Proceedings Fonetik 2009 - Institutionen för lingvistik

Proceedings, FONETIK 2009, Dept. of Linguistics, Stockholm Universitystst1086420-2-41086420-2-4medial rise (n=20)normalised timenormalised timemedial fall (n=17)initial rise (n=38) medial fall (n=17) final fall (n=16)Figure 1. Mean F0 contours of the three contentwords in the test sentence “Wallander förlänger tillnovember”; breaks in the curves symbolise wordboundaries; time is normalised (10 measurementsper word); semitones refer to an approximation ofindividual speakers’ base F0; adapted from Ambrazaitis(in press). Upper panel: two strategies offocus signalling on the medial word in a confirmation.Lower panel: Focus on the initial, medial, andfinal word in a confirmation; for medial and finalfocus, only the falling strategy is shown.That is, in confirmations with intended focus onthe medial or the final word, one strategy was toproduce a (non-focal) rise on the initial, and twofalling movements, one each on the medial andthe final word. As the lower panel in Figure 1shows, the mean curves of these two cases lookvery similar; moreover, they look similar to thepattern for initial focus, which was alwaysproduced with a rising focal accent. One possiblereason for this similarity could be that medialor final focus, in fact, were not marked at allin these confirmations, i.e. that the entire utterancewould be perceived as lacking any narrowfocus. Another possibility is that all patternsdisplayed in Figure 1 (lower panel) would beperceived with a focal accent on the initial word.Informal listening, however, indicates that inmany cases, an utterance-level prominence,indicating focus, can be perceived on the medialor the final word. Thus, future perception experimentsshould test whether focus can besignalled by the falling pattern found in confirmations,and furthermore, which acousticcorrelates of this fall serve as perceptual cues offocus in confirmations. Prior to that, the acousticcharacteristics of the falling pattern need to beestablished in more detail.It is known for a variety of languages thatprosodically focussed words in assertions arenot only marked tonally, i.e. by a pitch accent,but also temporally, i.e. by lengthening (e.g.Bruce, 1981, Heldner and Strangert, 2001, forSwedish; Cambier-Langeveld and Turk, 1999,for English and Dutch; Kügler, 2008, for German).Moreover, Bruce (1981) suggests thatincreased duration is not merely an adaptation tothe more complex tonal pattern, but rather afocus cue on its own, besides the tonal rise.The goal of this study is to examine the datafrom Ambrazaitis (in press) on focus realisationin assertions and confirmations in more detail asregards durational patterns. The results are expectedto provide information as to whetherduration should be considered as a possible cueto focus and to speech act in future perceptionexperiments. The hypothesis is that, if focus issignalled in confirmations, and if lengthening isa focus cue independent of the tonal pattern,then focal lengthening should be found, not onlyin assertions, but also in confirmations. Furthermore,it could still be the case that durationalpatterns differ in confirmations and assertions.MethodThe following two sections on the material andthe recording procedure are, slightly modified,reproduced from Ambrazaitis (in press).MaterialThe test sentence used in this study was “Wallanderförlänger till november” (‘Wallander iscontinuing until November’). In the case of aconfirmation, the test sentence was preceded by“ja” (‘yes’). Dialogue contexts were constructedin order to elicit the test sentence with focus onthe first, second, or third content word, in eachcase both as an assertion and as a confirmation.These dialogue contexts consisted of a situationalframe context, which was the same forall conditions (‘You are a police officer meetinga former colleague. You are talking about retirementand the possibility to continue working.’),plus six different context questions, one73