Proceedings Fonetik 2009 - Institutionen för lingvistik

Proceedings, FONETIK 2009, Dept. of Linguistics, Stockholm UniversityFocal lengthening in assertions and confirmationsGilbert AmbrazaitisLinguistics and Phonetics, Centre for Languages and Literature, Lund UniversityAbstractThis paper reports on duration measurements ina corpus of 270 utterances by 9 StandardSwedish speakers, where focus position is variedsystematically in two different speech acts: assertionsand confirmations. The goal is to provideinformation needed for the construction ofa perception experiment, which will test thehypothesis that Swedish has a paradigmaticcontrast between a rising and a falling utterance-levelaccent, which are both capable ofsignalling focus, the falling one being expectedin confirmations. The results of the present studyare in line with this hypothesis, since they showthat focal lengthening occurs in both assertionsand confirmations, even if the target word isproduced with a falling pattern.IntroductionThis paper is concerned with temporal aspects offocus signalling in different types of speech acts– assertions and confirmations – in StandardSwedish. According to Büring (2007), mostdefinitions of focus have been based on either oftwo ‘intuitions’: first, ‘new material is focussed,given material is not’, second, ‘the material inthe answer that corresponds to the wh-constituentin the (constituent) question is focussed’(henceforth, ‘Question-Answer’ definition). Inmany cases, first of all in studies treating focusin assertions, there is no contradiction betweenthe two definitions; examples for usages of focusthat are compatible with both definitions areBruce (1977), Heldner and Strangert (2001), orLadd (2008), where focus is defined, more orless explicitly, with reference to ‘new information’,while a question-answer paradigm is usedto elicit or diagnose focus. In this study, focus isbasically understood in the same sense as in, e.g.Ladd (2008). However, reference to the notionof ‘newness’ in defining focus is avoided, sinceit might seem inappropriate to speak of ‘newinformation’ in confirmations. Instead, the‘Question-Answer’ definition is adopted, however,in a generalised form not restricted towh-questions. Focus signalling or focussing isthen understood as a ‘highlighting’ of the constituentin focus. Focus can refer to constituentsof different size (e.g. individual words or entirephrases), and signalled by different, e.g. morphosyntactic,means, but only narrow focus (i.e.focus on individual words) as signalled byprosodic means is of interest for this paper.For Swedish, Bruce (1977) demonstratedthat focus is signalled by a focal accent – a tonalrise that follows the word accent gesture. In theLund model of Swedish intonation (e.g. Bruce etal., 2000) it is assumed that focal accent may bepresent or absent in a word, but there is noparadigmatic contrast of different focal accents.However, the Lund model is primarily based onthe investigation of a certain type of speech act,namely assertions (Bruce, 1977). This paper ispart of an attempt to systematically includefurther speech acts in the investigation ofSwedish intonation.In Ambrazaitis (2007), it was shown thatconfirmations may be produced without a risingfocal accent (H-). It was argued, however, thatthe fall found in confirmations not merely reflectsa ‘non-focal’ accent, but rather an utterance-levelprominence, which paradigmaticallycontrasts with a H-. Therefore, in Ambrazaitis(in press), it is explored if and how focus can besignalled prosodically in confirmations. To thisend, the test sentence “Wallander förlänger tillnovember.” (‘Wallander is continuing untilNovember.’) was elicited both as an assertionand as a confirmation, with focus either on theinitial, medial, or final content word. An examplefor a context question eliciting final focus ina confirmation is ‘Until when is Wallandercontinuing, actually? Until November, right?’.As a major result, one strategy of signalling aconfirmation was by means of a lowered H- riseon the target word. However, another strategywas, like in Ambrazaitis (2007), to realise thetarget word with a lack of a H- rise, i.e. withfalling F0 pattern (cf. Figure 1, upper panel).The initial word was always produced with arise, irrespective of whether the initial worditself was in focus or not. Initial, pre-focal riseshave been widely observed in Swedish and receiveddifferent interpretations (e.g. Horne,1991; Myrberg, in press; Roll et al., 2009). Forthe present paper, it is sufficient to note that aninitial rise is not necessarily associated withfocus.72