Text and Speech-based Tunisian Arabic Sub-Dialects Identification

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 6405–6411 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 6405 Text and Speech-based Tunisian Arabic Sub-Dialects Identification Najla Ben Abdallah1, Sam´eh Kchaou2, Fethi BOUGARES3 1FLAH Mannouba , 2Sfax Universit´e, 3Le Mans Universit´e Tunis-Tunisia, Sfax-Tunisia, Le Mans-France benabdalahnajla@gmail.com, samehkchaou4@gmail.com, fethi.bougares@univ-lemans.fr Abstract Dialect IDentification (DID) is a challenging task, and it becomes more complicated when it is about the identification of dialects that belong to the same country. Indeed, dialects of the same country are closely related and exhibit a significant overlapping at the phonetic and lexical levels. In this paper, we present our first results on a dialect classification task covering four sub-dialects spoken in Tunisia. We use the term ’sub-dialect’ to refer to the dialects belonging to the same country. We conducted our experiments aiming to discriminate between Tunisian sub-dialects belonging to four different cities: namely Tunis, Sfax, Sousse and Tataouine. A spoken corpus of 1673 utterances is collected, transcribed and freely distributed. We used this corpus to build several speech- and text-based DID systems. Our results confirm that, at this level of granularity, dialects are much better distinguishable using the speech modality. Indeed, we were able to reach an F-1 score of 93.75% using our best speech-based identification system while the F-1 score is limited to 54.16% using text-based DID on the same test set. Keywords: Tunisian Dialects, Sub-Dialects Identification, Speech Corpus, Phonetic description 1. Introduction Due to historical and sociological factors, Arabic-speaking countries have two main varieties of languages that they use in daily life. The first one is Modern Standrad Arabic (MSA), while the second is the dialect of that country. The former (i.e. MSA) is the official language that all Arab countries share and commonly use for their official communication in TV broadcasts and written documents, while the latter (i.e. dialect) is used for daily oral commu- nication. Arabic dialects have no official status, they are not used for official writing, and therefore do not have an agreed-upon writing system. Nevertheless, it is necessary to study Arabic dialects (ADs) since they are the means of communication across all the Arab World. In fact, in Arab countries, MSA is considered to be the High Variety (HV) as it is the official language of the news broadcasts, official documents, etc. The HV is the standardized written variety that is taught in schools and institutions. On the other hand, ADs are considered to be the Low Variety. They are used for daily non-official communication, they have no con- ventional writing system, and they are considered as being chaotic at the linguistic level including syntax, phonetics and phonology, and morphology. ADs are commonly known as spoken or colloquial Arabic, acquired naturally as the mother tongue of all Arabs. They are, nowadays, emerging as the language of informal communication on the web, including emails, blogs, forums, chat rooms and social media. This new situation amplifies the need for consistent language resources and processing …

Content truncated to 500 words