Some articulatory and cerebral correlates of prosodic focus in French

International audienceAmong the different prosodic phenomena, prosodic focus, which is very often used in speech communication, deserves particular attention. It attracts the listener/viewer's attention to relevant elements in the discourse. It is a means to designate, identify or select an element (see e.g., Halliday, 1967; Gussenhoven, 1983; Selkirk, 1984; Nølke, 1994; Birch & Clifton, 1995; Ladd, 1996). In this sense, prosodic focus is related to pointing or deixis. Pointing is a universal ability which orients the attention of another person so that an object/person/direction/event becomes the shared focus of attention. It serves to single out, to individuate what will become the referent. A pointing gesture is most often performed with the index finger and arm extended in the direction of the interesting object and with the other fingers curled inside the hand (Butterworth, 2003). But pointing can also be vocal or linguistic. Linguists define deixis as the way language expresses reference to points in time, space or events (Fillmore, 1997). In its narrow sense, deixis refers to the contextual meaning of deictic words, which include pronouns, place deictics, time deictics and demonstratives. In its broad sense, it provides a means to highlight relevant elements in the discourse, to designate, identify or select an element. In French, as in many languages, broad sense deixis can be conveyed by syntactic extraction or prosodic focus (Berthoud, 1990). Syntactic extraction involves the use of a cleft presentation form such as in the example below: C'est Madeleine qui m'amena (It's Madeleine who brought me along.) Deixis can also be conveyed by contrastive prosodic focus, i.e. by using a specific intonational contour on the pointed item, such as in the example below: MADELEINEF m'amena. (MADELEINEF brought me along.) The effect of this intonational contour is to highlight the pointed item ("Madeleine"), the rest of the utterance bearing a flat, post-focal contour. Compared with a broad focus or non-deictic rendition of an utterance, prosodic focus involves precise acoustic and articulatory modifications. It has been shown that speakers use reliable proprioceptive strategies to organize phonation and articulation adequately in order to convey prosodic focus (see e.g. Dohen et al., 2006). On the perception side, visual only and auditory-visual perception tests carried out in our department (Dohen & Lœvenbruck, 2005; in press) have shown that these phonatory and articulatory patterns are not just productive habits but that they are recovered by listeners/viewers and used for prosodic focus detection. It should be mentioned that syntactic extraction may or not be accompanied by prosodic focus. The use of a specific (stored) linguistic construction can convey a pointing meaning, without the need for specific somatosensory (acoustic and articulatory) change, with respect to a non-deictic utterance. Interestingly, pointing is involved at several stages of human communication development. Ocular pointing (or deictic gazes, at 6 - 9 months) and, later, index finger-pointing (deictic gestures, at 9 - 11 months) have been shown to be two key stages in infant cognitive development that are correlated with stages in oral speech development. Pointing seems to be one of the first communicative tools used by babies (e.g. Bates et al., 1975; Bruner, 1975; Hewes, 1981; Kita, 2003; Tomasello et al., 2007). It is a key part of the shared attention mechanism in child-adult interaction. It seems to emerge spontaneously, in stages, and in association with oral productions. The crucial role of index-finger pointing in language development suggests that vocal pointing may share features in common with manual pointing, and typically may be grounded in close cerebral tissues. Our aim is to better understand the cerebral correlates of the production of prosodic and syntactic focus in French. We make the hypothesis that integrated representations (acoustic, articulatory, proprioceptive) may be needed in order to produce prosodic pointing adequately and that these multisensory representations may be formed via the activation of associative cerebral areas. Syntactic extraction may not necessitate such a network. A first preliminary fMRI study of the production of prosodic focus and syntactic extraction is presented. It shows that prosodic pointing seems to recruit a left temporo-parieto-frontal network whereas grammaticalized syntactic pointing mainly involves frontal regions. The exploratory results of a recent study of multimodal pointing (digital, ocular, prosodic and syntactic) are then presented, which provides convergent results. Altogether these findings suggest that speakers may use multisensory representations in order to produce prosodic focus (pointing) adequately, just like they do to produce a manual or ocular gesture. These representations may be formed via the activation of associative parietal areas. The lack of parietal activation in syntactic pointing could be due to the absence of on-line multisensory construction in syntactic pointing, compared with prosodic pointing. Syntactic pointing uses a grammaticalized "frozen" construction and may not need integrated representations to be formed. These fMRI study therefore shed light on the complementary relationship between syntax and prosody, with syntactic focus conceived as more "stored" and prosodic focus as more "built on-line"