Social and Linguistic Factors Conditioning the Glottal Stop in Nicaraguan Spanish

Whitney Chappell, Ph.D. (currently at University of Texas at San Antonio)

This dissertation investigates the word-final, intervocalic glottal stop [ʔ] in Nicaraguan Spanish, a highly understudied variant in an equally understudied dialect of Spanish. The glottal stop occurs most frequently in western Nicaragua as an allophone of word-final, intervocalic /s/ (the V/s/#V environment), e.g. [ma.ʔal.to] for más alto ‘higher’, but the glottal stop also surfaces between vowels at the word boundary where there is no underlying segment (the V#V environment), e.g. [la.ʔo.la] for la ola ‘the wave’. The studies to date on the glottal stop in dialects of Spanish, i.e. Argentinian/Paraguayan (Sanicky 1989, Thon 1989), Yucatan (Lope Blanch 1987), Philipino (Lipski 2000), and Puerto Rican (Valentín- Márquez 2006), propose a contact hypothesis: when Spanish is in close contact with another language and the glottal stop is used in that language, bilingual speakers transfer the sound into their Spanish. However, the contact hypothesis falls short in Managua, Nicaragua, as no other languages are in contact with Spanish nor have other languages been in contact with Spanish for some time. Instead, I propose that the glottal stop is a language internal innovation in western Nicaragua, motivated by phonological, phonetic, and social factors.

After collecting data in an informal sociolinguistic interview, a guided image identification task, and a more formal reading task from 36 Managuan participants, I conducted an acoustic analysis of 3,701 variants in the V/s/#V environment and 3,431 variants in the V#V environment and fit binomial logistic regression models with random effects to the data. The results indicate that both linguistic and social factors condition the glottal stop. In the V/s/#V environment, the glottal stop is significantly more likely to occur before a following stressed vowel, between high frequency word pairings (e.g. determiner-noun strings), following words with a high number of syllables, and in the more formal reading task. The effect of social factors on the glottal stop’s likelihood is task-specific: the youngest speakers are more likely to use the glottal stop in the informal sociolinguistic interview than the oldest speakers, but the oldest speakers are actually more likely to use the glottal stop in the formal reading task than the youngest speakers. The most educated speakers are less likely to use the glottal stop in the more formal tasks than the lower education groups, and these less educated speakers are also most likely to insert glottal constriction in the V#V environment where there is no underlying segment in the formal reading task.

Linguistically, the glottal stop in the V/s/#V environment is used to 1) saliently mark the underlying segment 2) strengthen strong prosodic positions, i.e. stressed, word-initial vowels, and 3) maximally demarcate between adjacent vowels to resolve postlexical hiatus in environments where deletion rates are the highest, i.e. following long words. The less educated speakers’ extension of the glottal stop into the V#V environment is not surprising phonetically: even though there is no underlying segment, the glottal stop in the V#V environment continues to strengthen strong prosodic positions and maximally demarcate between adjacent vowels. Socially, the glottal stop is the most ‘neutral’ variant, or the variant least imbued with social meaning, which allows the glottal stop to serve the aforementioned linguistic purposes without evoking strong social correlates like the other variants in the V/s/#V environment.

Finally, I propose that the phonological system has been reanalyzed for speakers of Nicaraguan Spanish: instead of underlying coda /s/, Managuans operate with underlying coda /h/. This argument is supported by high rates of glottal frication in the data and the likelihood of the glottal stop in strong environments. This distribution suggests that the glottal stop is a strengthened variant of /h/, providing greater and more salient gestural closure while maintaining the glottal place of articulation with no oral gesture. Further substantiating this argument is the fact that some speakers never use or incorrectly use coda sibilance, suggesting that sibilance simply does not correspond with these speakers’ underlying representation. I conclude that coda sibilance is not a local variant for Nicaraguan speakers, but speakers with more recent or greater exposure to prescriptive Spanish norms, i.e. younger and more educated speakers, learn to approximate coda sibilance in formal speech as a global formality strategy. However, speakers with less exposure to prescriptive Spanish norms, i.e. older and less educated speakers, instead use the glottal stop in formal speech, hyperarticulating the /h/ with a local formality strategy.”