| Summary: | This paper presents an analysis of the Bolognese 3ms.nom clitic, which deviates from the expected alternations found in other Romance languages. It appears as [al] preconsonantally and [l] prevocalically, but it surprisingly has an apparent third allomorph, [a], which occurs only (and sometimes optionally) when preceding dat, acc, or neg clitics. For example, [a=t=ˈdiːz] ‘he says to you.s’ seems to show a sequence of 3ms.nom [a] and 2s.dat [t]; the expected preconsonantal [al] is replaced by [a]. We argue that constructions of this sort involve not a string of clitics but instead a “duplex” clitic [at] that combines 3ms.nom with 2s.dat. This approach explains why the apparent [a] surfaces only before certain clitics: it is actually the first half of a larger clitic that is available only in the presence the appropriate feature combinations (such as 3ms.nom and 2s.dat). We formalize this proposal in Optimality Theory using the framework of Lexical Selection. This analysis accounts for the puzzling behavior of the 3ms.nom clitic and necessitates refinements to the Lexical Selection formalism.
|
|---|