The widespread pattern of "noun-less" variants of DPs (e.g. two/these [nP/NumPo] vs. two/these cars) might lead to the expectation that the definite determiner should be able to head DPs with a null complement. The best candidate for representing the structure [Ddef [nP/NumPo]] are 3rd person personal pronouns (Elbourne, 2005; a.o), an analysis supported by evidence for NP-ellipsis (or N-anaphora) in pronouns. However, equating pronouns with THE+[nP/NumPo] faces a number of issues: the different behavior of pronouns and other definite descriptions with respect to binding; the differences in syntactic and semantic features between pronouns and DPs headed by THE (e.g. gender); the one-to-many structure-form relationship between the single structure THE+[nP/NumPo] and different series of pronominal forms (strong and weak forms); and, a gap in the attested combinations of THE and [No]. Our account addresses these issues. We propose that the D used in pronouns, which we label Dpron, has the semantics of THE, but THE and Dpron differ in their formal features. We posit that a null NP complement needs to be licensed by features on the determiner the NP merges with, and that Dpron bears such features. Thus, we capture the fact that typical THE does not surface in cases where there is no overt element in the complement of D (i.e. what we call total emptiness) via the absence of these features on THE. Additionally, to further explore the differences between strong and weak forms in Romanian, we present the results of a corpus study on these forms.
