Abstract (eng)
Until recently there has been a big debate going on regarding the acquisition and learning of language in children.
The aim of the work is to study the syntactical feature of irregular verb morphology in Williams syndrome in an interdisciplinary framework. The present study has been motivated by conducting a small case study to find out, if there is either a developmental delay in the acquisition of Williams syndrome language or rather if there is a deficit in the lexicon. Thus , it comes down to the research question:
Is irregular verb morphology selectively vulnerable in German speaking WS children?
In case this question is true, the second research question was:
Are there are any over-generalization effects that occur in irregular verb inflection?
Apart from this, three tests were conducted that served as a basis for matching. The controls in the present study were matched on chronological-age, verbal mental age and an included digit span.
The results showed that German speaking WS children show a rather weak performance on irregular inflected verbs, as it has been observed by Clahsen and Almazan for English speaking WS children.
In addition, we conducted a connectionist study to simulate two experimental conditions. In a connectionist network, a WS model and a normal model were implemented and trained to learn the participle and past perfect form. Additionally, the learning algorithm used was not only back propagation but an algorithm named ART (Adaptive Resonance Theory). It has the advan-tage that it simulates learning either very rapidly or slowly. Besides, it is biologically plausible, because of the existence of a short-term memory and a long-term memory compound.
The results suggest that the ART algorithm is better suitable for modelling irregular verb inflection in a normal model, whereas back propagation simulates better a delay in irregular verb flexion.
acquisition. In conclusion to that, computational modelling, such as back propagation and ART algorithm give a perfect opportunity to simulate developmental disorders and normal develop-ment of language depending on the given framework.