Mirative

Although the Maidu 'evidential' -wéw (Shipley 1964:45) might correspond somewhat to the Washo visual, and the (unexemplified) Sierra Miwok 'circumstantial evidence' marker taˀ, tat, ˀiš- (Freeland 1951:169) may correspond to the Washo inferential (mirative).

In Archi mirativity is grammaticalized as part of the verbal category of evidentiality, so the study of the mirative in Nakh-Daghestanian languages might help to identify the meaning of exclamatives more precisely.

The Tarms Quechua Mirative often refers to information that is withheld from the addressee until the speaker sees fit to reveal it, a frequent strategy in narratives of which the unexpected outcome is reserved for the end. By consequence, the speaker him/herself need not be under the impact of surprise any longer when using the Mirative. […] Characteristically, actions performed during one's sleep or in a state of unconsciousness are expressed in the Mirative[…]. The Mirative can also be used in recounting dreams[…].

[T]he speaker had heard on the radio that a bear had attacked a woman. From the description of her wounds on the radio, he infers, using the mirative/inferential particle lą̄ą̄, that she was dragged by the bear. […] This particle also implies that the speaker was surprised at the event. Bear attacks are uncommon in Arizona, and the woman was someone the speaker was acquainted with. In fact, lą̄ą̄ is more fundamentally a mirative than an inferential, […]

He [Timothy Jowan Curnow] points out that miratives are very rare with first person, more common with second, and most common with third person. In all cases, however, non-miratives are more common than miratives.

The prefinal suffix -áʔyiʔ Mirative indicates that the speaker knows of the action described by the verb, not having observed it occur, but only inferentially from observation of its effects. It thus commonly conveys an emotion of surprise.

In the first in-depth analysis of a mirative construction which I am aware of, discussing the Turkish "evidential" perfect, Slobin & Aksu (1983; see also Aksu-Koç & Slobin 1986) give a unified characterization of the category as marking that the speaker's mind was "not prepared" for the information which is now being relayed. […] In this paper I adopt the older term "mirative" for the marked category, leaving the unmarked category unlabelled.

This is why a noneyewitness evidentiality specification in a two-term system and an inferential evidential in a three-term system may acquire a mirative extension. In Quechua […] the reported evidential can be used in a mirative sense.