Length as strength: a new account of Raddoppiamento Fonosintattico

In this paper, I propose a new analysis of Raddoppiamento Fonosintattico based on phonological strength (Smolensky and Goldrick 2016). RF is a sandhi process that affects initial consonants in Tuscan and Standard Italian. I consider RF as a type of “strengthening” that gives rise to strong segments in the phonological representation, which are then realized as long by the phonetics. The core evidence for this claim is provided by phonetic data that has been neglected by all previous analyses of RF. The duration of RF-segments is gradiently located between the duration of singleton and geminates and that their gestural profile is similar to those of singletons, rather than to those of geminates (Payne 2005, Campos-Astorkiza 2014). I propose that RF is a process of association of strength to a segment, given the presence of floating strength in the linguistic representation, rather than a gemination process. The analysis is couched in Gradient Symbolic Representations, which is the sole appropriate framework to capture these gradient effects.