Cliticization as Unselective Attract

Abstract

The purpose of this article is to provide an explanatory account of the divide between enclisis and proclisis in pronominal clitic constructions in Romance and Semitic languages.The analysis is based on two fundamental assumptions: (i) clitics do not target designated prelabelled positions, but take maximal advantage of the available categorial structure; (ii) cliticization patterns are tightly dependent on the inflectional properties of the language, more specifically, on the feature content of the two functional categories, Infl and v.We show that the various asymmetries in clitic behavior can elegantly be explained in terms of the minimalist theory of movement, combined with certain formal hypotheses about the building of phrase structure and about the relation of morphology to syntax.Relying on certain ideas about uninterpretable features, Attract and Agree, we argue that cliticization patterns can be made to follow from the strategies made available by U.G. to check the uninterpretable feature of the category Infl and from the derivational origin of the tense and person-number features.A principle, the Unselective Attract Principle, is introduced according to which an uninterpretable feature is a potential attractor for all the features which are of the same type as the one which it selectively attracts.In Romance and in Semitic, clitic phi-sets are unselectively attracted by Infl.Two additional principles, the Priority Principle and the Single Licensing Condition, insure that at some point in the derivation a clitic can incorporate into Infl only if Infl doesn't already host an attracted inflectional morpheme.This idea holds the key for the enclisis/proclisis divide.Enclisis, i.e. clitic incorporation into Infl, is disallowed in Romance finite clauses where the uninterpretable feature of Infl selectively attracts the person-number agreement phi-set; it is legitimate in Semitic and European Portuguese finite clauses in which the same feature is checked through Agree.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 22 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.