How do you pronounce this in a local language? A digital tool for learning and disseminating the pronunciation of local languages

International audience Encountering written words and names in one of the almost 30 Kanak languages of New Caledonia is very common whether on road signs, in newspapers or in other media. It is challenge to know how to pronounce these words, especially since the grapheme-phoneme (letter-to-sound) correspondences differ – often drastically – from language to language.To help address this challenge, we are developing a web application called LaPasserelle.nc (“the gateway”), which is conceived and designed as a tool to help people in properly pronouncing words en langue (in local language) [Welby et al. 2023].For this talk, we will describe and give a demo of the web application. We will also offer a perspective on the use of semantic web (also called web of data) technologies and concepts, which are useful in terms of interoperability and knowledge dissemination [Tomaszuk and Hyland-Wood 2020].The main aim of this phonetizer is to provide an idea of how to pronounce a word written in Oceanic languages, especially Kanak languages. The user enter text in the input box. This input is processed by using a grapheme – phoneme correspondences, and a pronunciation is displayed as output in either: 1/ the International Phonetic Alphabet (IPA) or, since IPA is not widely understood, 2/ franétique (> français + phonétique), a pronunciation respelling based on French (the language shared by almost all Caledonians), enhanced with multimodal pedagogical tips. [Kartushina and Martin 2019].The user is “taken by the hand” thanks to a clear association between input and output. Each grapheme is visually connected to its pronunciation. When a phoneme is not used in the output language, a multimodal pedagogical tip explains how to use mouth and tong to properly pronounce what is written.Work is in progress to reference the grapheme-phoneme associations into a database designed with these modeled associations, which can be queried using standardized languages dedicated to the semantic web. An application programming interface (API) is under construction and can be interrogated by the web application as well as by other machines wishing to feed on or use the modeled knowledge for other purposes.The phonetization is based on the proposed writing systems of the Académie des Langues Kanak [ALK 2013, Sam 2009], as well as current research. A beta version of the phonetizer handles the Drehu and Paicî languages, and we plan to extend it to other languages spoken in New Caledonia, if the different communities so desire, including other Kanak languages, but also possibly Polynesian languages such as Wallisian, and Tahitian. The web of data database will be an asset for the dissemination and re-use of the knowledge acquired as a part of this project.For a neighboring Pacific country, Papua New Guinea, Kik et al. (2021) report that “ethnobiological knowledge is closely correlated with indigenous language skills” and that both are “declin[ing] precipitously” [Kik et al. 2021]. With that in mind, and fully aware that correlation does not imply causation, a broader goal of the LaPasserelle.nc project is to support and promote linguistic diversity in New Caledonia, by supporting language learning and the preservation and transmission of cultural knowledge more generally.Key words (max 5): phonetization, under-resourced languages, Kanak languages, New Caledonia, Oceania ReferencesAcadémie des Langues Kanak, “Propositions d’écriture du paicî”, 2013. Nouméa: ALK.Kartushina N., Clara. "Talker and acoustic variability in learning to produce nonnative sounds: Evidence from articulatory training", Language Learning, 2019, 69, 71–105.Kik A., Adamec M., Aikhenvald A. Y., Bajzekova J., Baro N., Bowern C., Colwell R. K., Drozd P., Duda P., Ibalim S., Jorge L. R., Mogina J., Ruli B., Sam K., Sarvasy H., Saulei S., Weiblen G. D., Zrzavy J., Novotny V. 2021. Language and ethnobiological skills decline precipitously in Papua New Guinea, the world's most linguistically diverse nation. Proceedings of the National Academy of Sciences (USA). doi: 10.1073/pnas.2100096118. PMID: 34039709; PMCID: PMC8179190.Sam L. D. “Aqane cinyihanyi qene drehu; Propositions d’écriture de la langue drehu”, 2009. Nouméa: Académie des Langues Kanak.Tomaszuk D., Hyland-Wood D. "RDF 1.1: Knowledge Representation and Data Integration Language for the Web", Symmetry, 2020, 12, 84; doi:10.3390/sym12010084Welby P., Bigi B., Corral A., Wacalie F., Wattelez G. "A visit to the Cliffs of Jokin: A role for phonetizers in second language pronunciation and word learning, with an example from the under-resourced languages of New Caledonia", SIGUL (Special Interest Group on Under-resourced Languages), "2nd annual meeting of the Special Interest on Under-resourced Languages" (SIGUL session 2 : Resources, Pronunciation & Dialect), 2023.