Between feature mapping and thematic prominence: Old english se-demonstratives and pronouns in discourse

Abstract Old English se-demonstratives (which usually trace less salient referents) and personal pronouns (usually continuing previous topics) have frequently been taken to share a common pronominal property (e.g. Breban 2012; Epstein 2011; van Gelderen 2013, 2011; Kiparsky 2002; Howe 1996). This assumption holds despite their non-overlapping distribution which still remains a puzzle (cf. van Gelderen 2013; Los and van Kemenade 2018). In this paper, we argue that this distributional discrepancy stems from the lack of syntactic and formal affinities between the two forms. Se-demonstratives are either dependent (introducing full DPs) or independent (usually labeled as “pronominal”), but still instances of the same lexical item. As a D-category, they necessarily license their NP complements regardless of their being lexical or empty, thereby entering into tight formal and semantic relations with their nominal antecedents. In doing so, they rely on the working of their gender- and case-features, the two carrying semantic import and mapping onto the specific reference [+ref/spec]-property in the semantic module(s). Being bundles of case- and/or φ-features, pronominals lack the complex syntactic structure of se-demonstratives. Their formal and semantic relations with nominal antecedents are thus less intimate, holding due to interpretable person- and number-features.