Discovery of Words: Towards a Computational Model of Language Acquisition

The computational model presented here shows that words (and word-like entities) can be discovered without the need for a lexicon that is already populated. This discovery mechanism uses two very general learning principles that also play a role in language acquisition: the repetitive character of infant-directed speech on the one hand, and crossmodal associations in the speech and visual input on the other hand.

The computational model presented here shows that words (and word-like entities) can be discovered without the need for a lexicon that is already populated.