Multiple Measures of Language Development in Autism: Understanding Language Ability and Usage across Contexts

Background: Toddlers with autism demonstrate highly variable language abilities (Kjelgaard & Tager-Flusberg, 2001), and as such it has been recommended that each language phase of early development in autism be assessed using multiple sources of information (Tager-Flusberg et al., 2010). Although automated measures of child vocalizations has been found to be stable and valid for predicting future spoken language outcomes (Woynaroski, Oller, Kaysili, & Yoder, 2016), these measures do not provide specific detail as to why a child is communication or whom the communication is directed. Additionally, many measures across development are not predictive of skills 5 years later in children with autism. However, early observations of non-verbal communicative acts are significantly associated with later language, social, and communication outcomes (Charman et al., 2005).

Objectives: Coupled with the current recommendations for assessment of expressive language benchmarks in early development of autism, we consider 4 primary measures of language: child vocalizations, non-verbal communication observed during a semi-structured observation, vocabulary as reported by parents, and pragmatics observed during a naturalistic language sample. (1) How do language measures relate to and diverge from one another in toddlers with autism? (2) What profiles of autism emerge across measures?

Methods: A total of 46 toddlers (mn=33, sd=6 months; 40% minority race; Mullen T-score mean=26.9, sd=8.7) with autism have been assessed for baseline characteristics and language abilities across contexts. Toddlers were recruited in the Chicagoland area and were included as part of a larger randomized control trial based on the following characteristics: a) between 18 and 36 months of age, b) live with their biological mother and she was willing to participate in the study, c) a confirmed diagnosis of autism, d) no other developmental or sensory impairment diagnosis including hearing impairment, uncorrected vision impairment, identified genetic disorder, or seizure disorder, and English as the primary language spoken at home. Children were assessed across multiple contexts. Parents reported the number of different words the child understood and said (MCDI, Fenson et al., 2007). Total child vocalizations was extracted from day-long recordings at home using LENA^TM technology (LENA Research Foundation).

Results: All measures of early language and communication were moderately and significantly correlated (r=0.34-0.75; p<0.00), except for the relationships between observed non-verbal communicative behaviors and all other measures of language and communication (r=0.03-0.23, p>0.1). Across participants, 4 profiles of toddlers emerge: high performing toddlers across all measures (8%), low performing toddlers across all measures (24%), toddlers who perform low across measures except demonstrate many different gestures (35%), and toddlers who present with mixed profiles (39%). These profiles illustrate the added benefit of including gestures and vocalizations in language measures.

Conclusions: Due to the discrepancy between non-verbal communication and verbal communication measures, it is critical to consider both forms of communication in early intervention research with this population, especially due to the known relationship between non-verbal communication (above other indicators) and long-term language outcomes in autism (Charman et al., 2005).