Risk for ASD is known to be elevated in both parents and siblings of affected children (Bailey et al., 1995; Dawson, Estes, Munson, Schellenberg, Bernier & Abbott, 2007; Folstein & Rutter, 1977, 1988; Landa, Piven, Wzorek, Gayle, Chase, & Folstein, 1992; Piven, Palmer, Jacobi, Childress & Arndt, 1997; Santangelo & Folstein, 1999). In addition, Le Couteur, Bailey, Goode, Pickles, Robertson, and Gottesman (1996) found that, in twin pairs discordant for autism, concordance for language impairment and a broader autism phenotype (BAP) was much greater in monozygotic pairs than dizygotic pairs, consistent with a strong genetic component. Although definitions are still evolving, Micali, Chakrabarti, & Fombonne (2004) present a definition of BAP based on epidemiological findings, which includes social and communication impairments, as well as repetitive and stereotyped behaviors. Social abnormalities have been found to consist of impaired friendships, impaired social play, odd behavior and impaired conversation. Developmental language abnormalities are seen particularly in pragmatic language areas (Landa et al., 1992). Obsessional and repetitive behaviors can include circumscribed interests, rigidity, obsessions/compulsions and repetitive interests and activities. Micali et al. (2004) define BAP as a constellation of these abnormalities at a subthreshold level of severity that would qualify an individual for neither ASD nor another diagnosis, such as developmental delay or anxiety disorder.

Recent research on ASD has included ongoing monitoring of infants with an older sibling with ASD. This research focus stems from the literature reviewed above, which suggests that genetic factors play a significant role in the incidence of ASD (Bailey et al., 1995; Folstein & Rutter, 1988; Le Couteur et al., 1996; Rutter, 2005). In addition, awareness of risk for disorders associated with BAP, even in the absence of the full-blown syndrome (Szatmari et al., 2000), contributes to this trend. Studies documenting infant siblings’ development are aimed at earlier identification of these potential problems, and earlier provision of treatment in order to maximize outcome potential.

In the present report, we describe findings on early vocal behaviors in a group of infant siblings with this risk, and compare these behaviors to those of age-mates from low-risk families. In doing so, we would like to argue that the direct and detailed analysis of vocal behaviors in infants at risk for ASD will provide richer information than scores on standard tests, such as Mullen (1995), or than retrospective parent report of early vocal behavior. First, these analyses can tell us not only that high risk infants are delayed relative to peers, but also in precisely what aspects of prespeech behavior, whether it be the acquisition of specific consonants, syllable shapes, or prosodic contours. This level of detail may be illuminating in better understanding how the path to spoken language is affected in high risk infants. Second, although parents are sensitive to certain milestones of speech development (first canonical syllables [Oller, Eilers, Neal, & Cobo-Lewis, 1998], first words), they are not generally attuned to the specific consonants their children are using or the relative distribution of consonant and syllable types in infant production. Finally, retrospective parent reports, which have been used in other studies of early speech development (e.g., Gernbacher, Sauer, Geye, Schweigert, & Goldsmith, 2008) are subject to errors of recall and bias. By directly examining the maturity of prelinguistic vocalizations, we test the hypothesis that infant siblings of children with ASD will show precursors to delays in expressive language that are consistent with the development of ASD or BAP at higher rates than will low risk infants, and we examine the relation of these delays in the first year of life to the presence of autistic symptomotology in the second year.

Second, differences in vocal production in the first year of life were associated with outcomes in terms of autistic symptomotology in the second year for the HR group.

Differences seen in prespeech vocal production of HR infants cannot be ascribed to a general decrement in developmental level, as their MSEL Visual Reception scores do not differ from those of LR peers at any time point measured. Rather, they appear to reflect a specific delay in acquiring age-appropriate sounds and syllable shapes in the first year, and in transitioning from the diverse range of sound production typical of all the young babies to a more restricted range dominated by speech-like sounds at the first year’s end. These delays are captured not only by the detailed analysis of sound production reported here; they are hinted at by MSEL Expressive Language scores, which are significantly lower than those of the LR group throughout the first year, although these scores are within the normal range. Moreover, Expressive Language scores on the MSEL correlate with vocal production of consonants at 12 months. This is likely a reflection of the fact that most items used to score Expressive Language on the MSEL at the 12 month level relate to consonant production. It is also noteworthy, however, that fine motor skills are also related to consonant production at this age.

These findings can be interpreted to suggest that pre-speech sound development is a sensitive indicator of the rate and degree to which infants at risk for ASD are following the developmental path to language acquisition, and, by extension, to expressive communication. Motor delays cannot be ruled out as a source of the pattern seen here, especially in light of the correlation between Fine Motor scores on the MSEL and consonant production at 12 months. Despite this correlation, though, findings from the MSEL suggest that, apart from an apparent mild delay in Fine Motor performance at 6 months, no significant motor delays persist into the second half of the first year.

Does this detailed analysis of vocal production in infants at risk for ASD contribute any information beyond what is measured by the more general MSEL language scales? Potential answers to this question may be seen in the changing pattern of relationship between the vocal performance of the HR and LR groups throughout the first year of life, as well as the findings of the discriminant function analyses. That is, detailed description of vocal production reveals that the areas of difference consistently reflect slower progress on the part of the HR infants in moving from one phase of pre-speech development to the next. By 9 months -- when LR children are developing a range of tools necessary to begin the transition from pre-speech to first speech production -- HR children produce, on average, a lower rate of production of canonical syllables, the kind that will eventually be used as first words, and a smaller repertoire of consonants, particularly the Late group, those just emerging at this developmental level. By 12 months, the participants with HR, as a group, are not significantly different from LR peers in production of the consonant and syllable types their LR peers learned some months earlier (although they continue to trend toward less frequent production) but they continue, as they have throughout the first year, to produce a higher proportion of other kinds of vocalizations that do not contribute to their transition to first words (See Fig. 5). As we have argued previously (Schoen et al., 2008) LR infants, tend, toward the end of the first year of life, to match their productions more and more closely to ambient speech models. HR infants, on the other hand, show less winnowing of the other kinds of sounds characteristic of the first year of life. Thus the pattern of pre-speech development seen in the HR group suggests both slower transition from one phase of pre-speech development to the next, and a reduced tendency to hone sound production increasingly closely to models produced by others in the environment as the child nears the first birthday. This more detailed examination suggests that low MSEL Expressive Language scores have different sources at different ages: slow acquisition of pre-speech behaviors during the first year, and a slower transition away from non-speech vocal production near the first birthday. Moreover, although MSEL Expressive Language scores at 6 mo. appear to be somewhat sensitive to differences between children who will and will not go on to show ASD symptoms at 24 mo., the same is not true at 12 mo., even though MSEL Expressive Language is generally correlated with consonant production at this age. It would appear that although the MSEL Expressive Language score captures some of the decreased prespeech behavior present at 6 mo. in HR infants who go on to develop symptoms, it is less sensitive to the transitional behaviors at 12 months that are associated with later symptom appearance.

It is important to recall that the high prevalence of autistic symptoms in the HR does not necessarily mean that all symptomatic children with remain on the autism spectrum. Thus HR/LR group differences reported here may be driven by the children who have significant vulnerabilities for ASD. The discriminant function analyses are telling in this regard. At each age level, the metric most strongly associated with placement in the outcome group with or without autistic symptoms is the production of newly emerging speech behaviors. That is, at six months it is the frequency of middle consonant production, the consonants just emerging at this age, that predicts group placement; at 9 months, it is the frequency of late consonants, again those at the leading edge of prespeech development. At 12 months, even though group differences in consonant production are no longer significant, it is the total number of consonants, which can be interpreted to reflect the amount of word-like babble, that is associated with outcome.

What might this pattern suggest about spoken language development in children at risk for ASD? It would seem, first, to indicate that it is not only the development of communicative intention and language function that is vulnerable in the HR population; the acquisition of basic vocal “tools” for the development of speech is also affected, even before they are recruited into the service of linguistic communication. Prior to the onset of language, babbling is used in typical development both as a means of “practicing” the motor patterns associated with speech production and as an early form of interaction, through back-and-forth babbled “conversations” with family members (Locke, 1989). Perhaps vocal development is vulnerable in HR infants because they are less apt to “practice” sound-making in order to match it more closely to the sounds they hear others in their environment producing, as well as because they are less captivated by the turn-taking exchanges in which babbled productions allow them to participate. Thus, the differences observed here may be a result of vulnerabilities for social disability, and not their source. It is not the case, however, that HR infants are simply “quieter” than LR peers, since the total number of vocalizations produced never differed between the two risk groups at any age. Differences appear not in sound production, but in the kinds of sounds produced, with LR children and HR children who do not develop ASD symptoms producing more vocal behavior at the “cutting edge” of development at each phase.