Project 8: Prosodic structure and FPD: segmental-suprasegmental interaction
The aim of this project is to elucidate how disparate acoustic parameters covary to cause language-specific interpretations of prosodic properties and conversational functions. S2S partners differ widely in their methods. Intonologists use laboratory phonology to produce formal phonological descriptions of f0. Conversation Analysts study functional roles of correlated phonetic parameters in conversations. The S2S groups have expertise in virtually all paradigms used to explore prosodic structure and its interaction with segments, and will use pairs of languages which are more closely related (Dutch-English, French-Italian) or less closely related (initially Czech, extended if possible to other S2S languages). Intonation systems differ in each of these pairs to a greater or lesser extent. Cross-linguistic comparisons will show to what extent the dependencies are language-specific.
Production studies will identify correlations between acoustic parameters reflecting tempo, timing, articulatory setting, voice quality and intonation, and compare functionally- and formally-oriented analyses. There are 3 foci: (1) How systematic variation in FPD influences perceptual parsing of speech into longer linguistic units e.g. intonational phrases. (2) Dependencies in FPD in various types of questions. (3) Covarying parameters that modify categorical interpretations of intonation contours when alignments between segmental and suprasegmental properties change.
Perception studies will have two aspects. (1) To test hypotheses formulated from the combined results of the production analyses. Various linguistic and psycholinguistic paradigms will be used to test the naturalness of resynthesized stimuli and changes in the interpretation of utterances as a function of FPD parameters. E.g. parameters will be systematically manipulated in brief utterances and placed in their original conversational contexts to assess whether listeners’ interpretation changes. (2) Interactions between acoustic and visual cues to prosodic information will be assessed in noise and in quiet; visual cues are expected to aid comprehension in noise.
Materials Groups will use each others’ data, which includes much spontaneous speech (conversational and other e.g. dictations, interviews), as well as controlled read sentences. At least three corpora include many different styles (Czech: 400 speakers; English: IViE: 126 speakers; Norwegian: 50 hrs). Most perception experiments will manipulate natural speech via resynthesis (PROCSY, Residual-LPC-based concatenative synthesis, PRAAT PSOLA resynthesis).
Young researchers One ESR (Cangemi) is based at Aix, and the other ESR (Zellers) is at Cambridge. They have strong backgrounds in some of the following, and are learning about the others: CA, intonational phonology, acoustic phonetics, (audio-visual) experimental design/analysis. They will make appropriate visits to: Sheffield/York for CA & audio-visual; Nijmegen, Prague or elsewhere for intonation, data collection and perception experiments. If feasible, Prague’s ER will help and mentor both ESRs.
Links Project 8 addresses severe fragmentation within four areas of linguistics: CA, intonational phonology, prosodic phonetics, segmental phonetics. Thus it has direct links to Theme I, Project 5 (perceptual coherence) and Project 7.
Working on this project: » Dr Brechtje Post » Prof MariaPaola D'Imperio » Prof Carlos Gussenhoven » Prof Noël Nguyen » Prof Zdena Palková » Meg Zellers » Francesco Cangemi » Prof David House
