We discuss semantic agreement, cases in which agreement between a head and a DP seems to drive, or be driven by, interpretive properties of this DP rather than its syntactic properties. This discussion centers mostly on the case of subject/verb number and person agreement (but its conclusions would extend to other cases of agreement between a head and a DP). In such a configuration, semantic agreement occurs in cases in which the feature values on T do not seem to match the feature values of its DP subject. More specifically, we conclude that features on agreeing heads can and sometimes must be semantically interpretable. In such cases, the values of the features on T can target the denotational properties of its DP subject, instead of its phi (ϕ)-feature (values): they can or must trigger a presupposition about this DP subject's denotation.
Charnavel et al. (Mon,) studied this question.