Minimal wordlist size for a phonological profile: New evidence from Kra-Dai languages

Abstract

This study replicates, and extends to Kra-Dai languages, earlier work on minimal wordlist size needed to make a phonological profile of a language. Previous  work on Australian languages recommended approximately 400 randomly sampled 
words to comprise a minimally complete profile in terms of reliably capturing every phoneme, and with accurate distribution. We survey 55 Kra-Dai languages to show that a longer minimal list is necessary, which we attribute to typological differences 
like larger consonant/vowel inventories. Given the widespread use of short wordlists in fieldwork, these results hold significance for designing language documentation surveys as well as projects that use legacy wordlist data.