Diversifying language models for lesser-studied languages and language-usage contexts: A case of second language Korean

Abstract

This study investigates the extent to which currently available morpheme parsers/taggers apply to lesser-studied languages and languageusage contexts, specifically focusing on second language (L2) Korean. We pursue this inquiry by ( 1) training a neural-network model (pretrained on first language [L1] Korean data) on varying L2 datasets and (2) measuring its morpheme parsing/tagging performance on L2 test sets from both the same and different sources of L2 training sets. The results show that the L2 trained models generally excel in L2 domainspecific parsing and tagging tasks compared to the L1 pre-trained baseline model. Interestingly, increasing the size of L2 training data does not lead to improving model performance consistently.