MultiMASC: An Open Linguistic Infrastructure for Language Research

This paper describes MultiMASC, which builds upon the Manually Annotated Sub-Corpus (MASC) (Ide et al., 2008; Ide et al., 2010) project, a community-based collaborative effort to create, annotate, and validate linguistic data and annotations on a broad-genre open language data. MultiMASC will extend MASC to include comparable corpora in other languages that not only represent the same genres and styles, but also include similar types and number of annotations represented in a common format. Like MASC, MultiMASC will contain only completely open data, and will rely on a collaborative community-based effort for its development. We describe the possible ways in which additional corpora for MultiMASC can be collected and annotated and consider the dimensions along which “comparability” for MultiMASC corpora can be determined. Because it is unlikely that all language-specific MultiMASC corpora can be comparable along every dimension, we also outline the measures that can be used to gauge comparability for a number of different