A Longitudinal Analysis of Communication Repair Skills across Three Neurodevelopmental Disabilities

Abstract

Background:

This study is a longitudinal follow-up to prior work examining the important pragmatic skill of communication repair (i.e., the ability to effectively respond to a request for clarification of an unclear message) across three neurodevelopmental disabilities in which language skills are impaired: fragile X syndrome with and without autism spectrum disorder (FXS-ASD; FXS-O), idiopathic ASD (ASD-O), Down syndrome (DS), and controls with typical development (TD). Prior work examining communication repair skills at younger ages indicated impairments in boys with FXS-ASD and ASD-O, with females performing comparably to each other across groups.

Aims:

To characterize communication repair skills in young individuals with FXS-ASD, FXS-O, ASD-O, DS, and TD, across groups and over development. A secondary aim included documenting sex differences in FXS (with and without ASD) and DS.

Methods & Procedures:

Sixty young individuals with FXS-ASD (49 males, 11 females), 38 with FXS-O (13 males, 25 females), 38 with ASD-O (males only), 42 with DS (21 males, 21 females), and 41 with TD (21 males, 20 females) participated, with a subsample reported on here who were retested at a second time point 2.7 years later on average. Participants completed a structured, picture-based task designed to assess the ability to repair breakdowns in communication. Participants’ responses were compared across groups and sexes at the second time point, and interpreted with respect to previously published (time one) findings.

Outcomes & Results:

Key findings included that, with age, male groups (including those with FXS-ASD and ASD-O, who showed difficulty at time one) performed more comparably, decreasing their use of inappropriate responses, in spite of relatively little change observed in general cognitive or structural language abilities in the clinical groups. However, girls with FXS and DS became more non-responsive with age, and differences between boys and girls with FXS-ASD emerged over time as well.

Conclusions & Implications:

Findings suggest that impairments in a critical pragmatic skill–the ability to repair communication breakdown–show significant change with age across three neurodevelopmental disabilities, with important sex-specific patterns. These developments were often observed in spite of a relative plateau in cognitive and language growth, suggesting that repair skills may be more malleable and therefore an excellent target for intervention. Findings not only inform the nature of pragmatic impairment across groups, but can importantly inform clinical practice, suggesting that clinicians should monitor pragmatic skills such as repairs throughout development and also consider the role of sex in clinical efforts.

Introduction

Successful communicative interactions rely on fluid exchange of information between communication partners, and critically, the ability to revise and clarify a message when breakdowns in communication occur. When a message is misunderstood, the listener typically requests clarification (e.g., Huh? What? I don’t understand). The speaker must then repair this breakdown to restore a mutual understanding between communication partners. The ability to repair a communication breakdown is essential for effective social communication, and social interactions more broadly. In typical development, repair skills are learned early on, and with increasing language and cognitive abilities are deployed with greater frequency and sophistication to navigate successful communicative exchanges. For instance, children as young as 15 months of age are able to provide clarification by repeating their utterances (Anselmi et al. 1986, Gallagher 1977) and by school age, are able to repair messages through revisions in syntactic structure (e.g., “The nest was made by the birds” to “The birds made the nest”), and adding novel information (e.g., “The bird flew” to “The birds flew to their nest”; Brinton et al. 1986). As more sophisticated repair strategies are mastered, children also learn to deploy these strategies more flexibly (Brinton et al. 1986).

Communication repair abilities are often impaired in neurodevelopmental disabilities impacting language, and where breakdowns in communication often occur more frequently due factors such as delays in cognition and structural language abilities, broader pragmatic difficulties, and problems with speech intelligibility (Halle et al. 2004, Keen 2005). Fragile X syndrome (FXS), autism spectrum disorder (ASD), and Down syndrome (DS) are three of the most prevalent neurodevelopmental disorders linked with communication impairment, with significant pragmatic challenges documented in each (Abbeduto and Hesketh 1997, Martin et al. 2017, Rice et al. 2005). Investigation of repair skills in these groups, particularly through a developmental lens, is especially important in order to characterize the pragmatic profiles and potentially unique intervention needs of each group. In prior work, we showed evidence of complex profiles of repair skills relative to broader pragmatic language abilities in each of these groups, documenting areas of strength, weakness, and some important areas of overlap that may point towards some common underlying mechanisms with implications for intervention (Barstein et al. 2018). The current study is a direct follow-up to this prior work and examines the longitudinal change in repair skills in these same groups, examining a second time point later in development.

Communication Repair Skills in FXS, ASD, and DS

Studies of communication repair ability in FXS, ASD, and DS have identified both strengths and weaknesses. For instance, studies of individuals with ASD-O have shown that while this group can often utilize advanced repair strategies, such as adding novel information, when prompted, they also often employ ineffective strategies, including off-topic or vague responses (Geller 1998, Paul and Cohen 1984, Volden 2004). In FXS, caused by a mutation of the FMR1 gene, up to 75% of individuals meet diagnostic criteria for ASD (making it the most common known genetic cause of ASD) and exhibit pragmatic language impairments characteristic of ASD-O and distinct from the pragmatic profiles in FXS-O (Klusek et al. 2014a, 2014b, Lee et al. 2016, Losh et al. 2012). A prior study of repair skills in FXS (Barstein et al. 2018) showed that boys with FXS (both with and without ASD) were more responsive to requests for clarification than boys with ASD-O. However, boys with FXS-ASD demonstrated more resistance to the task instructions (e.g., “Let’s do the next thing”), a type of inappropriate response, compared to boys with ASD-O and DS. Thus, whereas both ASD groups (idiopathic and syndromic) showed difficulty with communication repair, the specific type of problem differed. Understanding the specific overlap between ASD-O and FXS-ASD is important for clarifying how FMR1 may play a role in shared ASD symptomatology, and for informing tailored clinical efforts when symptoms differ.

Individuals with DS, the leading genetic cause of intellectual disability, appear to engage in communication repair as frequently as younger children with TD (Johnston and Stansfield 1997, Laws and Bishop 2004, Price et al. 2018). Furthermore, individuals with DS have demonstrated appropriate use of nonverbal strategies and even increased use of the suprasegmental properties of speech (e.g., volume, stress) compared to boys with ASD-O and FXS-ASD for communication repair (Abbeduto et al. 2008, Barstein et al. 2018).

A pragmatic skill closely related to communication repair is non-comprehension signaling, or the ability to indicate misunderstanding of a message (i.e., the ability to request a repair). Both skills are critical for managing communication breakdowns but from reverse perspectives (i.e., repair from the speaker’s perspective, and non-comprehension signaling from the listener’s perspective). A few studies have found that individuals with FXS and DS show difficulty with this skill (Abbeduto et al. 2008, Martin et al. 2017, Thurman et al. 2017), with little change over development for males and females with FXS (Thurman et al. 2017).

Sex Differences

Sex differences in pragmatic language have been well documented in typical development, with girls found to initiate less and respond less often to questions (Berghout Austin et al. 1987, Cook et al. 1985, Leaper 1991, Martin et al. 2018) but also to respond less inappropriately (e.g., fewer off-topic responses) to requests for clarification than boys (Barstein et al. 2018). However, girls tend to be underrepresented in neurodevelopmental disabilities research, and in studies of pragmatics in particular (Mesinger et al. 2015, Rinehart et al. 2011, Thompson et al. 2003). In FXS, sex differences are expected given that girls carry an unaffected copy of the FMR1 gene on their second X chromosome (and therefore usually show less severe symptoms than males overall) (Hagerman and Hagerman 2002). Mixed findings have been described for non-comprehension signaling, with two studies suggesting greater difficulty for males (Abbeduto et al. 2008, Thurman et al. 2017) and another study reporting no sex differences (Martin et al. 2017a). Girls with FXS-ASD also responded less to questions than their male counterparts in another recent study using a semi-structured conversational interaction (Martin et al. 2018). In DS, mixed findings have been reported for off-topic language, with one study reporting more inappropriate topic changes for girls versus boys (Lee et al. 2017) and another finding no sex differences for off-topic language (Martin et al. 2018).

Only one study of repair abilities (Barstein et al. 2018) has included boys and girls with FXS-O, FXS-ASD, and DS. This study found that boys with FXS-ASD demonstrated reduced use of gestures and more resistance to the task compared to females with FXS-ASD. As described earlier, boys with FXS-ASD in this study also responded more inappropriately than boys with DS, although this difference was not found between girls with FXS-ASD and DS. This suggests more difficulty with communication repair for boys with FXS-ASD than for girls, at least at younger ages, although longitudinal tracking of sex overlap and differences over time is needed.

Summary and Present Study

In summary, the existing literature on communication repair ability in FXS, ASD, and DS suggests both strengths and weaknesses, with particular challenges noted for boys with ASD (both those with and without FXS) and a strength observed for children with DS, along with some evidence to suggest sex differences in the subgroup of individuals with FXS-ASD. Given rapid development of communication repair skills observed in typical development, and the important role of repair skills to broader social and communicative functioning, it is important to understand how communication repair skills might grow, or otherwise change with development across these different neurodevelopmental disabilities.

The current longitudinal study examined developmental changes in communication repair skills in young individuals with FXS-ASD, FXS-O, ASD-O, DS, and TD. This study builds on prior work described by Barstein et al. 2018 by examining changes in repair skills across two time points. Our goal was to understand overlapping and distinct patterns of the development of communication repair strategies across populations and sexes to inform the nature of communication skills in each group, and highlight potential targets for clinical assessment and intervention to support this key pragmatic skill. Specifically, the present investigation addressed these aims: (1) to assess overlap and differences across groups over time, (2) to document the impact of ASD status in FXS and overlap with idiopathic ASD over time, and (3) to examine sex differences in FXS and DS over time. The longitudinal nature of this study was addressed in two ways: 1) by conducting cross-sectional comparisons at two time points, and 2) by using hierarchical linear models (HLM) that included all data and examined the role of chronological age in repair skills.

Methods

Participants

Participants included 60 individuals with FXS and ASD (and and boys and girls FXS-ASD; 49 males, 11 females), 38 with FXS only (FXS-O; 13 males, 25 females), 38 with ASD only (ASD-O; males only), 42 with DS (21 males, 21 females), and 41 with TD (21 males, 20 females). The TD group was chronologically younger, but more similar in cognitive and structural language skills which are necessary skills for pragmatic language (see Martin et al. 2017b review for importance of controlling for these skills). Children were recruited through schools, research registries, genetic and developmental clinics, and physician’s offices. All participants were tested for repair ability at the first time point of a three-time-point longitudinal study of various aspects of pragmatic language; a subsample was retested for repair ability at the third time point (hereafter referred to as Time 1 and Time 2, respectively). Communication repair skills were not assessed between these two time points. The reason for some children having less than 2 time points was primarily a function of “rolling enrollment” where subjects enrolled later in the course of the study did not have the opportunity for multiple visits, since the study ended before this was possible. At Time 1, there were no significant differences detected for chronological age, model covariates, or dependent variables between those participants who only completed the first time point and those who completed both (ps > .05). The only marginal difference was for the resistance to task variable (p = .059). This was driven by boys with FXS-ASD; those who completed both assessments resisted the task more often than those who only completed the first assessment (p = .098). Time 1 data were previously reported in Barstein et al. 2018. The average number of years across groups between Time 1 and Time 2 was 2.7 years. The range in length of time (in years) between the first and second time points for each group are as follows: for males, FXS-ASD=2.2–3.8, FXS-O=2.5–3.8, ASD-O=2.1–3.6, DS=2.2–3.6, and TD=2.1–4.1, and for females, FXS-ASD=2.1–3.8, FXS-O=2.2–3.8, DS=2.1–3.1, and TD=1.9–2.8. There were no significant differences in length of time for males (p = .205) or females (p = .259). Institutional Review Boards at the University of North Carolina at Chapel Hill and Northwestern University reviewed and approved the study protocol. Parents or guardians provided informed consent for all participants.

All participants spoke English as the first and primary language in the home, and used three or more words in an utterance. Participants were excluded for a hearing threshold greater than 30 dB HL across 500; 1,000; 2,000; and 4,000 Hz in the better ear, and participants with DS and TD could not have ASD. Table 1 provides participant characteristics by group; mirroring Barstein et al. 2018, in an attempt to control for the impact of general delays on communication repair skill development, all analyses controlled for nonverbal mental age (Leiter International Performance Scale-Revised [Leiter-R]; (Roid and Miller 1997), receptive vocabulary age equivalent (Peabody Picture Vocabulary Test-Third Edition [PPVT-III]; Dunn and Dunn 1997), expressive vocabulary age equivalent (Expressive Vocabulary Test [EVT]; Williams 1997), and mean length of utterance in morphemes (MLU). MLU was obtained from language samples derived from the Autism Diagnostic Observation Schedule (ADOS; Lord et al. 2001) which were transcribed with the Systematic Analysis of Language Transcripts software (SALT; Miller and Iglesias 2008). Please note that age equivalent scores were used in analyses because of floor effects in EVT standard scores. For information on standard scores for these covariates, please see table 2.

Table 1.

Group Characteristics at Times 1 and 2

Time 1
Group	Chronological Age M (SD) Range	Nonverbal Mental Age1 M (SD)	PPVT AE2 M (SD)	EVT AE3 M (SD)	MLU4 M (SD)	ADOS Severity5 M (SD) Range
FXS-ASD: Males (n = 49)	10.6 (2.4) 6.6–15.1	5.0 (0.5)	5.7 (1.4)	5.0 (1.0)	3.5 (0.7)	6.6 (1.6) 4.0–10.0
FXS-ASD: Females (n = 11)	9.1 (4.0) 4.9–15.9	5.2 (0.7)	6.4 (2.0)	5.8 (1.3)	4.0 (1.3)	6.8 (1.6) 4.0–9.5
FXS-O: Males (n = 13)	9.7 (3.3) 6.1–15.0	5.5 (1.0)	6.5 (2.6)	5.4 (1.5)	4.0 (0.7)	2.4 (1.0) 1.0–3.5
FXS-O: Females (n = 25)	9.2 (3.7) 4.2–14.9	7.0 (2.3)	8.3 (3.4)	8.4 (3.9)	4.9 (1.1)	2.1 (0.8) 1.0–3.5
ASD-O: Males (n = 38)	9.2 (2.6) 4.2–13.3	7.1 (3.3)	6.8 (3.0)	6.3 (2.2)	4.3 (1.1)	7.6 (1.9) 2.0–10.0
DS: Males (n = 21)	11.0 (2.1) 6.8–14.9	5.3 (0.8)	5.2 (1.4)	5.4 (1.3)	3.1 (0.8)	1.4 (0.6) 1.0–3.0
DS: Females (n = 21)	9.1 (2.2) 6.0–14.2	5.0 (0.7)	4.8 (1.9)	4.7 (1.5)	3.3 (1.0)	1.7 (0.6) 1.0–3.0
TD: Males (n = 21)	4.8 (1.1) 3.3–7.0	5.3 (1.2)	6.0 (1.6)	5.7 (1.6)	4.8 (0.7)	1.6 (0.7) 1.0–3.0
TD: Females (n = 20)	5.4 (2.4) 3.2–11.8	6.1 (2.7)	6.4 (3.0)	6.1 (2.4)	5.3 (1.5)	1.4 (0.7) 1.0–3.0
Time 2
Group	Chronological Age M (SD) Range	Nonverbal Mental Age M (SD)	PPVT AE M (SD)	EVT AE M (SD)	MLU M (SD)	ADOS Severity M (SD) Range
FXS-ASD: Males (n = 29)	13.3 (2.5) 9.2–17.9	5.2 (0.6)	6.6 (1.4)	5.6 (1.3)	3.6 (0.7)	6.4 (1.6) 4.0–9.5
FXS-ASD: Females (n = 11)	10.6 (3.1) 7.2–17.1	6.1 (1.0)	7.9 (3.1)	7.7 (3.7)	4.4 (1.8)	6.0 (1.8) 4.0–9.5
FXS-O: Males (n = 5)	11.6 (2.9) 8.7–16.4	5.1 (0.8)	7.4 (0.8)	6.6 (1.4)	4.6 (1.6)	2.5 (1.0) 1.0–3.5
FXS-O: Females (n = 19)	11.9 (3.6) 7.3–17.2	8.4 (3.7)	10.6 (3.2)	9.8 (3.0)	5.3 (0.9)	2.1 (0.8) 1.0–3.5
ASD-O: Males (n = 13)	12.0 (1.7) 10.1–15.3	7.1 (1.8)	7.8 (1.7)	7.0 (2.3)	4.9 (1.6)	7.5 (1.5) 5.0–9.0
DS: Males (n = 14)	14.2 (2.6) 9.7–17.9	6.0 (1.3)	6.7 (1.7)	6.2 (1.3)	3.4 (0.8)	1.3 (0.3) 1.0–1.67
DS: Females (n = 13)	10.6 (1.7) 8.3–13.4	5.1 (0.9)	5.2 (2.1)	5.0 (1.4)	3.3 (1.0)	1.8 (0.5) 1.0–3.0
TD: Males (n = 8)	7.2 (1.2) 6.2–9.4	7.3 (1.0)	8.5 (1.5)	7.8 (1.2)	5.3 (1.2)	1.5 (0.5) 1.0–2.0
TD: Females (n = 6)	8.1 (2.6) 5.0–11.4	9.7 (3.3)	9.8 (2.8)	9.8 (3.4)	5.8 (1.4)	1.4 (0.5) 1.0–2.0

Note. M = mean; SD = standard deviation; AE = age equivalent; FXS-ASD = fragile X syndrome with autism spectrum disorder; FXS-O = FXS only; ASD-O = ASD only; DS = Down syndrome; TD = typical development.

¹Leiter International Performance Scale-Revised

²Peabody Picture Vocabulary Test-3rd edition

³Expressive Vocabulary Test

⁴Mean Length of Utterance in Morphemes

⁵Average Autism Diagnostic Observation Schedule severity score (across all available time points)

Table 2.

Standard Scores for IQ, Receptive Vocabulary, and Expressive Vocabulary at Times 1 and 2

Time 1
Group	Nonverbal IQ1 M (SD)	PPVT SS2 M (SD)	EVT SS3 M (SD)
FXS-ASD: Males (n = 49)	52.0 (12.2)	62.1 (15.7)	50.9 (15.1)
FXS-ASD: Females (n = 11)	72.6 (22.4)	83.0 (20.7)	74.9 (20.8)
FXS-O: Males (n = 13)	65.1 (14.7)	75.4 (15.4)	64.4 (16.4)
FXS-O: Females (n = 25)	88.6 (21.4)	95.1 (15.1)	94.6 (17.3)
ASD-O: Males (n = 38)	82.0 (25.4)	79.8 (21.0)	74.8 (21.1)
DS: Males (n = 21)	52.8 (10.1)	55.6 (12.6)	51.6 (15.2)
DS: Females (n = 21)	60.2 (9.7)	60.4 (11.9)	52.1 (11.4)
TD: Males (n = 21)	113.1 (10.3)	114.4 (10.5)	111.2 (9.0)
TD: Females (n = 20)	117.9 (10.8)	111.0 (11.5)	110.1 (9.7)
Time 2
Group	Nonverbal IQ M (SD)	PPVT SS M (SD)	EVT SS M (SD)
FXS-ASD: Males (n = 29)	44.8 (8.6)	59.7 (14.2)	47.5 (13.1)
FXS-ASD: Females (n = 11)	65.5 (13.4)	80.0 (21.0)	75.5 (22.7)
FXS-O: Males (n = 5)	50.2 (9.9)	74.0 (7.5)	61.8 (8.0)
FXS-O: Females (n = 19)	79.8 (23.1)	94.1 (14.1)	89.4 (17.8)
ASD-O: Males (n = 13)	66.2 (17.1)	72.9 (13.5)	63.5 (20.7)
DS: Males (n = 14)	51.9 (11.5)	56.9 (14.7)	48.8 (9.8)
DS: Females (n = 13)	49.5 (6.1)	54.9 (14.6)	47.6 (11.3)
TD: Males (n = 8)	106.6 (12.0)	111.8 (7.0)	106.0 (8.0)
TD: Females (n = 6)	119.3 (5.9)	112.7 (11.0)	112.2 (9.7)

Note. M = mean; SD = standard deviation; SS = standard score; FXS-ASD = fragile X syndrome with autism spectrum disorder; FXS-O = FXS only; ASD-O = ASD only; DS = Down syndrome; TD = typical development.

¹Leiter International Performance Scale-Revised

²Peabody Picture Vocabulary Test-3rd edition

³Expressive Vocabulary Test

Group differences on the covariates at Time 1 were reported in Barstein et al. 2018. At Time 2, boys with ASD-O and TD were higher in mental age than boys with FXS-ASD and FXS-O (ps < .05). Boys with ASD-O had a higher MLU than boys with FXS-ASD (p = .011) and DS (p = .013). Boys with TD scored higher in expressive vocabulary (p = .008) and receptive vocabulary (p = .030) than boys with FXS-ASD, and higher in MLU than boys with FXS-ASD (p = .002) and DS (p = .002). Girls with FXS-O and TD were higher in mental age, receptive vocabulary, expressive vocabulary, and MLU than females with DS (ps < .05). Girls with FXS-ASD were higher in mental age (p = .019) than boys with FXS-ASD; girls with FXS-O scored higher in receptive (p = .001) and expressive vocabulary (p = .030) than boys with FXS-O; and boys with DS scored higher in nonverbal cognition, receptive vocabulary, and expressive vocabulary than females with DS (ps < .05).

Confirmation of ASD in the ASD-O group and determination of ASD in the FXS group was conducted with the ADOS, a gold-standard measure. All available ADOSs were considered, and the ADOS severity score (Gotham et al. 2009) averaged, to determine ASD status given the longitudinal nature of the project (see table 1). Please note that one boy with ASD-O was not classified as having ASD based on the ADOS, but was included because he met criteria for autism on the Autism Diagnostic Interview–Revised (Lord et al. 1994) and had a clinical ASD diagnosis. At Time 1, boys with FXS-ASD and ASD-O had a higher average ASD severity score than boys with FXS-O, DS, and TD (ps < .001), and boys with ASD-O were higher than boys with FXS-ASD (p = .001). At Time 1, girls with FXS-ASD were higher in average ASD severity than girls with FXS-O, DS, and TD (ps < .001), and girls with FXS-O were also higher than girls with TD (p = .02). For the smaller Time 2 subset, all of the significant differences for males remained and no additional significant differences were detected. For females at Time 2, the difference between FXS-O and TD was no longer significant. Girls with DS also had a higher average severity score than boys with DS at Time 2 (p = .003).

Procedure

This longitudinal follow up to an earlier report Barstein et al. 2018. included procedures identical to those described previously. Assessments, which took place in a research setting or the child’s school or home, were audio- and video-taped.

General Cognition and Language.

Nonverbal mental age was assessed with the Brief IQ composite of the Leiter-R (Roid and Miller 1997), which includes subtests that require a participant to recognize an item from separate parts, select the next item in a sequence, and give the missing part of a pattern. Receptive vocabulary age was asssessed with the PPVT-III (Dunn and Dunn 1997), which requires a test-taker to point to a picture out of an array based on the word said by an examiner. Expressive vocabulary age was asssessed with the EVT (Williams 1997), which requires the participant to provide a word in response to pictures. Morphosyntactic complexity was assessed with MLU in morphemes (Brown 1973). MLU was derived from ADOS (Lord et al. 2001) language samples. Language samples (module 2 or 3 ADOS free play and chosen non-play contexts) were transcribed by trained research assistants from audio and video (see Martin et al. 2018 for further details on procedure and reliability).

Communication Repairs Task.

Participants were asked to describe silly (nonsensical) pictures (Webber 1998) to an examiner who could not view the picture (e.g., an octopus cutting a man’s hair, a horse riding a man, a woman with a bird’s nest in her hair). Eight pictures were experimental and ten were control. The examiner accepted the child’s description for control pictures and requested no clarification. For experimental pictures (Brinton et al. 1986), the examiner acted as if she did not understand the child’s description, using a series of three clarification requests (i.e., Huh? What? I don’t understand). Responses for experimental conditions were coded (see table 3) based on definitions provided by Brinton et al. (1986) and Volden (2004). Variables used in analyses were proportions, based on experimental pictures, averaged across requests. The No Response variable was the proportion of non-responses out of total requests made by the examiner. For other codes, proportions were out of all usable responses. The maximum number of usable responses (denominator) was 24 (8 experimental pictures X 3 responses), although this could be lower considering the exclusion of No Response and unintelligible responses. As described in Barstein et al. 2018, data were not included in the study if a participant had fewer than five intelligible responses on the repairs task. This was the case for one girl with DS at Time 2 and one girl with FXS-O at Time 2. For repairs data included in the study, the mean percentages of intelligible responses by group for both time points are reported in table 4; proportions were used in analyses to account for the differences in intelligibility across groups. For each usable response, only one code was given (except when a supplemental code of suprasegmental or gesture was appropriate, which would be coded in addition to the primary code). If more than one type of response was used, a hierarchy was applied in which they received credit for the more sophisticated response (where Inappropriate was considered least sophisticated and Metacomment was considered most sophisticated; see table 3 for order of sophistication). Coding utilized transcriptions of the repairs task as well as the video. Coders were first trained to reliability on gold standard files. After achieving 90% agreement or higher on gold standard files, 10% of the data were randomly selected and independently coded for reliability, and the intraclass correlation coefficient (ICC) across groups, prompts, and signal types was .982 at Time 1 (as reported in Barstein et al. 2018) and .922 at Time 2.

Table 3.

Coding Categories

Code	Example (C: Child, E: Examiner)
No Response: Does not respond within 5 seconds of request for clarification
Inappropriate (includes 3 subtypes)
Look: Tells examiner to look at picture (against rules of task)	C has a hotdog on it. E what? C look right there.
Off-Topic: Says something unrelated to the picture or the task	C the girl is sleeping E I don’t understand. C uh the igloo.
Resistance: Resists task instructions	E what? C next.
Repetition: Repeats one or multiple words in a previous response without adding information or using a different grammatical form	C a jet. E huh? C a jet.
Revision: Adjusts response without adding information, including use of different grammatical form or synonym	C and he’s trying to catch the fish but the fish caught a hotdog. E what? C the fish caught a hotdog but he’s trying to catch a fish.
Addition: Changes previous response by adding new and appropriate information	C the woman is tanning outside with snow. E I don’t understand. C she’s laying on the snow in a bathing suit.
Background: Offers context in which previous response can be understood	C she’s laying on a towel and she’s wearing a bathing suit. E I don’t understand. C she thought it was sunny and she is outside in the winter.
Metacomment: Talks about the repair process	E I don’t understand. C there’s not a way to explain it now.
Supplemental Strategies (coded in addition to codes above)
Suprasegmental: Becomes louder, slows down, emphasizes words, or articulates more precisely	C a horse ride cowboy (emphasizes horse)
Gesture: Uses nonverbal behaviors such as pantomiming or pointing	C go fishing (pantomimes fishing)

Table 4.

Percentage of Intelligible Responses by Group at Times 1 and 2

Time 1
Group	Percent Intelligible Responses
FXS-ASD: Males	84.0%
FXS-ASD: Females	94.0%
FXS-O: Males	89.0%
FXS-O: Females	97.0%
ASD-O: Males	93.0%
DS: Males	84.0%
DS: Females	80.0%
TD: Males	95.0%
TD: Females	96.0%
Time 2	Percent Intelligible Responses
Group
FXS-ASD: Males	92.0%
FXS-ASD: Females	94.0%
FXS-O: Males	97.0%
FXS-O: Females	98.0%
ASD-O: Males	98.0%
DS: Males	92.0%
DS: Females	75.0%
TD: Males	97.0%
TD: Females	99.0%

Note. FXS-ASD = fragile X syndrome with autism spectrum disorder; FXS-O = FXS only; ASD-O = ASD only; DS = Down syndrome; TD = typical development.

Analysis Plan

To characterize group differences at different time points, a series of one-way analyses of covariance (ANCOVA) models, followed by pairwise comparisons using Bonferroni correction to account for multiple comparisons, were conducted for males and females separately at each time point for each category of verbal and supplemental strategies, controlling for nonverbal mental age, expressive vocabulary age, receptive vocabulary age, and MLU, with diagnostic group as the independent variable. ANCOVAs, controlling for identical variables, were also used to examine sex differences within the FXS, DS and TD groups, with sex as the independent variable. Please note that Time 1 analyses were previously reported in Barstein et al. 2018, and so a summary of those findings are presented here. For Time 2 group comparisons, effect sizes (Cohen’s d) were also calculated since sample sizes were smaller for this time point, where 0.2 was deemed small, 0.5 medium, and 0.8 large (Cohen 1988).

To investigate changes in the use of strategies over time, hierarchical linear models (HLM) were conducted for males and females separately with chronological age (the marker of change over time) nested within individual participant. Covariates included nonverbal mental age, receptive vocabulary age, expressive vocabulary age, and MLU. Covariates were grand-mean centered across both time points in order to reduce collinearity amongst covariates and also to ensure that main effects would be estimated at the means of these variables. In order to examine changes within each group by sex, HLMs were conducted for each diagnostic group, with sex as an independent variable. Random intercepts and slopes were initially included in the models for all strategies. When random slopes and/or intercepts were inestimable, the models were run with only fixed effects included (i.e., revision for males and females, inappropriate for males, resistance for males).

Results

Group and Sex Comparisons of Communication Repair Skills

Time 1.

Significant findings at the first time point are reviewed here; please refer to Barstein et al. 2018 for statistical details. Boys with ASD-O were more non-responsive to requests for clarification than boys with FXS-ASD and FXS-O, boys with FXS-ASD were more resistant to the task (a type of inappropriate response) than boys with ASD-O and DS, and boys with DS used more suprasegmental repairs than males with FXS-ASD and ASD-O. No significant differences were found for the female groups at Time 1. Sex differences were detected for the FXS-ASD and TD groups; boys with FXS-ASD used gestures less often than their female counterparts, and boys with TD used more inappropriate responses than girls.

Time 2.

See table 5 (boys) and table 6 (girls) for results of ANCOVAs for each repair variable, along with effect sizes for pairwise comparisons. No significant differences between male or female groups were detected at Time 2. Sex comparisons indicated that boys with FXS-ASD used more inappropriate strategies than girls (F(1,34) = 8.33, p = .007, d = 1.02), which was driven by off-topic responses in particular (F(1,34) = 7.86, p = .008, d = 0.99). No other sex differences were significant (ds 0.00–1.02), although the difference between males and females with FXS-ASD on non-responsiveness was marginal (p = .069; d = 1.02), with girls being more non-responsive than boys.

Table 5.

Group Comparisons at Time 2: Boys

Variable	ANCOVA F	ANCOVA p-value	FXS-ASD M (SE) Range	FXS-O M (SE) Range	ASD-O M (SE) Range	DS M (SE) Range	TD M (SE) Range	Effect Size1 Range
No Response	F(4,60) = 2.09	p = .093	.03(.02) .00-.33	.03(.05) .00-.04	.03(.03) .00-.17	.06(.03) .00-.30	.15(.04) .00-.50	0.00–0.78
Inappropriate	F(4,60) = 0.31	p = .874	.06(.01) .00-.23	.07(.03) .00-.18	.04(.02) .00-.20	.04(.02) .00-.25	.05(.03) .00-.14	0.00–0.00
Look	F(4,60) = 1.39	p = .250	.001(.002) .00-.04	.00(.01) .00-.00	.00(.004) .00-.00	.00(.003) .00-.00	.01(.01) .00-.08	0.00–0.40
Off-Topic	F(4,60) = 1.72	p = .158	.04(.01) .00-.18	.06(.02) .00-.18	.02(.01) .00-.15	.01(.01) .00-.13	.003(.02) .00-.04	0.00–0.50
Resistance	F(4,60) = 0.40	p = .811	.02(.01) .00-.10	.004(.02) .00-.04	.02(.01) .00-.08	.03(.01) .00-.14	.03(.02) .00-.14	0.00–0.00
Repetition	F(4,60) = 0.52	p = .720	.23(.04) .00-.68	.12(.10) .00-.35	.29(.07) .00-.63	.28(.06) .00-.90	.25(.08) .00-.58	0.00–0.00
Revision	F(4,60) = 0.82	p = .515	.39(.03) .13-.67	.39(.07) .23-.64	.31(.05) .25-.46	.41(.04) .10-.79	.29(.06) .08-.65	0.00–0.00
Addition	F(4,60) = 0.88	p = .479	.31(.03) .04-.60	.31(.06) .22-.33	.29(.04) .04-.58	.23(.04) .00-.41	.33(.05) .15-.54	0.00–0.00
Background	F(4,60) = 1.09	p = .371	.02(.01) .00-.26	.01(.02) .00-.05	.02(.01) .00-.09	.004(.01) .00-.05	.04(.02) .00-.10	0.00–0.12
Metacomment	F(4,60) = 1.60	p = .187	.01(.02) .00-.10	.11(.04) .00-.35	.07(.03) .00-.17	.05(.03) .00-.43	.09(.04) .00-.33	0.00–0.36
Suprasegmental	F(4,60) = 1.80	p = .142	.26(.04) .00-.71	.13(.09) .00-.46	.15(.06) .00-.83	.21(.06) .00-.71	.01(.08) .00-.30	0.00–0.62
Gesture	F(4,59) = 0.54	p = .708	.27(.05) .00-.88	.13(.11) .00-.30	.21(.08) .00-.63	.20(.07) .00-.63	.18(.09) .00-.56	0.00–0.00

Note. M = adjusted mean; SE = standard error; FXS-ASD = fragile X syndrome with autism spectrum disorder; FXS-O = FXS only; ASD-O = ASD only; DS = Down syndrome; TD = typical development.

¹Computed as Cohen’s d. Ranges are based on raw scores.

Table 6.

Group Comparisons at Time 2: Girls

Variable	ANCOVA F	ANCOVA p-value	FXS-ASD M (SE) Range	FXS-O M (SE) Range	DS M (SE) Range	TD M (SE) Range	Effect Size1 Range
No Response	F(3,41) = 2.63	p = .063	.13(.03) .00-.42	.07(.03) .00-.79	.09(.04) .00-.29	-.04(.05) .00-.05	0.00–1.10
Inappropriate	F(3,41) = 1.13	p = .347	-.002(.02) .00-.05	.05(.02) .00-.20	.02(.02) .00-.38	.03(.03) .00-.05	0.00–0.29
Look	F(3,41) = 1.54	p = .219	-.001(.003) .00-.00	.006(.002) .00-.05	-.002(.003) .00-.00	.002(.004) .00-.00	0.00–0.32
Off-Topic	F(3,41) = 0.97	p = .418	-.01(.02) .00-.00	.04(.02) .00-.15	.02(.02) .00-.38	.03(.03) .00-.05	0.00–0.19
Resistance	F(3,41) = 0.10	p = .959	.004(.01) .00-.05	.01(.01) .00-.06	.01(.01) .00-.11	.001(.01) .00-.00	0.00–0.00
Repetition	F(3,41) = 1.04	p = .384	.21(.05) .00-.48	.11(.04) .00-.46	.21(.05) .00-.58	.18(.07) .00-.29	0.00–0.13
Revision	F(3,41) = 0.82	p = .493	.40(.05) .05-.65	.34(.04) .09-.63	.46(.06) .05-.78	.40(.07) .29-.61	0.00–0.09
Addition	F(3,41) = 0.87	p = .464	.39(.05) .13-.67	.42(.04) .13-.75	.31(.05) .10-.43	.36(.07) .14-.63	0.00–0.05
Background	F(3,41) = 0.51	p = .679	.02(.02) .00-.13	.04(.01) .00-.22	.03(.02) .00-.05	.01(.02) .00-.09	0.00–0.00
Metacomment	F(3,41) = 1.09	p = .364	.01(.04) .00-.21	.08(.03) .00-.70	-.003(.04) .00-.29	.04(.05) .00-.05	0.00–0.12
Suprasegmental	F(3,41) = 0.43	p = .734	.15(.07) .00-.65	.20(.06) .00-.86	.14(.08) .00-.86	.29(.11) .00-.68	0.00–0.00
Gesture	F(3,40) = 1.90	p = .145	.46(.09) .10-.91	.21(.07) .00-.70	.33(.09) .00-.78	.17(.12) .00-.29	0.00–0.52

Note. M = adjusted mean; SE = standard error; FXS-ASD = fragile X syndrome with autism spectrum disorder; FXS-O = FXS only; DS = Down syndrome; TD = typical development.

¹Computed as Cohen’s d. Ranges are based on raw scores.

Figures 1 and 2 display adjusted means and standard errors of overall proportions of each response type for males and females at each time point, controlling for cognition and language. These figures show Time 1 and Time 2 means next to each other for each group.

Figure 1.

Adjusted Means and Standard Errors at Times 1 and 2: Males. FXS-ASD = fragile X syndrome with autism spectrum disorder; FXS-O = FXS only; ASD-O = ASD only; DS = Down syndrome; TD = typical development. Significant differences at Time 1 (p < .05): no response, ASD-O > FXS-ASD, FXS-O; resistance, FXS-ASD > ASD-O, DS; suprasegmental: DS > FXS-ASD, ASD-O. No significant differences at Time 2 (p < .05).

Figure 2.

Adjusted Means and Standard Errors at Times 1 and 2: Females. FXS-ASD = fragile X syndrome with autism spectrum disorder; FXS-O = FXS only; DS = Down syndrome; TD = typical development. No significant differences at Times 1 or 2 (p < .05).

Changes in Communication Repair Skills with Age (HLM analysis)

Main effects of the HLM for group are not reported in the text below as these effects were assessed by ANCOVAs reported above. Instead, we present below the main effects of age and the group by age interactions, which are novel to the HLM analysis. However, please see tables 7 (males) and 8 (females) for a comprehensive list of effects.

Table 7.

Fixed Effects of HLM Models for Males

Effect	F-Value	p-value
Non-Response
Group	2.871*	.025
Chronological Age	.230	.632
Group*Chronological Age	1.657	.162
Inappropriate
Group	1.198	.313
Chronological Age	8.211*	.005
Group*Chronological Age	1.100	.358
Look^
Group	.660	.620
Chronological Age	.013	.909
Group*Chronological Age	.548	.701
Off-Topic^
Group	.807	.523
Chronological Age	10.405*	.001
Group*Chronological Age	1.923	.109
Resistance^
Group	2.963*	.021
Chronological Age	.847	.358
Group*Chronological Age	.492	.742
Repetition
Group	.770	.546
Chronological Age	.015	.904
Group*Chronological Age	.338	.852
Revision
Group	.212	.931
Chronological Age	.226	.635
Group*Chronological Age	1.025	.396
Addition
Group	.431	.786
Chronological Age	.254	.615
Group*Chronological Age	1.302	.271
Background
Group	.610	.656
Chronological Age	.721	.397
Group*Chronological Age	.887	.473
Metacomment
Group	1.499	.205
Chronological Age	3.943*	.048
Group*Chronological Age	2.539*	.041
Suprasegmental
Group	3.292*	.013
Chronological Age	7.738*	.006
Group*Chronological Age	1.846	.122
Gesture
Group	1.023	.398
Chronological Age	.063	.802
Group*Chronological Age	1.949	.104

Note.

^*Indicates p < .05.

^{^}Included in Inappropriate.

Table 8.

Fixed Effects of HLM Models for Females

Effect	F-Value	p-value
Non-Response
Group	3.211*	.028
Chronological Age	4.556*	.035
Group*Chronological Age	3.385*	.021
Inappropriate
Group	1.222	.308
Chronological Age	.591	.444
Group*Chronological Age	.827	.482
Look^
Group	.668	.575
Chronological Age	1.727	.192
Group*Chronological Age	1.384	.252
Off-Topic^
Group	.535	.660
Chronological Age	.228	.634
Group*Chronological Age	1.053	.373
Resistance^
Group	.855	.469
Chronological Age	2.723	.103
Group*Chronological Age	1.932	.130
Repetition
Group	.817	.489
Chronological Age	6.405*	.013
Group*Chronological Age	.842	.474
Revision
Group	.478	.698
Chronological Age	.352	.554
Group*Chronological Age	1.059	.369
Addition
Group	.197	.898
Chronological Age	.005	.945
Group*Chronological Age	1.491	.222
Background
Group	.605	.614
Chronological Age	.295	.588
Group*Chronological Age	.257	.856
Metacomment
Group	1.806	.152
Chronological Age	10.059*	.002
Group*Chronological Age	1.612	.191
Suprasegmental
Group	.200	.896
Chronological Age	.716	.400
Group*Chronological Age	1.072	.365
Gesture
Group	2.065	.112
Chronological Age	1.512	.222
Group*Chronological Age	.511	.676

Note.

^*Indicates p < .05.

^{^}Included in Inappropriate.

Boys.

There was a significant main effect of age for the use of inappropriate strategies in boys (F(1,195) = 8.21, p = .005), with all groups decreasing their use of inappropriate strategies with increasing age (slopes ranging from −.02 to −.004; see figure 3). This was driven by a significant main effect of age for the use of off-topic strategies in boys (F(1,189.93) = 10.41, p = .001), with all groups decreasing their use of off-topic strategies with increasing age (slopes ranging from −.08 to −.01). For metacomments, there was also a significant main effect of age (F(1,193.82) = 3.94, p = .048), as well as a significant group by age interaction (F(4,185.96) = 2.54, p = .041); with age, metacomments increased for boys with FXS-O, ASD-O, DS, and TD (slopes ranging from .001 to .02), whereas boys with FXS-ASD decreased their use of metacomments (slope = −.01). There was also a significant main effect of age for suprasegmentals (F(1,190.03) = 7.74, p = .006); all male groups decreased their use of suprasegmentals with age (slopes ranging from −.08 to −.002).

Figure 3.

Inappropriate Responses by Chronological Age and Diagnosis: Males; FXS-ASD = fragile X syndrome with autism spectrum disorder; FXS-O = FXS only; ASD-O = ASD only; DS = Down syndrome; TD = typical development

Girls.

For girls, there was a significant effect of age in the use of repetition (F(1,95.72) = 6.41, p = .013), with all groups decreasing use of repetition with increasing age (slopes ranging from −.05 to −.02; see figure 4). A significant main effect of age was also found for metacomments (F(1, 103.95) = 10.06, p = .002), with all groups increasing their use of metacomments with age (slopes ranging from .01 to .02). For non-responsiveness, there was a significant main effect of age (F(1,97.46) = 4.56, p = .035), and also a significant group by age interaction (F(3,101.55) = 3.39, p = .021); whereas girls with TD decreased in non-responsiveness with age (albeit only slightly; slope = −.003), all other groups were more non-responsive with age (slopes ranging from .004 to .03).

Figure 4.

Repetition by Chronological Age and Diagnosis: Females; FXS-ASD = fragile X syndrome with autism spectrum disorder; FXS-O = FXS only; DS = Down syndrome; TD = typical development

Sex Differences.

For non-responsiveness in FXS-ASD, there were significant main effects of sex (F(1, 61.26) = 12.76, p = .001) and age (F(1,79.02) = 19.94, p < .001), and a significant sex by age interaction (F(1,83.61) = 19.42, p < .001). Whereas females became more non-responsive with age (slope = .03), males remained relatively stable (slope = .00). For FXS-ASD, there was also a significant main effect of sex in the use of inappropriate strategies (F(1, 91.00) = 5.25, p = .024), metacomments (F(1, 63.88) = 5.74, p = .020), and gestures (F(1, 65.20) = 4.12, p = .046). There was also a significant sex by age interaction in the use of inappropriate strategies in FXS-O (F(1,46.39) = 4.67, p = .036). Whereas males became less inappropriate with age (slope = −.02), females remained relatively stable (slope = .00). Significant main effects of age additionally occurred for the use of addition (F(1, 46.57) = 6.03, p = .018) and non-responsiveness (F(1, 47.00) = 4.93, p = .031) in individuals with TD, as well as for gestures in individuals with DS (F(1, 58.03) = 8.11, p = .006).

Discussion

This longitudinal study examined the development of communication repairs skills across two time points, nearly three years apart, and across three neurodevelopmental disabilities where language development is impacted– fragile X syndrome (FXS), autism spectrum disorder (ASD-O), and Down syndrome (DS). We also compared trajectories in boys with FXS with and without ASD (FXS-ASD, FXS-O, respectively), and examined potential sex differences in all but the ASD-O group, where data were not available from girls.

Group Differences

Boys.

Group differences for boys at Time 1, previously reported in Barstein et al. 2018, indicated that boys with ASD-O were more non-responsive than boys with FXS-ASD and FXS-O, whereas boys with FXS-ASD responded more inappropriately than boys with ASD-O and DS. With age, the male groups performed more comparably. In fact, though non-significant, the largest effect size for non-responsiveness was between ASD-O and TD groups (d = 0.78), with the TD group being more non-responsiveness than the ASD-O group. All male groups decreased their use of inappropriate and suprasegmental strategies with increasing age, and all groups but FXS-ASD also demonstrated an increase in use of the more advanced metacommenting strategy of referencing the actual communication repair process. Of note, these improvements in repair skills among the male clinical groups were observed in spite of limited growth in structural language and general cognitive abilities, which remained relatively stable over time points for these groups (e.g., MLU in males with FXS-ASD was 3.5 at time one, and 3.6 at time two) and were controlled for in analyses. This pattern is potentially promising in its suggestion that communication repair strategies are not necessarily bound by limitations in language and cognition over the course of development, but may be more malleable and thus a good target for interventions.

As previously reported, at Time 1, boys with DS used more suprasegmental repairs (e.g., repeating the same sentence but adding emphasis to a word with increased pitch and/or loudness) than boys with FXS-ASD and ASD-O, which we interpreted as possibly related to poor speech intelligibility in DS where suprasegmental cues may serve as an adaptive strategy to aid listeners (Barstein et al. 2018). However, this pattern did not appear to persist later in development, as male groups did not differ in suprasegmental use at Time 2, and HLM analysis indicated that all male groups decreased their use of suprasegmentals with age. Thus, whereas suprasegmental cues may be important at younger ages, it appears that individuals with DS are less reliant on this strategy over time.

Girls.

There were no significant differences between female groups at Time 1 or Time 2. Of note, although not significant, a large effect size (d = 1.10) indicated that girls with FXS-ASD were more non-responsive to requests for clarification than girls with TD at Time 2. Since sample sizes were smaller for Time 1 than Time 2, this suggests that we may have found a significant difference for non-responsiveness at Time 2 with a larger sample. Results from our HLM analysis support this possibility. Girls with TD slightly decreased in non-responsiveness with age, whereas girls with FXS-ASD, FXS-O, and DS became more non-responsive with age. Together, results suggest that girls with FXS and DS, and girls with FXS-ASD in particular, may show more impairment in communication repair ability over time. It also may be that the repairs task employed in this study is less effective at eliciting contextually appropriate repair responses among older individuals such as the participants who comprised the female groups at Time 2. Future studies should examine longitudinally changes in repair skills over this age range using more naturalistic, conversationally based tasks.

All female groups also decreased their use of repetition with increasing age, consistent with existing literature indicating that repetition is an early developing, less sophisticated strategy that is replaced by more varied and advanced repair strategies at older ages (Anselmi et al. 1986, Brinton et al. 1986, Gallagher 1977). All female groups, like all male groups except for boys with FXS-ASD, also increased their use of metacomments with age.

Overlap of FXS and ASD

As reported above, at Time 1, boys with ASD-O were more non-responsive than boys with FXS-ASD and FXS-O, whereas boys with FXS-ASD responded more inappropriately than boys with ASD-O and DS. Thus, while both groups of boys with ASD showed difficulties responding to requests for clarification, they exhibited unique types of difficulty. Further, while FXS-ASD and FXS-O male groups did not differ significantly on any variables, it was only the FXS-ASD group who showed a relatively high level of inappropriate responses (compared with ASD-O and DS groups), suggesting a negative impact of ASD status on repair ability in boys with FXS at younger ages. Similarly, whereas the FXS-O, FXS-ASD, and ASD-O male groups performed more comparably with age overall, the FXS-ASD failed to demonstrate an increase in metacommenting with age that was observed in all other groups. These findings together suggest areas of weakness in repair ability over time that are specific to boys with FXS-ASD, underscoring the important impact of ASD status on pragmatic abilities, and repair skills in particular, in FXS.

As reported above, although there were no differences between girls with FXS-ASD and girls with FXS-O at either timepoint, a non-significant, large effect size suggested that girls with FXS-ASD were more non-responsive than girls with TD at Time 2. Therefore, although girls with FXS-ASD and FXS-O did not differ from each other, only girls with FXS-ASD showed some evidence of difficulty, suggesting that the negative impact of ASD status in FXS is observed in both males and females. Inclusion of girls with ASD-O in future studies will be important to clarify further the potential overlap of FXS and ASD in girls.

Sex Differences

This was one of the first studies to examine sex differences in pragmatic language in children with these different neurodevelopmental disabilities, as the vast majority of previous studies included boys only. At Time 1, boys with FXS-ASD used significantly fewer gestures than girls with FXS-ASD. At Time 2, boys with FXS-ASD used significantly more inappropriate responses (and off-topic responses in particular) than girls, and marginally fewer non-responses with a large effect size (d = 1.02). Results of the HLM analysis also indicated that girls with FXS-ASD became more non-responsive with age, with boys remaining stable. Together, results suggest that sex differences in the FXS-ASD group may continue to emerge over time. At Time 1, boys with TD also used more inappropriate responses than girls (but not at Time 2), suggesting of course that this is a typical pattern in young boys and girls that fades with age. The lack of sex differences in the FXS-O and DS groups is also important, suggesting that sex is not a significant factor in repair ability for these groups. Again, including girls with ASD-O in a future study would help to increase our understanding of sex differences in idiopathic ASD.

Study Strengths and Limitations

There are several strengths of the current investigation. We examined communication repair skills in three neurodevelopmental disabilities characterized by impairments in language, and examined repair abilities longitudinally, allowing side-by-side comparison of strengths and weaknesses in each group over development. The study’s inclusion of both boys and girls in the FXS and DS groups also builds on the relatively small literature examining pragmatics in girls across FXS and DS, and together with longitudinal characterization, contributes to a richer understanding of developmental pragmatic language profiles across conditions.

There are also some limitations of this work. We used a structured task, and examination of repair skills in a conversational context may yield some different results and should therefore be the focus of future studies. We also do not have data on the type or intensity of interventions that participants may have received between the two time points, and thus do not know whether some changes may have been the result of successful intervention. Girls with ASD-O should also be included in future studies, especially given our findings for sex differences in FXS-ASD. Similarly, children with comorbid DS and ASD should be included in future research to examine the impact of ASD in DS on communication repair and other pragmatic abilities. Further, unintelligibility was characteristic of the speech of some groups, especially girls with DS. While we only coded repair strategy type for intelligible responses, this means that repair skill was characterized based on fewer responses for unintelligible children. Finally, while our samples sizes were large overall compared with previous research, sample sizes were smaller for the second than the first time point. As such, group differences for Time 2 alone should be interpreted cautiously, and in the context of the HLM analysis which included all data. Relatedly, as described earlier, there was a marginal difference between boys with FXS-ASD who only completed the first time point and those who completed both on the resistance to task variable (a type of inappropriate response). Boys with FXS-ASD who completed both assessments resisted the task more often than those who only completed the first assessment. This marginal difference suggests that boys who produced the most resistant responses at Time 1 were more likely to be followed longitudinally. At Time 1, boys with FXS-ASD showed more resistance to task instructions than boys with ASD-O and DS, yet there were no differences at Time 2. Thus, this discrepancy–though not ideal for a longitudinal investigation–provides further support for our overall conclusion that boys with FXS-ASD show less impairment over time, as no group differences were detected for this (or any other variable) for boys at Time 2.

Summary and Clinical Implications

Communication repair ability is a key pragmatic skill for maintaining successful communicative interactions. Longitudinal findings from this study suggest that profiles of group and sex differences in the groups studied here may change with age, indicating potentially different areas of focus for assessment and intervention. Overall, findings expand our knowledge of pragmatic phenotypes in FXS, ASD, and DS by showing that, with age, male groups performed more similarly. This is especially important for FXS-ASD and ASD-O groups, who showed notable difficulties at the first time point. For boys with FXS-O and DS, our longitudinal findings continue to show that repair skills are not an area of pragmatic deficit in these groups. Similarly, female groups did not differ significantly from each other at either time point, although results of our HLM analyses, and a large effect size between girls with FXS-ASD and TD in particular, suggest that girls with FXS and DS may become more non-responsive to clarification requests over time. Finally, sex differences in FXS-ASD were observed over time, most notably with boys using more inappropriate responses but fewer non-responses than girls, suggesting that some phenotypic sex differences in this group may emerge over time.

Taken together with our prior study of repair ability in these groups at an earlier time point (Barstein et al. 2018), findings suggest that allocating additional focus on communication repair skills during assessment and intervention is warranted for boys with FXS-ASD and ASD-O at younger ages (the two groups who showed notable difficulty at Time 1). However, as boys with FXS-ASD and ASD-O no longer showed differences from other male groups in inappropriate responding and non-responsiveness, respectively, at Time 2, repair skills along these domains may increase with age and be less likely to be the focus of intervention at older ages. Of course, as noted above, these changes could also be the result of successful intervention, and this remains an important question for future work. Results also suggest that clinicians should consider sex-specific profiles in children with FXS-ASD, especially as children grow older, given our findings that some sex differences emerged only with age (e.g., inappropriate responses and non-responsiveness). Boys with FXS-ASD were the only group in the present study not to increase their use of metacomments with age; in fact, they decreased. Therefore, the use of metacomments and perhaps metacognitive skills in general may be a fruitful focus of intervention for this group. Finally, girls with FXS-ASD, FXS-O, and DS became more non-responsive with age, and a large (though non-significant) effect size indicated that girls with FXS-ASD were more non-responsive than TD controls at the second time point, suggesting that clinicians may monitor responsiveness to clarification requests as these girls grow older.

What This Paper Adds.

What is already known on this subject

The ability to repair breakdowns in communication is a critical pragmatic skill that can impact broader communicative abilities, particularly among individuals with neurodevelopmental disabilities impacting language. Previous studies of communication repair ability in fragile X syndrome (FXS), autism spectrum disorder (ASD), and Down syndrome (DS) have identified both strengths and weaknesses, with particular challenges noted for boys with idiopathic ASD and boys with comorbid FXS and ASD. Although this skill has been shown to develop over time in children with typical development, change over time had not yet been examined in these clinical groups. Research on girls with neurodevelopmental disabilities is also lacking.

What this study adds

The longitudinal findings presented in this study demonstrate that profiles of group and sex differences in these groups may change with age. In particular, boys with comorbid FXS and ASD and boys with idiopathic ASD showed significant changes with development, where marked difficulties were noted at the first time point but performance was more comparable to the other male groups (DS, FXS without ASD, and typical development) at a second time point approximately 3 years later. Whereas girls across groups did not differ significantly from each other at either time point, girls from the clinical groups became more non-responsive to clarification requests with age. Further, sex differences in boys and girls with comorbid FXS and ASD emerged over time.

Clinical implications of this study

Findings indicate that repair ability may be less of a clinical focus for boys with comorbid FXS and ASD and boys with idiopathic ASD as they grow older. On the other hand, girls with FXS and DS may become less responsive to clarification requests over time, together indicating potentially different areas of clinical focus across time and for boys versus girls.