Prosodic boundaries are marked in speech by modifications to dimensions such as F0, duration, and segmental quality. The experiment reported here tests the hypothesis that modifications at the end of a prosodic domain may be amplified or attenuated depending on the type of sentence (statement or question). The prosodic modifications investigated here for French are sentence-final lengthening and vowel devoicing, which is related to changes in voice quality.
Six native speakers of French read 10 matched sets of sentences, which included both statements and questions. Measurements were made of the last vowel in each test sentence and of sentence-medial vowels in control sentences. Statement-final vowels were less periodic and more often devoiced than vowels at the end of questions or sentence-medial vowels, but question-final vowels were lengthened more than statement-final vowels. These results occurred in questions both with and without a final pitch rise, suggesting that sentence type is more important than final intonation contour in determining how prosodic finality is marked. The difference between statements and questions thus appears to be in the relative importance of lengthening and voice quality modification in marking finality, not in the overall extent to which it is marked.
There is by now a large body of data supporting the view that the production of a spoken utterance reflects its prosodic structure in many ways. Both laryngeal and supralaryngeal articulations are subject to prosodically-governed patterns, which are organized at different levels of prosodic structure, all the way from the syllable and word to the utterance and up to complete discourse units. Prosodic modifications to articulation of course have consequences for the resulting acoustic signal, and these effects have often been identified from their acoustic consequences, rather than directly in the articulations. Although perhaps the most obvious, prosodic structure and phonetic context are by no means the only influences on production. A number of studies have identified other aspects of grammatical structure that can alter the production of speech sounds. Some of these "grammatical" factors include the topic structure of a discourse (e.g., Swerts & Geluykens, 1994), whether a word has previously occurred in a discourse or not (e.g., Fowler & Housum, 1987), and the lexical category of a word (Sereno & Jongman, 1995; Yaeger-Dror, 1992). Such evidence implies that many different components of linguistic structure contribute to determining the exact manner in which a sound is produced. This has led many linguists to posit a separate prosodic component of the grammar, which mediates these various effects. One such model, proposed by Shattuck-Hufnagel and Turk (1996, p. 237), has a prosody component as the interface between components such as syntax, pragmatics, speaking rate, and the phonetics and segmental phonology. The present paper explores the effect of a "grammatical" factor, sentence type, on some segmental properties of speech sounds. Greater understanding of the role of this nonprosodic factor in speech production should clarify the organization of the various components of the grammar which are involved...