|Title:||Doubling and resumptive pronouns in Tyrolean wh-extraction|
|Comment:||Paper will be submitted to the Proceedings of the Workshop on Syntactic Doubling in European Dialects, March 16 - March 18, 2006, Meertens Institute, Amsterdam.|
|Abstract:||This paper analyzes long extraction of relative and interrogative pronouns in the Tyrolean dialect of Meran, a Southern Bavarian variety. In this variety, long extraction is characterized by the presence of doubling of pronouns in intermediate [Spec,CP] positions both in relatives and interrogatives (1), and apparent optionality between the doubling structure and a resumptive pronoun structure limited to relatives in certain contexts (s. (1a). vs. (2); cf. McCloskey 1990, 2002, Rouveret 2002 and Adger & Ramchand 2005 for the discussion of similar structures in the Gaelic languages):
Doubling of this type is an instance of repetition of a semantically superfluous element. In particular, it cannot be reinterpreted as a 'spare-movement' strategy, as an instantiation of two items with different function, or as agreement (cf. other phenomena discussed at the Workshop on Syntactic Doubling in European Dialects, March 16 - March 18, 2006, Meertens Institute, Amsterdam). The existence of doubling structures in wh-movement thus shows that doubling as the repetition of semantically empty elements does exist as a phenomenon in need of explanation.
The core of my proposal, cast in the framework of Optimality Theory, (Prince & Smolensky 1993, Legendre et al. 1995, Grimshaw, J. 1997, Legendre, G., P. Smolensky & C. Wilson 1998) is that doubling is triggered by a constraint requiring the base position of extraction to be traceable. This constraint is best understood as a processing-optimizing strategy. When other constraints force a violation of the constraint, the resumptive pronoun strategy is employed, as in (2), where the verb introducing the lower clause selects the complementizer dass, which is incompatible with a relative pronoun in [Spec, CP].
Interpreting doubling in wh-movement as triggered by a functional requirement of this type explains why doubling does not occur in A-movement: A-movement chains are short and no long-distance processing is required. An analysis along these lines furthermore explains why doubling is found extensively in dialect systems, though much less so in standard languages. Standard languages are to a large extent, sometimes exclusively, used as written languages, whereas dialects are almost always used oraly. Processing a complex sentence is arguably more difficult in oral than in written parsing, hence the predominance of structures facilitating processing in dialect systems. Finally, the factorial typology of the proposed constraints shows that a structure where both doubling and a resumptive pronoun cooccur is harmonically bounded, a welcome result since there are no languages exhibiting this pattern, as far as I know.
A final piece of data discussed in the present paper concerns the extraction of full XPs. In relatives or interrogatives where a whole phrase has been extracted, the preferred structure is one where neither doubling nor resumptive pronouns occur. I attribute this fact to a constraint banning the repetition of whole phrases. When dominant, the constraint will ban both doubling structures and resumptive pronoun structures since in both types of structure semantically empty elements are repeated.