Linguistic and social conditioning factors
The social and linguistic factors that have been shown to shape variation in Spanish /s/ production are numerous and well-documented. Here we provide a list of the most frequently investigated factors and a brief description of the nature of their covariation with /s/ reduction. These descriptions are not meant to be the final word on any of these factors, and we direct the reader interested in a more thorough and nuanced discussion to one of the many summaries of the literature along these lines (Mason 1994; Lynch 2009; Erker 2017).
- • Syllable position. Most studies find that /s/ is more likely to be reduced syllable-finally compared to syllable-initially, which is consistent with crosslinguistic research that finds syllable coda to be an intrinsically reduction-promoting position (Ohala & Kawasaki 1984; Recasens 2004). There are, however, also well-established patterns of Spanish Isl reduction in syllabic onset position (Brown & Torres Cacoullos 2002; Brown 2005; Brogan & Bolyanatz 2018). Note also that while most studies consider syllable position to be defined lexically, others assume that in continuous speech a lexical coda Is/ can be resyllabified in word-final prevocalic position.
- • Following sound. Studies vary in terms of how they operationalize descriptions of following phonetic context. The widest generalization is that following consonants favor Is/ reduction compared to following vowels and pauses. Among following consonants and vowels, certain classes (i.e. obstruents) as well as individual segments are more reduction- favoring than others.
- • Preceding sound. Similarly to following sounds, preceding phonetic context is operationalized at varying levels of detail across studies, for example the identification of individual segments, sound classes, and, in the case of vowels, acoustic-phonetic measurements. The broadest generalization with respect to this factor is that greater articulatory distance between hi and the preceding sound favors reduction, though this factor is routinely weaker in effect than is the following phonetic context.
- • Speech rate. Once again, studies vary in terms of the details of how this factor is operationalized, but a clear trend emerges: Faster talking promotes reduction.
- • Stress. Reduction of Is/ is more likely to occur in unstressed syllables.
- • Word position. Reduction of hi is more likely in word-final compared to word internal-position, for example hablas “you speak” vs. mismo “same.”
- • Lexical frequency. Reduction is more likely in higher frequency words.
- • Length of carrier word. Whether counted in terms of number of syllables or number of phones, there is evidence that increased word length correlates with reduction.
- • Morphological role. Of all the linguistic factors listed here, this is the most controversial. While some studies find that morphemic hi is less likely to be reduced than non-morphemic Is/ (e.g. the hi of lutblas vs. that of entonces), many other studies have found an opposite pattern or have failed to find any evidence that this factor significantly covaries with reduction.
- • Individual. As with all variable phenomena, individual speakers vary in comparison to other members of their communities. To account for this, most recent research includes Speaker or Individual as a (random) variable in multivariate quantitative analysis.
- • Sex. Men are routinely reported to have higher rates of reduction than women in the same communities.
- • Age. Younger speakers are more likely to weaken hi than older speakers (though see Brogan & Bolyanatz 2018 for the opposite pattern).
- • Regional origin. Higher rates of reduction are typically reported for communities in Caribbean locales, as well as in coastal and lowland areas of Mainland Latin America compared to residents of the interior and highlands of Latin America. Among Spanish-speaking Iberians, rates of hi reduction are typically higher among those who reside in Southern as opposed to Central Northern regions.
- • Socioeconomic status. Whether measured through general social class measures, levels of education, occupation, or area/location of residence, the vast majority of studies that have examined this factor report higher
rates of reduction among speakers of lower socioeconomic status.
• Formalitv/Register. Rates of reduction are higher in less formal contexts.