Speech technology as documentation for endangered language preservation: The case of Irish.
Ailbhe Nı́ Chasaide, Neasa Ní Chiaráin, Harald Berthelsen, Christoph Wendler, Andrew Murphy · ICPhS · 2015
Developing speech technology such as text-tospeech (TTS), requiring as it does a raft of phonetic and linguistic resources, can provide a powerful way to document endangered languages. Drawing on the experience of the ABAIR initiative, developing such resources for Irish [1], we illustrate how both the technology and the underpinning resources can be exploited in a variety of ways that can contribute to the preservation and revitalisation of these languages. By enabling new avenues of application, they can further help address the particular challenges that face the language users and learners. To maximise the immediate and downstream impact, resource development should ideally involve linguistically transparent, rule-based approaches, rather than the machine learning approaches typical of the commercially driven TTS systems for major world languages.