the children had had with their classmates throughoutthe year.

Multiple regression analyses were conducted to determinethe relative contribution of age, IQ, and language skills inpredicting the observed variance in POSNOM scores. Be-cause none of the variables were significant predictors ofNEGTOT, regression analyses were not conducted with thatvariable. The variables entered in the analyses were age,KABC-PRC, PPVT-R, REY-REC, and REY-EXP. Rawscores were used for the language measures to adjust for thelanguage differences among children of varying ages withsimilar standard scores. The GFTA scores were not enteredbecause they were not available for the ESL children andwithout these children the total number of subjects wouldhave been unacceptably low.

In an initial analysis utilizing stepwise variable selection,PPVT-R was the single best predictor of positive nominations[R2 = .35, F(1, 29) = 15.36, p < .001; see Table 4]. Tofurther evaluate the independent contributions of languageability to social status, when controlling for age and IQ, twoadditional regression analyses were conducted, alternatingthe order of forced entry of the predictor variables. When ageand IQ were entered simultaneously, they accounted for 25%of the variance in total positive nominations [R2 = .25, F(2,27) = 4.74, p < .05]. However, two language measuresaccounted for unique variance in the positive nominationmeasure. PPVT-R and Reynell-Expressive accounted for anadditional 12% and 9% of the variance, respectively, when

920 Journal of Speech and Hearing Research

TABLE 4. Multiple regression equations: Variance accounted for In positive nominations.

Positive nominations

Steps in regression R2 R2 change F-value Adjusted R2

Stepwise variable selection1. PPVT-R .35 15.36*** .32

Forced variable entry1. Age/lO .25 4.74* .202. PPVT-R .37 .12* .303. Reynell-Receptive .37 .00 .284. Reynell-Expressive .41 .03 .29

2. Reynell-Receptive .30 .05 .223. Reynell-Expressive .35 .05 .254. PPVT-R .41 .06 .29

2. Reynell-Expressive .35 .09t .273. PPVT-R .38 .04 .294. Reynell-Receptive .41 .02 .29

Forced variable entry1. PPVT-R/Rey-Rec/Rey-Exp .38 5.58** .312. Age .39 .01 .303. IQ .41 .01 .29

2. Q10 .39 .01 .303. Age .41 .01 .29

Note: ***p < .001, **p < .01, *p < .05, tp = .06.

entered on Step 2 (or 10% and 7% of the variance afteradjusting for the degrees of freedom; see Table 4). It isimportant to note that the variance in these two measuresoverlapped considerably, in that once one of these variableswas entered, the remaining language measures did notcontribute any unique variance to the explanation of thepositive nomination measure. On the other hand, when thethree language measures were entered simultaneously, theyaccounted for 38% of the variance in positive nominations[R2 = .38, F(3, 27) = 5.58, p < .01]. Age and IQ did notprovide a significant unique contribution to the variance inpositive nominations beyond that contributed by the lan-guage measures. Considered as a set, the regression anal-yses indicated that, in this small sample, language abilitycontributed significantly to the observed variability in thechildren's social status, and did so independent of age or IQlevel. Although these findings are congruent with the line ofinterpretation explored here, it must also be noted that asingle sample regression analysis, especially with a smallsample size, needs to be supported with other samplesbefore one can conclude that the general model holds.

Discussion

The results of this investigation indicate that limited lan-guage ability is associated with lower levels of social accept-ance among peers. The children with language limitationswere the least likely to be identified as preferred peerplaymates, at least when the verbally demanding activity ofdramatic play was used as the context to elicit peer nomina-tions. Furthermore, in this sample of children, languageability was a better predictor of peer status than age or

intelligence. These findings suggest that communicationabilities must be carefully considered in any attempt toexplicate the formation of peer relationships. Children withcommunication limitations are less well equipped to uselanguage to establish and maintain friendships in earlychildhood than are children with normally developing lan-guage.

The mean scores on the positive nomination measureindicated that the children in the ND group were liked morethan the children in the ESL or S/LI groups. Furthermore,only three children, all with normally developing languageabilities, achieved above-average popularity as indicated bya clear placement in the Liked cell of Black and Hazen's(1990) social status chart above one standard deviation onthe positive dimension and below one standard deviation onthe negative dimension. In contrast, only three childrenappeared to be overtly disliked by their classmates, two fromthe S/LI group and one from the ESL group.

In the group comparisons of this study, it is possible toconsider whether or not subtle social deficits of a group ofchildren contribute to lower peer status. For example, if theS/LI children were somehow less motivated to interact withpeers or less able to negotiate peer group entry because oflimitations in social competence (cf. Craig & Washington,1993), they could conceivably score lower than the ESLgroup. Thus, the ESL children provide an important compar-ison group in this regard. These children are normal lan-guage learners with average to above-average intelligence,whose parents are highly educated and encouraging of theirchildren's development. Although these youngsters' back-grounds are ethnically diverse, there is no reason to believethat they have intrinsic deficits in social competence, such as

37 913-923 August 1994

Gertner et al.: Peer Preferences 921

low motivation for peer interaction, that would contribute totheir low peer status. Instead, the equivalent peer status ofthe ESL and S/LI groups implicates the shared characteristic,that of limited communicative competence. Ethnic influencesare also ruled out as key determinants of low status in thissetting, in that the S/LI group did not demonstrate the ethnicdiversity of the ESL group. Further implication of the role ofcommunicative competence is noted if the English profi-ciency of the ESL group is taken into consideration. If theESL children are divided into subgroups on the basis ofEnglish proficiency, the mean scores on the positive nomi-nation are clearly related to English proficiency. The meanrating for the five children in the High English proficiencygroup was 3.00 (cf. ND group mean = 4.22) and for the Lowproficiency group, 2.20 (cf. S/LI group mean = 2.33). In thesetting observed in this study, and with this sample of ESLchildren, communication limitations seem to be the commonfactor for low peer status.

It is important to note that this line of interpretation doespresuppose a certain distinction between social and linguisticcompetence. This can be confusing with regard to thecultural influences that are also inextricably operative in thecase of children from different cultures. They do, of course,bring their cultural expectations with them, and many of thoseexpectations include conventions for verbal interaction, suchas the timing and form of initiations. In this sense, then,sociocultural and linguistic competencies are intimately inter-related, and likely to vary across cultural groups. Recall thatin LAP, the children have diverse cultural backgrounds, soone could expect diversity among the ESL children's peerstatus, to the extent that culturally driven factors exert astrong influence on peer status. On the other hand, atanother level of social cognition, there are similarities acrosschildren that seem to be less tightly tied to verbal or,arguably, cultural conventions. These include such basicthings as an interest in playing and interacting with otherchildren, understanding the daily schedule of activities in apreschool classroom, knowing the local rules for sharing toysand carrying out play activities, understanding the communi-cative intent underlying other children's verbal behavior, andappreciating the fact that verbal interactions are operative formeeting social needs, such as negotiating for toy objects ororganizing favorite activities.

In the case of S/LI children, their lack of verbal interactionswith peers could seem to be related to a lack of appreciationfor such fundamental social insights. This level of basic socialcognition is not, however, thought to be unavailable to ESLchildren (although they do certainly have to learn the localclassroom "culture" for carrying out social interactions).What we believe is that both groups of children have thefundamental levels of social cognition necessary for thesocial underpinnings of peer relationships, and that thegroups are also similar in their limited control of the formallinguistic system of English. The two groups are also alike intheir low social status among their peers. This, to our way ofthinking, points in the direction of verbal ability as a mediatorof peer status. It is, at the same time, impossible to rule outthe nonparsimonious possibility that the low peer status ofthe two groups derives from different sources, such that theESL children experienced social difficulties because of cul-

TABLE 5. Number of mutual friends by group.

None 1 2

ND 1 3 5ESL 2 4 4S/LI 4 7 1

turally mediated adjustments in their interactions with peers,adjustments that operated in a way that had a similar impactirrespective of the diversity in cultural backgrounds.

What we wish to emphasize here is that there is value indifferentiating among fundamental levels of children's socialcognition that are necessary for the foundation of peerrelationships, levels of acquisition of the formal linguisticcodes that are sufficient for the development of verbalinteractions among peers, and an awareness of the socio-cultural conventions that govern the operation of the first twolevels. There is no doubt that all three are closely associatedas children develop their peer relationships. What we wish tohighlight is the potentially important role of the linguisticsystem, and the need to identify ways in which children'slinguistic competencies influence their social development.

Another way to approach the findings is with regard tomutual friendships. As noted in the introduction, children'sfriendships serve as important incubators for the develop-ment of social skills. In order to learn about the mutualfriendships evident in these groups of children, the data wereinspected for the pairs of children who nominated each otheras friends. The number of mutual friends by group is reportedin Table 5. It is clear that children with speech/languageimpairments are more likely to lack reciprocal friendshipsthan children in the other groups. Only one child in the S/LIgroup had more than one mutual friend, and in this case thefamilies of the children were known to socialize togetheroutside of class. On the other hand, four of the S/LI childrenwere without any mutual friendships.

Among the children in the S/LI group, general receptivelanguage appeared to be a discriminating factor between theindividuals who fared well in regard to social acceptance andthose who did not. Despite speech and expressive languagelimitations, children with age-appropriate receptive skills re-ceived fewer negative nominations than average (NegativeZ-score < 0). On the other hand, the children with significantreceptive language deficits (see Table 1) received more (orthe average number of) negative nominations, with only oneexception (i.e., S/LI-5). Thus, the profile of speech/languageimpairment seemed to play a role in peer relationships.

The finding that receptive language abilities differentiatedamong children within the S/LI group accords well with otherstudies that have examined social-conversational skills inthis population. Craig and Evans (1989) identified differencesin appropriate turn exchange behaviors for children withdiffering receptive and expressive profiles. Similarly, Craigand Washington (1993) found receptive language abilities tobe a critical factor influencing children's success in joining theongoing play of two peers. In their study, three of the fivechildren with specific language impairment did not gainaccess into the peers' play. Receptive language deficitsdifferentiated between those children who achieved access

922 Journal of Speech and Hearing Research

and those who did not. One reason for the lower social statusevident among the children with limited receptive abilities inthis study may be that they had difficulty in joining theongoing group play in the LAP classroom.

The fact that the children who had speech impairmentsonly were found on the border between the Liked and LowImpact cells and not in the Disliked cell does not aligncompletely with the literature suggesting that children withspeech impairments are less popular than their peers whohave normal articulation skills (Crowe Hall, 1991; Madison,1992; Silverman & Falk, 1992). The difference in findingsmay be attributable to age differences in the samples ob-served, in that the children were of school age and older inthe studies that found a negative social status for speechimpairments. This might suggest that school-age and olderchildren with speech impairments may encounter more neg-ative attitudes from their peers than is the case for children ofpreschool age. One explanation for these differences may bethat many preschool-age children have not yet mastered anadult-like phonological system. Thus, children's developmen-tally appropriate speech sound substitutions (and even moreexcessive use of phonological simplifications in the case ofS/LI-7), did not influence the negative nominations. In thisstudy, as long as receptive language skills were intact,children with both mild and severe articulation problems wereliked by their peers.

There is one interesting finding that suggests that peer statusmay be affected by some fairly discrete consequences oflimited language skill. It was noted during data collection thatsome of the children were unable to name a few or several oftheir peers even though data were collected toward the end ofthe school year. In particular, those children with speech and/orlanguage impairments were less likely to know the names oftheir peers. The more unfamiliar names of the ESL children andthe names of the quiet, more passive children seemed to be thenames that were most difficult for these children to remember.But the fact that a child did not know a peer's name did notmean he or she would not pick the peer as someone the childliked, or even did not like, to play with. What is suggested is thatthe inability to address someone by name could have implica-tions for being liked by one's peers. Children who can useproper names have a better chance of establishing joint atten-tion and interpersonal focus than those children who have only"hey, you" or other awkward social devices at their disposal.Such small social gestures could influence children's likability.

Although it is the argument of this study that languagelimitations can significantly and negatively influence a child'spopularity rating, and do so to some interesting extentindependent of age, it is also possible that factors notmeasured in this study may have a profound effect on howwell liked a child is by his or her peers. For instance,personality traits surely play a role in a child's popularity.Denham and Holt (1993) report that for preschool childrenthe qualities of friendliness, cooperativeness, and tractabilityare associated with likability, whereas aggressiveness is not.In the subjects who participated in this study, some werenoteworthy for qualities such as compliance, dominance,aggressiveness, and creativity. There is some reason tospeculate that these qualities may account for why someindividual children did not conform to group trends in their

peers' ratings. What is of the most interest is whether or notsuch personal qualities interact in systematic ways withcommunicative competence and, if so, how. The data col-lected here cannot adequately address this question, and itremains for future studies to do so. The unresolved keyquestion is if such an interaction holds, how it might play outfor children's peer relationships.

Collectively, these findings support the idea that verbalabilities contribute to peer acceptance. Children with lan-guage limitations are indeed at risk for the negative conse-quences of unpopularity. For example, children who are notaccepted by their peers often do not develop sufficientamounts of self-esteem. Children's perception of their lowsocial status may also contribute to future adjustments inpatterns of conversational interaction. Children who recog-nize that some children don't particularly like them may bemore reticent to approach new peers than those children whomake friends easily. This reluctance could have seriousdrawbacks during times of change such as the transition intokindergarten. And finally, a child's peers are not the onlygroup of people to be swayed by evidence of the child'slanguage limitations. Adults, including teachers, may per-ceive that children with language difficulties are less sociallymature and less intellectually competent (Rice, Hadley, &Alexander, 1993). If adults inappropriately interpret thesource of children's language limitations, they may not en-courage these children to achieve academically. This inevi-tably limits their options for employment and influences thesocial circles they will participate in later in life.

Therefore, it is understandable why the preschool yearsare a very important time to begin facilitating social friend-ships among children. If children are not liked by their peers,it will be difficult for them to establish and maintain friend-ships. One factor contributing to limited peer acceptance islimited language competency. To the extent that this factor isoperative, it is likely that children with language limitationswill not have a circle of friends with whom to experiencesocioverbal interactions. This may confound their languageproblem as they will not have social partners with whom toconverse, to practice interactive verbal skills, or to listen to.This will make it even more difficult for their language skills todevelop and thrive. Thus, it is essential that preschoolteachers and language interventionists become sensitized tothe potential social consequences of language limitations,and develop intervention strategies designed to facilitatechildren's conversational skills in the context of peer-relatedsocial tasks.

Acknowledgments

Data collection was completed in partial fulfillment of the require-ments for the degree of Master of Arts completed by the first authorand was supported by the Department of Education, Office of SpecialEducation under award #HO24U80001 to the Kansas Early Child-hood Research Institute for which Rice is principal investigator. Weappreciate the participation of the children and their parents in theLanguage Acquisition Preschool and express special thanks to ourcolleagues Betty Bunce and Kim Wilcox for their helpful suggestions.Portions of this study were presented at the Biennial Meeting of theSociety for Research in Child Development, New Orleans, LA, March1993.

37 913-923 August 1994

Gertner et al.: Peer Preferences 923

References

American National Standards Institute (1970). Specifications foraudiometers (ANSI S3.6-1969). New York: ANSI.

Black, B., & Hazen, N. L. (1990). Social status and patterns ofcommunication in acquainted and unacquainted preschool chil-dren. Developmental Psychology, 26, 379-387.

Brlnton, B., & Fujikl, M. (1993). Language, social skills, andsocioemotional behavior. Language, Speech, and Hearing Serv-ices in Schools, 24, 194-198.

Craig, H. K. (1993). Social skills of children with specific languageimpairment: Peer relationships. Language, Speech, and HearingServices in Schools, 24, 206-215.

Craig, H. K., & Evans, J. L (1989). Turn exchange characteristicsof SLI children's simultaneous and nonsimultaneous speech.Journal of Speech and Hearing Disorders, 54, 334-346.

Craig, H. K., & Washington, J. A. (1993). Access behaviors ofchildren with specific language impairment. Journal of Speech andHearing Research, 36, 322-337.

Crowe Hall, B.J. (1991). Attitudes of fourth and sixth graderstoward peers with mild articulation disorders. Language, Speech,and Hearing Services in Schools, 22, 334-340.

Denham, S.A., & Holt, R. W. (1993). Preschoolers' likability ascause or consequence of their social behavior. DevelopmentalPsychology, 29, 271-275.

de Villiers, J. G., & de Villiers, P. A. (1973). A cross-sectional studyof the acquisition of grammatical morphemes. Journal of Psycho-linguistic Research, 2, 267-278.

Dunn, L. M., & Dunn, L M. (1981). Peabody Picture VocabularyTest-Revised. Circle Pines, MN: American Guidance Service.

Gallagher, T. (1991). Language and social skills: Implications forassessment and intervention with school-aged children. In T. M.Gallagher (Ed.), Pragmatics of language: Clinical practice issues(pp. 11-41). San Diego, CA: Singular Publishing Group, Inc.

Gallagher, T. (1993). Language skill and the development of socialcompetence in school-age children. Language, Speech, and Hear-ing Services in Schools, 24, 199-205.

Goldman, R., & Frlstoe, M. (1986). The Goldman-Fristoe Test ofArticulation. Circle Pines, MN: American Guidance Service.

Guralnick, M. J. (1981). The social behavior of preschool children ofdifferent developmental levels: Effects of group composition. Jour-nal of Experimental Child Psychology, 31, 115-130.

Hadley, P. A., & Rice, M. L (1991). Conversational responsivenessof speech- and language-impaired preschoolers. Journal ofSpeech and Hearing Research, 34, 1308-1317.

Hazen, N. L, & Black, B. (1989). Preschool peer communicationskills: The role of social status and interaction context. ChildDevelopment, 60, 867-876.

Hemphill, L., & Sipersteln, G. (1990). Conversational competenceand peer response to mildly retarded children. Journal of Educa-tional Psychology, 82, 128-134.

Howes, C. (1983). Patterns of friendship. Child Development, 54,1041-1053.

Howes, C. (1988). Peer interaction of young children. Monographsof the Society for Research in Child Development, 53 (1, Serial No.217).

Kaufman, A. S., & Kaufman, N. L (1983). Kaufman AssessmentBattery for Children. Circle Pines, MN: American Guidance Ser-vice.

Madison, C. (1992). Attitudes toward mild misarticulation-disorderedpeers. Language, Speech, and Hearing Services in Schools, 23,188.

Miller, J. (1981). Assessing language production in children. Balti-more, MD: University Park Press.

Place, K. S., & Becker, J.A. (1991). The influence of pragmaticcompetence on the likability of grade-school children. DiscourseProcess, 14, 227-243.

Reynell, J. K., & Gruber, C. P. (1990). Reynell DevelopmentalLanguage Scales - U.S. Edition. Los Angeles, CA: WesternPsychological Services.

Rice, M. L. (1993). Don't talk to him; he's weird: A social conse-quences account of language and social interactions. In A. P.Kaiser & D. B. Gray (Eds.), Enhancing children's communication:Research foundations for intervention (pp. 139-158). Baltimore,MD: Brookes Publishing Company.

Rice, M. L., Hadley, P. A., & Alexander, A. L. (1993). Social biasestoward children with specific language impairment: A correlativecausal model of language limitations. Applied Psycholinguistics,14, 443-471.

Rice, M. L., Sell, M. A., & Hadley, P. A. (1991). Social interactionsof speech- and language-impaired children. Journal of Speechand Hearing Research, 34, 1299-1307.

Rice, M. L, & Wilcox, K. A. (Eds.) (in press). Language AcquisitionPreschool. Baltimore, MD: Brookes Publishing Co.

Silverman, F., & Falk, S. (1992). Attitudes of teenagers towardpeers who have a single articulation error. Language, Speech, andHearing Services in Schools, 23, 187.

Received August 2, 1993Accepted February 24, 1994

Contact author: Mabel L. Rice, Child Language Program, Univer-sity of Kansas, 1082 Dole Center, Lawrence, KS 66045. e-mail:Mabel@ukanvm.bitnet