A Diachronic Account of τ–Features and of Their Output as Vocabulary Items: On the Limits to the Vocabulary Item Ø

Assuming basic tenets of Distributed Morphology and likewise the minimalist framework of Agree, it is argued that the segmentation into Vocabulary Items (VIs) of the Past forms of verbs in Present Day English is as in e.g. deem-ed rather than deem-ed-Ø. The generalized position in the literature is for the Ø-VI to be subject to the Elsewhere condition, which entails that the proper form is deem-ed, that is the form with non-exponence after -ed. The main purpose of the discussion is to give evidence of the Elsewhere condition, and I propose to do so by taking a diachronic perspective and tracking down the relevant changes affecting verbal morphology in the language. It is argued that there are three types of τ–features in Old English and that the specific τ–feature that has as output the VI´s that are commonly referred to as subject agreement endings, which are those among which the Ø-VI steadily imposes itself from the end of the Old English period, is a τ–feature that combines φ– and τ–interpretation. The feature is labelled here [+/–past]AgrT and its τ–interpretation is identified as [morphological distinctiveness between Present and Past relative to Agreement]. The progressive imposition of the Ø-VI entails that the specific content of the cited [morphological distinctiveness…] varies in time, which variation is given diverse formulations throughout the discussion with the help of the Subset Principle requirements. The ultimate formulation is reached after analyzing the differences and similarities between English and Danish–Swedish being another case in point–as regards morpho-phonological loss and the connection with V-to-T movement. The cited formulation entails that the Ø-VI is not available if it is the only VI realizing a given formal feature (note the Elsewhere condition). A corollary of the account is for Present Day English, or rather from the English language from the eighteenth century onwards, not to rely on one binary feature like [+/–past] but on two privative features, each of a different type .