Variations in sentence duration due to lexical content, semantic emphasis, and text structure
Arvin Levine · The Journal of the Acoustical Society of America · 1977
Research on the relative durations of words within sentences leads to an algorithm that assigns fractions of the lexical duration, dependent, in a neutral context, on the grammatical function of each word. This algorithm accords, in general, with work by Klatt [J. Acoust. Soc. Am. 59, 1208–1221 (1976)]. One important difference is that we utilize duration information from the recorded isolation form. The durational effect of semantic emphasis can be accounted for in terms of word duration adjustments and additional silences surrounding specified words. Another extension of this algorithm predicts some of the deviation of an entire sentence (or utterance from the neutral context duration which we predict for it, by using elementary text structures (such as lists) that can be generally applied to paragraphs and other text units. Comparisons of predicted durations with observed durations, in utterances which were independently recorded for use in the Institute's computerized “Introduction to Logic” course [W.R. Sanders and R. Laddaga, J. Acoust. Soc. Am. 60 S76–S77(A) (1976)], are examined. [Work supported by NSF grant SED77-09698.]