A large-scale investigation of vowel co-occurrence patterns in the world’s lexicons

This paper explores whether there are universal trends for vowels that co-occur to share featural properties. The existence of various productive featural vowel harmony systems across the world’s languages suggests that the factors underlying harmony may be universal. An empirical prediction that follows from this proposal is that vowel co-occurrence in the world’s lexicons should be featurally organized. In corpus analyses of two cross-linguistic, phonologically transcribed lexicons —92 lexicons in the XPF corpus and 107 lexicons in the NorthEuraLex corpus— we find a cross-linguistic tendency for languages to over-represent pairs of identical vowels but no universal preference for height or backness harmony. We do, however, find some evidence that identity affects some vowels systematically more than others across languages, which indicates that vowel co-occurrence is sensitive to the phonetic properties of vowel categories in cross-linguistically generalizable ways. Ultimately, the lack of featural harmony and the over-representation of identity is consistent with the notion that the phonological organization of lexicons is subject to factors beyond local assimilation or phonetic and phonological well-formedness.