Duration of Spanish [s] in three contexts: Citation forms versus fluent reading

Abstract

Consonants in Spanish rhymes are typically resyllabified to the onset position of the following syllable in all but the most careful speech styles. There is, however, a lack of consensus regarding the temporal characteristics in the production of two identical phones at word boundaries (e.g., las salas, ‘‘the rooms’’). Malmberg (1971) states that the two segments are reduced to a single sequence, with a slightly longer duration than one phone. According to Navarro Tomás (1968), there is reduction from two segments to one, with no increase in duration. In the present study, adult native Mexican Spanish speakers produced three sets of phrases containing the sequences [s]+vowel, vowel+[s], and [s]+[s] in both citation forms and fluent reading. Preliminary findings support Malmberg’s position regarding the citation form productions and Navarro Tomás’s position in the fluent reading productions. This suggests that Spanish-language speech synthesis systems, in the absence of allophonic variation for a given sequence, can vary the duration of rhyme plus onset sequences at word boundaries to achieve a citation vs. fluent reading speech style.