In this paper we examine the relation between the loss of formal gender and Case features on simple demonstratives and the topic shifting property they manifest. The examination period spans between Old English and Early Middle English. While we argue that this loss has important discourse-pragmatic and derivational effects on demonstratives, we also employ the Strong Minimalist Hypothesis approach (Chomsky 2001) and feature valuation, as defined in Pesetsky & Torrego (2007), to display how their syntactic computation and pragmatic properties have come about. To account for the above innovations yielding the Early Middle English ϸe ('the'), we first discuss the formal properties of the Old English demonstratives which distinguish number, gender, and Case features. This inflectional variety of forms allows the Old English demonstratives to be used independently and to show the anaphoric and discourse-linking properties of topics. Crucially, the same properties characterise also German and Dutch demonstratives that manifest Case and/or gender morphology overtly, which shows that the syntactic distribution of LIs and their morphological richness should be considered as intertwined. The above properties are then confronted with the determiner system in Early Middle English, whose forms undergo inflectional levelling producing the invariant ϸe/ðe form that loses its distributional independence and acquires the article status. The levelling process in question is argued to stimulate the shift of the [+ref/spec] feature from the formal to the semantic pole. This suggests that the Early Middle English ϸe form no longer counts as an appropriate anaphor in topic shift contexts owing to its indeterminacy of Case, gender, and φ-features, which means that it cannot satisfy the Full Interpretation requirement at the interfaces.