The enigma of the origin of non-canonical subject marking in the world’s languages has been met with two competing hypotheses: the Object-to-Subject Hypothesis and the Oblique Subject/Semantic Alignment Hypothesis (cf. Eythórsson and Barðdal, 2005). The present article argues in favor of the Oblique Subject/Semantic Alignment Hypothesis, presenting five sets of cognate predicates in the early/archaic Indo-European daughter languages that occur in the Oblique Subject Construction. These cognate sets have not figured in the earlier literature. Not only are they stem cognates, but they also occur in a cognate compositional predicate and argument structure construction, with a dative subject, the verb ‘be’ and an adjective, a noun, or an adverb. The discovery of these cognate data sets immediately invalidates the axiomatic assumption that non-canonical subject marking must originate in an earlier object status of these arguments. The data, moreover, form the input of a correspondence set, on which basis we reconstruct predicate-specific oblique subject constructions, a partial predicate-specific oblique subject construction, as well as a more abstract schematic dative subject construction for Proto-Indo-European, using the formalism of Sign-based Construction Grammar. The evidence presented here thus suggests that oblique subjects are inherited from an early proto-stage and do not represent an individual development in the Indo-European daughter languages.
Stored only when both papers are in the corpus — 52 citations exist upstream.