Many if not all evidential languages have a mirative evidential: an indirect evidential that can, in some contexts, mark mirativity (the expression of speaker surprise) instead of indirect evidence. We address several questions posed by this systematic polysemy: What is the affinity between indirect evidence and speaker surprise? What conditions the two interpretations? And how do mirative evidentials relate to other mirative markers? We propose a unified analysis of mirative evidentials where indirect evidentiality and mirativity involve a common epistemic component. (...) A mirative interpretation requires a close temporal proximity between the speech event and the event of the speaker's learning the at-issue content. (shrink)

In recent years, much attention has been given to the puzzling relationship between tense and evidence type found in languages where a single morpheme appears to encode both reference to time and to the evidential source for the assertion. In natural language, _tense_ has long been understood as serving to locate the time at which the proposition expressed by the sentence holds. The two main theories of _evidentials_ both agree that these morphemes serve to identify the type of evidence the (...) speaker has for their assertion. In languages with evidential-tense morphology, these two categories of meaning are intertwined in ways that are unexpected given our understanding of both phenomena. Specifically, these evidential-tense morphemes appear to encode reference to a time that is linked to the situation in which the speaker gains evidence for their assertion. Two competing approaches have emerged in the literature as to whether these evidential-tense morphemes make crucial reference to the _time_ evidence was acquired (Lee 2013 ; Smirnova 2013 ) or to the _time and place_ of the speaker with respect to the event (Faller 2004 ; Chung 2007 ). This paper examines the temporal and evidential properties of the Mvskoke (or Creek) _graded_ past tense system and finds novel support for the view in which evidential-tenses encode _Evidence Acquisition Time_ (EAT). Mvskoke is shown to have three evidential-tenses which form part of its graded tense system, comprising recent, middle, and distant past. The main proposal is a formalization of EAT as a moment of belief-state change, i.e., the moment the speaker comes to believe the proposition. It is shown that Mvskoke’s evidential-tenses are compatible with a range of evidence types, and this distribution is explained through interactions with viewpoint aspect. (shrink)

This paper focuses on the Korean suffix -te, which has been variously analyzed as a marker of tense, aspect, tense–aspect, mood, mood–tense, or evidentiality. I argue against all of these approaches and propose instead that -te is a spatial deictic past tense, which triggers an evidential environment. It refers to a certain past time when the speaker either observed an event or some evidence of the event within his (her) perceptual field. Thus, the denotation of -te is ‘overlap’, not between (...) the speaker's perceptual field and the event itself, but between the speaker's perceptual field and the evidence of the event at the past reference time. To account for this denotation, propose an ‘evidence trace’ function as well as a ‘speaker's perceptual trace’ function (cf. M. Faller, J Semantics 21:45–85, 2004). My analysis shows that suffixes like -ess (which is traditionally analyzed as a perfect) play two roles, as an indirect evidential and a perfect, depending on whether they appear with the spatial deictic tense -te or with a simple deictic tense. I argue that in Korean two distinct tense systems—the regular tense–aspect system and the spatial deictic tense–evidential system—exist in parallel. Thus the proposed analysis allows evidentials to be subsumed under the formal theory of tense, aspect, and mood. (shrink)

This paper explores how the meaning of evidentiality is temporally constrained, by investigating the meaning of Korean evidential sentences with –te. Unlike evidential sentences in languages that have previously been formally analyzed , e.g. Cuzco Quechua and Cheyenne, Korean evidential sentences with –te are compatible with both direct and indirect evidence types. In this paper, I analyze –te as an evidential that lexically encodes the meaning of a ‘sensory observation’. I account for the availability of both direct and indirect evidential (...) readings in terms of the variable temporal relation between relevant eventualities. I show that this temporal relation is compositionally determined by the interaction between –te and tense, and that it in turn constrains possible (direct vs. indirect) evidence types. I also provide empirical evidence for the modal meaning contributions of –te sentences, and develop a formal analysis in terms of Kratzer’s modal theory. The paper concludes by discussing the empirical and theoretical improvements of the proposed analysis over earlier analyses by Chung, and the implications for crosslinguistic studies of evidentials. (shrink)