Bridging or reproducing inequity? Machine translation tools as disruptive participants in clinical interactions

Abstract

Language differences are a critical social determinant of health that can limit healthcare access, leading to adverse health outcomes when interpreters are unavailable. Machine translation tools (MTTs) may offer a pragmatic solution, but little is known about their interactional consequences. This exploratory study adopts an interactional analysis informed by Conversation Analysis (CA) and Interactional Sociolinguistics (IS) to explore two Neural MTTs (Google Translate, Microsoft Translator) in simulated South African case history interviews mediating English, isiZulu and Sesotho. Analysis reveals how MTTs may function as disruptive third participants. Technical failures and inaccuracies force participants into a time-consuming ‘MTT management project’ with interactional work organised across turns to manage the tool. This competing project recurrently disrupts the medical information gathering of the ‘case history project’, reorganising interactional architecture, halting progressivity and undermining intersubjectivity. Redirecting orientation to device management erodes patient-centred communication and risks reinforcing linguistic inequity. However, observed adaptive strategies highlight potential avenues for managing communication breakdowns when professional interpreters are unavailable.