Analyzing Code-mixing in Linguistic Corpora Using Kratylos

Abstract

Code-switching, code-mixing, and, more generally, multilingualism pose technological challenges for language documentation, the sub-discipline of linguistics that deals with the annotation and basic analysis of field recordings and other primary data. We focus here on a case study involving code-mixing in the endangered Koda language, which poses special problems for morphosyntactic analysis. We offer a robust approach to multilingual annotations that involves a combination of the popular open source software FieldWorks Language Explorer (FLEx) with Kratylos, a web-based corpus tool for display and query. Kratylos exposes linguistic data from various formats to powerful regular-expression queries that can exploit tier structure and other aspects of interlinear glossed text. We show how Kratylos can target mixed structures in our FLEx database of Koda that cannot be easily identified within the original FLEx software itself.