Lexical Stress Effects in Italian Spoken Word Recognition

Lexical Stress Effects In Italian Spoken Word Recognition Lara Tagliapietra (tagliapi@psico.units.it) Department of Psychology, Via S. Anastasio 12 Trieste, Italy Patrizia Tabossi (tabossi@units.it) Department of Psychology, Via S. Anastasio 12 Trieste, Italy Abstract presented with a visual target (e.g., MUSIC) after a one syllable long fragment. Responses were faster when the prime’s stress matched the target word (e.g., mu from music) than when it did not (e.g., mu from museum). However, the mismatching prime produced faster responses than the control prime. Facilitation in the mismatching condition disappeared only when one-syllable long priming fragments were replaced by two-syllable long fragments. In Dutch, facilitation for a visual target (e.g., OKTOBER, October) after one syllable long fragment was observed only with a matching prime (e.g, ok from ok’tober, October), but not after a mismatching one (e.g, ‘oc from ‘octopus, octopus). Furthermore, after disyllabic fragments facilitation in the matching condition as well as inhibition in the mismatching condition have been recently reported (Van Donselaar et. al., 2005). The only evidence in a Romance language replicates the Dutch results with disyllabic primes (Soto-Faraco, Sebastian-Galles & Cutler, 2001). Somewhat less clear are results of associative priming studies. Cutler (1986) reported that in English minimal pairs such as forebear (N) and forbear (V) behave as homophones, both facilitating the recognition of targets associated to either meaning (e.g., ANCESTOR and TOLERATE). On the contrary, in Dutch Donselaar et al. (2005) obtained lexical stress effects using associative priming. In this study, responses to a visual target (e.g., HEMEL, heaven) were facilitated, respect to the control, when preceded by a fragment prime corresponding to the initial part of an associate word (para from para’dijs, paradise). Differently, responses were not facilitated when the fragment was the initial part of a word differently stressed (pa’ra from pa’rade, parade). However, facilitation was clearly observed only when priming fragments (e.g. kas’ta) uniquely identified a word (e.g., kastanjie). Altogether, these findings suggest that English and Dutch speakers are sensitive to lexical stress information from the early stages of word recognition and are consistent with the view that spoken word recognition involves competition between the phonological representations of multiple activated candidates. Phonological representations wholly or partially matching the input are initially activated and compete for recognition inhibiting each other. The inhibitory process depends on the level of activation of the phonological representation, which varies with the degree of correspondence between the input and the phonological representation and the number of the competing candidates. Two cross-modal priming experiments tested whether Italian listeners rely on lexical stress information in spoken word recognition. In both experiments participants performed a lexical decision task on visual targets; however in the trials where there was a relation between prime and target, the relation was one of identity in Experiment 1, whereas it was a semantic association in Experiment 2. In Experiment 1, the match condition (e.g., the prime ['go.mi] from gomito, elbow, for the target GOMITO), produced facilitation relative to the control condition, and the mismatch condition (e.g., the prime [go.’mi] from gomitolo, clew, for the target GOMITO) produced inhibition. In Experiment 2, responses were faster in the related (e.g, the prime ['go.mi] from gomito, elbow, for the target BRACCIO, arm) than in the control condition. This, however, did not differ from the unrelated condition (e.g, [go.’mi] from gomitolo, clew, for the target BRACCIO, arm). The methodological and theoretical implications of the results are discussed. Introduction In this study we investigated the role of lexical stress during spoken word recognition in Italian. The paradigm typically used in this line of work is the cross-modal priming paradigm, which is assumed to tap the early phases of lexical processing: participants have to perform a lexical decision or naming task on a visual target that is preceded by a spoken prime. Response times to the same target as a function of the tests or the unrelated control primes are measured (Swinney, Onifer, Prather, & Hirshkowitz, 1979). In the spoken language literature, this paradigm is normally associated with the lexical decision task that occurs in two versions, depending upon the nature of the relation between the test prime and the target: the identity and the associative priming (Van Donselaar, Koster & Cutler, 2005). In the former, the relation is one of identity (e.g., elbow and ELBOW) 1 , whereas in the latter it is a semantic association (e.g., elbow and ARM). Both techniques are employed in current research on the effects of lexical stress in spoken language processing. Identity priming studies showed that the paradigm is sensitive to lexical stress effects. Cooper, Cutler and Wales (2002) conducted a study in English. Participants were Spoken primes are in italics, whereas visual targets are in capital letters.