Annette Hohenberger - Evidentiality, Theory of Mind, and Argumentation: A Cognitive Science Perspective from Turkey

In this talk, I will present findings from four developmental studies on the relation between Turkish evidential markers and Theory of Mind (ToM) understanding in Turkish children from 4-11 years as well as from one study with adults on the role of evidential markers in argumentation, from a Bayesian perspective.

Turkish has a three-way evidential system that encodes the source of some information morpho-syntactically in the form of different past-tense suffixes on the verb stem: (1) Direct, perceptual evidence (-DI), (2) indirect evidence from hearsay (-mIş), (3) indirect evidence from inference (-mIş).

In our developmental studies (Bayramoglu, Özoran, Gözenman, Arslan), previous expectations that the use of evidential markers might facilitate ToM understanding were not confirmed. The expectation is based on the assumption that the evaluation of the source of some information may allow the child to take an explicit attitude towards the information content conveyed in the proposition, thus facilitating false belief understanding. In our studies testing 1^st order ToM, we adapted Wellman & Liu’s (2004) Theory of Mind scale to Turkish and cast the ToM-stories in three different evidential versions: (1) direct evidence ( -DI), (2) hearsay (-mIş) and (3) neutral (present progressive). In the study testing 2^nd order ToM we used two stories developed by Flobbe et al. (2008). At younger ages (4-5 years) ToM scores were consistently higher for direct evidence as compared to both indirect evidence and the neutral condition; however, at older ages (> 6-7 years), no such difference was found anymore. A control study with 4-6-year-old children on the understanding of narratives (requiring no ToM reasoning) with the same three evidential versions showed exactly the same pattern, indicating that the facilitatory effect of –DI found in the ToM studies was rather the result of facilitated narrative understanding, due to the simpler semantics and clearer use of –DI as compared to –mIş whose use is more varied and therefore more ambiguous.

In the adult study, the role of Turkish evidential markers in argumentation was studied in a Bayesian framework (Bayindir). The same three versions, (1) (-DI), (2) (-mIş) and (3) neutral were used. Besides evidentiality, three other factors were tested: prior belief, polarity of the argument, and reliability (of the information source). Subjects had to indicate how strongly a story character would be convinced of an argument in a given combination of the four experimental factors. As a result, evidentiality, prior belief and reliability showed significant main effects. Furthermore, the significant interaction between evidentiality and reliability showed that these factors – despite considerable conceptual overlap – convey distinctive information to native Turkish speakers and make individual as well as joint contributions to reasoning processes in argumentative contexts.

The positive and negative findings of our series of studies are discussed in the light of the literature, with a special focus on the irritating irrelevance of evidential markers in (the development of) ToM.