Speech Units Workshop

This talk will discuss the methodology needed to discover the phonemic inventory of an underdocumented language. Carrying out phonemic analysis requires an iterative process whereby the researcher identifies both (1) minimal pairs, suggesting certain segmental contrasts; and (2) predictable phonological properties derived from context, suggesting phonological processes that derive allophones. The former requires a large enough sample of lexical items, while the latter requires detailed observations of alternations within and across morphemes, as well as knowledge of phonetically natural phonological processes and phonological typology. I will be drawing on case studies from my own phonological analysis conducted with speakers of Panãra (Jê) and Kawaiwete (Tupí-Guaraní), two underdocumented Amazonian Indigenous languages of Central Brazil. Crucially, these case studies will show how determining the phonological analysis of a language may require years of work, spanning multiple areas of study of the grammar of a language, and may at other sometimes remain inconclusive. I will discuss common pitfalls encountered during the analytical process, and provide tips and trick for identifying minimal pairs.