Paula Menéndez-Benito (Göttingen)

Reasoning and evidence: Sources and direction

Joint work with Keir Moulton (Simon Fraser University)

Natural languages have constructions that indicate that a claim is based on reasoning from evidence. Some of these constructions encode a particular directionality of evidence (e.g., Davis and Hara 2014, Winans 2016). The phenomenon can be illustrated with the examples in (1) and (2) (after Davis and Hara 2014). While epistemic "must" expresses conclusions that follow from a piece of evidence (1) as well as conclusions about what might have caused the evidence (2), "seem"-reports are only possible if the embedded claim is assumed to be cause of the available evidence (a "Reasoning Back" (RB) effect, as in (2)). Other constructions that have been shown to convey a RB effect are (i) a sub-class of evidential inferential elements (see Krawczyk 2012 and references therein), and (ii) the presentational "this" construction (Winans 2016).

(1) Reasoning Forward from Evidence.
We see, from the 20th floor, rain pouring down but we cannot see the street.
a. The sidewalks must be soaked.
b. #The sidewalks seem to be soaked / It seems that the sidewalks are
soaked.

(2) Reasoning Back from Evidence.
We see, on a security camera that shows only the sidewalks, that they are soaking wet.
a. It must be pouring rain.
b. It seems that it?s pouring rain.

In this talk, which reports on work in progress, we show that the class of constructions that display a RB effect is larger than previously thought, and its members are surprising (they include bare assertions, and some canonical doxastic attitudes like "believe" or "think"). We tentatively suggest that
there are two possible sources for the RB back effect: (i) evidential and epistemic items might contribute RB lexically (as Davis and Hara 2014 argued for the Japanese evidential particle "youda"); (ii) for bare assertions and belief reports, RB comes about via a default predication relation that holds between propositions and topic situations (building on Winan's (2016) proposal for presentational "this" constructions). We illustrate this by showing how the RB effect is sensitive to Question Under Discussion manipulations (Roberts 1998/2012, Schwarz 2009).

References

Davis, C. and Y. Hara. 2014. Evidentiality as a Causal Relation: A Case Study from Japanese youda. Empirical Issues in Syntax and Semantics 10, ed. Christopher Piñón, 179?196.

Krawcyk, E. A. 2012. Inferred Propositions and the Expression of the Evidence Relation in Natural Language Evidentiality in Central Alaskan Yup'ik Eskimo and English. Ph.D. dissertation, Georgetown University.

Roberts, C. 2012. Information structure in discourse: Towards an integrated formal theory of pragmatics. Semantics and Pragmatics Volume 5, Article 6, 1-69. [Based on a 1998 amendation of Roberts 1996]

Roberts, C. 1996. Information structure in discourse: towards an integrated formal theory of pragmatics. In Jae-Hak Yoon & Andreas Kathol (eds.) Papers in semantics (Working Papers in Linguistics 49). The Ohio State University. http://linguistics.osu.edu/files/linguistics/workingpapers/ osu_wpl_49.pdf. [Amended in 1998.

Schwarz, F. 2009. Two Types of Definites in Natural Language, Ph.D. dissertation, UMass Amherst.

Winans, Lauren. 2016. Inferences of "Will". Ph.D. dissertation, UCLA.