Abstract

Increasingly, it is agreed that a robust theory of the syntax-phonology interface must account for variation within and between speakers in the distribution of prosodic boundaries, because the variation is not random, but is instead observed to vary within fixed limits, which a grammar can and should account for (Post, 2000; Myrberg, 2013; Feldhausen, 2014).

This paper explores the possibility that the theory must also account for variation within and between speakers in the cues used to mark prosodic boundaries. This further step is necessary if the observed variation in boundary cues can be shown to vary within fixed limits and can thus also be accounted for by the grammar, as has been shown previously for data in German (Truckenbrodt, 2002, 2004, 2007). This paper makes two claims: i) firstly, that a similar pattern of cue variation (that is, within fixed limits) is observed in Egyptian Arabic (Hellmuth, 2012), and that the set of languages in which such variation is expected to occur might be predicted from other features of prosodic typology (cf. Jun, 2014); and, ii) secondly, that speaker variation in boundary cues interacts with speaker variation in the distribution of boundaries (Frazier, Carlson, & Clifton, 2006), opening up the possibility of a principled account of the observed variation. The claims are supported empirically from experimental read speech data in colloquial Egyptian Arabic, analysed in Match Theory (Selkirk, 2009, 2011), and preliminary results from parallel experimental data in Jordanian Arabic and from a corpus of broadcast Modern Standard Arabic.