Agenda
Jan Odijk
Semi-Automatic Analysis of Spontaneous Language
The analysis of spontaneous language samples is considered an important tool in assessing language development disorders and aphasia. The analysis of transcripts of such spontaneous language samples is currently done mostly by hand, and requires specific knowledge, expertise and a significant amount of effort. Automation or partial automation of this process is therefore highly desirable.
I will present results of the SASTA application for the automatic grammatical analysis of transcripts of spontaneous Dutch language that is currently being developed. The application can deal with three language assessment methods commonly in use in the Netherlands: TARSP, STAP and ASTA. I will also report on results of modules of SASTA to deal automatically with deviating language that often occurs in these transcripts of spoken language of young children and persons with (suspected) language disorders. The correct analysis of such deviating language not only identifies such deviancies, which is required by some of the methods and may benefit the assessment, but also improves the automatic grammatical analysis.