Production and perception of vowel harmony: Phonological predictors of ratings and on-line adaptations of Russian vowels in Yakut (Sakha)

Abstract

This dissertation investigates phonological input predictors of the Russian vowels /a, e, i, o, u/ in on-line adaptations among modern Russian-Yakut bilingual speakers in the Republic of Sakha (Yakutia) in the Russian Federation, with a particular focus on the production and perception of vowel harmony. Vowel harmony in Yakut involves backness harmony that requires complete agreement of vowels within words in backness, and rounding harmony, where rounded vowels spread their roundedness feature from left to right, with restrictions regarding the height of the vowels involved. Unlike Yakut, Russian does not have vowel harmony. Previous research indicates that earlier loanwords, adapted when most Yakuts were monolingual, underwent complete vowel harmony and were fully nativized. Thus, considering the context of increasing bilingualism among Yakuts and distinct vowel systems between the source (Russian) and borrowing (Yakut) languages, I examine the most-frequent adaptations of the five Russian vowels and also investigate phonological characteristics of input vowels and words that lead to vowel harmony application and Yakut-likeness in bilingual speakers’ on-line adaptations of input disyllabic borrowed, nonce, and un-borrowed words that contain different combinations and stress variations of the five Russian vowels. Thirty-seven Russian-Yakut bilinguals spontaneously adapted the input words in a frame sentence on-line after reading each word silently in the production task. In addition, the bilingual speakers, as well three Yakut monolinguals, rated Yakut nonce words on Yakut-likeness after hearing each word in recorded audio files. I hypothesized that the input characteristics of vowels and words that include backness and roundness, height, the quality of an individual vowel input, stress, vowel sharedness between the languages (shared and unshared vowels), vowel reduction and the syllables where these features occur, are significant in driving adaptations of the input vowels. I tested this primary hypothesis using linear mixed-effects model analyses to reveal the input predictors of vowels and words that predict vowel harmony and Yakut-like adaptations. The main findings suggest that vowel harmony is robust in adaptations, especially backness harmony, which appeared in 81.06% of produced words. The phonological input predictors backness, roundness and height are shown to be significant in driving vowel harmony and Yakut-like adaptation: Russian back vowels (which are also rounded) and non-high vowels were better overall triggers of vowel harmony than front and high vowels, respectively. Moreover, harmony was better achieved in harmonic input words sharing the same features than in disharmonic input. Consistent with the native phonology’s rightward directionality of vowel harmony spreading, a first syllable input vowel’s features were important in predicting more uniform vowel adaptations and a general rightward propagation of vowel harmony. The speakers were also generally efficient in judging grammatical and non-grammatical nonce words in perception, however, rounding harmony violations were more tolerated than violations of backness harmony. The most-frequent adaptations of the five input vowels /a, e, i, o, u/ in their unstressed forms as /ɑ, ɛ, i, ɒ, u/ show categorical approximation in adaptation, and the stressed vowels, adapted as /ɑː, iɛ, iː, uɒ, uː/, suggest consistent reflection of stressed vowels’ phonetic characteristics. I conclude that phonological characteristics of input vowels and words are essential in predicting vowel harmony and Yakut-likeness in on-line adaptations produced by Russian-Yakut bilinguals. An Optimality Theory (OT) analysis using alignment and faithfulness constraints is shown to account well for the harmonic outputs, but is not necessarily reflective of the less frequent disharmonic adaptations that occurred systematically, contingent on the characteristics of input vowels and words.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 81 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.