The emergence of "weak" definite articles

Abstract

This paper investigates the emergence and distribution of weak and strong definite articles in German. While reduced weak forms are already attested in Old High German, their use was initially rare and largely restricted to noun phrases that are very likely to have a singleton extension in any situation. Over time, weak articles expanded to a wider range of noun phrases, yet systematically failed to extend to anaphoric environments with an indefinite antecedent, where strong forms remain obligatory in Modern German. We aim to explain this diachronic trajectory by combining evidence from historical corpora with models from computational pragmatics and agent-based simulations using reinforcement learning.