Gradient morphophonology: Evidence from Uyghur vowel harmony

Abstract

For the Structuralists and early Generativists (e.g. Bloomfield 1933; Chomsky & Halle 1968), all grammatical knowledge was by definition discrete and categorical. Since phonetic patterns are gradient, early work argued that phonetics was extra-grammatical. However, a significant body of work has since shown that phonetic patterns are language-specific and must constitute part of a speaker’s knowledge about their language (e.g. Keating 1985). As a result, linguistic knowledge is not ontologically categorical. For other areas of the grammar, though, much work continues to assume that linguistic knowledge is categorical. In this paper, I investigate the categoricality of phonological patterns using acoustic vowel harmony data from Uyghur. By comparing subphonemic variation in Uyghur with attested patterns of phonetic reduction and interpolation, I demonstrate that gradience is not always derivable from phonetic forces. On these grounds I argue that vowel harmony, and by extension phonology, may be gradient. Furthermore, the claim that gradience plays a larger role in linguistic representations is also supported by a number of descriptive works, which suggest that gradient harmony may occur in a wide range of languages. Building on experimental, and typological evidence, I thus contend that gradience isn’t restricted to phonetics, but pervades both the phonological and phonetic modules of the grammar.

Issues in Uyghur phonology

“…Both vowel reduction and umlaut have been less carefully studied from a theoretical perspective than vowel harmony. Preliminary analyses of vowel reduction are presented in McCollum (2019) and Mayer (2021, appendix E). To our knowledge, there has been no phonological analysis to date of umlaut in Uyghur.…”

Vowel harmony and positional variation in Kyrgyz

“…In Kyrgyz, Washington (2006 finds that affix vowels are less distinct than root-internal vowels. Elsewhere in the family, McCollum (2015McCollum ( , 2019bMcCollum ( , 2019c argues that positional reduction is asymmetric in Kazakh and Uyghur. McCollum reports that back vowels undergo non-initial fronting without any comparable effects among the front vowels.…”

Phonetically Identical Forms can Have Different Phonological Behaviors

“…Under such a theory, the assumption of categorical phonological representation in traditional formal phonology is completely dropped, and fine-grained gradient information is allowed inside phonology. Although a consensus has not been reached by previous studies (Lionnet, 2017;Pierrehumbert et al, 2000;Silverman, 2006;Tucker & Warner, 2010; inter alia) on how to incorporate gradience inside formal phonology, as McCollum (2019) pointed out, at least some form of continuously valued variables should be allowed. To apply this theory in German final devoicing, phonology should not only direct a underlyingly voiced segment to devoice, but also demand to what the degree the devoicing process should occur to distinguish the derived voiceless segment from its underlying counterpart.…”

“…One solution that is gaining increasing attention to the puzzle posted by incomplete neutralization is to introduce gradience into phonology (McCollum, 2019). Under such a theory, the assumption of categorical phonological representation in traditional formal phonology is completely dropped, and fine-grained gradient information is allowed inside phonology.…”