Immer schön der Reihe nach: Sequenzialisierung in konnektionistischen Sprachproduktionsmodellen

Abstract

Local connectionist models are built up from networks of connected nodes. In these models, each node represents exactly one item to be represented. In language production models, these items are words, morphemes, phonemes, and so on. Connectionist models incorporate parallel processing. Thus, they are ideal models for both the production of slips of the tongue and aphasic speech. With respect to aphasia, the model incorporates specific impairments. However, connectionist models are not ideal for the production of sequences of linguistic units. In this article we first explain why sequentialisation is that problematic at all. Then, we present a solution for our model, which is based on control chains on the one hand and lateral inhibition on the other hand.