1 FEATURE RELATIVIZATION AND BARE NOUN AGREEMENT IN BRAZILIAN PORTUGUESE Danniel Carvalho UFBA/CAPES ABSTRACT The aim of this paper is to promote a unified analysis for the licensing of what Foltran and Rodrigues (2013) called abstract entities, i.e., subjects of copular predicates which do not trigger agreement marking. To do so, we propose that what is assumed by the author to be agreement failure (Wechsler, 2011) can be reinterpreted as a mechanism of feature relativization (Preminger, 2014), whose core idea is to deal with φ-agreement obligatoriness even in contexts which it seems not to occur. Relativized minimality (Rizzi, 1990) predicts a unified analysis of the data presented here, whose proposal points out different structures for the different clause interpretations, and, in addition, may predict generalizations which capture default agreement as a predictable phenomenon in grammar and not a last resort strategy. Thus, I assume that number and gender agreement in such structures can be seen as the satisfaction of the DP internal requirements, as predicted by Borer (2005). KEY WORDS Relativized agreement. DP syntax. Number/Gender features. Bare nominal. Introduction Phi-features are known to be crucially involved in agreement relations, such as the one between grammatical subjects and its predicate (e.g. the finite verb in a sentence or adjectival predicates (Kerstens, 1993)). Features such as person, number and gender are commonly associated to triggering these relations in such contexts. According to Corbett (1991, p.1), gender is the most enigmatic of grammatical categories. It not only leaves open many questions about its behaviour cross-linguistically but also intra-linguistically/within a