Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Is It Navajo? Accurate Language Detection in Endangered Athabaskan Languages

Ivory Yang, Weicheng Ma, Chunhui Zhang, Soroush Vosoughi

TL;DR

This work addresses the lack of NLP support for Navajo and related Native American languages by evaluating Google's LangID and proposing a targeted Random Forest classifier trained to distinguish Navajo from languages LangID misidentifies. The approach leverages two benchmark datasets and 5,000-feature representations, achieving 97% overall accuracy with Navajo precision of 1.00 and recall of 0.99, and demonstrates generalizability to Athabaskan relatives. The findings highlight the viability of decentralized, language-specific NLP tools to aid preservation and revitalization, while revealing limitations when extending to closely related languages. Overall, the study underscores the urgency of inclusive language technologies that respect linguistic diversity and cultural heritage in the digital age.

Abstract

Endangered languages, such as Navajo - the most widely spoken Native American language - are significantly underrepresented in contemporary language technologies, exacerbating the challenges of their preservation and revitalization. This study evaluates Google's Language Identification (LangID) tool, which does not currently support any Native American languages. To address this, we introduce a random forest classifier trained on Navajo and twenty erroneously suggested languages by LangID. Despite its simplicity, the classifier achieves near-perfect accuracy (97-100%). Additionally, the model demonstrates robustness across other Athabaskan languages - a family of Native American languages spoken primarily in Alaska, the Pacific Northwest, and parts of the Southwestern United States - suggesting its potential for broader application. Our findings underscore the pressing need for NLP systems that prioritize linguistic diversity and adaptability over centralized, one-size-fits-all solutions, especially in supporting underrepresented languages in a multicultural world. This work directly contributes to ongoing efforts to address cultural biases in language models and advocates for the development of culturally localized NLP tools that serve diverse linguistic communities.

Is It Navajo? Accurate Language Detection in Endangered Athabaskan Languages

TL;DR

This work addresses the lack of NLP support for Navajo and related Native American languages by evaluating Google's LangID and proposing a targeted Random Forest classifier trained to distinguish Navajo from languages LangID misidentifies. The approach leverages two benchmark datasets and 5,000-feature representations, achieving 97% overall accuracy with Navajo precision of 1.00 and recall of 0.99, and demonstrates generalizability to Athabaskan relatives. The findings highlight the viability of decentralized, language-specific NLP tools to aid preservation and revitalization, while revealing limitations when extending to closely related languages. Overall, the study underscores the urgency of inclusive language technologies that respect linguistic diversity and cultural heritage in the digital age.

Abstract

Endangered languages, such as Navajo - the most widely spoken Native American language - are significantly underrepresented in contemporary language technologies, exacerbating the challenges of their preservation and revitalization. This study evaluates Google's Language Identification (LangID) tool, which does not currently support any Native American languages. To address this, we introduce a random forest classifier trained on Navajo and twenty erroneously suggested languages by LangID. Despite its simplicity, the classifier achieves near-perfect accuracy (97-100%). Additionally, the model demonstrates robustness across other Athabaskan languages - a family of Native American languages spoken primarily in Alaska, the Pacific Northwest, and parts of the Southwestern United States - suggesting its potential for broader application. Our findings underscore the pressing need for NLP systems that prioritize linguistic diversity and adaptability over centralized, one-size-fits-all solutions, especially in supporting underrepresented languages in a multicultural world. This work directly contributes to ongoing efforts to address cultural biases in language models and advocates for the development of culturally localized NLP tools that serve diverse linguistic communities.
Paper Structure (9 sections, 5 figures, 2 tables)

This paper contains 9 sections, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Identification for NatAm Languages
  4. Benchmark Construction
  5. Model Specifics
  6. Experimental Results
  7. Model Generalizability
  8. Conclusion and Future Work
  9. Aligned Words Across Native American Languages

Figures (5)

  • Figure 1: Google's LangID does not currently support any Native American languages, including Navajo and other Athabaskan languages. Our model addresses these challenges effectively.
  • Figure 2: Visual representation of some erroneously suggested languages by Google LangID, along with their frequency counts.
  • Figure 3: Classification results for Navajo and 20 other languages, presented as a confusion matrix. The intensity of the color indicates the frequency of classifications. Labels on x and y axes correspond to the languages listed in Table \ref{['tbl:classification_metrics']}.
  • Figure 4: Family Tree for Athabaskan Languages
  • Figure 5: Aligned Words from apache.