Coarticulation between Aspirated-s and Voiceless Stops in Spanish: An Interdialectal Comparison Francisco Torreira University of Illinois at Urbana-Champaign 1. Introduction In a large number of Spanish dialects, representing many of the world’s Spanish speakers, /s/ is reduced to or deleted entirely in word-internal preconsonantal position (e.g. /este/ [ehte], este ‘this’), in word-final preconsonantal position (e.g. /las#toman/ [lahtoman], las toman ‘they take them’), and/or in prepausal position (e.g. /komemos/ [komemo(h)], comemos ‘we eat’). In those dialects considered the most phonologically innovative (such as the Spanish of Andalusia, Extremadura, Canary Islands, Hispanic Caribbean, Pacific coast of South America), /s/ debuccalization is also found in prevocalic environments word-finally (e.g. /las#alas/ [lahala], las alas ‘the wings’) and even word-internally, this phenomenon being less common (e.g. /asi/ [ahi], así ‘this way’). When considered in detail, the manifestations of aspirated-s can be very different depending on dialectal, phonetic and even sociolinguistic factors. Only in preconsonantal position, for example, different s-apirating dialects behave differently; and, even within each dialect and in preconsonantal position, different manifestations arise as a result of the consonant type following aspirated-s. In this study, I show that aspirated-s before voiceless stops has different phonetic characteristics in Western Andalusian, on one part, and Puerto Rican and Porteño Spanish on the other. While Western Andalusian exhibits consistent postaspiration and shorter or inexistent preaspiration, Puerto Rican and Porteño display consistent preaspiration but no postaspiration. Moreover, Andalusian Spanish voiceless stops in /hC/ clusters show a longer closure than voiceless stops in other conditions, a contrast that does not apply for Porteño and Puerto Rican Spanish. In the discussion, I will argue that a precise timing of some of the gestures involved in the sound sequences studied here is relevant for the sound structure of each dialect. Finally, I will consider an application of the Articulatory Phonology syllabic model to a hypothetical sound change in Western Andalusian Spanish. 2. Correlates of aspirated-s The most obvious acoustic correlate of debuccalized /s/ is, of course, a period of glottal frication. In Andalusian Spanish, this correlate is typically present in utterance-medial word-final position, such as in the utterance las alas [lahala]. However, when we consider multiple phonetic contexts and dialectal differences, the phonetic manifestations of aspirated-s appear as being quite varied and finely detailed. For this reason, in this introduction we will only refer to the context studied here, that is utterance-medial /s/ aspiration before voiceless stops. In a perception study on Miami Cuban Spanish carried out with natural stimuli, Hammond (1978) found that speakers were unable to detect utterance-final /s/, while they consistently distinguished minimal pairs such as costa ~ cota (the word costa being always pronounced without a full sibilant). It follows from this that, when confronted with las costas ‘the coasts’, speakers relied on cues in the medial part of the utterance in order to distinguish it from la costa, since utterance-final /s/ was categorically lost. After analyzing the acoustic differences among the stimuli used in the experiment, it was found that the vowel preceding aspirated /s/ (e.g. /a/ in gasto) was considerably longer than vowels in open syllables (e.g. gato). This effect was recently replicated for Puerto Rican Spanish in a study sharing the same design and test-words as Hammond (1978) (Figueroa, 2000). Gerfen (2002) also found that for words sharing this pattern Eastern Andalusian speakers also tended to produce © 2006 Francisco Torreira. Selected Proceedings of the 9th Hispanic Linguistics Symposium, ed. Nuria Sagarra and Almeida Jacqueline Toribio, 113-120. Somerville, MA: Cascadilla Proceedings Project.