![[ Visit Client Website ]](images/banner.JPG)
Friday, May 16, 2008
Champagne Terrace/Bordeaux (Novotel London West)
Background: Delays in syntactic language development are found in verbal individuals with autism, with proposals that a third demonstrate comorbid Specific Language Impairment (SLI) (Kjelgaard & Tager-Flusberg, 2001). While interventions informed by linguistic theory can assist people with SLI in achieving longer utterances and enhanced communication (Ebbels, Van del Lely & Dockrell, 2007), it is unclear if such interventions can help learners with autism. Most current language interventions for autism focus on general communicative improvement, often assessed by vocabulary growth, rather than explicitly addressing syntactic delays.
GrammarTrainer is a text-based linguistic software program that teaches children to produce sentences of increasing grammatical complexity. It requires learners to actively construct phrases word by word, and provides principled feedback for ungrammatical utterances. As the program tests and corrects students, it keeps a running record of language errors.
Objectives: One reason for a de-emphasis on grammatical remediation is that the syntactic limitations for people with autism are not well documented. We examine an online database created by users of grammarTrainer in order to target the intervention to the needs of the users.
Methods: A database of 22 individuals with autism using grammar trainer was coded for type (syntactic/semantic/pragmatic) and frequency of errors. This error analysis was compared with a subset of the group, 6 individuals, who completed 7 week-long lessons.
Results: For complex utterances such as questions, 34% had syntactic errors, while 55% had pragmatic errors. Common syntax errors include leaving out the auxiliary, verb morphology, and failing to use subj-verb inversion for WH-questions. Pragmatic errors often involved use of the wrong question word. Individuals who completed 7 lessons had a 21% syntax error rate, and 8.3% semantic/pragmatic errors. Posttest syntactic errors less globally affected comprehensibility compared to pretest.
Conclusions: Syntactic and semantic/pragmatic errors are prevalent, and may be responsive to linguistically informed remediation techniques.
GrammarTrainer is a text-based linguistic software program that teaches children to produce sentences of increasing grammatical complexity. It requires learners to actively construct phrases word by word, and provides principled feedback for ungrammatical utterances. As the program tests and corrects students, it keeps a running record of language errors.
Objectives: One reason for a de-emphasis on grammatical remediation is that the syntactic limitations for people with autism are not well documented. We examine an online database created by users of grammarTrainer in order to target the intervention to the needs of the users.
Methods: A database of 22 individuals with autism using grammar trainer was coded for type (syntactic/semantic/pragmatic) and frequency of errors. This error analysis was compared with a subset of the group, 6 individuals, who completed 7 week-long lessons.
Results: For complex utterances such as questions, 34% had syntactic errors, while 55% had pragmatic errors. Common syntax errors include leaving out the auxiliary, verb morphology, and failing to use subj-verb inversion for WH-questions. Pragmatic errors often involved use of the wrong question word. Individuals who completed 7 lessons had a 21% syntax error rate, and 8.3% semantic/pragmatic errors. Posttest syntactic errors less globally affected comprehensibility compared to pretest.
Conclusions: Syntactic and semantic/pragmatic errors are prevalent, and may be responsive to linguistically informed remediation techniques.