Functions of prosodyChunking
The division of the speech stream into chunks is obtained by prosodic phrasing. Words are grouped together (grouping) and/or set apart (demarcation) through intonational means, together with timing patterns and variations in segmental realization. Languages and language-varieties may show particular preferences in intonational grouping, as well as in the cues that signal phrase breaks. This is shown by the examples in (5) and (6).
Unlike in (5) and (6), however, there are cases where chunking constrains the meaning of the utterance. In such cases, prosodic phrasing is critical to what is being said. In (7), there are three different words (porta, chaves e leite 'door, keys and milk'), whereas in (8) there are only two words (porta-chaves e leite, 'Keychain and milk').
Importantly, the organizational structure of spoken utterances, i.e., the prosodic phrasing, is not entirely predictable from their (morpho)syntactic structure, depending on the phonological grammar of the language and the particular choices made by the speaker among the possibilities available for a given utterance (as shown in Nespor & Vogel 2007 and Wagner & Watson 2010, among many others). For the chunking or phrasing of the speech stream in Portuguese, see Vigário 2003, 2010 and Vigário & Fernandes-Svartman 2010 for prosodic word level chunks, and Tenani 2002, Frota & Vigário 2007, Elordieta et al. 2005, Fernandes-Svartman 2007, Cruz & Frota 2013, Frota 2014, Barros & Frota 2015 and Frota & Moraes 2016, for phrase level chunks.