Exploring Lexical Network Development in Second Language Learners.

Abstract

This study explores how neural network models can simulate word production in second language (L2) learners. A neural network was trained to simulate L2 word production using a variety of word properties related to connectionist networks (hypernymy, polysemy, concreteness, and meaningfulness). The study demonstrates that a neural network can produce words to a similar degree as L2 learners. The findings are important for theories of L2 lexical growth and production.