Tibetic languages are known for their complex systems of evidentiality with markers for ego­phoric/par­ti­cipatory knowledge, coding highest epistemic authority of the epistemic origo (speaker in statements, addressee in information-seeking questions), setting this apart from mere observation (direct knowledge) of other persons. While there are discussions about whether egophoric knowledge is part of evidentiality, pragmatic variation or the attested high flexibility of the Tibetic systems is not accounted for, nor linked to the discourse function of egophoric marking.

I shall demonstrate that egophoric marking in Ladakhi is closely linked to the origo’s right to present a situation as exclusively personally known or as having greater epistemic authority than the addressee (see, however, Hanks 2014:5, who states “the social warrant for knowing is apparently the least likely to be encoded in particles or paradigms (I know of no cases)”). This right can be forsaken in order to distance oneself from a situation, using markers for mere observation. This right must be forsaken in the case of “mutual knowledge”, particularly for generic facts, when shared observations, and in some inclusive pural contexts:

(1) ŋaʧa-a ʒiŋ ɲis jot.

we.excl-aes field 2 have.egophoric

‘We have two fields.’ (Private)

(2) ŋaʧa-a ʒiŋ ʧikʧig-aŋ mi-ruk.

we.excl-aes field single-foc neg-have.visual

‘We don’t have any field (and it’s a shame)!’ (Distanced)

(3) ladaks-la mi maŋpo-a ʒiŋ jot-einok.

Ladakh-all people many-aes field have-epistemic

‘Many people in Ladakh have fields.’ (Generic)

(4) daŋ oɣo ɲēke alu tōn. / *tōn-pen.

yesterday we.incl both.erg potato extract.observed / *extract-egophoric

‘Yesterday, we both (you and me) took out the potatoes.’ (Inclusive)

Other pragmatic variations have to do, i.a., with the presence in hearing distance of the person talked about (respectful upgrading to egophoric) or with relationships between speaker and addressee, speaker and topic, and addressee and topic, that is, S may present information about P as egophoric to A, iff

S knows P sufficiently well, so that P belongs to S’s “Territory of Information” &

A does not know (much) about P &

the relationship between S and A is not closer than the relationship between S and P;

e.g., when talking to a family member about a neighbour one personally knows well, the
egophoric marker is not used.

methods: elicitation, participatory observation, philology, no particular linguistic theory.

Hanks, W.F. 2014. Foreword. Evidentiality in social interaction. In: J. Nuckolls & L. Michael (eds.), Evidentiality in Interaction. Amsterdam/ Philadelphia: Benjamins:1–12.

We explore how artists and carer-companions coordinate instruction sequences in arts-oriented activities organised for people living with dementia (PwD). Managing participant engagement in arts activities involves a wide range of interactional competences from the workshop facilitators. These can be topic specific - for example an ability to impart aesthetic appreciation and genre knowledge of the particular art form; but also an ability to organise active participation in the activities - for example in teaching the use of relevant tools and materials, or enabling participants to achieve particular outcomes. One feature of activities such as these is that they are at times instruction heavy, with the workshop leader eliciting particular actions from the workshop participants, while drawing on shared knowledge of what these should entail.

As a condition of dementia progresses, an afflicted person may however experience emerging gaps in what is oriented to as mundane, shared knowledge. The developing fractured knowledge ecology of the PwD presents a particular challenge to them and to the people with whom they interact. For the PwD, they must work to understand what is being required of them, and what knowledge is assumed to be shared. For their interactional partners, these must work to identify what knowledge is still accessible, and therefore what complexity of instruction a person is expected to be able to process.

Although instruction is a social activity with a great deal of complexity built into it, once acquired as a social practice, it can appear straightforward, and unproblematic for all involved. In the settings we study, however, following instruction can be very challenging. It requires of participants that they understand how the action they are being instructed to perform is implicated in the wider activity of the arts workshop, including knowing what target the activity is to achieve, knowing what is the instructed action, and knowledge of the formatting features that make up an instruction, including, for example, separating out what politeness conventions are being used, and what sociolinguistic formats for delivering a directive.

Where trouble occurs, this triggers diagnostic work on the part of the co-participants, who must identify that there is trouble, what the trouble might be, and what remedial work could address this. Our analysis unpacks how such remedial action is produced in a step-wise fashion, monitoring the PwD’s response at each new instruction format, and developing the instruction in response to the kind of response they receive.

The growing ubiquity of AI-based devices in everyday life and professional settings induces newly motivated scrutiny and an incentive to “describe how AI features in the world as it is” [3]. AI technology defines radiomics, the new field of medical imaging analysis based on extracting large-scale quantitative features by machine-learning algorithms [6]. One of the challenges in adopting radiomics for medical practice is the limited interpretability of the resulting models, leading to low confidence in the proposed diagnosis and treatment. While there have been notable improvements over the last decade, little is known about the physicians’ and researchers’ conduct while assessing and producing radiomics models. In this context, we focus on the situated use of an online radiomics platform QuantImage v2 [1] in collective trial sessions conducted at Centre hospitalier universitaire vaudois (CHUV), where pairs of novice users work with the platform together with a co-present expert. We employ ethnomethodological and conversation analytic (EM/CA) video analysis, which is a long-established approach to study healthcare and hospital settings [2]. We aim to specify the shared epistemic grounds that constitute common knowledge in radiomics’ work, as well as the contingencies of members’ activities [4] while they encounter troubles of substantive (in radiomics more generally) and procedural (in the platform specifically) nature.

Our findings show that knowledge of radiomics as a field and knowledge of the specific platform are deeply intertwined. QuantImage is featured as a representative instance of radiomics and learning how to use the platform is tied to learning about radiomics. Absence of fundamental background knowledge (e.g., of statistical performance measures) can make it impossible for the participants to proceed with the task on their own. Common-sense knowledge (e.g., about meanings of colors) can interfere with the visualization standards established in the professional domain. Participants’ skills in using the platform and their knowledge of radiomics are routinely displayed in assessments of resulting machine-learning models, in monitoring the platform’s pace of operation for possible problems, and in ascribing independent actions to the platform (e.g., to its algorithms). Reformulating confidence in AI as “practical trust” [5] that is at the very basis of social interaction can aid in identifying and explicating features of the socio-technical systems in which epistemic issues of understanding, explaining, and transparency are naturally embedded and situated as members’ matters.

[1] Abler, D. et al. 2023. QuantImage v2. https://doi.org/10.1186/s41747-023-00326-z

[2] Barnes, R. K. 2019. Conversation analysis of communication in medical care. https://doi.org/10.1080/08351813.2019.1631056

[3] Brooker, P. et al. 2019. The new ghosts in the machine. https://doi.org/10.5281/zenodo.3459327

[4] Garfinkel, H. 2022. Studies of Work in the Sciences. Routledge.

[5] González-Martínez, E., & Mlynář, J. 2019. Practical trust. https://doi.org/10.1177/0539018419890565

[6] Guiot, J. et al. 2022. A review in radiomics. https://doi.org/10.1002/med.21846