Are neutral roots in Uyghur really neutral? Evaluating a covert phonemic contrast

Abstract

This paper looks at the case of so-called neutral roots in Uyghur (Turkic: China), whose idiosyncratic behavior with respect to the backness harmony system has been analyzed as stemming from a covert vowel contrast. Based on considerations of the structural properties of the language and the results of an experimental study, we suggest that an analysis based on lexical exceptionality is more parsimonious than the traditional analysis, unifying the treatment of neutral roots with other cases of exceptionality in the harmony system and accounting for a relationship between the patterning of roots and their frequency. We close by discussing implications for covert contrast analyses in general.

Issues in Uyghur phonology

“…The treatment of neutral roots in Uyghur also bears on the nature of lexical representations: under the analyses proposed by Lindblad (1990) and Hahn (1991b), speakers represent these harmonically ambiguous roots with abstract phonemes that are never realised on the surface. Mayer (2021) and Mayer et al (2022) suggest that speakers' lexical representations correspond more closely to the surface realisations of these roots, with cases of idiosyncratic harmonising behaviour represented as lexical exceptions. These two hypotheses bear directly on the division of labour between phonology and morphology.…”

“…More research is needed to address this issue. Mayer et al (2022) proposes an alternative account of neutral roots based on lexical exceptionality. Given a covert phonemic contrast, one surprising property of Uyghur is the total absence of homophonic neutral roots that differ in the backness of suffixes they take (i.e., underlying minimal pairs between /i/-/ɯ/ and /e/-/ɤ/), even though such pairs are plentiful for other vowel pairs.…”

“…The closest thing is a small set of noun-verb pairs that contain the same roots, but differ in their harmonising behaviour (e.g. [iz-lɑr] 'trace-PL' vs. [izli-maek] 'search-INF', derived from /iz/ plus the verbalising suffix /-lA/ and the infinitival suffix /-mAK/; see Mayer et al, 2022 for more details). In addition, many cases of phonological neutralisation have resulted in subtle phonetic distinctions that reflect the original contrast, but are typically not perceived by speakers (e.g., Labov, 1994;Yu, 2007).…”

“…In addition, many cases of phonological neutralisation have resulted in subtle phonetic distinctions that reflect the original contrast, but are typically not perceived by speakers (e.g., Labov, 1994;Yu, 2007). Acoustic analyses of neutral roots in Uyghur, however, find no clear correlation between vowel F2 in neutral roots and whether such roots take front or back suffixes (e.g., Mayer et al, 2022;McCollum, 2021). Mayer et al propose that the majority of neutral roots take the default [+back] suffixes, and the small number of roots that take [−back] suffixes are represented as lexical exceptions.…”

“…Connor Mayer https://orcid.org/0000-0003-2710-3475 ENDNOTES 1 Abdulla et al (2010) suggest that /χ/ also serves as a back trigger, but evidence for this is equivocal, as it is not an undergoer of harmony in the same way as /q/ or /ʁ/. The velar nasal /ŋ/ does not appear to participate in the harmony process, though it does have an allophonic backing effect on adjacent vowels (Mayer et al, 2022). 2 The dative suffix may also surface as [-qae] when attached to a root with a front vowel that ends in a voiceless uvular, as in [χaelq-qae] 'people-DAT' (cf.…”