TweetTaglish: A Dataset for Investigating Tagalog-English Code-Switching

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 2090–2097 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 2090 TweetTaglish: A Dataset for Investigating Tagalog-English Code-Switching Megan Herrera, Ankit Aich, and Natalie Parde Natural Language Processing Laboratory Department of Computer Science University of Illinois at Chicago {mherre42, aaich2, parde}@uic.edu Abstract Deploying recent natural language processing innovations to low-resource settings allows for state-of-the-art research findings and applications to be accessed across cultural and linguistic borders. One low-resource setting of increasing interest is code- switching, the phenomenon of combining, swapping, or alternating the use of two or more languages in continuous dialogue. In this paper, we introduce a large dataset (20k+ instances) to facilitate investigation of Tagalog-English code-switching, which has become a popular mode of discourse in Philippine culture. Tagalog is an Austronesian language and former official language of the Philippines spoken by over 23 million people worldwide, but it and Tagalog-English are under-represented in NLP research and practice. We describe our methods for data collection, as well as our labeling procedures. We analyze our resulting dataset, and finally conclude by providing results from a proof-of-concept regression task to establish dataset validity, achieving a strong performance benchmark (R2=0.797–0.909; RMSE=0.068–0.057). Keywords: Tagalog, code-switching, Tagalog-English 1. Introduction Bilingualism and multilingualism are common phe- nomena in the Philippines, a country with around 181 languages actively spoken within its borders (Bravante and Holden, 2020). Standardized Tagalog (also re- ferred to, and sometimes formally distinguished from, Filipino) is the native language of approximately 23% of the Philippines’ population (Bravante and Holden, 2020), comprising the largest cultural-linguistic group in the country.1 Despite Tagalog’s cultural importance within the Philippines, many cornerstones of modern natural language processing (e.g., syntactic and depen- dency parsers) are unavailable or underdeveloped for this widely spoken language (Aquino and de Leon, 2020). Tagalog data remains scarce (Manguilimotan and Matsumoto, 2011; Samson, 2018), and the lan- guage’s typological differences from higher-resourced Austronesian languages such as Indonesian (Greenhill et al., 2009; Reid, 2018) hinder its further computa- tional exploration. Mixing of Tagalog and English in a single utterance or conversation is part of a sociolinguistic phenomenon often called code-switching (CS) or code-mixing (Jain and Bhat, 2014). Code-switching between the two lan- guages (Taglish) is extremely common in the Philip- pines, and serves as a cultural and social tool. For ex- ample, Taglish is seen as the language of the “educated, middle- and upper-class urbanites of the Philippines,” 1Filipino and Tagalog are mutually intelligible languages and sometimes no distinction is made between them (Cornell University, 2022); mixing them is common and considered “the normal acceptable conversational style of speaking and writing” (Goulet, 1968). Throughout this paper, we refer to the language as Tagalog since that is how most speakers of the language refer to and use it in colloquial, informal settings. and speakers may code-switch to create or lessen dis- tance from that association (Lesada, 2017). Thus, ef- fectively processing language in real-world Tagalog ap- plications may require not …

Content truncated to 500 words