How to Make a Result Evidential

Patrick Muñoz
The University of Chicago
Fri April 30th 2021, 1:30 - 3:00pm
Online (Zoom)

Email for Zoom link


Many authors (e.g. de Haan 2001) have treated ‘inference’ or ‘results’ evidentiality – grammaticized marking signaling that the eventuality associated with a clause has been learned of by perceiving not the eventuality itself, but rather some ‘concrete result’ of it – as a core type of evidentiality, the latter considered a sui generis semantic category marking ‘information source.' I instead entertain the hypothesis that results evidentiality does not specify a value for a primitive semantic category of information source, but rather arises through the interaction of multiple functional categories pertaining to time and learning: results readings surface when some combination of tense, aspect, epistemic modality, or 'learning operators’ that introduce secondary learning events (Koev 2017) combine to require that (i) the evidential origo has learned of the ‘main’ eventuality associated with a clause by perceptual means, and (ii) the time of the learned eventuality is prior to the time of learning. As such, results evidentials can be composed in multiple ways, and so have no uniform semantic origin, and even single languages may have more than one ‘results’ construction, if their inventory of temporal and learning operators allows for it. I illustrate this with cross-linguistic examples showing how these evidentials can arise through (i) the alignment of primary and secondary tense, (ii) the stacking of perceptual learning operators on perfect or completive aspect, and (iii) the restriction of past-oriented epistemic modals to bases that include only propositions learned perceptually. I finally consider what prospects there are for the hypothesis that some languages additionally encode results evidentiality as a sui generis category.