Durational Relations between Connected Text, List Sentences, and Carrier Sentences

Abstract

For experimental convenience, it is customary in speech studies to use lists of independent short sentences, or of words substituted into a repeated carrier phrase. Little is known, however, about the degree to which such .material is or is not representative of continuous text. This paper reports a comparison of temporal effects in three forms of material. Two speakers read first a list of 86 test sentences, disordered and mixed with irrelevant sentences; then a list of 252 words taken from the test sentences and embedded in the carrier phrase “say — instead”; and finally, the test sentences in their original form, a coherent narrative. Measurements of the data show that the durations of whole sentences and of individual content words average 7% longer in listed sentences than in connected text. Words in the carrier sentence range from the same duration to twice as long as those in continuous text, with an average increase of 45% or 140 msec. The report presents details of these differences at the phoneme and word levels, and attempts to interpret them in terms of pause, intonation, and stress changes and at higher levels of structure.