In this work we have tried to show that: 1) the agreement with embedded NP is a (syntactic) parametric option in some languages; 2) the distribution of agreement with the embedded NP is sensitive to different syntactic environments and that the semantics of the nouns involved is not the main responsible of the pattern of distribution of agreement. Our conclusion will suggest that minimalist assumptions on agreement should include a different notion of phase to account for the present data and that agreement with an embedded NP with a θ role interpretation is a syntactic parameter not involving fine - grained semantic distinctions within the class of noun.
