Harnessing Indigenous Tweets: The Reo Māori Twitter corpus

Area of Science:

• Computational linguistics
• Indigenous language documentation
• Social media analysis

Background:

• Te reo Māori is a vital part of Aotearoa New Zealand's cultural heritage.
• Limited digital resources exist for linguistic analysis and NLP development for te reo Māori.
• Social media platforms offer a growing space for Indigenous language use.

Purpose of the Study:

• To construct the Reo Māori Twitter (RMT) Corpus, a significant collection of te reo Māori tweets.
• To provide a valuable dataset for linguistic research on Māori language use online.
• To facilitate the development and enhancement of Natural Language Processing (NLP) tools for te reo Māori.

Main Methods:

• Collected 79,000 Māori-language tweets using computational methods.
• Identified Māori-language users via the Indigenous Tweets website.
• Pre-processed data and filtered out non-Māori content and noise.

Main Results:

• The Reo Māori Twitter (RMT) Corpus is the largest publicly available collection of exclusively Māori text on social media.
• Tweet volume peaked in 2014, with active users peaking in 2017.
• Over 600 users remained active in 2020, indicating continued online engagement.

Conclusions:

• The RMT Corpus is a unique and valuable resource for linguists, NLP developers, and Indigenous researchers.
• The corpus can enhance the development of Māori-language NLP applications.
• Increased visibility and use of te reo Māori online is supported by this resource.