Year: 2006
Type: article
Abstract: If bilingualism is a common phenomenon, research on bilingualism using recurrent networks is not.Few models using recurrent neural networks were developed in the past years in an attempt to address, among others, the question: is the lexical knowledge about two (or more) languages kept apart?French's (1998) results have shown that there are two separate regions in the lexicon, each one corresponding to one specific language.The present study extends French's model by including syntactic differences between the languages in addition to the exclusively lexical ones in the original simulation.Our network not only kept the lexicons apart, but also represented more fine grained structural differences between the languages in the hidden layer activation space.
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W2624561030