LESSONS LEARNED AFTER DEVELOPMENT AND USE OF A DATA COLLECTION APP FOR LANGUAGE DOCUMENTATION (LIG-AIKUMA)

Laurent Besacier (1, 2, 3, 4) , Elodie Gauthier (2) , Sylvie Voisin (5)

1 GETALP - Groupe d’Étude en Traduction Automatique/Traitement Automatisé des Langues et de la Parole
2 LIG - Laboratoire d'Informatique de Grenoble
3 UGA [2016-2019] - Université Grenoble Alpes [2016-2019]
4 IUF - Institut universitaire de France
5 DDL - Dynamique Du Langage

Lig-Aikuma is a free Android app running on various mobile phones and tablets. It proposes a range of different speech collection modes (recording, respeaking, translation and elicitation) and offers the possibility to share recordings between users. More than 250 hours of speech in 6 different languages from sub-Saharan Africa (including 3 oral languages in the process of being documented) have already been collected with Lig-Aikuma. This paper presents the lessons learned after 3 years of development and use of Lig-Aikuma. While significant data collections were conducted, this has not been done without difficulties. Some mixed results lead us to stress the importance of design choices, data sharing architecture and user manual. We also discuss other potential uses of the app, discovered during its deployment: data collection for language revitalisation, data collection for speech technology development (ASR) and enrichment of existing corpora through the addition of spoken comments.

Mots clés: Mobile applications; Data collection; Language documentation; Speech recording; Eliciting speech

Domaines: Informatique et langage [cs.CL]; Linguistique

Laurent Besacier, Elodie Gauthier, Sylvie Voisin. LESSONS LEARNED AFTER DEVELOPMENT AND USE OF A DATA COLLECTION APP FOR LANGUAGE DOCUMENTATION (LIG-AIKUMA). International Congress of Phonetic Sciences ICPhS 2019, Aug 2019, Melbourne, Australia. ⟨hal-02264418⟩