The Bahrain Corpus: A Multi-genre Corpus of Bahraini Arabic

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 2345–2352 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 2345 The Bahrain Corpus: A Multi-genre Corpus of Bahraini Arabic Dana Abdulrahim, Go Inoue†, Latifa Shamsan, Salam Khalifa‡, Nizar Habash† University of Bahrain, †New York University Abu Dhabi, ‡Stony Brook University {darahim,lshamsan}@uob.edu.bh {go.inoue,nizar.habash}@nyu.edu salam.khalifa@stonybrook.edu Abstract In recent years, the focus on developing natural language processing (NLP) tools for Arabic has shifted from Modern Standard Arabic to various Arabic dialects. Various corpora of various sizes and representing different genres, have been created for a number of Arabic dialects. As far as Gulf Arabic is concerned, Gumar Corpus (Khalifa et al., 2016) is the largest corpus, to date, that includes data representing the dialectal Arabic of the six Gulf Cooperation Council countries (Bahrain, Kuwait, Saudi Arabia, Qatar, United Arab Emirates, and Oman), particularly in the genre of “online forum novels”. In this paper, we present the Bahrain Corpus. Our objective is to create a specialized corpus of the Bahraini Arabic dialect, which includes written texts as well as transcripts of audio files, belonging to a different genre (folktales, comedy shows, plays, cooking shows, etc.). The corpus comprises 620K words, carefully curated. We provide automatic morphological annotations of the full corpus using state-of-the-art morphosyntactic disambiguation for Gulf Arabic. We validate the quality of the annotations on a 7.6K word sample. We plan to make the annotated sample as well as the full corpus publicly available to support researchers interested in Arabic NLP. Keywords: Arabic, Gulf Arabic, dialect, morphology, corpus 1. Introduction There is an abundance of corpora of various sizes, types, and representing various genres, that cater pri- marily to Modern Standard Arabic (MSA). Dialectal Arabic (DA), on the other hand, has only recently re- ceived attention in the field of corpus linguistics, in particular, and the field of natural language processing (NLP), in general. According to Habash (2021), there are numerous chal- lenges that face NLP researchers of DA. Most notewor- thy is the large number of forms of Arabic words, re- sulting from the rich morphology of Arabic language (Holes, 2004). Lexical items in Arabic can be marked for different inflectional features such as number, per- son, gender, aspect, mood, case, voice, etc. All of that in addition to pronouns and particles (e.g. prepositions, conjunctions, the definite article, etc.) that can attach to the lexical item. Dialects of Arabic do not all share the same exact set of inflectional affixes or clitics (neither with each other, nor with MSA). Due to the significant differences that exist between MSA and DA, MSA pro- cessing tools that have been previously developed have proved to be quite insufficient in the attempt of pro- cessing DA data (e.g. Khalifa et al. (2016)). Addition- ally, lack of a standard orthography for DA results in inconsistent spelling of lexical items in DA (Habash, 2021). For instance, speakers of Bahraini Arabic would provide the following different spellings for the phrase �� …

Content truncated to 500 words