Not always about you: Prioritizing community needs when developing endangered language technology

Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 3933 - 3944 May 22-27, 2022 c⃝2022 Association for Computational Linguistics Not always about you: Prioritizing community needs when developing endangered language technology Zoey Liu ∗ Boston College zoey.liu@bc.edu Crystal Richardson (Karuk) ∗ University of California, Davis cjrichardson@ucdavis.edu Richard Hatcher Jr University at Buffalo rjhatche@buffalo.edu Emily Prud’hommeaux Boston College prudhome@bc.edu Abstract Languages are classified as low-resource when they lack the quantity of data necessary for training statistical and machine learning tools and models. Causes of resource scarcity vary but can include poor access to technology for developing these resources, a relatively small population of speakers, or a lack of urgency for collecting such resources in bilingual pop- ulations where the second language is high- resource. As a result, the languages described as low-resource in the literature are as different as Finnish on the one hand, with millions of speakers using it in every imaginable domain, and Seneca, with only a small-handful of flu- ent speakers using the language primarily in a restricted domain. While issues stemming from the lack of resources necessary to train models unite this disparate group of languages, many other issues cut across the divide between widely-spoken low-resource languages and en- dangered languages. In this position paper, we discuss the unique technological, cultural, prac- tical, and ethical challenges that researchers and indigenous speech community members face when working together to develop lan- guage technology to support endangered lan- guage documentation and revitalization. We report the perspectives of language teachers, Master Speakers and elders from indigenous communities, as well as the point of view of academics. We describe an ongoing fruitful col- laboration and make recommendations for fu- ture partnerships between academic researchers and language community stakeholders. 1 A thought experiment Say that we have three speech communities, which we will refer to as Elephant, Ocelot, and Coy- ote. Each community has their own language, which is commonly characterized by language tech- nology researchers as “low-resource” or “under- resourced”. Elephant’s language is spoken by *Equal contribution. around 10 million first language (L1) speakers and millions more second language (L2) speakers, and it has a standard widely accepted orthography. Eth- nologue (Lewis et al., 2015; Eberhard et al., 2021) classifies this language as “used in education, work, mass media, and government at the national level.” Ocelot’s language has about 500,000 L1 speakers and three different orthographic representations. This language is noted in Ethnologue as “in vig- orous use, with literature in a standardized form being used by some though not yet widespread or sustainable”, and is described as an “indigenous” language. In comparison, the language of Coy- ote has only a handful of speakers that could be considered fluent L1 speakers, most of whom are elders. The language currently lacks a standard orthography, and in Ethnologue it is classified as “endangered” (Meek, 2012). Elephant wants to develop spoken language tech- nology primarily to support the use of speech- enabled applications and smartphone …

Content truncated to 500 words