International Meeting for Autism Research: Advantages of CAI for Single Word Comprehension and Grammar Production Training

Advantages of CAI for Single Word Comprehension and Grammar Production Training

Thursday, May 20, 2010
Franklin Hall B Level 4 (Philadelphia Marriott Downtown)
2:00 PM
F. Hurewitz , Department of Psychology, Drexel University, Phila, PA
K. Boser , Individual Differences in Learning, Inc, Ellicott City, MD
Background:   Several controlled studies have recently demonstrated increased learning benefits from computer aided instruction (CAI) for children with autism in areas such as functional, play and pragmatic skills. Potential benefits of CAI in language instruction include data automatization, modeling, ease of information presentation and immediate feedback delivery for more individually directed learning.  The social-communicative nature of language highlights the importance of research regarding the advantages of 'individualized' instruction in using CAI compared to more natural language learning contexts.

We report two experiments that examine individualized benefits and efficacy of CAI for language instruction for 1) single word comprehension  and 2) sentence level production in  children with autism

  Experiment 1) reports a 10-week study using Cosmo’s Learning System (CLS) software (from Anthrotronix, Inc.) with six 'low-verbal' children with autism, aged 6-12 years. Subjects received two half hour sessions a week focused on developing identification of auditory and written words (e.g., prepositions, shape, color) and also following directions and discrimination of  size/ amount.  CLS included leveled cueing, rewarding animations and automatized progress through levels. Experiment 2) reports a 6 week study using GrammarTrainer(GT) software to instruct eleven 9-17 year-old individuals with autism. Training was conducted daily for 20 minutes. GT teaches syntax production through progressive exposure to written grammar  increasing in complexity from small phrases to complex sentences. Training begins with easier recognition-style responses and increases in difficulty to full elicited production of  constructions. All responses are 'keyed' in via word buttons.  Students were verbal, but showed weaknesses in language production, particularly syntax and morphology.

  Experiment 1) A computerized assessment was created to evaluate initial skills, progress on CLS  and generalization to new questions. All subjects improved and generalized on post-test in all areas tested except color identification. Lower functioning subjects focused on easier problems and completed less content. More verbal subjects improved an average of 20% in comprehension of prepositions,  following directions and with written words. Relative size/amount improved but students had difficulty understanding the concept ‘same’. All subjects showed an increase in the number of trials completed and required less prompting in later training sessions. Experiment  2) All students improved across time and in pre-, progress and post testing of novel items. As hypothesized, the students using GT showed significant improvement on the morphemes subtest of the Comprehensive Assessment of Spoken Language (CASL), but did not improve on measures of vocabulary (e.g. PPVT-IV), or on the pragmatics subtest of the CASL.  This shows that writing-based delivery of CAI may generalize to spoken language syntax.

CAI may be one way to improve necessary language exposure and feedback through increased repetition otherwise difficult to implement in more 'natural' learning settings for children with autism. These experiments support generalization of specific language skills in CA instruction, but CAIs must be well-matched to student profiles and skills.
See more of: Treatment
See more of: Clinical & Genetic Studies