The Domain of Final Lengthening in the Production of Dutch

Speakers have numerous ways at their disposal to mark the prosodic structure of their speech. The cues signalling this structure are potentially useful to the listener. One way for speakers to signal prosodic structure is by marking the depth of a boundary between two constituents, using a variety of prosodie features. For example, the end of an intonational phrase is typically marked by a pause, a boundary-marking pitch movement and/or lengthening of the preboundary segments (Downing 1970). Which prosodie features are used to mark a certain boundary may depend on the language and on the type of boundary involved, but is also to some degree variable, e.g. across speakers (Sanderman 1996). We will refer to the lengthening of preboundary segments as final lengthening. In the past, final lengthening was thought to indicate syntactic boundaries (Klatt 1975). With the development of prosodie phonology, it turned out that the application of final lengthening is triggered by prosodie boundaries (just like other phonetic processes) rather than by syntactic ones. A final lengthening effect at utterance boundaries has been firmly established by numerous phonetic studies (Crystal and House 1988). Other prosodie boundaries are marked by final lengthening as well (Beckman and Edwards 1990). Furthermore, the amount of final lengthening has been shown to be related to the depth of the following boundary, both as produced by the speaker and as required by the listener (Nooteboom and Doodeman 1980, Gussenhoven and Rietveld 1992). Thus, a deeper prosodie boundary is, and should be, marked by more final lengthening. However, very little attention has been paid to the question which segments are affected; both production and perception experiments have generally been concerned with lengthening within a fixed domain (such as the vowel in the final syllable, or the final rhyme). So, previous research has shown that at least in the word-final rhyme the amount of lengthening is related to boundary depth, but we do not know whether boundary depth has any effect on the size of the domain that is lengthened. This leads us to the present research question: