Unifying agreement relations: A Minimalist analysis of Berber.

Abstract

This dissertation investigates three seemingly disparate ranges of empirical facts, namely Subject-verb agreement, Clitic-doubling, and Negative Concord, which have received different analytical treatments, and proposes that they are all forms of agreement derived under the same mechanism, namely Agree (Chomsky 2000, 2001, 2004). I demonstrate how argument extraction, whether the argument is a subject, a clitic-doubled DP or a Negative Concord, is possible only if conditions for feature valuation such as locality are met. This dissertation also addresses a fundamental question in recent Minimalism with regard to how subject-verb agreement is obtained and proposes a new analysis of the so called Anti-Agreement Effect (Ouhalla (1993, 2005). Chomsky (2004, 2005, and 2006) hypothesizes that T inherits its phi-features from C, and then enters into a Probe-Goal Match relation with the subject; as a result subject verb agreement is obtained. Given Chomsky's hypothesis that C is first merged bearing the phi-features, I propose that this hypothesis should allow three logical possibilities which I label: (a) DONATE: C transfers its Phi-Features to T and does not keep a copy of these features. (b) KEEP: C does not transfer its Phi-Features to T at all. (c) SHARE: C transfers its Phi-Features to T and keeps a copy. Given strong empirical evidence from languages such as Berber, I argue that all these options are possible, and that they are ordered naturally under principles of efficient computation i.e. economy and Minimal Search, with DONATE being the most economical, and KEEP being the last resort and least economical. I extend the same analysis to negative clauses and propose a novel treatment of the role of negation within an approach where C selects Neg, which in turns selects T. Given such an approach, I address the question of how T inherits phi-features from C with the presence of an intervening Neg-head. I propose that the phi-features are transferred from C to T via Neg and show empirical evidence, such as the Irrealis verb form in Berber and inflected negation in Finnish, which results from such transfer.