Abstract
Research on Automatic Speech Recognition (ASR)-assisted interpreting has concentrated on simultaneous conference interpreting, leaving remote public service settings underexplored. This article reviews methodological approaches to ASR-assisted interpreting research and presents two pilot studies on remote legal and medical interpreting. In the legal pilot, two professional interpreters completed ASR and non-ASR tasks in simultaneous and consecutive modes; ASR was associated with higher NTR accuracy and lower weighted error scores, especially in consecutive tasks, while interviews showed selective transcript use alongside style- and attention-related trade-offs. In the medical pilot, four trainee interpreters completed four ASR display conditions; full transcripts and ASR-fed summaries were associated with lower perceived workload, while the eye-tracking component exposed design issues requiring refinement. Together, the pilots show how domain-specific conditions shape research design, user responses, and the methodological challenges of studying AI-assisted interpreting in Public Service Interpreting.