Statistic Rough Draft (2 pages)

JSLHR

Article

Effects of Sampling Context on Spontaneous Expressive Language in Males With Fragile

X Syndrome or Down Syndrome Sara T. Kover,a Andrea McDuffie,b Leonard Abbeduto,b and W. Ted Brownc

Purpose: In this study, the authors examined the impact of sampling context on multiple aspects of expressive language in male participants with fragile X syndrome in comparison to male participants with Down syndrome or typical development. Method: Participants with fragile X syndrome (n = 27), ages 10–17 years, were matched groupwise on nonverbal mental age to adolescents with Down syndrome (n = 15) and typically developing 3- to 6-year-olds (n = 15). Language sampling contexts were an interview-style conversation and narration of a wordless book, with scripted examiner behavior. Language was assessed in terms of amount of talk, mean length of communication unit (MLCU), lexical diversity, fluency, and intelligibility. Results: Participants with fragile X syndrome had lower MLCU and lexical diversity than did participants with typical development.

Participants with Down syndrome produced yet lower MLCU. A differential effect of context among those with fragile X syndrome, Down syndrome, and typical development emerged for the number of attempts per minute, MLCU, and fluency. For participants with fragile X syndrome, autism symptom severity related to the number of utterances produced in conversation. Aspects of examiner behavior related to participant performance. Conclusion: Sampling context characteristics should be considered when assessing expressive language in individuals with neurodevelopmental disabilities.

Key Words: language sampling, conversation, narrative, fragile X syndrome, mean length of utterance (MLU)

O f the methods used to assess expressive languagein children and adolescents with intellectual dis-abilities, standardized tests and spontaneous language samples are the most often used (Abbeduto, Kover, & McDuffie, 2012). Although scores on standard- ized assessments and language samples tend to be corre- lated, each provides unique information (Condouris, Meyer, & Tager-Flusberg, 2003; Ukrainetz & Blomquist, 2002). Standardized measures of expressive language offer a relatively quick evaluation of performance relative

to age expectations. However, most standardized assess- ments yield a single summary score for expressive language ability, which precludes the possibility of identi- fying patterns of relative strength or weakness across domains (e.g., vocabulary, syntax), and can mask clinically meaningful differences among individuals. When assess- ing individuals with intellectual disabilities, standard- ized language tasks are also prone to floor effects (Mervis & Robinson, 2005). Spontaneous language samples avoid these limitations by providing contextualized data on spe- cific aspects of ability and, in this way, are well suited to establishing expressive language profiles (Westerveld, Gillon, & Miller, 2004). The present study was designed to further explore the utility of language-sampling proce- dures for individuals with intellectual disabilities.

In using language samples to characterize any population with disabilities, it is important to recognize that the nature of the language-sampling context can af- fect the language produced by the child and, consequently, the conclusions reached by the researcher or clinician. In this study, we focused on the ways in which sam- pling context impacts the characterization of expressive

aWaisman Center, University of Wisconsin—Madison bM.I.N.D. Institute, University of California, Davis cNew York State Institute for Basic Research in Developmental Disabilities, Staten Island

Correspondence to Sara T. Kover: kover@wisc.edu

Editor: Janna Oetting Associate Editor: Elizabeth Crais

Received March 28, 2011 Revision received September 6, 2011 Accepted November 22, 2011 DOI: 10.1044/1092-4388(2011/11-0075)

Journal of Speech, Language, and Hearing Research • Vol. 55 • 1022–1038 • August 2012 • D American Speech-Language-Hearing Association1022

language in individuals with specific neurodevelopmen- tal disabilities associated with intellectual disability. In particular, we assessed language in two contexts, conversation and narration, in male adolescents with fragile X syndrome, male adolescents with Down syn- drome, and boys with typical development at similar levels of nonverbal cognitive development.

Contexts for Sampling Spontaneous Language

Research on children with typical development or language impairment suggests that sampling contexts vary in the extent to which they elicit the upper bound of an individual’s linguistic ability (Southwood & Russell, 2004; Thordardottir, 2008; Wagner, Nettelbladt, Sahlen, & Nilholm, 2000). Interview-style conversations, for instance, elicit more utterances and utterances with higher mean length of utterance (MLU) relative to free-play contexts (Evans & Craig, 1992; Southwood & Russell, 2004); however, narrative contexts may be opti- mal for assessing syntactic ability because they elicit longer and more complex sentences than either conver- sation or free play, on average (MacLachlan & Chapman, 1988; Westerveld et al., 2004). In terms of disabilities, high-functioning children with autism have been shown to demonstrate a weakness in complex syntax rel- ative to typically developing children during personal narratives but not when telling a story from a wordless book, demonstrating differences between contexts, even for children in the normal range of cognitive functioning (Losh & Capps, 2003).

Context effects also have been described for individ- uals with intellectual disabilities. Abbeduto, Benson, Short, and Dolish (1995), for example, examined the language of children and adolescents with intellec- tual disability of unspecified etiology and found more communication attempts per minute in a conversation than in narration, whereas language produced in narra- tion had a higher mean length of communication unit (MLCU) than in conversation. Abbeduto et al. suggested that narration more fully engages the syntactic abilities of individuals with intellectual disability because narra- tive revolves around content more likely to be encoded in multiclause utterances (e.g., cause and effect, psycholog- ical states). Likewise, the visual scaffolding provided by a wordless storybook in a narrative context can free up cognitive resources, such as working memory, which might also support the production of more complex sen- tences (Miles, Chapman, & Sindberg, 2006). Thus, vari- ation in the characteristics of the sampling context can affect conclusions reached about an individual with in- tellectual disability. From a clinical perspective, such findings suggest the need for multiple sampling contexts and to understand the advantages and disadvantages of

each. In this study, we were interested in the possible differential effects of context on the language of individ- uals with intellectual disabilities.

Consistency Within Language- Sampling Contexts

Difficulties with the interpretation of individual or diagnostic group differences in expressive language profiles can arise from inadequate consistency of proce- dures within a given context. In particular, performance can be affected by variations in materials, partners, and experimenter behavior (Dollaghan, Campbell, & Tomlin, 1990; Hansson, Nettelbladt, & Nilholm, 2000). In con- versation, for example, the amount of talk and rate of questioning by a partner can influence both the child’s amount of talk and MLU (Cowan, Weber, Hoddinott, & Klein, 1967; Evans & Craig, 1992; Johnston, 2001; Johnston, Miller, Curtiss, & Tallal, 1993). Thus, failure to consistently structure the sampling context and script examiner behavior across participants could result in uninterpretable differences across individuals, diag- nostic groups, and studies because differences in lan- guage ability cannot be distinguished from variations attributable to the sampling context. In the present study, we created procedures and scripts for consis- tently structuring critical aspects of our two language- sampling contexts, thereby allowing unambiguous interpretation of differences in performance across diag- nostic groups.

Although it is important to structure and script the language-sampling context, this “standardization” must be flexible enough to allow the examiner to adapt his or her behavior to the skill level and interaction style of the child. Consequently, even within the constraints of a structured procedure, variation in the amount and com- plexity of examiner talk can emerge as the examiner attempts to engage the individual in the language sam- pling task. For example, an examiner’s syntactic complex- ity, as reflected in complex sentence use, has been found to be positively correlated with child MLU, lexical diver- sity, and morphosyntactic complexity in school-age chil- dren with language impairments (Dethorne & Channell, 2007). This correlation was found despite the fact that for each child, the language sample was 15 min long, elicited with a standard set of toys, and followed Leadholm and Miller’s (1992) recommendations for language sam- pling (e.g., use of open-ended questions). Although it is difficult to determine causality in such a pattern of con- current relationships, it is important to document them. In the present study, we used structured procedures that were highly prescriptive with respect to examiner behavior, but allowed for responsiveness and adapta- tion to the participant. We then evaluated the relation- ship between examiners’ language and the language

Kover et al.: Context Effects 1023

produced by participants with fragile X syndrome, who were the focus of this study, and those with Down syn- drome or typical development, as points of comparison.

Fragile X Syndrome Fragile X syndrome is the leading cause of inherited

intellectual disability. It is caused by a cytosine-guanine- guanine (CGG) trinucleotide expansion beyond 200 re- peats on the FMR1 gene of the X chromosome (Verkerk et al., 1991). In addition to intellectual disabilities, ap- proximately one third of males with fragile X syndrome meet diagnostic criteria for autism, with many of the re- mainder displaying symptoms of autism, including ste- reotyped language (Bailey et al., 1998; Hagerman, 2008; Hagerman, Jackson, Levitas, Rimland, & Braden, 1986; Rogers, Wehner, & Hagerman, 2001).

Delays in vocabulary and syntax are present in both receptive and expressive language for the majority of males with fragile X syndrome regardless of the extent of autism symptoms (Abbeduto, Brady, & Kover, 2007). Expressive language is usually delayed relative to chro- nological age in males with fragile X syndrome, although it may be commensurate with nonverbal cognition when assessed with standardized measures (Finestack & Abbeduto, 2010; Roberts, Price, et al., 2007). Relatively few studies, however, have thoroughly described the spontaneous expressive language abilities of individuals with fragile X syndrome.

Spontaneous expressive language in fragile X syn- drome. Spontaneous expressive language in males with fragile X syndrome generally has been character- ized as rapid, repetitive, disfluent, and unintelligible (Belser & Sudhalter, 2001; Ferrier, Bashir, Meryash, Johnston, & Wolff, 1991; Hanson, Jackson, & Hagerman, 1986; Paul, Cohen, Breg, Watson, & Herman 1984; Van Borsel, Dor, & Rondal, 2008); however, early studies have produced some conflicting results. For example, some studies of males with fragile X syndrome pointed to a specific delay in MLU relative to nonverbal mental age (e.g., Paul et al., 1984), whereas others failed to find weaknesses in MLU compared with cognitive-level expectations (e.g., Ferrier et al., 1991).

In a recent series of studies, Roberts and colleagues (Roberts, Hennon, et al., 2007; Roberts, Martin, et al., 2007; Price et al., 2008; Barnes et al., 2009) analyzed lan- guage samples collected during the Autism Diagnostic Observation Schedule (ADOS; Lord, Rutter, DiLavore, & Risi, 1999). Participants included boys between the ages of 3 and 16 years who were categorized as having either only fragile X syndrome or comorbid fragile X syn- drome and autism spectrum disorder (ASD), as deter- mined by the ADOS. In one study, boys with only fragile X syndrome were found to perform more poorly on lexical

diversity, MLU, and the Index of Productive Syntax (IPSyn; Scarborough, 1990) relative to typically develop- ing boys after controlling for nonverbal mental age, in- telligibility, and maternal education (Roberts, Hennon, et al., 2007). In a follow-up study, Price et al. (2008) found that boys with fragile X syndrome with and with- out comorbid ASD and boys with Down syndrome had lower MLU and IPSyn scores than typically developing boys after controlling for nonverbal mental age and ma- ternal education.

Additional analyses by Roberts and colleagues focused on other aspects of spoken language using these same language samples. Boys with fragile X syn- drome, with and without comorbid ASD, produced fewer intelligible words in connected speech than typically de- veloping boys and did not differ from those with Down syndrome in this regard, controlling for nonverbal men- tal age (Barnes et al., 2009). In terms of pragmatics, boys with comorbid fragile X syndrome and ASD displayed more noncontingent talk than boys with fragile X syn- drome only, Down syndrome, or typical development in utterances that were collected equally from free-play versus other ADOS activities. In addition, boys with fragile X syndrome produced more perseverative lan- guage than boys with Down syndrome or typical devel- opment regardless of ASD status (Roberts, Martin, et al., 2007).

Limitations of previous research. Although previous studies in which spontaneous expressive language in fragile X syndrome has been examined are informative, most have been based on small samples, poorly struc- tured sampling contexts, or language samples collected in a single context. The studies of Roberts and colleagues, for example, included a wide age range of males and in- corporated a protocol that was not designed specifically for the elicitation of language samples. Although the ADOS provides a standard set of activities, the nature of the activities, the amount of time spenton each activity, and the level of examiner prompting might vary within and across modules as well as across participants. For example, conversation and telling a story from a book are probed in Modules 2 and 3, but not Module 1; free play appears in Modules 1 and 2, but not Module 3. Due to potentially important variations in materials and examiner behavior, the ADOS may not provide the ideal sampling context from which to draw conclusions about language profiles within and across groups of in- dividuals with neurodevelopmental disabilities, although it is considered the gold standard for assessing autism symptoms.

In addition, most research has addressed the potential impact of autism symptoms on language performance in individuals with fragile X syndrome from a categorical perspective by either (a) comparing participants with

1024 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1022–1038 • August 2012

or without a comorbid autism or ASD diagnosis (e.g., Price et al., 2008) or (b) excluding participants with comorbid autism from analyses (e.g., Roberts, Hennon, et al., 2007). Despite the contributions of this research, there is a need to move beyond dichotomizing the fragile X syndrome phenotype according to cut offs for diagnos- ing individuals with idiopathic autism. Such cut offs, although displaying adequate levels of sensitivity and specificity for individuals with idiopathic autism, might not be suited to characterizing variation among individuals with neurodevelopmental disabilities other than autism, particularly given that autism symptoms are present to some degree in most males with fragile X syndrome (Hall, Lightbody, Hirt, Rezvani, & Reiss, 2010; Moss & Howlin, 2009). Instead, including partici- pants with all levels of autism symptom severity and considering the incremental impact of increased autism symptomology on language performance is likely to be more informative with respect to the full range of the phenotype in fragile X syndrome (McDuffie, Kover, Abbeduto, Lewis, & Brown, 2012; Moss & Howlin, 2009). Thus, we included all participants with fragile X syndrome in group comparisons regardless of their potential autism-diagnostic status and examined vari- ability within fragile X syndrome by evaluating the relationship between language performance and a con- tinuous scale of autism symptom severity derived from the ADOS.

Down Syndrome Down syndrome is the leading genetic cause of intel-

lectual disability and, in most cases, results from a third copy of chromosome 21 (Chapman & Hesketh, 2000). Autism can occur in individuals with Down syndrome, but it is less common than in fragile X syndrome, with prevalence estimates of approximately 10% (Hepburn, Philofsky, Fidler, & Rogers, 2008; Kent, Evans, Paul, & Sharp, 1999). Language is significantly impaired in Down syndrome compared with nonverbal cognition, with expressive language more delayed than recep- tive language, and syntax delayed beyond vocabulary (Chapman & Hesketh, 2001). Individuals with Down syndrome provide a useful comparison to those with fragile X syndrome because their distinct impair- ments may differentially affect their ability to meet the demands of any given language-sampling task, thereby yielding information about etiological differ- ences in the effects of sampling context on expressive language.

Spontaneous expressive language in Down syn- drome. Research on expressive language in Down syn- drome has yielded more consistent results than studies on fragile X syndrome, particularly in terms of a relative delay in syntactic complexity (Laws & Bishop, 2003).

Roberts and colleagues, as described above, found that males with Down syndrome had lower MLU and IPSyn scores when controlling for nonverbal cognition and maternal education than males with fragile X syn- drome or typical development (Price et al., 2008).

Even for individuals with Down syndrome, however, language-sampling contexts can have differential effects on expressive language. Miles and colleagues (2006) assessed interview-style conversation and narrative performance in 14 adolescents with Down syndrome, 12–21 years of age, relative to 14 typically developing children who were matched on receptive language abil- ity. Although the MLU of the typically developing children did not differ across the conversation and narration contexts, the narrative context yielded sig- nificantly higher MLU for participants with Down syndrome. In fact, MLU did not differ between par- ticipants with Down syndrome and typically devel- oping children in the narrative context. This work illustrates the nuanced characterization of expressive language ability that can emerge from the comparison of language elicited in multiple, carefully structured language-sampling contexts. Whether similar effects of sampling context extend to other neurodevelopmental disabilities, such as fragile X syndrome, remains to be determined.

Comparisons Between Fragile X Syndrome and Down Syndrome

Little research has compared the expressive lan- guage profiles of adolescents with fragile X syndrome and Down syndrome using structured sampling contexts designed to elicit spontaneous language, with the excep- tion of a series of studies by Abbeduto and colleagues. Two studies examined performance only in a narrative context and demonstrated that adolescent and young adult males and females with fragile X syndrome used more grammatically correct and complex utterances than did participants with Down syndrome (Finestack & Abbeduto, 2010; Keller-Bell & Abbeduto, 2007). Examining both a conversation and a narrative context, Kover and Abbeduto (2010) reported several effects of context (e.g., greater MLCU and fluency, but poorer lexical diversity, in narration than in conversation) for older adolescents with fragile X syndrome or with Down syndrome, but lacked a comparison group of typi- cally developing participants, making the expressive language profiles more difficult to interpret. We extend the findings of these studies in the present study by eval- uating the effect of context (i.e., conversation vs. narra- tion) on multiple aspects of expressive language ability, including the extent to which younger male adolescents with fragile X syndrome or Down syndrome successfully completed the language-sampling tasks.

Kover et al.: Context Effects 1025

Research Aims The present study was designed to examine the

effects of language-sampling context on the expressive language profiles of males with fragile X syndrome using structured conversation and narration sampling contexts. The performance of males with fragile X syn- drome was compared with that of adolescents with Down syndrome and typically developing boys at similar levels of nonverbal cognitive development to assess the possibility that sampling context differentially affects individuals with different neurodevelopmental disabil- ities. For participants with fragile X syndrome, we also explored the impact of autism symptom severity on per- formance. Finally, the relationship of examiner behavior to the language produced was examined to probe the potential benefits and limitations of these sampling contexts.

Method Participants

Participants (N = 57) were boys drawn from a large longitudinal project on language development. Youth with fragile X syndrome were recruited nationally (see McDuffie et al., 2010); those with Down syndrome or typ- ical development were recruited primarily locally. Parti- cipants with typical development were not receiving special education services and had no significant sensory or motor impairments according to parent report. All participants were native English speakers and were reported by a parent to regularly use three-word phrases. Approximately 95% of the participants were Caucasian. Approximately 85% of typically developing boys had mothers with a college degree or higher com- pared with approximately 50% of participants with frag- ile X syndrome or Down syndrome. Although several sibling pairs participated, only one child per family was included in the analyses reported here. This sample overlaps with that of McDuffie et al. (2010) and Pierpont et al. (2011), although the primary measures differ. The project was Institutional Review Board approved.

Each participant contributed one conversation sam- ple and one narration sample from a single time point. Exclusions of participants from the present analyses were made on the basis of incomplete conversations or narrations or level of nonverbal cognitive ability. Three participants with Down syndrome from the larger proj- ect were not included in the present sample because they did not complete one or both language-sampling tasks at any visit. One participant with fragile X syndrome from the larger project was not included in the sample be- cause his conversation was only 5 min in duration. Par- ticipants were included in the present analyses only if

they completed the narrative task in a meaningful way—that is, the child produced an utterance relevant to the storybook for at least 12 of the 16 pages. Three par- ticipants with fragile X syndrome and two participants with Down syndrome were not included in the present analyses because they failed to meet this criterion. The analyzed language samples were collected during the first annual assessment for all but five participants with fragile X syndrome and three participants with Down syndrome, who failed to engage in the language- sampling tasks at previous annual visits. Because these participants had not successfully engaged in the tasks during prior visits, data for all participants reflect their first completion of the language-sampling activi- ties. Finally, four participants with typical development and six participants with fragile X syndrome were ex- cluded because their nonverbal mental ages were higher than would allow valid comparisons across all three groups, as described below. The foregoing exclusions resulted in the following samples for the present analy- ses: 10- to 17-year-olds with fragile X syndrome (n = 27) or Down syndrome (n = 15) and 3- to 6-year-old boys with typical development (n = 15).

Participants with fragile X syndrome or Down syn- drome had a confirmed genetic diagnosis. Of those with fragile X syndrome, all had molecular genetic test results indicating the full mutation or mosaicism (i.e., full and premutation cells; n = 7). Karyotypes confirming trisomy 21 were available to the project for 13 of the 15 participants with Down syndrome; however, for the remaining two participants, Down syndrome was listed as the diagnosis in educational or medical reports and corroborated by parent report.

We administered the Leiter-R Brief IQ Screener (Roid & Miller, 1997), which yields both a standard score (nonverbal IQ) and an age-equivalent score (non- verbal mental age). Nonverbal mental age was obtained by averaging the subtest age equivalents associated with the earned raw scores over all completed subtests. To ensure that group comparisons were not confounded by differences in nonverbal mental age, participants were selected such that there were no significant group differences in nonverbal mental age. To achieve this groupwise matching while also maximizing sample sizes, comparisons of the participants with fragile X syn- drome to those with Down syndrome or typical develop- ment were limited to participants whose nonverbal mental age scores ranged between 3.1 and 6.54, yielding the 57 participants included in the analyses described here. Participant characteristics are shown in Table 1. Participants with fragile X syndrome had Leiter Brief IQs between 36 and 62; participants with Down syn- drome had Leiter Brief IQs between 36 and 54, and typ- ically developing boys had Leiter Brief IQs between 87 and 127.

1026 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1022–1038 • August 2012

All but one participant with fragile X syndrome received the ADOS (n = 26), usually at the first annual assessment. Examiners were trained to research reli- ability. Because variability in the Down syndrome phe- notype due to autism symptomology was not the focus of the present study, individuals with Down syndrome were not screened for autism. Autism severity scores, ranging from 1 (nonspectrum) to 10 (autism, highest possible severity score), were assigned to participants with fragile X syndrome according to the chronological age, module administered, and total algorithm score from the ADOS (Gotham, Pickles, & Lord, 2009). Of the 26 participants with fragile X syndrome who received the ADOS, 10 fell in the nonspectrum range (i.e., severity scores ranging from 1 to 3), one fell in the ASD range (i.e., severity scores of 4 or 5), and 15 fell in the autism range (i.e., severity scores ranging from 6 to 10). The mean se- verity score was 5.81, with a standard deviation of 3.19.

Measurement of Expressive Language Abilities

Expressive language abilities were assessed by eliciting spontaneous speech in two distinct sampling contexts, as described by Abbeduto et al. (1995): an interview-style conversation and a narration of a word- less picture book.

In the conversation, each participant talked with one of several female examiners for a target time of 10 min, during which the examiner’s goal was to keep her talk to a minimum. The examiner said that she would like to get to know the participant better and asked open-ended questions, while avoiding yes/no ques- tions as much as possible. Consistency across partic- ipants was ensured with a scripted order of topics and follow-up questions; however, the amount of time spent on a topic was based on the participant’s interests. The topics included school, teachers, pets, and so forth, and were introduced in a broad manner, such as “Do you have any pets? Tell me about them.” Follow-up probes were also broad (e.g., “Tell me what you like about your pet”). All but two conversations elicited from participants

with fragile X syndrome and two from participants with typical development reached the 10-min target. Because of logistical constraints or participant engagement, these four conversations were nearly 8 min.

In the narration activity, participants were shown one of two wordless picture books, Frog Goes to Dinner (Mayer, 1974) or Frog on His Own (Mayer, 1973), alter- nating between participants in the larger study. No sig- nificant differences in performance between books was found for any of the seven dependent variables of inter- est described below (all ps > .18). In this sample of 57 participants, Frog Goes to Dinner was told by 59% of participants with fragile X syndrome, 53% of partici- pants with Down syndrome, and 47% of participants with typical development. The participant was told that he would look at the book and then be asked to tell the story. The examiner turned the pages of the book one by one, allowing the participant to look at each for about 10 s on the initial viewing. The partici- pant then was asked to tell everything about the story for each page. This time through, the experimenter turned to the next page 5 s after the participant had fin- ished narrating a page. The examiner was restricted to scripted prompts if the participant did not respond to the first page (e.g., “What about the boy? What’s he doing, thinking, and feeling?”). If the participant did not talk on subsequent pages, the examiner’s scripted prompts were more limited (e.g., “What’s happening in this part of the story?”). Narratives ranged from 1 to 7 min for the participants with fragile X syndrome, from 3 to 8 min for participants with typical development, and from 3 to 11 min for participants with Down syndrome.

The language samples were recorded onto audio tapes and transcribed using Systematic Analysis of Lan- guage Transcripts software (Miller & Iglesias, 2006) according to research conventions. An experienced pri- mary transcriber first transcribed each sample. A second trained transcriber then listened to the tape and marked suggested changes or perceived discrepan- cies on every transcript. The primary transcriber then verified or updated the transcript for each language sample before the data were analyzed. The first 10 min

Table 1. Participant characteristics.

Participant group n

Chronological age Nonverbal IQa Nonverbal age equivalent

M (SD) M (SD) M (SD)

Fragile X syndrome 27 12.70 (1.84) 44.15 (6.72) 5.08 (0.73) Down syndrome 15 12.99 (2.02) 41.33 (6.03) 4.73 (0.75) Typical development 15 4.67 (1.12) 107.80 (12.25) 4.95 (1.03)

Note. The sample of boys with fragile X syndrome (n = 27) includes participants regardless of autism symptom severity. aDue to a missing subtest, one participant with fragile X syndrome lacked a nonverbal IQ score.

Kover et al.: Context Effects 1027

of each conversation were transcribed, whereas each narrative was transcribed in its entirety. For 17 partici- pants from the larger project, conversation and/or nar- ration samples were independently transcribed again in the manner described to assess interrater agreement. Intertranscriber agreement including utterance seg- mentation and morpheme-level variables averaged 90% across contexts and groups (see Table 2).

Expressive language variables. All speech was seg- mented into communication units (C-units). A C-unit is defined as an independent clause and any of its modi- fiers, including dependent clauses (Loban, 1976). Seg- menting speech into C-units, as opposed to utterances, avoids overestimating language abilities for long but simple utterances combined with coordinating conjunc- tions (Abbeduto et al., 1995; Scarborough, Rescorla, Tager-Flusberg, Fowler, & Sudhalter, 1991). The C-unit is the unit of analysis in the present study.

The amount of talk was assessed in terms of (a) the total number of C-units produced, (b) the number of complete and intelligible C-units produced, and (c) the number of C-units attempted per minute (including incomplete C-units). Language ability was assessed in terms of MLCU, lexical diversity, fluency, and intelligi- bility. MLCU was defined as the mean number of mor- phemes per complete and intelligible C-unit. Lexical diversity was defined as the total number of lexical word roots in the language sample, reflecting range of vocabulary use. Fluency was calculated as the propor- tion of C-units containing mazes, such as filled pauses and partial or full repetitions. Intelligibility was defined as the proportion of C-units partially or fully unintelligi- ble during transcription.

Results Comparison of Fragile X Syndrome, Down Syndrome, and Typical Development

We assessed profiles of expressive language abilities in a comparison of all participants with fragile X syndrome

(n = 27), Down syndrome (n = 15), and typical develop- ment (n = 15). Participants with fragile X syndrome and those with Down syndrome did not differ in terms of chronological age, t(40) = 0.53, p = .600, or nonverbal IQ, t(40) = 1.04, p = .305. The three groups did not differ in nonverbal mental age, F(2, 54) = 0.91, p = .410; how- ever, because the p ≥ .50 matching criteria suggested by Mervis and Robinson (1999) was not reached, nonverbal mental age was included as a covariate in these analy- ses. As such, we conducted repeated measures analyses of covariance to test the effects of context, group, and their interaction. We conducted planned comparisons following omnibus tests with Shaffer’s (1986) postomni- bus procedure.

Amount of talk. The unadjusted scores for the amount of language produced are presented in Table 3. The total number of C-units produced was higher in conversation than in narration, F(1, 53) = 5.44, p = .024, hp

2 = .09. There was neither an effect of group, F(2, 53) = 2.09, p = .133, nor an interaction between context and group, F(2, 53) = 1.82, p = .173, for the number of C-units produced. The number of complete and intelligi- ble utterances was higher in conversation than in narra- tion, but just failed to reach significance, F(1, 53) = 3.88, p = .054. There was also no effect of group, F(2, 53) = 0.58, p = .562, or Context × Group interaction, F(2, 53) = 0.50, p = .612, for complete and intelligible utterances.

For the number of C-units attempted per minute, the interaction between context and group was signifi- cant, F(2, 53) = 5.99, p = .005, hp

2 = .18. The differential effect of context across groups was significant for fragile X syndrome versus Down syndrome, t(53) = 3.41, p = .001, and Down syndrome versus typical develop- ment, t(53) = 2.43, p = .018, but not fragile X syndrome versus typical development, t(53) = 0.69, p = .492. Partici- pants with Down syndrome showed the largest effect of context with few C-units attempted during narration compared with conversation relative to participants with typical development and participants with fragile X syndrome. Neither the main effect of context, F(1, 53) = 3.37, p = .072, nor the main effect of group was sig- nificant, F(2, 53) = 2.83, p = .068.

Language ability. Results for expressive language ability are presented in Table 4. For MLCU, the main ef- fect of context was not significant, F(1, 53) = 1.26, p = .267, but there was an effect of group with the partic- ipants with typical development outperforming those with fragile X syndrome, who in turn scored higher than those with Down syndrome, F(2, 53) = 17.55, p < .001, hp

2 = .40, one-tailed ps < .05. There was also an interaction between context and group for MLCU, F(2, 53) = 4.20, p = .020, hp

2 = .14. The differential effect of context across groups was significant for fragile X syn- drome versus typical development, t(53) = 2.87, p = .006, and just failed to reach significance for Down syndrome

Table 2. Transcription reliability.

Participant group

Conversation Narration

n % agreement n % agreement

Fragile X syndrome 6 91 6 90 Down syndrome 5 87 5 86 Typical development 5 92 5 94

Note. Percentage (%) agreement is averaged over utterance segmentation, intelligibility, mazes, overlaps, pauses, abandonment, word identification, number of morphemes and words, and ending punctuation.

1028 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1022–1038 • August 2012

versus typical development, t(53) = 1.91, p = .061. The effect of context on MLCU did not differ between partic- ipants with fragile X syndrome and Down syndrome, t(53) = 0.69, p = .496. Boys with typical development showed little effect of context on MLCU compared with the other groups, with those with fragile X syndrome producing higher MLCU in narration than in conversation.

For lexical diversity, range of vocabulary use dif- fered among groups, F(2, 53) = 4.32, p = .018, hp

2 = .14, with better performance by the typically developing boys than participants with fragile X syndrome or Down syn- drome (ps < .05), who did not differ (p = .211). There was no effect of context, F(1, 53) = 0.10, p = .757, or interac- tion, F(2, 53) = 2.54, p = .089.

Fluency differed among groups, F(2, 53) = 3.90, p = .026, hp

2 = .13, and was a strength of the participants with fragile X syndrome relative to those with Down syn- drome (p = .008). There was also an interaction between

group and context, F(2, 53) = 7.03, p = .002, hp 2 = .21. The

effect of context differed across all groups: fragile X syn- drome versus Down syndrome, t(53) = 2.04, p = .046; fragile X syndrome versus typical development, t(53) = 2.19, p = .033; and Down syndrome versus typical devel- opment, t(53) = 3.75, p < .001. Participants with fragile X syndrome showed similar fluency in both contexts; par- ticipants with Down syndrome were more fluent (i.e., produced a smaller proportion of C-units with mazes) during conversation than narration, whereas partici- pants with typical development were more fluent during narration. The main effect of context was not significant, F(1, 53) = 0.41, p = .526.

Intelligibility differed among groups, F(2, 53) = 9.10, p < .001, hp

2 = .26, such that it was a weakness of parti- cipants with Down syndrome relative to those with frag- ile X syndrome and those with typical development (ps < .002). Participants with fragile X syndrome were less intelligible than those with typical development,

Table 4. Language performance in conversation and narration.

Participant group n

MLCU Lexical diversitya Fluencyb Intelligibilityc

M SD Range M SD Range M SD Range M SD Range

Conversation Fragile X syndrome 27 3.75 1.45 1.32–7.05 143.85 70.45 38–313 18 13 0–42 12 8 2–37 Down syndrome 15 3.06 0.80 1.84–4.32 115.00 30.69 59–171 23 17 1–66 20 12 9–57 Typical development 15 5.52 1.26 3.30–7.05 173.73 50.80 72–236 27 10 11–48 6 4 1–13

Narration Fragile X syndrome 27 4.85 1.36 2.17–7.37 62.26 22.34 28–127 17 10 2–34 10 8 0–28 Down syndrome 15 3.94 1.23 1.40–6.00 50.27 21.63 14–79 31 19 4–68 20 18 3–67 Typical development 15 5.59 0.93 4.07–7.11 67.20 18.35 38–100 18 10 3–38 8 5 2–18

Note. MLCU = mean length of communication unit. aLexical diversity reflects the number of different word roots produced in the language sample. bFluency reflects the percentage of C-units with mazes. cIntelligibility reflects the percentage of unintelligible C-units.

Table 3. Amount of talk produced in conversation and narration.

Participant group n

Total C-units Complete and intelligible C-units attempted

per minute

M SD Range M SD Range M SD Range

Conversation Fragile X syndrome 27 126.04 39.22 65–202 107.07 36.93 42–187 12.79 3.85 6.50–20.20 Down syndrome 15 131.73 33.17 70–174 100.87 28.23 38–143 13.17 3.32 7.00–17.40 Typical development 15 107.33 25.64 70–148 95.60 23.25 61–140 11.03 2.86 7.50–16.80

Narration Fragile X syndrome 27 43.56 16.89 21–76 36.41 13.09 19–60 11.32 4.27 4.30–22.14 Down syndrome 15 41.33 14.22 18–67 31.13 10.79 10–50 7.30 3.04 3.17–14.11 Typical development 15 39.87 10.07 20–56 34.93 10.41 17–52 8.65 2.61 4.48–12.81

Note. C-units = communication units.

Kover et al.: Context Effects 1029

but not significantly so (p = .151). Neither the effect of context, F(1, 53) = 0.02, p = .900, nor the interaction, F(2, 53) = 0.71, p = .495, was significant for intelligibility.

Variability Within Fragile X Syndrome

We examined the impact of autism symptom sever- ity on the amount of talk and language performance of participants with fragile X syndrome (n = 26), control- ling for nonverbal mental age. For each sampling con- text, we entered autism severity and nonverbal mental age into separate regressions predicting (a) the number of complete and intelligible C-units, (b) the number of C-unit attempts per minute, and (c) MLCU. We predicted that autism symptom severity would negatively relate to language production and, therefore, used one-tailed p values for testing the regression coefficients in this exploratory analysis; nondirectional p values were used to test overall models. Unstandardized coefficients are reported (see Table 5).

In conversation, autism severity scores were nega- tively related to the number of complete and intelligible C-units produced, controlling for nonverbal cognitive ability (b = –4.36), t(23) = –2.02, p = .028, one-tailed, semipartial r = –.37. The overall model predicting the number of attempts per minute in conversation was not significant (p = .079); however, severity scores did re- late negatively to the number of attempts per minute, controlling for nonverbal cognitive ability (b = –0.44), t(23) = –1.91, p = .034, one-tailed, semipartial r = –.36. Nonverbal cognitive ability was positively related to MLCU in conversation (b = 0.92), t(23) = 2.53, p = .010, one-tailed, semipartial r = .46; autism severity was not (b = –0.07), t(23) = –0.81, p = .212, one-tailed.

In the narrative context, only the regression predict- ing MLCU was significant (p = .007). As in conversation, nonverbal mental age was positively related to MLCU (b = 0.99), t(23) = 3.06, p = .003, one-tailed, semipartial r = .52, whereas autism severity was not (b = –0.10), t(23) = –1.30, p = .103, one-tailed. The overall models predicting the number of complete and intelligible C-units or the number of C-units attempted failed to reach significance for narration (ps > .18).

Examiner Behavior We examined the extent to which examiner behavior

was related to the language performance of participants with fragile X syndrome using separate regressions for conversation and narration. As a point of comparison, we repeated these regressions for the adolescents with Down syndrome and the boys with typical development. Of course, these analyses cannot tease apart the direc- tion of causation in the relationship between examiner and child language, but they do provide guidance for future research. Number of C-unit attempts per minute, MLCU, and proportion of questions by the examiner were all expected to be negatively related to amount of participant language and performance (Kover, Abbeduto, Schroeder, Giles, & Richmond, 2008). We tested these three aspects of examiner behavior as simultaneous predictors of participant language output and performance using one-tailed p values. We chose participant number of C-unit attempts per minute as a dependent variable to index language output because it reflects the clinical utility and efficiency of language samples. We chose participant MLCU as a dependent variable to index language performance because it is widely used as an estimate of spoken language ability.

Table 5. Results of regressions for variability among participants with fragile X syndrome (n = 26).

Dependent variable

Full model

Individual predictors

NVMA Autism severity

F R2 b t b t

Conversation Complete and intelligible C-units 3.48* .23 13.49 1.44 –4.36 –2.02y

C-units attempted per minute 2.84 .20 1.16 1.18 –0.44 –1.91y

MLCU 3.82* .25 0.92 2.53y –0.07 –0.81 Narration Complete and intelligible C-units 1.85 .14 6.46 1.86y 0.57 0.71 C-units attempted per minute 1.25 .10 1.77 1.51 0.18 0.65 MLCU 6.11* .35 0.99 3.06y –0.10 –1.30

Note. NVMA = nonverbal mental age. yp < .05, one-tailed. *p < .05.

1030 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1022–1038 • August 2012

For conversation, the overall model predicting the number of C-units attempted per minute was not signif- icant for participants with fragile X syndrome (p = .080) or Down syndrome (p = .151). The model was significant for participants with typical development, F(3, 11) = 3.69, p = .047, for whom examiner MLCU was significantly neg- atively related to C-units per minute (b = –2.25), t(11) = –2.46, p = .016, one-tailed, semipartial r = –.52 (see Table 6). When predicting participant MLCU, the over- all model was significant for participants with fragile X syndrome, F(3, 23) = 11.67, p < .001, with examiner number of C-units per minute (b = –0.24), t(23) = –4.88, p < .001, one-tailed, semipartial r = –.64, and examiner MLCU (b = –0.66), t(23) = –3.01, p = .003, one-tailed, semipartial r = –.40, emerging as negative predictors (see Table 7). The model predicting MLCU was also signif- icant for typically developing participants, F(3, 11) = 5.67, p = .014, with only examiners’ number of C-unit attempts per minute negatively relating to MLCU (b = –0.26), t(11) = –1.83, p = .047, one-tailed, semipartial r = –.35. The over- all model predicting MLCU was not significant for par- ticipants with Down syndrome (p = .065).

For narration, the model predicting the number of C-units per minute was significant for participants with fragile X syndrome, F(3, 23) = 5.64, p = .005. A higher proportion of questions asked by the examiner was associated with fewer C-units per minute (b = –0.26), t(23) = –4.10, p < .001, one-tailed, semipartial r = –.65. The models were not significant for participants with Down syndrome or typical development (ps > .18). The model predicting MLCU was not significant for par- ticipants with fragile X syndrome (p = .238), but it was for participants with Down syndrome and typical de- velopment, F(3, 11) = 6.99, p = .007; and, F(3, 11) = 3.77, p = .044, respectively. For participants with Down syndrome, the number of attempts per minute by the

examiner was significantly associated with lower MLCU (b = –0.50), t(11) = –3.69, p = .002, one-tailed, semipartial r = –.65. For participants with typical development, exam- iner MLCU was negatively related to MLCU (b = –0.44), t(11) = –1.89, p = .043, one-tailed, semipartial r = –.40.

Discussion In this study, we sought to examine the differential

effects of language-sampling context on the spontaneous expressive language profiles of male adolescents with fragile X syndrome and Down syndrome. The inclusion of a comparison group of typically developing boys also allowed conclusions to be drawn about the extent of delay in each neurodevelopmental disability relative to nonverbal cognitive ability.

Comparison of Fragile X Syndrome, Down Syndrome, and Typical Development

Effects of context among groups. Performance on the conversation and narration language-sampling tasks was assessed in terms of the amount of talk produced and in terms of language ability. Within both of these areas, differential effects of context on performance were found among participants with fragile X syndrome (regardless of autism symptom severity), Down syn- drome, and typical development, controlling for nonver- bal mental age.

In terms of the amount of language produced, a sig- nificant Group × Context interaction revealed that adolescents with Down syndrome differed from both adolescents with fragile X syndrome and boys with typ- ical development in the number of C-units attempted per minute in conversation relative to narration. Although all

Table 6. Aspects of examiner behavior as predictors of participant C-unit attempts per minute.

Participant group n

Full model

Individual predictors

Ex. attempts Ex. MLCU Ex. questions

F R2 b t b t b t

Conversation Fragile X syndrome 27 2.56 .25 –0.09 –0.52 –1.75 –2.20y 0.03 0.34 Down syndrome 15 2.16 .37 0.56 2.10 0.08 0.06 –0.13 –1.44 Typical development 15 3.69* .50 –0.45 –1.23 –2.25 –2.46y 0.18 2.11

Narration Fragile X syndrome 27 5.64* .42 0.08 0.26 0.65 1.14 –0.26 –4.10y

Down syndrome 15 0.77 .17 –0.29 –0.55 0.06 0.06 –0.10 –1.05 Typical development 15 1.94 .35 –0.01 –0.01 –0.75 –0.99 –0.07 –0.83

Note. Ex. = Examiner. yp < .05, one-tailed. *p < .05.

Kover et al.: Context Effects 1031

groups tended to have higher rates of C-unit attempts in conversation than narration, adolescents with Down syndrome had particularly low rates of C-units at- tempted per minute in the narrative context relative to the conversation context. Adolescents with fragile X syn- drome showed the smallest difference in number of attempts per minute across contexts, but this effect did not differ from the boys with typical development. Miller (1996) has suggested that the amount of talking during a language sample is an important quantifiable aspect of performance that might reflect an individual’s linguistic competence. Narration may be a particularly challeng- ing task for adolescents with Down syndrome relative to their nonverbal cognitive abilities and relative to those with other sources of intellectual disability, such as fragile X syndrome. Given the constraints and de- mands of the content of the wordless book on the lan- guage produced, we speculate that participants with Down syndrome may have required additional process- ing time in formulating their expressive responses, leading to fewer C-units per minute. In contrast, the demands of reciprocal social interaction in conversation might be more of a challenge for adolescents with fragile X syndrome, resulting in a smaller difference between conversation and narration language production for this population.

In terms of expressive language ability, the effect of context on MLCU differed across groups. Relative to the typically developing boys, for whom MLCU was similar in conversation and narration, adolescents with fragile X syndrome demonstrated higher MLCU in narration than in conversation. Adolescents with Down syndrome also produced higher MLCU in narration than conversa- tion, although the comparison with boys with typical de- velopment for the effect of context just failed to reach significance. Previous studies have also suggested that MLU elicited during narration tends to be higher than

conversation for adolescents and young adults with neu- rodevelopmental disabilities, including fragile X syn- drome, Down syndrome, and unspecified etiology (Abbeduto et al., 1995; Kover & Abbeduto, 2010; Levy, Gottesman, Borochowitz, Frydman, & Sagi, 2006; Miles et al., 2006). The narrative task provides opportu- nities to describe the actions and mental states of a pro- tagonist in relation to other characters as well as to linguistically encode event sequences, all of which are best accomplished using multiclause constructions (Heilmann, Nockerts, & Miller, 2010). The visual sup- port of narrating a book also may allow production of more advanced syntax (Levy et al., 2006). Miles and col- leagues (2006), for example, found that MLCU for partic- ipants with Down syndrome was higher in narratives obtained using wordless picture books compared with narratives embedded within an interview-style conver- sation. Thus, our findings reinforce that narration is par- ticularly well suited to eliciting the upper bounds of syntactic ability in individuals with neurodevelopmen- tal disabilities and that weaknesses might be more likely identified in conversation samples, in which MLCU was lower for participants with fragile X syndrome and Down syndrome, than in narration, in which the discrepancy between the groups with intellectual disability and typ- ical development was minimized (Abbeduto et al., 1995).

There was also a differential effect of context across groups for fluency, which was indexed by the proportion of C-units with mazes (i.e., filled pauses and repetitions). Fluency was found to be better (i.e., fewer mazes) in nar- ration than conversation for typically developing boys, relative to participants with Down syndrome who were more fluent in conversation than narration. Adolescents with fragile X syndrome showed comparable fluency ac- ross contexts relative to the other groups. Research on adolescents with Down syndrome and children with typ- ical development has shown poorer fluency in conversation

Table 7. Aspects of examiner behavior as predictors of participant MLCU.

Participant group n

Full model

Individual predictors

Ex. attempts Ex. MLCU Ex. questions

F R2 b t b t b t

Conversation Fragile X syndrome 27 11.67* .60 –0.24 –4.88y –0.66 –3.01y 0.01 0.29 Down syndrome 15 3.22 .47 –0.07 –1.12 0.19 0.62 –0.04 –1.98y

Typical development 15 5.67* .61 –0.26 –1.83y –0.16 –0.45 –0.02 –0.71 Narration Fragile X syndrome 27 1.51 .17 –0.12 –1.04 –0.20 –0.90 –0.02 –0.65 Down syndrome 15 6.99* .66 –0.50 –3.69y –0.20 –0.81 –0.02 –0.84 Typical development 15 3.77* .51 –0.26 –1.69 –0.44 –1.89y 0.01 0.46

yp < .05, one-tailed. *p < .05.

1032 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1022–1038 • August 2012

than narration (Miles et al., 2006), although some re- search on children with language problems has indicated the opposite (MacLachlan & Chapman, 1988; Wagner et al., 2000). Becoming familiar with a story before nar- rating likely leads to mental rehearsal, and thus a fluent narrative, whereas the dynamic demands of conversation decrease the opportunities to plan words and phrases in advance (Kover & Abbeduto, 2010). The demands of the narrative task relative to nonverbal cognitive ability for adolescents with particular neurodevelopmental dis- abilities, however, might outweigh the benefits of pre- viewing a story, leading to particular difficulty with fluency during narration.

The finding that fluency differed less between con- versation and narration for participants with fragile X syndrome relative to those with Down syndrome or typical development is surprising, as one might expect that the well-documented social anxiety associated with fragile X syndrome would lead to poorer fluency in con- versation. However, relative to the boys with typical de- velopment, the adolescents with fragile X syndrome had increased age and life experiences that may have sup- ported performance in the conversation task. It is also possible, however, that individuals with fragile X syn- drome relied more on rote or repetitive (i.e., highly re- hearsed) linguistic contributions, thereby reducing the processing demands of the task. In fact, Roberts, Price, et al. (2007) found that boys with fragile X syndrome had more perseverative talk than typically developing chil- dren and that boys with fragile X syndrome with comor- bid ASD had more noncontingent talk than those with fragile X syndrome only. Future research will need to an- alyze the content of utterances produced by individuals with fragile X syndrome in relation to the other aspects of their language performance. Although the content and form of communicative attempts might be appropri- ate targets for intervention, producing those attempts fluently appears to be a strength for adolescents with fragile X syndrome.

Of the dimensions of performance examined, there was one significant main effect of context. Participants produced more C-units in conversation than in narra- tion. These results are consistent with those of Abbeduto et al. (1995), who studied youths with intellectual disabil- ity of unspecified etiology. The context effect for the ab- solute number of C-units might be accounted for simply by the fact that the conversation sample was structured to last at least 10 min, whereas the child had the option to say as much or as little as he chose for each page of the storybook in the narrative context. Because the number of C-units attempted per minute also tended to be greater in conversation than narration, these results suggest that conversation is ideal for assessing some aspects of ex- pressive language because of the increased likelihood of efficiently eliciting a larger sample of utterances.

Taken together, the effects of context observed in the present study for participants with fragile X syndrome and Down syndrome highlight the need for language- sampling procedures that include multiple structured contexts to fully understand the profile of language abilities in individuals or groups of individuals with developmental disabilities (Abbeduto et al., 2012). Although the ADOS has been recommended as one structured context in which language can be sampled (Tager-Flusberg et al., 2009), it is unknown whether the ADOS provides a sufficiently consistent context with stable expectations of language performance, even for typically developing individuals. Given the dif- ferential effects of context across diagnostic groups, one might expect different conclusions to be drawn about language abilities in neurodevelopmental disabilities depending on the properties of the language-sampling task.

Extent of delay in fragile X syndrome and Down syn- drome. Several differences in performance emerged be- tween adolescents with fragile X syndrome (irrespective of autism symptoms) or Down syndrome and boys with typical development, controlling for nonverbal mental age.

Adolescents with fragile X syndrome demonstrated lower MLCU than typically developing boys, indicating that syntactic complexity is an area of particular weak- ness, as shown in previous studies (Roberts, Hennon, et al., 2007). Participants with Down syndrome pro- duced lower MLCU than both the boys with typical de- velopment and the adolescents with fragile X syndrome. Price et al. (2008) found the same pattern of results in a younger sample of boys using the ADOS as a language- sampling context: Participants with typical develop- ment outperformed those with fragile X syndrome, who outperformed those with Down syndrome. In a nar- rative context, Finestack and Abbeduto (2010) found that participants with fragile X syndrome without comorbid autism also outperformed participants with Down syndrome on a measure of overall grammatical ability. Children with other neurodevelopmental dis- abilities, including idiopathic autism, also have deficits in MLU beyond nonverbal expectations (e.g., IQ; Eigsti, Bennetto, & Dadlani, 2007), reinforcing the notion that syntax should be a focus of intervention for a wide range of children with neurodevelopmental disabilities. How- ever, in agreement with findings of previous research, the presentstudy demonstrates that syntactic complexity is a syndrome-specific area of weakness in need of partic- ular attention in adolescents with Down syndrome.

Lexical diversity also was found to be an area of weak- ness for adolescents with fragile X syndrome or Down syn- drome. Both groups of participants with intellectual disability used a smaller range of vocabulary than the boys with typical development, although participants with fragile X syndrome and Down syndrome did not

Kover et al.: Context Effects 1033

differ. Although the interaction between context and group failed to reach significance in the present study, the impact of sampling context on lexical diversity should continue to be considered in future research on neurodevelopmental disabilities. Descriptively, partici- pants with Down syndrome seem to use a particularly restricted range of vocabulary during narration, and previous studies have shown that conversation is more likely than narration to assess the scope of vocabulary available to a participant (Kover & Abbeduto, 2010).

Adolescents with fragile X syndrome or Down syn- drome did not differ from boys with typical development in fluency; however, those with fragile X syndrome pro- duced a smaller proportion of C-units with mazes (i.e., were more fluent) compared with participants with Down syndrome. In general, participants with fragile X syndrome were the most fluent, which might be expected if these participants were the most likely to use repetitive (i.e., well-practiced) language. The overall group difference between fragile X syndrome and Down syndrome is likely driven by the particularly poor fluency of the adolescents with Down syndrome during the nar- rative task, suggesting a syndrome- and, to some extent, context-specific area of weakness.

Participants with Down syndrome were less intel- ligible than both participants with fragile X syndrome and those with typical development. Problems with intelligibility for individuals with Down syndrome have been well documented in previous research, in which even single-word production was impaired relative to typically developing boys and boys with fragile X syndrome with similar nonverbal cognitive abilities (Roberts, Price, et al., 2007). The present results are also generally in line with Barnes et al. (2009), who found that boys with Down syndrome produced a smaller percentage of intelligible words than those with typical development; however, boys with fragile X syndrome did not differ from those with Down syndrome in the Barnes et al. study. In a sample of older adolescents, Kover and Abbeduto (2010) also failed to find differences between participants with Down syndrome and those with fragile X syndrome with or without autism. The in- consistent findings for differences in intelligibility be- tween individuals with fragile X syndrome and Down syndrome may be attributable to varying age ranges of participants, procedures for assessing intelligibility, or language-sampling contexts.

In summary, combining words into developmentally progressive grammatical utterances and expanding range of vocabulary are likely to be important interven- tion targets for adolescents with fragile X syndrome. For adolescents with Down syndrome, extending syntactic complexity and vocabulary, in addition to enhancing flu- ency and intelligibility during connected speech, are likely to be areas in need of remediation.

Autism Symptom Severity We examined the relationship between a continuous

metric of autism symptom severity in males with fragile X syndrome and their language production in the con- versation and narrative contexts, controlling for nonver- bal mental age. Our results suggest that individuals with more severe autism symptoms are less able to engage with the demands of conversation, as reflected in the production of fewer complete and intelligible utterances. We found no relationship between autism symptom severity and language production in narra- tion; however, nonverbal cognitive level accounted for variability in MLCU in both narration and conversation. It seems reasonable that males with fragile X syndrome who display more symptoms of autism could find the so- cial demands of conversation more challenging given the pressure to engage in a social give-and-take, whereas engaging in the narrative task—and perhaps producing grammatically complex utterances in general—is more dependent on nonverbal cognitive skills. Producing con- tingent talk in conversational contexts has been identi- fied as an area of weakness for children with idiopathic autism as well (Tager-Flusberg & Anderson, 1991).

Examiner Behavior To our knowledge, no previous studies have exam-

ined the ways in which examiner behavior relates to spontaneous language in youths with fragile X syndrome or Down syndrome. Even within the constraints of our structured procedures, relationships were found be- tween variability in examiners’ talk and the language performance of participants. In conversation, partici- pants’ MLCU was negatively related to both examiner MLCU and number of C-unit attempts per minute for participants with fragile X syndrome and was related to C-unit attempts per minute for participants with typ- ical development. Examiner behavior was not related to the performance of participants with Down syndrome in conversation. In the narrative, number of C-unit attempts per minute by participants with fragile X syn- drome was negatively related to the examiner’s propor- tion of questions, whereas MLCU for participants with Down syndrome or typical development was negatively predicted by examiners’ C-unit attempts per minute and MLCU, respectively.

For a group of language-impaired children, Dethorne and Channell (2007) interpreted correlations between clinician behavior and child language as resulting from the clinician adjusting her talk in relation to the talk produced by the children. It would be expected that a child who is reticent might elicit more coaxing from an examiner, and the structure provided by the examiner as an interlocutor should not be discounted. Similarly,

1034 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1022–1038 • August 2012

the most likely explanation of our results is that the rela- tionships observed reflect variability in response to the language-sampling tasks by participants and the efforts of the examiner to maintain participant engagement; however, it is noteworthy that the pattern of relationships varied across participants with fragile X syndrome, Down syndrome, and typical development. Such relationships should be taken into account in designing language- sampling protocols that are flexible, yet structured, to meet the needs of particular individuals and diagnostic groups. Future research should probe the direction of causality between examiner and child language.

Limitations The results presented here are based on language

samples of relatively short duration and a relatively small number of utterances. Although longer language samples are in some ways ideal (e.g., 175 utterances; Rice et al., 2010), there is evidence that shorter (e.g., 3-min) language samples can provide reliable estimates of language ability (Heilmann et al., 2010). In this sense, the results are optimally interpretable from a clinical perspective, given that it might be unrealistic to collect larger language samples to assess youths with neurode- velopmental disabilities in a clinical setting. This is par- ticularly true for those with fragile X syndrome, for whom inattentiveness or hyperactivity could impact participation in assessment procedures. Nonetheless, the fact that the language samples analyzed here contained a limited number of utterances, particularly for the narrative samples, should be considered in terms of the generalizability of the findings. Only nine participants with fragile X syndrome, four participants with Down syndrome, and three participants with typi- cal development produced 50 or more C-units during narration. A potential advantage of the ADOS as a language-sampling context is the likelihood of obtain- ing a larger sample during the 30- to 60-min assessment.

The participants in the present study were limited to those who scored within a restricted range of nonver- bal cognitive ability, in order to allow simultaneous com- parison of those with fragile X syndrome and typical development, who tended to have higher developmental levels, to those with Down syndrome, who tended to have slightly lower developmental levels in the larger sample from which these data are drawn. Having con- trolled for nonverbal mental age, this analysis strategy allowed for unambiguous interpretation of group differ- ences within this developmental range but limits the generalizability of the findings to other individuals along the range of the highly variable phentoypes of fragile X syndrome and Down syndrome. Investiga- tion of the impact of sampling contexts on the language of individuals with neurodevelopmental disabilities

across a range of nonverbal cognitive ability levels is warranted.

Finally, the present study did not address the con- tent of the utterances produced by participants in terms of repetitiveness, perseveration, or contingency. This will be an important area for future research, espe- cially with respect to the impact of autism symptoms on spontaneous expressive language in adolescents with fragile X syndrome. Indeed, behavioral and pharmaceu- tical interventions that seek to improve expressive lan- guage in fragile X syndrome are likely to target not only the amount of talk and the grammatical complexity of that talk but also its communicative function by assess- ing and documenting reduction in perseverative or non- contingent utterances.

Conclusions We have reported findings that draw attention to

the importance of considering the context in which lan- guage samples are elicited for individuals with neurode- velopmental disabilities. These results suggest that the demands of the language-sampling context could have implications for targets selected for intervention and conclusions drawn about language profiles.

Acknowledgments This work was supported by National Institutes of Health

(NIH) Grants R01 HD024356 and P30 HD003352, as well as three fellowships awarded to the first author: the Waisman Center’s Anderson Hoffman Wisconsin Distinguished Grad- uate Fellowship, the Michael Vincent and Harriet Frisbie Eastabrooks O’Shea Fellowship, and NIH F31 DC010959 National Research Service Award Individual Predoctoral Fellowship. Portions of these data were presented at the 43rd Gatlinburg Conference on Research and Theory in Intel- lectual and Developmental Disabilities in Annapolis, Maryland. We offer our sincere appreciation to all of the families who par- ticipated in this research. We thank Susen Schroeder for her dedication in supervising the transcription of language samples and Pamela Lewis for contributing to the autism evaluations.

References Abbeduto, L., Benson, G., Short, K., & Dolish, J. (1995). Effects of sampling context on the expressive language of children and adolescents with mental retardation. Mental Retardation, 33, 279–288.

Abbeduto, L., Brady, N., & Kover, S. T. (2007). Language development and fragile X syndrome: Profiles, syndrome- specificity, and within-syndrome differences. Mental Retar- dation and Developmental Disabilities Research Reviews, 13, 36–46.

Abbeduto, L., Kover, S. T., & McDuffie, A. (2012). Study- ing the language development of children with intellectual

Kover et al.: Context Effects 1035

disabilities. In E. Hoff (Ed.), Guide to research methods in child language (pp. 330–346). Oxford, England: Wiley-Blackwell.

Bailey, D. B., Mesibov, G. B., Hatton, D. D., Clark, R. D., Roberts, J. E., & Mayhew, L. (1998). Autistic behavior in young boys with fragile X syndrome. Journal of Autism and Developmental Disorders, 28, 499–508.

Barnes, E., Roberts, J., Long, S. H., Martin, G. E., Berni, M. C., Mandulak, K. C., & Sideris, J. (2009). Phonological accuracy and intelligibility in connected speech of boys with fragile X syndrome or Down syndrome. Journal of Speech, Language, and Hearing Research, 52, 1048–1061.

Belser, R. C., & Sudhalter, V. (2001). Conversational char- acteristics of children with fragile X syndrome: Repetitive speech. American Journal of Mental Retardation, 106, 28–38.

Chapman, R., & Hesketh, L. (2000). Behavioral phenotype of individuals with Down syndrome. Mental Retardation and Developmental Disabilities Research Reviews, 6, 84–95.

Chapman, R., & Hesketh, L. (2001). Language, cognition, and short-term memory in individuals with Down syndrome. Down Syndrome Research and Practice, 7, 1–7.

Condouris, K., Meyer, E., & Tager-Flusberg, H. (2003). The relationship between standardized measures of langauge and measures of spontaneous speech in children with au- tism. American Journal of Speech-Language Pathology, 12, 349–358.

Cowan, P. A., Weber, J., Hoddinott, B. A., & Klein, J. (1967). Mean length of spoken response as a function of stimulus, experimenter, and subject. Child Development, 3, 191–203.

Dethorne, L. S., & Channell, R. W. (2007). Clinician-child interactions: Adjustments in linguistic complexity. American Journal of Speech-Language Pathology, 16, 119–127.

Dollaghan, C. A., Campbell, T. F., & Tomlin, R. (1990). Video narration as a language sampling context. Journal of Speech and Hearing Disorders, 55, 582–590.

Eigsti, I.-M., Bennetto, L., & Dadlani, M. B. (2007). Beyond pragmatics: Morphosyntactic development in autism. Jour- nal of Autism and Developmental Disorders, 37, 1007–1023.

Evans, J., & Craig, H. (1992). Language sample collection and analysis: Interview compared to freeplay assessment con- texts. Journal of Speech and Hearing Research, 35, 343–353.

Ferrier, L. J., Bashir, A. S., Meryash, D. L., Johnston, J., & Wolff, P. (1991). Conversational skills of individuals with fragile-X syndrome: A comparison with autism and Down syndrome. Developmental Medicine and Child Neurology, 33, 776–788.

Finestack, L. H., & Abbeduto, L. (2010). Expressive lan- guage profiles of verbally expressive adolescents and young adults with Down syndrome or fragile X syndrome. Journal of Speech, Language, and Hearing Research, 53, 1334–1348.

Gotham, K., Pickles, A., & Lord, C. (2009). Standardizing ADOS scores for a measure of severity in autism spectrum disorders. Journal of Autism and Developmental Disorders, 39, 693–705.

Hagerman, R. (2008). Etiology, diagnosis, and development in fragile X syndrome. In J. E. Roberts, R. S. Chapman, & S. F. Warren (Eds.), Speech and language development and inter- vention in Down syndrome and fragile X syndrome (pp. 27–49). Baltimore, MD: Brookes.

Hagerman, R. J., Jackson, A. W., Levitas, A., Rimland, B., & Braden, M. (1986). An analysis of autism in fifty males with the fragile X syndrome. American Journal of Medical Genetics, 23, 359–374.

Hall, S. S., Lightbody, A. A., Hirt, M., Rezvani, A., & Reiss, A. (2010). Autism in fragile X syndrome: A category mistake? Journal of the American Academy of Child & Adolescent Psychaitry, 49, 921–933.

Hanson, D. M., Jackson, A. W., & Hagerman, R. J. (1986). Speech disturbances (cluttering) in mildly impaired males with the Martin-Bell/fragile X syndrome. American Journal of Medical Genetics, 23, 195–206.

Hansson, K., Nettelbladt, U., & Nilholm, C. (2000). Con- textual influence on the language production of children with speech/language impairment. International Journal on Language and Communication Disorders, 35, 31–47.

Heilmann, J., Nockerts, A., & Miller, J. F. (2010). Lan- guage sampling: Does the length of the transcript matter? Language, Speech, and Hearing Services in Schools, 41, 393–404.

Hepburn, S., Philofsky, A., Fidler, D. J., & Rogers, S. (2008). Autism symptoms in toddlers with Down syndrome: A descriptive study. Journal of Applied Research in Intellec- tual Disabilities, 21, 48–57.

Johnston, J. R. (2001). An alternate MLU calculation: Mag- nitude and variability of effects. Journal of Speech, Lan- guage, and Hearing Research, 44, 156–164.

Johnston, J., Miller, J., Curtiss, S., & Tallal, P. (1993). Conversations with children who are language impaired: Asking questions. Journal of Speech and Hearing Research, 36, 973–978.

Keller-Bell, Y. D., & Abbeduto, L. D. (2007). Narrative de- velopment in adolescents and young adults with fragile X syndrome. American Journal on Mental Retardation, 112, 289–299.

Kent, L., Evans, J., Paul, M., & Sharp, M. (1999). Comor- bidity of autistic spectrum disorders in children with Down syndrome. Developmental Medicine and Child Neurology, 41, 153–158.

Kover, S. T., & Abbeduto, L. (2010). Expressive language in male adolescents with fragile X syndrome with and without comorbid autism. Journal of Intellectual Disability Research, 54, 246–265.

Kover, S. T., Abbeduto, L., Schroeder, S. M., Giles, N. D., & Richmond, E. K. (2008, March). Effects of sampling context and examiner behavior on expressive language in adolescents with fragile X or Down syndrome. Poster presented at the 41st annual Gatlinburg Conference on Research and Theory in Intellectual and Developmental Disabilities, San Diego, CA.

Laws, G., & Bishop, D. V. (2003). A comparison of language abilities in adolescents with Down syndrome and children with specific language impairment. Journal of Speech, Lan- guage, and Hearing Research, 46, 1324–1339.

Leadholm, B. J., & Miller, J. F. (1992). Language sample analysis: The Wisconsin guide. Madison: Wisconsin Depart- ment of Public Instruction.

Levy, Y., Gottesman, R., Borochowitz, Z., Frydman, M., & Sagi, M. (2006). Language in boys with fragile X syndrome. Journal of Child Language, 33, 125–144.

1036 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1022–1038 • August 2012

Loban, W. (1976). Language development: Kindergarten through grade twelve. Urbana, IL: National Council of Teachers of English.

Lord, C., Rutter, M., DiLavore, P., & Risi, S. (1999). Autism Diagnostic Observation Schedule–Generic. Los Angeles, CA: Western Psychological Services.

Losh, M., & Capps, L. (2003). Narrative ability in high- functioning children with autism or Asperger’s syndrome. Journal of Autism and Developmental Disorders, 33, 239–251.

MacLachlan, B. G., & Chapman, R. S. (1988). Communica- tion breakdowns in normal and language learning-disabled children’s conversation and narration. Journal of Speech and Hearing Disorders, 53, 2–7.

Mayer, M. (1973). Frog on his own. New York, NY: Dial Books for Young Readers.

Mayer, M. (1974). Frog goes to dinner. New York, NY: Dial Books for Young Readers.

McDuffie, A. M., Abbeduto, L., Lewis, P., Kover, S. T., Kim, J.-S., Weber, A., & Brown, W. T. (2010). Autism spectrum disorder in children and adolescents with fragile X syn- drome: Within-syndrome differences and age-related changes. American Journal on Intellectual and Develop- mental Disabilities, 115, 307–326.

McDuffie, A. S., Kover, S. T., Abbeduto, L., Lewis, P., & Brown, W. T. (2012). Profiles of receptive and expressive language abilities in males with comorbid fragile X syn- drome and autism. American Journal on Intellectual and Developmental Disabilities, 117, 18–32.

Mervis, C. B., & Robinson, B. F. (1999). Methodological issues in cross-syndrome comparisons: Matching proce- dures, sensitivity (Se), and specificity (Sp). Monographs of the Society for Research in Child Development, 64, 115–130.

Mervis, C. B., & Robinson, B. F. (2005). Designing measures for profiling and genotype/phenotype studies of individuals with genetic syndromes or developmental language disor- ders. Applied Psycholinguistics, 26, 41–64.

Miles, S., Chapman, R., & Sindberg, H. (2006). Sampling context affects MLU in the language of adolescents with Down syndrome. Journal of Speech, Language, and Hearing Research, 49, 325–337.

Miller, J. F. (1996). Progress in assessing, describing, and de- fining child language disorder. In K. N. Cole, P. S. Dale, & D. J. Thal (Eds.), Assessment of communication and language (Vol. 6, pp. 309–324). Baltimore, MD: Brookes.

Miller, J., & Iglesias, A. (2006). Systematic Analysis of Lan- guage Transcripts (SALT Version 9) [Computer software]. Madison: Language Analysis Lab, University of Wisconsin- Madison.

Moss, J., & Howlin, P. (2009). Autism spectrum disorders in genetic syndromes: Implications for diagnosis, intervention and understanding the wider autism spectrum disorder population. Journal of Intellectual Disability Research, 53, 852–873.

Paul, R., Cohen, D. J., Breg, W. R., Watson, M., & Herman, S. (1984). Fragile X syndrome: Its relations to speech and lan- guage disorders. Journal of Speech and Hearing Disorders, 49, 328–332.

Pierpont, E., Richmond, E. K., Abbeduto, L., Kover, S. T., & Brown, W. T. (2011). Contributions of phonological and

working memory to language development in adolescents with fragile X syndrome. Journal of Neurodevelopmental Disorders, 3, 335–347.

Price, J. R., Roberts, J. E., Hennon, E. A., Berni, M. C., Anderson, K. L., & Sideris, J. (2008). Syntactic complexity during conversation of boys with fragile X syndrome and Down syndrome. Journal of Speech, Language, and Hearing Research, 51, 3–15.

Rice, M. L., Smolik, F., Perpich, D., Thompson, T., Rytting, N., & Blossom, M. (2010). Mean length of utterance levels in 6-month intervals for children 3 to 9 years with and with- out language impairment. Journal of Speech, Language, and Hearing Research, 53, 333–349.

Roberts, J. E., Hennon, E. A., Price, J. R., Dear, E., Anderson, K., & Vandergrift, N. A. (2007). Expressive language during conversational speech in boys with fragile X syndrome. American Journal on Mental Retardation, 112, 1–17.

Roberts, J., Martin, G. E., Moskowitz, L., Harris, A. A., Foreman, J., & Nelson, L. (2007). Discourse skills of boys with fragile X syndrome in comparison to boys with Down syndrome. Journal of Speech, Language, and Hearing Research, 50, 475–492.

Roberts, J., Price, J., Barnes, E., Nelson, L., Burchinal, M., Hennon, E., ... Hooper, S. R. (2007). Receptive vocabulary, expressive vocabulary, and speech production of boys with fragile X syndrome in comparison to boys with Down syn- drome. American Journal on Mental Retardation, 112, 177–193.

Rogers, S. J., Wehner, E. A., & Hagerman, R. J. (2001). The behavioral phenotype in fragile X: Symptoms of autism in very young children with fragile X syndrome, idiopathic au- tism, and other developmental disorders. Journal of Devel- openetal and Behavioral Pediatrics, 22, 409–417.

Roid, G., & Miller, L. (1997). Leiter International Performance Scale—Revised. Wood Dale, IL: Stoelting.

Scarborough, H. S. (1990). Index of productive syntax. Applied Psycholinguistics, 11, 1–22.

Scarborough, H. S., Rescorla, L. R., Tager-Flusberg, H., Fowler, A. E., & Sudhalter, V. (1991). The relation of utterance length to grammatical complexity in normal and language-disordered samples. Applied Psycholinguistics, 12, 23–45.

Shaffer, J. P. (1986). Modified sequentially rejective multiple test procedures. Journal of the American Statistical Associ- ation, 81, 826–831.

Southwood, F., & Russell, A. F. (2004). Comparison of con- versation, freeplay, and story generation as methods of lan- guage sample elicitation. Journal of Speech, Language, and Hearing Research, 47, 366–376.

Tager-Flusberg, H., & Anderson, M. (1991). The develop- ment of contingent discourse ability in autistic children. Journal of Child Psychology and Psychiatry, 32, 1123–1134.

Tager-Flusberg, H., Rogers, S., Cooper, J., Landa, R., Lord, C., Paul, R., . . . Yoder, P. (2009). Defining spoken langauge benchmarks and selecting measures of expressive language development for young children with autism spec- trum disorders. Journal of Speech, Language, and Hearing Research, 52, 643–652.

Thordardottir, E. (2008). Language-specific effects of task demands on the manifestation of specific language impairment:

Kover et al.: Context Effects 1037

A comparison of English and icelandic. Journal of Speech, Lan- guage, and Hearing Research, 51, 922–937.

Ukrainetz, T. A., & Blomquist, C. (2002). The criterior va- lidity of four vocabulary tests compared with a language sample. Child Language Teaching and Therapy, 18, 59–78.

Van Borsel, J., Dor, O., & Rondal, J. (2008). Speech fluency in fragile X syndrome. Clinical Linguistics and Phonetics, 22, 1–11.

Verkerk, A. J., Pieretti, M., Sutcliffe, J. S., Fu, Y.-H., Kuhl, D. P., Pizzuti, A., . . . Zhang, F. P. (1991). Identification of a gene (FMR-1) containing a CGG repeat coincident with a breakpoint cluster region exhibiting length variation in fragile X syndrome. Cell, 65, 905–914.

Wagner, C. R., Nettelbladt, U., Sahlen, B., & Nilholm, C. (2000). Conversation versus narration in pre-school children with language impairment. International Journal on Lan- guage and Communication Disorders, 35, 83–93.

Westerveld, M. F., Gillon, G. T., & Miller, J. F. (2004). Spo- ken language samples of New Zealand children in conver- sation and narration. Advances in Speech-Language Pathology, 6, 195–208.

1038 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1022–1038 • August 2012

Copyright of Journal of Speech, Language & Hearing Research is the property of American Speech-Language-

Hearing Association and its content may not be copied or emailed to multiple sites or posted to a listserv

without the copyright holder's express written permission. However, users may print, download, or email

articles for individual use.