The Roles of Vowel Harmony and Stress in Predicting Vowel-to-Vowel Coarticulation
Abstract
Similar phonetic and phonological processes often exist in predictable synchronic relationships across languages: when a process is phonologized, its phonetic predecessor is suppressed or limited in scope to resolve conflicting demands on the relevant set of acoustic cues (Cohn, 1990; Francis, Ciocca, Wong, & Chan, 2006). In the case of vowel harmony and vowelto-vowel (VV) coarticulation, the diachronic origins of harmony in VV coarticulation are wellsupported (Ohala, 1994b), but their synchronic relationship is not fully understood. Studies investigating VV coarticulation in harmonizing languages have found disparate patterns across languages. Beddor & Yavuz (1995) found that in Turkish, which has left-to-right vowel harmony, VV coarticulation is predominantly right-to-left, the opposite of the direction of harmony, while Dye (2015) investigated the harmonizing languages of Wolof and Pulaar and found that VV coarticulation was stronger in each language in the direction paralleling the native harmony process. A second factor known to influence magnitude of coarticulation in each direction is the location of stress; in general, stressed vowels coarticulate less, leading to the expectation that languages with word-final stress will prefer anticipatory coarticulation, while those with wordinitial stress will exhibit greater carryover coarticulation. This dissertation investigates coarticulatory directionality in two harmonizing languages with differing stress profiles, Tatar and Hungarian, and one non-harmonizing language with variable stress, Spanish, in order to better understand how stress and vowel harmony impact language-specific directional preferences in VV coarticulation. The data presented here on stress and coarticulation is the first of its kind in languages with backness harmony. In the Spanish study, the strongest coarticulation occurred in unstressed vowels, while stressed vowels inhibited coarticulatory magnitude, confirming the results of previous studies that found reduced coarticulation in stressed vowels (Beddor, Harnsberger, & Lindemann, 2002; Fowler, 1981; Recasens, 2015). Consequently, Tatar is expected to exhibit stronger anticipatory coarticulation due to its word-final stress, and Hungarian is expected to demonstrate stronger carryover coarticulation to accompany its word-initial stress, unless the direction of vowel harmony has an interfering influence on coarticulatory directional preferences. In the Tatar study, the dominant direction of coarticulation was anticipatory, and some evidence was found that Tatar harmony may be undergoing reanalysis, at least with regard to the marginalized lexical subset of orthographically disharmonic items. The finding of primarily anticipatory coarticulation suggests that (1) stress impacts coarticulatory direction as predicted in Tatar and (2) if the direction of vowel harmony has any impact on coarticulatory direction, it is to suppress coarticulation in the direction parallel to harmony. A more likely scenario is that the directions of harmony and coarticulation are synchronically divorced – that upon phonologization, their fates within the language become separated, as suggested by Beddor & Yavuz (1995). With regard to Hungarian, the results were mixed. The two target vowels exhibited stronger effects in opposite directions, and both carryover and anticipatory coarticulation were widely present, though anticipatory coarticulation appeared in a broader range of consonant and vowel conditions. The carryover coarticulation found in Hungarian is ascribed to the expected impact of word-initial stress, while other influences must have supported the operation of anticipatory coarticulation. The results of the three studies are situated within Hyman’s (2013; 1976) life-cycle model of phonologization, which recognizes that phonological processes arise from phonetic origins and are eventually lost, returning to the phonetic realm where they originated. With regard to vowel harmony and VV coarticulation, Hyman’s model allows for the existence of differing synchronic relationships between the two processes across languages. As vowel harmony progresses through the various stages of phonologization, greater variation across languages in the relationship between harmony and VV coarticulation becomes possible. Furthermore, the complexity of the coarticulatory results explored in this dissertation underscores the wide variety of factors influencing VV coarticulation and undermines global cross-linguistic predictions regarding VV coarticulatory direction.
Degree
Ph.D.
Advisors
Niepokuj, Purdue University.
Subject Area
Language|Acoustics|Linguistics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.