This study explores how the positioning of dialogue interpreters is shaped in mediatedinteraction through the combined investigation of two main units of analysis, i.e. assessments and gaze. The data used consists of a small corpus of authentic, videorecorded, mediated interactions in English and Italian. These encounters take place in pedagogical settings; in particular, the specific type of institutional talk analysed is that of mediated parent-teacher meetings, which represents uncharted territory for interpreting studies. An interdisciplinary approach encompassing conversation analysis and studies on non-verbal communication is adopted to explore how interactants orient to both verbal and non-verbal activities (mainly gaze) in the production and monitoring of each other's actions, in the initiation and maintenance of social encounters, and in the co-construction of meaning and participatory framework.As for the verbal dimension, this thesis focuses on assessments, given that evaluative talk characterises the interactions under scrutiny. In particular, some tendencies (namely upgrading and downgrading renditions) in the way interpreters handle utterances embedding evaluative assessments have been identified, explored and linked to issues of identity and epistemic authority. One of the most innovative aspects of this work lies in the exploration of how positioning is realised not only verbally, but also nonverbally, by accounting for non-verbal features in the analysis of verbal interaction. Although non-verbal features have been recognised as part and parcel of human social interaction as well as important vectors of meaning and co-ordination (e.g. Goodwin 1981; Kendon 1990), their sequential positioning in relation to the production of the ongoing flow of talk and their use by interpreters to complement/replace specific verbal features is uncharted territory for interpreting studies. Since the groundbreaking work by Lang (1976, 1978), little research has integrated gaze in the analysis of the interpreter's (and participants) verbal output (e.g. Wadensjö 2001; Bot 2005). To enable its investigation, gaze is systematically encoded alongside specific conversational cues via the ELAN software, which interfaces audio-video input in a user-friendlyhypertextual transcription. A specific gaze-encoding system has been developed for triadic interaction, building on Rossano's (2012) one for dyadic interaction. These symbols have been mapped onto the verbal transcript of specific sequences, with a view to investigating how gaze is used as an interactional resource in conjunction with verbal behaviour when producing such sequences. Through analysis of the actions performed via talk and gaze, the thesis investigates how displays of knowledge and epistemic authority are achieved and the impact of the interpreter's shifting positioning on the unfolding interaction. The micro-analysis of transcripts is placed within a macro-analytical framework to explore whether interpreters work as intercultural mediators when they display an engaged behaviour and act as ratified participants. Findings show that the specific moves isolated, although trying to establish a common ground with the mothers, do not seem to contribute to participants' empowerment and participation, thus suggesting the need for a more nuanced conceptualisation of intercultural mediation.