The Spring 2015 Philosophy Colloquium Series presents Tyler Peterson, University of Arizona Linguistics Dept. The title of Professor Peterson's talk is "Evidence, Knowledge And Inference: A Perspective From Grammatical Evidentiality." (Abstract below.)
Friday, Feb 27, 2015, 3-5pm, Chris Maloney Seminar Room, Social Sciences 224 (1145 E South Campus Drive 85721).
Grammatical evidentials provide the speaker the linguistic means to talk about states, events, or actions that she did not witnees, but rather that she has evidence for. In certain contexts modal auxiliaries such as 'must' behave like an evidential: in a context where the speaker observes people walking into a windowless room with wet shoes and umbrellas, she can assert 'It must be raining'. The use of 'must' expresses the inference based on this (observable) evidence. However many languages have specific words or morphemes that encode different kinds of evidence, while having a modal semantics. In this talk I will review the current theoretical approaches to analyzing evidentiality and epistemic modality. What this reveals is that these analyses rest upon the assumption that evidence is 'out there' in the world, but there is currently no explanation of what evidence actually is, nor how it connects to the theoretical representations we assign them. I explore a number of ideas in this talk including the claim that evidence is in fact a kind of knowledge (supporting Williamson 2000). However, further steps must be taken to implement this notion in explaining the use of an evidential (or epistemic modal). In taking steps towards this, I address the notion of 'awareness': a speaker can only perform an inference based on evidence if they are aware that this knowledge indeed counts as evidence for that inference--an observation occasionally made in the semantics literature, but not usually elaborated on. Using primary language data from the endangered indigenous language Gitksan (spoken in northwestern Canada), I show how a possible worlds analysis can account in general for the use of evidentials in that language, but that the inference encoded by the evidentials is complex: the acquisition of evidence (knowledge) in a particular context triggers what I call a 'stage-level inference' (i.e. the observations made in that context) and a 'kind-level inference' (what you know about how this can count as evidence). An analogy is drawn with predicates in the non-evidential/modal domain.