Assessing Digital Language Support on a Global Scale

Proceedings of the 29th International Conference on Computational Linguistics, pages 4299–4305 October 12–17, 2022. 4299 Assessing Digital Language Support on a Global Scale Gary F. Simons SIL International gary_simons@sil.org Abbey L. Thomas The University of Texas at Dallas abbey.thomas@utdallas.edu Chad K. White SIL International chad_white@sil.org Abstract The users of endangered languages struggle to thrive in a digitally-mediated world. We have developed an automated method for assessing how well every language recognized by ISO 639 is faring in terms of digital language sup- port. The assessment is based on scraping the names of supported languages from the web- sites of 143 digital tools selected to represent a full range of ways that digital technology can support languages. The method uses Mokken scale analysis to produce an explainable model for quantifying digital language support and monitoring it on a global scale. 1 Introduction The users of endangered languages struggle to thrive in a digitally-mediated world. The opportuni- ties afforded by digital technology differ drastically depending on the language being used. This has been dubbed the “digital language divide” (Mikami, 2008; Young, 2015; Soria, 2016; Matsakis, 2019). As digital modes of communicating and accessing information become increasingly necessary in daily life, lack of digital language support (DLS) for a language means that its speakers must use other languages to participate in the global information society or be left out. Linguists have been writing for decades about the role digital technology could play in lan- guage revitalization (Warschauer, 1998; Buszard- Welcher, 2001; Eisenlohr, 2004; Galla, 2009; Holton, 2011; Cru, 2016). Language technologists are recognizing the inequities facing the vast ma- jority of the world’s languages (Bird, 2020; Blasi et al., 2022) and are embracing the challenges of bringing greater equity in DLS (Joshi et al., 2019; Bapna et al., 2022; Edunov et al., 2022). However, in a world where most people are mul- tilingual and each language fits into its functional niche within an ecology of languages (Lewis and Si- mons, 2016), full digital support for every language is not a realistic goal nor what those multilingual individuals are necessarily looking for (Bird, 2022). The goal of our research is to develop a method for measuring DLS in every language, so that it will be possible to provide an empirical view of the digital state of the world’s languages and to observe the progress as so-called low-resource languages move toward crossing the digital language divide. 2 Related Work Our primary inspiration has been the seminal work by Kornai (2013) on developing a method for as- sessing the digital vitality of any language. He pro- poses a four-way classification of languages as dig- itally Thriving, Vital, Heritage, or Still, “roughly corresponding to the amount of digital communica- tion that takes place in the language.” His method harvests data from the Web, then uses supervised classification to automatically label all known lan- guages. In practice, he adds a fifth level, Borderline, to represent languages that show signs of crossing the gap from Still to Vital. He …

Content truncated to 500 words