Converging evidence on the autonomy and abstractness of the representation of lexical stress

The experiments reported in the thesis investigate the nature of word stress and its lexical representation. Focusing on Italian, I considered two research questions: How is lexical stress represented? How does this representation intervene in perceiving or producing a word? Italian is a polysyllabic language with free-stress position: Stress may appear on one of the last three-syllables (e.g., TAvolo, ‘table’, paROla, ‘word’, coliBRI, ‘hummingbird’, capitals indicate stress) and its position is not predictable by rules. Moreover, there is a large asymmetry in the distribution of the stress patterns, with about 80% of words bearing stress on the penultimate syllable (e.g., paROla, ‘word’). On the assumption that stress is a lexical feature and that the stress pattern of a word is part of the knowledge stored in the lexicon, three studies were designed in which a priming paradigm and a visual word paradigm were used. Specifically, we investigated lexical stress in two domains, i.e., spoken-word recognition (Chapter 2) and reading aloud (Chapters 4 and 5). The results shed new light on the nature of the stored prosodic knowledge about lexical stress and on what extent processing of lexical stress is similar in spoken-word recognition and reading aloud. In synthesis, the empirical evidence indicates that lexical stress is part of the abstract prosodic knowledge stored in the lexicon: It pertains to the suprasegmental level of word representation and it is dissociable from the information pertaining to the segmental level.