We compared the expressive syntax and vocabulary skills of 35 boys with fragile X syndrome and 27 younger typically developing boys who were at similar nonverbal mental levels. During a conversational speech sample, the boys with fragile X syndrome used shorter, less complex utterances and produced fewer different words than did the typically developing boys after controlling for their nonverbal MA, speech intelligibility, and mother's education. The boys with fragile X used less complex noun phrases, verb phrases, and sentence structure, but did not use fewer questions and negations. These findings suggest that the language difficulties in boys with fragile X reflect an overall expressive language delay and not a specific syntactic or vocabulary delay.
Males with fragile X syndrome generally have expressive language difficulties, with particular impairments in vocabulary, syntax, pragmatics, and speech intelligibility (Abbeduto & Hagerman, 1997; Hagerman & Hagerman, 2002; Mazzocco, 2000; Rice, Warren, & Betz, 2005). Expressive language appears to be more delayed than receptive language (Paul et al., 1987; Philofsky, Hepburn, Hayes, Hagerman, & Rogers, 2004; Roberts, Mirrett, & Burchinal, 2001). Most males with fragile X syndrome have moderate mental retardation and difficulties with social and adaptive skills (Cohen, 1988; Dykens, Hodapp, Ort, & Leckman, 1993, 2000; Hagerman & Hagerman, 2002; Reiss & Freund, 1992). Fragile X syndrome, the most common inherited genetic cause of mental retardation (Dykens, Hodapp, & Finucane, 2000; Hagerman & Hagerman, 2002), is caused by a mutation in the Fragile X Mental Retardation-1 (FMR-1) gene.
The mutation arises due to excessive repeats of the trinucleotide (CGG) sequence, which methylates the gene, causing a failure to produce the FMR-1 protein, which is thought to be critical for brain development (Hagerman & Hagerman, 2002; Imbert & Mandel, 1995; U.S. Department of Health and Human Services, 2003). In this study, we examined whether the language difficulties that occur in boys with fragile X syndrome are specific to vocabulary or syntax or represent more general expressive language difficulties evident in both vocabulary and syntax. Asynchronous development of syntax and vocabulary would suggest that different mechanisms underlie each domain and are differentially impaired in fragile X syndrome. Because this syndrome usually affects both overall development and communication development more severely in boys than girls (Abbeduto & Hagerman, 1997; Hagerman & Hagerman, 2002; Loesch et al., 2003; Reiss & Dant, 2003), all participants in this study were boys. Only a subgroup of females with fragile X syndrome have mental retardation (Hesl et al., 2001; Keysor & Mazzocco, 2002); therefore, inclusion of these females would not address the full range of cognitive and communicative variation among females.
The literature is inconsistent regarding whether the expressive vocabulary and syntax levels of males with fragile X syndrome are similar to their cognitive levels. Abbeduto et al. (2001) and Philofsky et al. (2004) examined overall expressive language without separating out syntax and vocabulary. Abbeduto and his colleagues reported that the expressive language of 18 children and adults with fragile X syndrome was similar to a mental age (MA) matched typically developing control group. However, Philofsky and colleagues found that a combined expressive language quotient was lower than the nonverbal quotient of 18 preschool boys with fragile X syndrome. In four other studies, researchers looked specifically at vocabulary and/ or syntax (Madison, George, & Moe schler, 1986;Paul, Cohen, Breg, Watson, & Herman, 1984; Paul et al., 1987; Sudhalter, Maranion, & Brooks, 1992). In a study of expressive syntax, Paul and colleagues (1984) reported three case studies of 10- to 13-year-old boys with fragile X syndrome whose age level for complex syntax and utterance length during conversational speech was lower than their nonverbal MA. In contrast, in a later study, Paul et al. (1987) found that 2 adult males with fragile X syndrome had expressive vocabulary, utterance length, and grammatical complexity levels that were commensurate with their nonverbal cognitive skills. Madison et al. reported that the expressive vocabulary scores of 4 adult males with fragile X syndrome in the same family were higher than their MAs, although their syntax was mostly comparable to their MAs. However, the 1 boy with fragile X syndrome studied had relatively similar delays in both expressive vocabulary and syntax relative to his MA. Sudhalter and col leagues found that 11 males (boys and adults) with fragile X syndrome who did not have autistic disorder had more semantic errors, but not more syntactic errors, on a sentence completion test than did younger typically developing girls and boys.
These studies present a complex picture with an uneven pattern of strengths and weaknesses re garding the expressive language in males with fragile X syndrome. Conflicting findings across the studies may be due to age or developmental differences of participants, differences in tasks used to measure expressive language skills, the inclusion of girls in one study (i.e., Abbeduto et al., 2001), and differences in research design, with some investigators comparing the expressive vocabulary and/or syntax of individuals with fragile X syndrome to younger typically developing children (Abbeduto et al., 2001; Paul et al., 1984; Philofsky et al., 2004; Scarborough, Rescorla, Tager-Flusberg, Fowler, & Sudhalter, 1991; Sudhalter et al., 1992) and others comparing expressive vocabulary and/or syntax levels to nonverbal cognitive levels within individuals (Madison et al., 1986; Paul et al., 1987).
Previous researchers examining factors influencing language development have emphasized the role of several child and environmental factors, such as nonverbal mental level, speech intelligibility, and mother's level of education (Chapman, Seung, Schwartz, & Kay-Raining Bird, 1998;Dollaghan et al., 1999; Roberts et al., 2001; Sigman & Ruskin, 1999). The relation between cognition and language is consistent with social interactionist perspectives (Bates & MacWhinney, 1982; Craig, 1995; McTear & Conti-Ramsden, 1992), which emphasize the bi-directional influence of linguistic and nonlinguistic development. Nonverbal mental level has been shown to be related to language skills in children with fragile X syndrome (Roberts et al., 2001) and in children with other developmental disabilities (see Rice et al., 2005, for a review). The role of speech difficulties in expressive language has been described in children with Down syndrome (Chapman et al., 1998) and has been implicated by others as a possible influence on language production in fragile X syndrome (Abbeduto & Hagerman, 1997; Wolf-Schein et al., 1987). A number of researchers studying typically developing children and children with developmental disabilities have demonstrated an association between maternal education and children's language development (Dollaghan et al., 1999; Fewell & Deutscher, 2003; Rice, Spitz, & O'Brien, 1999). In the current study, our focus was specifically on the relative deficits in vocabulary and syntax skills, not on the direct effects of these potential child and environmental effects. Therefore, we examined whether syntax and vocabulary deficits occurred in boys with fragile X syndrome relative to the typically developing boys after controlling for the boys' nonverbal mental level, their speech intelligibility, and their mothers' level of education.
We were also interested in following up the work of Sudhalter, Scarborough, and Cohen (1991) and Scarborough and colleagues (1991) regarding whether the syntactic complexity of males with fragile X syndrome is deviant or delayed at varying sentence lengths. Scarborough and colleagues found a high correlation (.93) between mean length of utterance (MLU) and the Index of Productive Syntax, a measure of grammatical complexity, for 10 males with fragile X syndrome between 5 and 19 years of age whose MLUs were below 3.0 (Scarborough et al., 1991). Yet, for 6 males with MLUs above 3.0, the correlation was .59, suggesting the use of more limited syntactic variation within longer utterances as compared to shorter utterances (Scarborough et al., 1991). Scarborough and colleagues concluded that the relationship of language complexity to utterance length in males with fragile X syndrome was generally similar to typically developing children; however, for males with fragile X syndrome at a higher linguistic level, MLU overestimated syntactic abilities, possibly suggesting a specific syntactic deficit. These results differed from the findings of Sudhalter and colleagues (1991) who reported on 19 boys and adult males with fragile X syndrome. They found that the relationship between MLU and Index of Productive Syntax did not differ for lower versus higher MLU levels and was highly correlated, r = .88, and similar in magnitude to that of reports of typically developing children (Scarborough, 1990), r = .88. Thus, results of Sudhalter et al.'s study suggest the presence of a language delay rather than a language deficit in males with fragile X syndrome. Given that both of these studies included small overall sample sizes, we were interested in examining the correlations between utterance length and grammatical complexity for boys with fragile X syndrome who were at lower versus higher linguistic levels.
In the current investigation, we had three goals. First, we compared the syntax and vocabulary skills of boys with fragile X syndrome to typically developing boys at similar MA levels. We sought to understand whether language difficulties that occur in boys with fragile X syndrome are specific to syntax or represent more general expressive language difficulties and are evident in both vocabulary and syntax. Further, we examined whether differences in the syntax and vocabulary skills between boys with fragile X syndrome and the typically developing group were present after controlling for three factors (boys' nonverbal cognitive skills, boys' speech intelligibility, and maternal education) that have been shown in the literature to be important for language development. Second, we were particularly interested in whether group differences would occur in overall measures of vocabulary and syntax and/or whether there would be differences in specific areas of syntactical complexity. We analyzed syntactical complexity using the Index of Productive Syntax (Scarborough, 1990) in order to look at syntax complexity for noun phrases, verb phrases, questions/negations, and sentence structure. Third, we examined whether the relationship between utterance length and syntactic complexity differed for boys with fragile X syndrome who used shorter versus longer utterance lengths. Although we did not include a disability comparison sample in this study, we examined several questions specific to the expressive language profile of individuals with fragile X syndrome and compared their performance to that of typically developing control participants of similar MA.
Boys with fragile X syndrome
Participants were 35 boys with fragile X syndrome recruited from a larger study examining language and speech development of young boys with fragile X syndrome (Roberts et al., 2005). All boys in the fragile X syndrome group had a confirmed diagnosis of full mutation fragile X syndrome, although we lacked information on mosaicism and FMRP. They were referred from genetic clinics in Maryland, Virginia, North Carolina, South Carolina, and Georgia. Each child with fragile X syndrome had begun to use word combinations in their spoken language, per parental report. We excluded boys whose hearing levels averaged greater than 25 dB HL in the better ear, as measured using pure tone thresholds for the frequencies 500, 1,000, 2,000, and 4,000 Hz. We also excluded boys who met the criteria for autism on the Autism Diagnostic Observation Schedule (Lord, Rutter, DiLavore, & Risi, 2001), which is described in the section below (Language Sampling). It was beyond our scope in this paper to examine the effects of co-morbid fragile X syndrome and autism on expressive language development. Excluding boys with fragile X syndrome and autism also allowed us to more readily compare our results with previous work in which boys with fragile X syndrome without autism were examined (e.g., Sudhalter, Maranion, & Brooks, 1992). A boy was excluded if his total scores on the Autism Diagnostic Observation Schedule met or exceeded the autism cutoff for three domains: Communication, Social Interaction, and combined Communication + Social Interaction. At the time of the assessment, the children ranged in age from 34 to 167 months (M = 109.4 months, SD = 34.7). Twenty-eight of the children were Caucasian, 4 were African American, and 2 were other. Nonverbal MA, as measured by the Leiter International Performance Scale–Revised—Leiter-R (Roid & Miller, 1997), averaged 58.7 months (SD = 12.3, range = 29 to 80). Leiter-R mean IQ was 60.5 (SD = 17.3). Nine of the mothers of children with fragile X syndrome had a high school education; 26 had attended college for at least one year.
Typically developing boys
These participants were 27 typically developing boys recruited from the same larger study as the boys in the fragile X syndrome group (Roberts et al., 2005). They ranged in age from 25 to 79 months (M = 53.6, SD = 14.5) and were recruited through preschools, day care centers, children's organizations, and word of mouth from towns in central North Carolina. Typically developing boys were excluded if, according to parental report, they had mental retardation, a developmental disability, autistic spectrum disorder, hearing loss, speech or language difficulties, or if they were receiving speech and language therapy. Prior to the initial speech and language assessment, we also screened typically developing boys using the Leiter-R, as described below, and excluded a boy who scored more than 1.5 SD above or below the mean. In subsequent testing, we excluded any child who scored in the autism range on the Autism Diagnostic Observation Schedule as well as those boys who scored more than 1.5 SD below the mean on any of the standardized language measures. Twenty of the children were Caucasian, 5 were African American, and 2 were other. Their Leiter-R mean age equivalent was 57.7 months (SD = 12.4, range = 34 to 85), and their Leiter-R mean IQ was 108.4 (SD = 7.3). They were recruited if their chronological age (CA) fell within the range of nonverbal developmental ages seen in the boys with fragile X syndrome. Mean nonverbal MA did not differ between the boys with fragile X syndrome and typically developing boys (boys with fragile X syndrome = 58.7 months, typically developing boys = 57.7 months). English was the primary language of all participants. Two of the mothers had a high school education; the remaining 25 mothers had one or more years of college.
Depending on parents' preferences, children were tested either in their homes, at their schools, or at a university research center. All sessions were audiotaped using a portable Digital Auditory Tape TASCAM (DA-P1) recorder with a Shure WBH headset microphone system and videotaped using a Sony Digital8 video camera (Model DCR-TVR27). All participants received the Leiter-R and language samples were obtained.
Leiter International Performance Scale–Revised
The Brief IQ composite from the Leiter-R measured nonverbal cognitive abilities (Roid & Miller, 1997). The Brief IQ composite takes 30 to 45 minutes to complete and assesses spatial reasoning, sequencing, and patterning skills. Adequate reliability and validity have been documented (Roid & Miller, 1997).
The conversational interaction between the examiner and child during the Autism Diagnostic Observation Schedule served as the language sample. This measure is a semi-structured play session designed to elicit language and social behaviors from children. It takes about 45 minutes to complete and includes a variety of developmentally appropriate social and toy-based interactions. Three modules of the Autism Diagnostic Observation Schedule based on each child's developmental and language levels were administered. Module 1 was given to children who were using single words to simple phrases; Module 2, for children who ranged from using three-word phrases to being verbally fluent; and Module 3, for children and young adolescents who were verbally fluent (Lord et al., 2001). Although all of the children were able to produce some simple phrases, 6 children with fragile X syndrome received Module 1 because they did not use three-word utterances regularly. No typically developing children received Module 1 because all of these participants were over 24 months of age and regularly used phrases of at least three words. Thirty-eight children received Module 2 (21 with fragile X and 17 typically developing). Module 3 was administered to 18 children (8 with fragile X syndrome and 10 typically developing). Regardless of the module administered, most child utterances were from play sessions. Although the physical objects used in the play sessions differed across modules (i.e., toys in Module 1 were larger in order to circumvent choking hazards), the social interactions were the same in that the examiner presented the toys to the child and encouraged the child to talk about his play and to interact with the examiner during play. For those children who received Module 1, most utterances were from free play and a few utterances were from a pretend birthday party, a snack break, and a balloon-based interaction. For those children given Module 2 or 3, most utterances were from the make-believe play and joint interactive play. A few utterances for Module 2 were taken from birthday, snack, and balloon interactions, which were also Module 1 activities. Adequate reliability and validity of the Autism Diagnostic Observation Schedule has been documented for measuring autistic disorders across a range of ages and severity levels (Lord et al., 2000).
Trained research assistants transcribed and coded all intelligible child and examiner utterances from videotapes using the Child Language Data Exchange System format— CHILDES (MacWhinney, 1995). Utterances were segmented when an examiner or child stopped talking and the other began speaking, based on speaker intonation (the rising or falling of intonation to determine the end of a thought) and a noticeable pause. A second research assistant reviewed the transcript while listening to the audiotape and marked any discrepancies. Finally, a third researcher verified the changes, checked all coding, and made any necessary corrections to the transcripts. For eight transcripts (12%: 4 fragile X and 4 typically developing), a second research assistant independently transcribed the language sample to check the percentage of agreement for the transcription of words. Reliability was computed using morpheme by morpheme agreement for children's utterances and was 93% (range from 87% to 98%).
The first 100 useable utterances for each child were used for all analyses. Following the guidelines specified by Scarborough (1990) for use of the Index of Productive Syntax, utterances were excluded from analyses if they were partially or fully unintelligible, an exact self-repetition (e.g., “I have a blue fire truck. I have a blue fire truck”), an imitation of a previous adult statement, a yes or no response to a question, or a portion of a routine (e.g., singing “Happy Birthday”). Five children with fragile X syndrome did not produce 100 useable utterances during the Autism Diagnostic Observation Schedule. However, because the Index of Productive Syntax analysis requires a sample of 100 useable utterances, transcribers reviewed tapes from other language assessments administered during the same visit as the Autism Diagnostic Observation Schedule, and spontaneous utterances produced by the child during other assessments were used to complete the samples. Child responses elicited by test questions or prompts were not used. For these 5 children, 7, 19, 21, 31, and 38 spontaneous utterances were collected from other assessments.
Measures of Speech and Language
Mean length of utterance
Mean length of utterance is a ratio of the total number of morphemes divided by the total number of utterances, as described by Brown (1973). We computed MLU using CHILDES (MacWhinney, 1995). The MLU is a well-established measure of syntactic development, with documented correspondence to the degree of grammatical complexity (Brown, 1973; Paul, 2001; Scarborough et al., 1991).
Number of different words
We computed the number of different words, defined as “the number of different word types used,” using CHILDES. This measure provided an index of the child's variety of vocabulary and has been implemented in previous studies as a measure of a child's lexicon (Chapman et al., 1998; Feldman, et al., 2005; Pan, Rowe, Singer, & Snow, 2005).
Index of Productive Syntax
We analyzed syntactic complexity using the Index of Productive Syntax (Scarborough, 1990). This instrument measures the development of syntax and morphology. Designed for analyzing language transcripts of children between 2 and 4 years of age developmentally, the Index of Productive Syntax scores for the presence of 56 syntactical and morphological forms, plus 4 items labeled “other.” Examples of forms include using a two-word phrase (“my sock”), an adverb (“spoke loudly”), and a direct object (“she has the book”). Each of these forms is given a maximum of 2 points for occurrence in the language sample. Zero occurrences of an item indicate that the child does not yet have this skill, one occurrence reflects emerging abilities with the skill, and two occurrences indicate productive use of the item (Scarborough, 1990). The total points are summed into four composite scores for Noun Phrases, Verb Phrases, Questions/Negations, and Sentence Structure, as well as an Index of Productive Syntax Total Score (the sum of the four composites). The maximum scores possible are 24 for Noun Phrases, 34 for Verb Phrases, 22 for Questions/Negations, 40 for Sentence Structure, and 120 for the total Index of Productive Syntax score. Good reliability and age sensitivity have been reported for children between 2 and 4 years of age (Scarborough, 1990). This instrument has been used with typically developing children as well as children with Down syndrome, fragile X syndrome, autistic disorder, and specific language impairment (Scarborough et al., 1991). A trained research assistant scored all Index of Productive Syntax transcripts, and 13% of those transcripts were independently scored by the second author to check for reliability of scoring. Interrater reliability for the Index of Productive Syntax scores was .82 (.80 for 5 boys with fragile X syndrome and .73 for 3 typically developing boys). Generally, kappa values above .75 indicate excellent reliability, and values between 0.4 to 0.75 denote good reliability (Landis & Koch, 1977; Rosner, 2000).
The percentage of utterances that were completely intelligible was also computed using CHLDES by dividing the total number of completely intelligible utterances by the total number of utterances in the transcript (i.e., fully intelligible plus partially or fully unintelligible utterances).
To address the first two research goals, we used a multivariate analysis of covariance (MANCOVA) to compare the syntax and vocabulary levels of the boys with fragile X syndrome and the typically developing boys. There were two overall outcome measures of syntax (Index of Productive Syntax Total and MLU), four specific outcome measures of syntax (Noun Phrases, Verb Phrases, Questions/Negations, and Sentence Structure), and one overall outcome measure of vocabulary (number of different words). We were interested in determining whether the expressive syntax and vocabulary outcomes were the same for the boys in both groups, while controlling for the child's nonverbal (Leiter-R) MA, speech intelligibility, and maternal education. The number of years of maternal education was computed using the highest level of education reported by the mother.
The MANCOVA was performed in two steps: first on the three general syntax and vocabulary outcomes (MLU, Index of Productive Syntax Total Score, total number of different words) and, second, on the four subscales of the Index of Productive Syntax (Noun Phrases, Verb Phrases, Questions/Negations, and Sentence Structure). The Wilks' Lambda multivariate F test was reported for both of these MANCOVAs. In all analyses, Leiter-R MA, the percentage of intelligible utterances, and the mother's years of education were used as covariates. Interactions between group (fragile X syndrome vs. typically developing) and Leiter-R MA, the percentage of intelligible utterances, and the mother's years of education were also included in the model. Because the interactions with group (for Leiter-R MA, speech intelligibility, or maternal education) were significant for only the Sentence Structure subscale of the Index of Productive Syntax, the interactions were removed from the analyses for Noun Phrases, Verb Phrases, and Questions/Negations. Adjusted means are reported where there was a significant group effect and Cohen's d (Cohen, 1988) was computed as a measure of effect size.
In a follow-up analysis, we explored whether language difficulties occurred for overall expressive language or were particular to expressive syntax. To test this, we examined two outcomes (MLU, Index of Productive Syntax Total Score), using a MANCOVA with the same covariate set as listed above, except we included the total number of different word types as an additional predictor.
To address the third research goal, examining the relationship between utterance length and syntactical complexity for boys with shorter versus longer utterances, we computed correlations between MLU and Index of Productive Syntax scores for boys with fragile X syndrome with an MLU above 3.0 and below 3.0.
Variability Among Children
Table 1 shows the descriptive statistics for the syntax and vocabulary outcomes and the covariates. Table 2 presents the correlations among the overall measures of expressive language (MLU, Index of Productive Syntax Total Score, and number of different words) and the covariates (nonverbal MA, speech intelligibility, and maternal education). Both groups of boys demonstrated high levels of correlation between the various measures of expressive language and nonverbal MA. Overall, MLU and Index of Productive Syntax Total Score were highly correlated both for boys with fragile X syndrome, r = .85, and for typically developing boys, r = .83. Table 3 shows the correlations among the subscales of the Index of Productive Syntax for boys with fragile X syndrome and typically developing boys. For boys with fragile X syndrome, all four areas of syntax were highly correlated. For typically developing boys, Sentence Structure, Verb Phrases, and Noun Phrases were correlated with each other, but Questions/ Negations were not correlated with any of the other measures. Further, the correlations between MLU and Index of Productive Syntax scores for boys with fragile X syndrome with an MLU above 3.0, r = .73, and below 3.0, r = .69, were similar in strength.
Overall multivariate analyses
The results of a MANCOVA on the global measures of syntax and vocabulary (i.e., MLU, Index of Productive Syntax Total Score, number of different words) indicated a significant overall effect for group and for MA (see Table 4). The boys with fragile X syndrome scored lower than did the typically developing boys across the more global measures, and boys with greater nonverbal mental abilities had higher scores.
Mean length of utterance
We found main effects for group and MA. The effect for group indicated that boys with fragile X syndrome lagged behind typically developing boys in MLU. Boys with fragile X syndrome produced sentences of an average length of 3.41 morphemes, whereas typically developing boys' sentences were 4.67 morphemes long on average. This represents a large effect, d = 1.69. The effect for MA indicated that there was a positive relationship between MA and length of utterance.
Total number of word types
There were significant effects for group and MA. The effect for group indicated that boys with fragile X syndrome did not use as many word types as did typically developing boys. They used an average of 130.3 different word types, whereas typically developing boys used 169.1 word types, on average. The difference is a large effect, d = 1.63. The effect for MA indicated that there was a positive relationship between MA and number of word types.
Index of productive syntax
There were significant effects for group and MA. The effect for group indicated that the boys with fragile X syndrome were delayed on their syntactic development when compared to typically developing boys. They had an average score of 67.53 on the Index of Productive Syntax Total, whereas typically developing boys had an average score of 82.65. The difference is a large effect, d = 1.68. The effect for MA indicated that there was a positive relationship between MA and Index of Productive Syntax Total.
Composite scores on the Index of Productive Syntax
The MANCOVA on the specific measures of syntax (i.e., Noun Phrases, Verb Phrases, Questions/ Negations, and Sentence Structure) indicated a significant overall effect for group and for MA across all measures (see Table 4). There were significant effects for group and MA on both the Noun Phrases and Verb Phrases composites, indicating that boys with fragile X syndrome were delayed in their use of noun phrases and verb phrases. Large effect sizes were also found on both the Noun Phrases and Verb Phrases, ds = 1.66 and 1.09, respectively) composites. For the Noun Phrases, Verb Phrases, and Sentence Structure composites, the effect for MA indicated a positive relationship between MA and the composite score. On the Question/Negation composite, there were no significant effects for group or MA, indicating that the boys with fragile X syndrome and typically developing boys scored similarly with respect to the use of questions and negations. On the Sentence Structure composite, there was a significant interaction between group and the percentage of intelligible utterances, indicating a differential effect for boys with fragile X syndrome and typically developing boys for the percentage of intelligible utterances (see Table 5 ). For boys with fragile X syndrome, the effect of the percentage of intelligible utterances was not significantly different from zero; therefore, it had no effect on the Sentence Structure score. For typically developing boys, for every percentage point above the mean on speech intelligibility, their Sentence Structure score increased by .55 points.
Using vocabulary as an additional covariate to predict MLU and Index of Productive Syntax
In this follow-up analysis, we added an additional covariate, the number of different words, to the three covariates and examined the outcomes of MLU and Index of Productive Syntax Total. In the results, the number of word types was the only significant predictor of both Index of Productive Syntax Total Score and the MLU, Fs(1, 61) = 57.82 and 58.10, respectively, ps < .001. No other main or interaction effects were significant either as individual effects or as multiple degree of freedom tests. This indicates that the observed delays for boys with fragile X syndrome represent an overall expressive language delay and not a specific syntactic delay.
In this study, we addressed three research goals. First, we examined whether the language difficulties of boys with fragile X syndrome are specific to syntax or represent more general expressive language difficulties. We found that these boys used shorter, less complex utterances and produced fewer different words than did the typically developing boys during a conversational speech sample, after controlling for their nonverbal MA, speech intelligibility, and mothers' education level. This finding suggests that boys with fragile X syndrome show a delay in their expressive language skills as compared to younger typically developing boys, and the delay is an overall expressive language delay and not specific to syntax or vocabulary. This synchrony between syntax and lexical development suggests that both of these linguistic domains may be driven by the same underlying mechanisms. Our second goal was to examine group differences in overall measures of vocabulary and syntax and/or determine whether these differences occur in specific areas of syntactic complexity. The boys with fragile X syndrome used less complexity in their noun phrases, verb phrases, and sentence structure, but not in their questions and negations, compared to the typically developing boys after controlling for the child and family background factors. The third goal was to explore the relation between utterance length and syntactical complexity for boys with fragile X syndrome who used shorter versus longer utterances. We found a similar relationship between syntax complexity and utterance length for these boys who had shorter and longer MLUs.
General Expressive Language Difficulties Not Specific to Syntax or Vocabulary in Fragile X Syndrome
Our findings suggest that boys with fragile X syndrome show an overall expressive language delay not specific to vocabulary or syntax. During the 100-utterance conversational speech sample, the boys with fragile X syndrome produced approximately 39 fewer words than did the typically developing boys after controlling for the children's cognitive and speech skills and maternal education. After controlling for the covariates, the boys with fragile X syndrome had a mean MLU of 3.4 and a total Index of Productive Syntax Score of 67.5, reflecting an approximate syntax level of 3 years, according to Brown's (1973) and Scarborough's (1990) data. The typically developing boys had an MLU of 4.7 and a total Index of Productive Syntax Score of 82.7, or an approximate syntax level of 4 years. Although the boys with fragile X syndrome differed from the younger typically developing boys, after controlling for the covariates, it is important to note that there was considerable individual variability, particularly for the boys with fragile X syndrome. For example, one 8-year-old boy with fragile X had an MLU of 2.8, or an age equivalent of about 37 months, and a total Index of Productive Syntax Score of 59, or age equivalent of 30 months. In comparison, another 8-year-old with fragile X had an MLU of 3.8 (age equivalent of about 50 months) and a total Index of Productive Syntax Score of 76 (or age equivalent of about 39 months).
Our findings are consistent with those of Sudhalter et al. (1992), who reported more semantic errors among 11 boys and adults with fragile X syndrome than in younger typically developing children as well as within-group comparisons that reported, as compared to their nonverbal mental skills, lower overall expressive language for 18 preschool boys with fragile X syndrome (Philofsky et al., 2004) and shorter utterances and less complex syntax for 3 boys with fragile X syndrome (Paul et al., 1984). However, our results differ from two studies in which researchers reported similar levels of expressive language in 18 children and adults with fragile X syndrome (Abbeduto et al., 2001) and a similar number of syntax errors in 11 boys and adult males with fragile X syndrome (Sudhalter et al., 1992) compared to typically developing samples. Our results also differ from within-group comparisons showing that nonverbal cognitive skills were at similar levels as expressive vocabulary, utterance length, and grammatical complexity in 2 adult males with fragile X syndrome (Paul et al., 1987); at mostly similar levels as syntax (although higher in expressive vocabulary) for 3 adult males with fragile X syndrome (Madison et al., 1986); and at similar levels for expressive vocabulary and syntax in 1 boy with fragile X syndrome (Madison et al., 1986).
Several factors may account for the differences in our findings versus previous findings. Our sample differed from those in previous studies in that it included more participants, contained only boys (unlike Abbeduto et al.'s, 2001, sample), and the sample was generally younger than those of other studies (i.e., Abbeduto et al., 2001; Madison et al., 1986; Paul et al., 1987; Sudhalter et al., 1992). In addition, we employed specific measures of syntax and vocabulary, whereas other researchers generally used global measures of expressive language. Moreover, as noted by Abbeduto and colleagues (2003), the within-group comparison of language and nonverbal cognition used by Madison and colleagues (1986) and Paul et al. (1987) is a problematic research design because different standardized tests (i.e., those assessing cognition and language) vary in their norming and are difficult to compare to each other. Given the sound design of our study and large number of participants, our findings lend strong support to the notion that syntax and vocabulary are both impaired compared to nonverbal MA expectations for boys with fragile X syndrome.
It is important to note that these study findings are not consistent with our previous report (including some of the same children) in which we found that the expressive vocabulary skills of boys with fragile X syndrome without autistic disorder did not differ from the younger typically developing boys as measured using a standardized expressive vocabulary test, after adjusting for nonverbal cognitive skills (Roberts et al., in press). It is possible that the sampling context of a conversational speech sample in the current study versus the standardized test used in our previous study (Roberts et al., in press) affected the results. Sampling context has been shown to affect the frequency and type of language a child uses (Bornstein, Painter, & Park, 2002; Eisenberg, 1997; Gazella & Stockman, 2003). Conversational samples are generally more useful in eliciting day-to-day knowledge, more of what the child has “ready access” to, may be more affected by processing capacity and production constraints, and possibly underestimate a child's linguistic knowledge. Standardized tests, on the other hand, may elicit the maximum possible skills, may be less taxing on processing and production, and may provide the opportunity for a child to demonstrate knowledge of a particular referent or form.
Differences in Overall Measures of Vocabulary and Syntax or Specific Areas of Syntactic Complexity
After controlling for the covariates, we found that the boys with fragile X syndrome used less complexity in their noun phrases, verb phrases, and sentence structure than did the typically developing boys, but the groups did not differ in their use of questions and negations. The boys with fragile X syndrome were skilled in their ability to ask questions and use negation. They were, for example, able to use routine questions to say “Who is that?” or “What are these?” as well as more complex forms, such as the inverting of an auxiliary, modal, or copula to say “How does it open?” The boys with fragile X syndrome also used simple negations, such as “no blocks” and more complex forms for negation of copula, modal, or auxiliary to say “It doesn't work.” In contrast, the boys with fragile X syndrome used many different nouns, pronouns, plurals, and two-word noun phrases, such as “the plate” or “his papa,” but not as many more complex noun structures involving three-word noun phrases, two-word phrases before verbs, or bound morphemes on nouns, such as “the little ladder” or “my dad eats.” The boys with fragile X syndrome used verb phrases with copula, present auxiliary, and present modal in utterances, such as “I'm fast” and “Dad is playing.” However, more complex verb structures, such as present and past tense verb markers (“-ed;t3” “-s” endings), or past tense copula, auxiliary, or modal, were not frequently used. Further, the boys with fragile X syndrome regularly used sentence structures of subject–verb–object (e.g., “I want more”) and simple conjunctions (e.g., “and that”), but did not often use more complex sentence structures, such as relative clauses (“it's something that you eat”) or bi-transitive predicates (“get the soccer ball for me”).
Relation Between Utterance Length and Syntactical Complexity for Boys With Fragile X Syndrome
We found a similar relationship between syntax complexity and utterance length for the boys with fragile X syndrome who had shorter and longer MLUs, supporting the findings of Sudhalter et al. (1991) but not those of Scarborough and colleagues (1991). The very high degree of correlation between the syntactic complexity as measured by the Index of Productive Syntax Total Scores and utterance length was similar to the correlation previously reported for typically developing children (Scarborough et al., 1991; Sudhalter, Cohen, Silverman, & Wolf-Schein, 1990). Our findings suggest that the development of expressive syntax is delayed but not deviant in boys with fragile X syndrome when MLU is controlled.
Possible Explanations for Shorter and Less Complex Utterances in Boys With Fragile X
What may account for the shorter, less complex utterances produced by the boys with fragile X syndrome? One possibility is that their expressive syntax is limited by their vocabulary abilities. This is consistent with our finding that the boys with fragile X syndrome produced significantly fewer different words than did the typically developing boys. Further, the group differences on syntax skills were not found once vocabulary skills were used as a predictor variable. That is, when we included the number of different words as a predictor, the boys with fragile X syndrome did not differ from the typically developing boys, and all boys with higher vocabulary scores had higher syntax scores. A second possibility is that other factors (e.g., working memory, cognitive processing skills) may be influencing both expressive vocabulary and syntax. Impairments in executive functioning, sequential processing, and working memory skills have been reported for males with fragile X syndrome (Cornish, Sudhalter, & Turk, 2004; Hodapp, Leckman, Dykens, Sparrow, Zelinsky, & Ort, 1992; Munir, Cornish, & Wilding, 2000). The production of more complex syntactic structures, such as three-word noun phrases and relative clauses, would potentially place more demands on working memory than would simpler constructions, such as simple noun or verb phrases, suggesting a link to the reported impairments in sequential processing and working memory (Cornish et al., 2004; Hodapp et al., 1992; Munir et al., 2000). Future research should investigate the relationship between cognitive abilities, such as working memory, and more advanced syntactic abilities.
We studied possible cognitive and speech constraints and environmental influences that could be related to the expressive language development of the boys with fragile X syndrome in this study. For the first covariate, nonverbal MA, we found that both young boys with fragile X syndrome and typically developing boys with higher nonverbal MAs had longer utterances, used more complex syntax, and used more different words during conversational speech. This finding is consistent with a previous study of the language skills of boys with fragile X syndrome (Roberts et al., 2001) and other children with developmental disabilities (Chapman et al., 1998; Sigman & Ruskin, 1999). For the second covariate, speech intelligibility, we found that typically developing boys who were more intelligible had higher scores on sentence structure but that speech intelligibility did not affect syntax or vocabulary scores for boys with fragile X syndrome. Although no researchers conducting studies with children who have fragile X syndrome have reported that speech intelligibility is related to the language of children with fragile X syndrome, the linkage has been described in one study of children with Down syndrome (Chapman et al., 1998) and has been suggested to be important for individuals with fragile X syndrome (Abbeduto & Hagerman, 1997; Wolf-Schein et al., 1987). For example, children may not be able to effectively use their language skills if intelligibility is poor (Chapman et al., 1998). For the third covariate, maternal education, it was surprising that we did not find that maternal education predicted children's expressive language because a number of studies have demonstrated an association between maternal education and children's language development (Campbell et al., 2003; Dollaghan et al., 1999; Fewell & Deutscher, 2003; Rice et al., 1999). In our own previous work using some of the same study children, we have found a relationship between maternal education and standardized measures of expressive as well as receptive vocabulary. It is possible that standardized measures of expressive vocabulary may be more sensitive to maternal education than is the language elicited during conversational samples, given the additional processing and production skills involved in conversation. In addition, the limited variability of maternal education in the current study may have prevented us from finding a relationship between maternal education and language.
Study Limitations and Strengths
The current project has several limitations that suggest directions for future research. First, only 35 boys with fragile X syndrome were included in this study. Given the variability in the language skills of the boys with fragile X syndrome, “mean” analyses may mask interesting subgroups and their developmental patterns. Future studies with more children who have fragile X syndrome would enable researchers to explore the variability inherent in fragile X syndrome. Second, only typically developing boys were included as a comparison group in the current study, so we cannot directly explore whether the expressive language delay we found in the current study is specific to fragile X syndrome or if it would be expected in other children with mental retardation. However, we note that an asynchrony in lexical and syntactic development has been consistently demonstrated in individuals with Down syndrome (Chapman et al., 1998; Laws & Bishop, 2003; Miller, 1988; Rice et al., 2005; Rosin, Swift, Bless, & Vetter, 1988); our data suggest a different profile for fragile X syndrome. Third, the mean developmental level of our fragile X syndrome sample was in the mild mental retardation range, as indicated by a mean IQ of 60, and is higher than the moderate to severe levels of mental retardation typically reported for males with fragile X syndrome (Bennetto & Pennington, 2002), which may limit the generalizability of our findings. Fourth, the relatively small sample size in the current study precluded our ability to statistically test specific syntactic structures; in future studies with larger samples, researchers could explore whether clusters of syntactic structures, such as negations or conjunctions, are more challenging for boys with fragile X syndrome. Fifth, in the current investigation we tested boys at a range of ages, but assessed each child only one time. Longitudinal studies would provide information on the patterns of language growth over time in boys with fragile X syndrome. Sixth, other characteristics often found in children with fragile X syndrome, such as oral apraxia (in which the child has difficulty coordinating his tongue and lips to produce speech), psychological characteristics (such as social anxiety that could affect conversation), and processing difficulties (such as working memory or word retrieval problems) could affect the boys' ability or willingness to produce complex expressive language, independently of their actual syntax or vocabulary skills, and warrant further study. Further, the role of the boys' genetic status (e.g., their FMRP levels) in their expressive language ability is important to examine. Finally, we used a semi-naturalistic conversational speech sample to assess syntax and vocabulary production. A few studies have demonstrated that narrative tasks, such as storytelling (Abbeduto, Benson, Short, & Dolish, 1995; Dollaghan, Campbell, & Tomblin, 1990), and elicited tasks (Eisenberg, 1997) may elicit more syntactically complex utterances than conversation.
There are also several strengths to this research. First, compared to previous studies, we had a relatively large sample of 35 boys with fragile X syndrome. Second, we eliminated boys with fragile X syndrome and autism so we could exclude the co-occurring effects of fragile X syndrome and autism. Third, in comparison to existing investigations of language development in fragile X syndrome, the children in the current project were fairly young (Abbeduto & Hagerman, 1997; Abbeduto et al., 2003; Hagerman & Hagerman, 2002; Mazzocco, 2000). This age difference allows for a better understanding of the early progression of language development in this syndrome. Fourth, while earlier investigators had to rely on cytogenetic techniques to verify fragile X syndrome status, the boys in the current study were diagnosed using molecular genetic testing, which is a more reliable diagnosis of full-mutation fragile X syndrome. Fifth, well-established measures were used to assess expressive language: MLU is the most commonly used measure of syntactic length, the Index of Productive Syntax is increasingly used for refined analysis of expressive spontaneous language, and number of different words is a frequently used measure of vocabulary diversity (Paul, 2001). Moreover, we conducted a refined analysis of more specific syntactic areas, such as noun and verb phrases. Finally, we addressed important questions regarding whether expressive language difficulties related to both syntax and vocabulary and investigated whether important child and environmental factors contributed to these skills.
These findings have several important implications for the assessment and intervention of boys with fragile X syndrome. First, intervention should address the expressive language skills of young boys with fragile X syndrome, focusing on both vocabulary and syntax skills. Increasing vocabulary skills in boys with fragile X syndrome may also increase syntax skills. Second, language interventions should target more complex noun, verb, and sentence structures, such as three-word noun phrases, two-word phrases before verbs, present and past tense verb markers, past tense copula, and relative clauses. Third, the finding that nonverbal cognitive skills predicted every measure of expressive language highlights the importance of facilitating nonverbal skills to improve expressive language. Finally, given the individual variability in expressive syntax and vocabulary development among boys with fragile X syndrome, each child should receive an in-depth assessment of his language and cognitive abilities and an intervention plan that builds on particular strengths in language and targets areas of weakness.
This research was supported by the National Institute for Child Health and Human Development Grants R03 HD40640, 1 R01 HD38819, 1 R01 HD044935, and T32-HD40127 and by the March of Dimes. We express our thanks to the children and families who participated in this study. We greatly appreciated the help of Elizabeth Barnes, Evelyn Boswell-Vilt, Siara Cowan, Anne Edwards, Cassia Gerolimatos, Julia Jurgens, Cheryl Malkin, Lauren Moskowitz, and Kathleen Neff in data-collection and Erica Bernau, Lauren Ellis, and Janet Soper for their assistance with transcription and coding. Sarah Henderson's assistance with manuscript preparation has been invaluable. Requests for reprints should be sent to Joanne Roberts, University of North Carolina at Chapel Hill, FPG Child Development Institute, 105 Smith Level Rd., CB# 8180, Chapel Hill, NC 27599-8180. joanne_ roberts@ unc. edu