Perception of intonational contours on given and new referents: a completion study and an eye-movement experiment

Two factors have facilitated advances in the study of the interplay between speech perception and phonology: first, we now have the benefit of various technological advances that allow for the collection and analysis of data on perception that were unavailable 20 years ago, and second, theoretical progress. In the framework of Optimality Theory (Prince & Smolensky 2004, McCarthy & Prince 1993), research has led to the formulation of perceptually grounded constraints that interact with constraints motivated by other modules of the linguistic system. See for instance Smolensky 1996, Steriade 1995, 1999, Flemming 1995, who have modeled in OT the influence of perception on the usual production grammar. A further progress is proposed by Boersma (1998 et seq.), who models perception itself in OT. In this paper, by presenting a completion experiment and an eye-tracking study and by implementing the results in an OT model, we rely on both the technological and the theoretical advances. Traditionally, phonological theory draws heavily on production, a fact which is explained by the ease with which articulatory facts, like the position of the tongue, can be analyzed as opposed to the difficulties encountered in the study of perception. It has been observed repeatedly, however, that the way we perceive contrasts in sound structure has an influence on phonological systems and sound changes, and vice-versa, and that we articulate sounds or sound sequences depending on how we perceive them (see for instance Ohala 1981, Lindblom 1990, Steriade 1999). To cite well-known examples, palatalization arises as a consequence of coarticulation between a coronal and a high vowel, and final devoicing as a consequence of a reduction of all final obstruents to plain voiceless ones. If neutralization arises as a result of palatalization or final devoicing, contrasts are lost. But clearly, the maintenance of a contrast may be in some cases more important than the reduction of articulatory effort, even when perception is not maximally guaranteed, such as in nonprominent syllables (see Hume 1994 and Cavar 2004, for instance). On a more abstract level, perception and production may not be separable into distinct cognitive modules, since these two components have necessarily evolved in parallel. The phonetic articulation of sounds and the intonation of melodies are achieved by manipulating the physical acoustic waves, and their perception is translated into auditory impressions by the auditory system. Phonological systems, on the other hand,