The emergence of Hebrew loydea / loydat (‘I dunno masc/fem’) from interaction

A general approach to discourse markers is sketched through an analysis which approaches grammar as emerging in interaction and coming into being through mundane language use (Hopper 1987, 2011). The study continues work on mental verb constructions in a variety of languages. By analyzing all 191 tokens of the (subj)-neg-pred construction of the Hebrew mental verb yada (‘know’) employed throughout audio-recordings of over 7.5 hours of Hebrew casual interactional data, I trace the route of this construction’s gravitation towards the discourse marker loydea / loydat (‘I dunno masc/fem’). I argue that employment of the construction is highly formulaic, not necessarily epistemic, and that its uses are closely tied to its prosodic, morphophonological and syntactic properties, to its position within the ongoing turn and sequence, and to the particular activities in which participants engage in interaction. Based on a mostly synchronic analysis of the data, I suggest two grammaticization paths leading to employment of this construction as a discourse marker. This is then supported with some diachronic evidence. The study of Hebrew loydea / loydat (‘I dunno masc/fem’) shows that the boundaries between the three categories of discourse marker, pragmatic marker, and modal particle can be rather blurred.