Assessing Grammar, Vocabulary, Syntactic Complexity and Pragmatics In Children with Autism Before and After STAR and TEACCH

Background: Language abilities in children with autism are one of the strongest predictors of behavioral and social outcomes (Rogers, 2005).  Interventions targeting language skills  generally conceptualize them very broadly; almost no larger studies offer detailed reporting of the intervention’s benefits for specific aspects of linguistic competence other than expressive or receptive vocabulary and language or communication composite scores. 

Objectives: To examine changes in grammar, vocabulary, pragmatics and syntactic complexity among children with ASD in response to a community implementation of evidence-based autism intervention, and whether intervention fidelity moderates linguistic change across the intervention period.  

Methods:  The Autism Instructional Methods Survey (AIMS) study (Mandell, Shin, Stahmer & Marcus, 2010), is a randomized field trial taking place in collaboration with the School District of Philadelphia.  AIMS compared two evidence-based interventions Strategies for Teaching based on Autism Research (STAR) and Treatment and Education of Autistic and Communication related handicapped Children (TEACCH) in 38 autism support classrooms in one urban school district.  STAR comprises three instructional strategies – discrete trial training, pivotal response training and teaching in functional routines – matched to a comprehensive curriculum. TEACCH comprises visual strategies and physical prompts to clarify and highlight important tasks the child has to accomplish.  Classroom staff received didactic training and in-school consultation in the intervention condition to which they were randomized.  Children were assessed in September and May of the intervention year using the Differential Abilities Scale and the Autism Diagnostic Observation Schedule.  The study also assessed teacher fidelity to the assigned intervention.

The sample for the current study included 91 of the 138 students enrolled in AIMS. These 91 had a confirmed diagnosis of autism, were between the ages 5.2 and  9.1 years, and were  assessed using the Module 2 or Module 3 of the ADOS.

             Naïve coders trained in linguistics transcribed > 200 utterances for each child using coding guidelines from the Child Language Data Exchange System (CHILDES).  These Natural Language Sample (NLS) Transcripts were analyzed in 4 areas, both pre- and post- intervention:  Productive Grammar (measured by Mean Length Utterance (MLU)), Vocabulary (assessed using number of distinct words divided by number of absolute words), Pragmatics (evaluated using the ratio of  responsive interactions versus interruptions), and syntactic complexity (assessed using the Index of Productive Syntax (IPSyn) (Scarborough, 1990)).

Results: Analysis, which is ongoing, will compare treatment groups stratified by teacher fidelity and intervention condition.

Conclusions:   While experts have emphasized that NLS is a crucial component of assessing functioning in children with autism (Tager-Flusberg, Rogers, Cooper, Landa, Lord, Paul et al., 2009), few intervention studies have reported using related measures of outcome and we know of no comparative intervention studies that provide detailed NLS-based language outcomes. The results can provide information on how different therapeutic methods contribute to generalizable linguistic skills in vocabulary, syntax, utterance length and pragmatics, and how these outcomes are affected by treatment fidelity.