Summary

Prosodic domains in Romance languages have been defined and investigated by means of various approaches and within influential frameworks. On the one hand, they have been directly derived from syntactic representations (see studies cast within the early Generative Theory, the Minimalist Theory, and the Direct Reference Hypothesis in general). On the other hand, a structure of prosodic constituents has been proposed as part of the phonological component of the grammar. In this case, two approaches have been followed.

A prosodic representation has been seen as mediating between syntax and phonetics-phonology, and prosodic domains have been identified on the basis of morphosyntactic information and phonological rules or patterns of prominence. Phenomena from various Romance languages have been described in order to support the proposal of specific domains, like the /s/ voicing assimilation in Spanish; the intervocalic fricativization and spirantization in Italian; various external sandhi phenomena involving resyllabification in Portuguese, Italian, and French; and stress clash resolution strategies in various Romance languages. As for the other languages, arguments in favor or against some constituents, especially the Phonological Phrase and the Clitic Group, are controversial, and proposals may even differ with reference to the same language; the existence of other constituents is more straightforward, as in the case of the Intonation Phrase (IP).

However, the prosodic structure and its domains have been identified by considering intonation and durational and segmental features independently from morphosyntactic information. In most Romance languages, two levels above the word are reported (the IP and the Intermediate Phrase [ip]), while in some languages a third level is found (the Accentual Phrase); rarely, only one level is reported (the IP). Even following this approach to defining prosodic constituents, evidence in favor of intermediate level domains, mainly the ip, is not always conclusive.

Romance languages offer many phenomena supporting the existence of prosodic domains, independently of the ways these domains are defined. Cross-language differences are acknowledged, and, interestingly, some authors suggest that constituents identified within different frameworks and adopting different criteria may actually correspond to the same unit.