English

Subword-Level Language Identification for Intra-Word Code-Switching

Computation and Language 2019-04-04 v1 Machine Learning

Abstract

Language identification for code-switching (CS), the phenomenon of alternating between two or more languages in conversations, has traditionally been approached under the assumption of a single language per token. However, if at least one language is morphologically rich, a large number of words can be composed of morphemes from more than one language (intra-word CS). In this paper, we extend the language identification task to the subword-level, such that it includes splitting mixed words while tagging each part with a language ID. We further propose a model for this task, which is based on a segmental recurrent neural network. In experiments on a new Spanish--Wixarika dataset and on an adapted German--Turkish dataset, our proposed model performs slightly better than or roughly on par with our best baseline, respectively. Considering only mixed words, however, it strongly outperforms all baselines.

Keywords

Cite

@article{arxiv.1904.01989,
  title  = {Subword-Level Language Identification for Intra-Word Code-Switching},
  author = {Manuel Mager and Özlem Çetinoğlu and Katharina Kann},
  journal= {arXiv preprint arXiv:1904.01989},
  year   = {2019}
}

Comments

NAACL-HLT 2019

R2 v1 2026-06-23T08:28:07.367Z