Theoretical study of lexical network structure in second language learning

This paper attempts to explore how neural network models can simulate word production in second language learners. Lexical Network Theory asserts that the semantic portion of the lexicon is best seen as a network of word senses, where each sense is connected to other semantically-related senses of the same word and indirectly to other words in the same semantic field. To this end, a neural network was trained to simulate L2 word production using a variety of word properties related to connectionist networks. Our purpose is to demonstrate how word properties can be used to simulate word production by second language learners.