On the role of prosody in syntactic and semantic disambiguation

In many ambiguous sentences, prosody is the only information through which syntactic and semantic ambiguities can be resolved. The same sequence of words can have two different syntactic and semantic representations. In the typical case of global syntactic ambiguities, a constituent can hold two different positions in the structural tree, while in semantic ambiguities the variable bound by a logical operator can be substituted by a lexical entry which occurs in the sentence in a position close or far from the operator. The two alternative readings of the sentence are conveyed to the listener through different prosodic structures. In this work we take into consideration both global syntactic ambiguities, in which prepositional phrases, adverbs and relative sentences can have two types of attachment ("high" or "low", restrictive and non-restrictive), and sentences with semantic ambiguities, in which the negation, the negative quantifier and the focus-sensitive operators even and only can have a wide or a narrow scope. The prosodic analysis of these data is carried out according to the principles and methods of the Autosegmental-Metrical Theory of Intonation (Beckman and Pierrehumbert, 1986) which combines a phonological approach (autoegmental and metrical phonology) with laboratory methods that allow a detailed phonetic analysis. Conrastively applied to in Italian and English sentences, the analysis allowed to highlight two central aspects of prosodic disambiguation: i) syntactic type ambiguities can be solved through variations of prosodic phrasing which are not exclusively conveyed by the presence of a pause, and semantic type ambiguities can be solved through variations in the distribution of phrasal prominences which only secondarily gives rise to phrasing differences; ii) disambiguation strategies are not universal but linguo-specific.