The Utility of Natural Language Samples for Assessing Communication and Language in Infants Referred with Early Signs of Autism.

Natural Language Sampling (NLS) offers clear potential for communication and language assessment, where other data might be difficult to interpret. We leveraged existing primary data for 18-month-olds showing early signs of autism, to examine the reliability and concurrent construct validity of NLS-derived measures coded from video-of child language, parent linguistic input, and dyadic balance of communicative interaction-against standardised assessment scores. Using Systematic Analysis of Language Transcripts (SALT) software and coding conventions, masked coders achieved good-to-excellent inter-rater agreement across all measures. Associations across concurrent measures of analogous constructs suggested strong validity of NLS applied to 6-min video clips. NLS offers benefits of feasibility and adaptability for validly quantifying emerging skills, and potential for standardisation for clinical use and rigorous research design.

Caregiver sensitivity predicts infant language use, and infant language complexity predicts caregiver language complexity, in the context of possible emerging autism.

While theory supports bidirectional effects between caregiver sensitivity and language use, and infant language acquisition-both caregiver-to-infant and also infant-to-caregiver effects-empirical research has chiefly explored the former unidirectional path. In the context of infants showing early signs of autism, we investigated prospective bidirectional associations with 6-min free-play interaction samples collected for 103 caregivers and their infants (mean age 12-months; and followed up 6-months later). We anticipated that measures of caregiver sensitivity/language input and infant language would show within-domain temporal stability/continuity, but also that there would be predictive associations from earlier caregiver input to subsequent child language, and vice versa. Caregiver sensitive responsiveness (from the Manchester Assessment of Caregiver-Infant interaction [MACI]) predicted subsequent infant word tokens (i.e., amount of language, coded following the Systematic Analysis of Language Transcripts [SALT]). Further, earlier infant Mean Length of Utterance (MLU; reflecting language complexity, also derived from SALT coding) predicted later caregiver MLU, even when controlling for variability in infant ages and clear within-domain temporal stability/continuity in key measures (i.e., caregiver sensitive responsiveness and infant word tokens; and infant and caregiver MLU). These data add empirical support to theorization on how caregiver input can be both supportive of, and potentially influenced by, infant capacities, when infants have social-communication differences and/or communication/language delays suggestive of possible emerging autism.