A speech corpus of Kom verbal arts and its applications

The move towards large-scale analysis of corpora in phonetics and phonology
 requires new approaches to data collection, both for African languages specifically
and for less-described languages more generally. We argue that short-form ver-
bal arts, such as proverbs and riddles, offer a convenient and culturally sensitive
way to build out suitable language resources in the African context, particularly
for less-documented languages. As a proof of concept that use of the verbal arts
to construct corpora is an effective strategy, we describe the construction of a cor-
pus of Kom speech (Grassfields Bantu, Cameroon) from more than 150 proverbs
(ngàynsɨ) and riddles (mɨtɨtɨ̂ ). We illustrate the potential uses of the corpus with
two case studies: vowel production under positional prominence and a variably ap-
plied phonological process of monophthongization which is sensitive to upcoming
segmental content. Throughout, we consider the affordances of proverbs and other
short verbal arts for corpus development, how these relate to the typical materials
for read-speech corpora, and the potential role of these corpora in the development
of language technology.