Effects of tonal alignment on lexical identification in Italian

In Neapolitan Italian, the LH nuclear accent rise is later in questions (L*+H) than in statements (L+H*). Furthermore, the H target is later in closed syllables than in open (penultimate) syllables, in both intonation modalities. As a consequence, the H target is differently located in lexical minimal pairs contrasting in consonantal length and syllable structure, such as nonno /'nnno/ ‘grandfather’ and nono /'nno/ ‘ninth’: it is aligned towards the coda consonant in nonno and around the end of the stressed vowel in nono. In this paper, we tested the hypothesis that Neapolitan listeners might employ such differences in alignment due to syllable structure in lexical identification. In Experiment I, listeners were asked to identify stimuli manipulated both in segmental length and tonal alignment as either nonno or nono. In Experiment II, listeners had to categorize stimuli manipulated in segmental length as well as for tonal alignment 1. as nonno or nono and, at the same time, 2. as questions or statements. Our results showed that, in stimuli ambiguous for segmental length, more nonno embedded in questions were perceived when the peak alignment was later within the coda consonant. This does not support models which predict that suprasegmental and segmental levels are independent from each other for the purpose of phonological contrast.