Crosslinguistic Phonetic Analysis
The phonetic realization of a given sound segment like /s/ or /a/ can differ considerably across speakers of a given language, as well as across languages. This variation, however, is not entirely unpredictable, but rather fairly structured: in other words, the pronunciation of /s/ is not independent of that of /z/, and this type of structure is not fully reducible to speaker anatomy. Only in the past few years has the crosslinguistic investigation of this variation and structure become feasible with dramatic increases in computational power, storage, and the availability of “massively multilingual speech corpora.” I present a novel methodological approach to processing such crosslinguistic phonetic resources and making them usable for the investigation of phonetic variation — its range and its limits – across the world’s languages.