Interpreter visibility in press conferences: a multimodal conversation analysis of speaker-interpreter interactions

The paper revisits the long-debated issue of interpreter visibility, originally problematized by Angelelli (2004), who contrasted professional norms of interpreter 'invisibility' with observable practices. Prior research has largely focused on visibility initiated by interpreters themselves across medical, legal, educational, and press conference contexts, while speaker-initiated visibility has been comparatively underexplored and mostly examined in simultaneous interpreting at the European Parliament. Addressing this gap, the study investigates how speakers make interpreters visible in a Chinese political press conference context using consecutive interpreting at the American Institute in Taiwan (AIT). It asks: (1) To what extent and in what circumstances do interpreters at AIT become visible? (2) How can speaker-interpreter interactions contribute to interpreter visibility and the communicative environment? The work underscores the importance of context and multimodality in understanding interpreter roles and interactional dynamics.

Research predating Angelelli (2004) had already questioned interpreter 'invisibility' (e.g., Anderson, 1976/2002; Wadensjö, 1998), with discourse-in-interaction studies examining turn-taking, meaning negotiation, and footing in institutional encounters. Angelelli (2004) defined visibility as interpreters establishing a distinct speaking position, observable when managing turns, filtering information, or aligning with parties. Subsequent debates (e.g., Ozolins, 2016; Downie, 2017) discussed professional norms versus actual interpreter practices. Three strands characterize interpreter-initiated visibility: (a) non-renditions used for coordination or, more controversially, advice/filtering, which can disrupt direct communication; (b) text ownership (partial/total) that varies visibility along a continuum; and (c) intercultural mediation, where interpreters handle 'rich points' and evaluative assessments to prevent misunderstanding, while acknowledging risks of overvisibility. Speaker-initiated visibility has been mainly reported in conference settings (Diriker, 2004; Duflou, 2012, 2016; Bartłomiejczyk, 2017), where speakers commend, critique, or otherwise reference interpreters, eliciting interpreters' third-person self-references or defensive non-renditions. Bartłomiejczyk (2017) identified six recurring topics in EP debates: appreciation, doubts, reminders of constraints, criticisms, announcements of difficulties, and apologies. The present study extends this line by focusing on consecutive interpreting in an Asian diplomatic/press context.

Design: Multimodal Conversation Analysis (CA) focusing on speaker references to interpreters and their output, integrating verbal and nonverbal resources (gaze, gesture, posture) following CA and multimodal conventions (Sacks et al., 1974; Goodwin, 1981; Mondada, 2007). Corpus and setting: Eight AIT press conferences (2006–2012) comprising approximately 9 h 24 min, 34,943 English words and 35,691 tokenized Chinese words. Three AIT Directors (Stephen Young, William Stanton, Christopher Marut) spoke in English; three professional female Taiwanese interpreters rendered consecutively into Chinese. The AIT context features diplomatic goals with a relatively relaxed press environment and side-by-side seating, enhancing physical and interactional visibility. Analytical steps: (1) Data delimitation: manually identified all speaker references to interpreters in videos (excluding moderator’s introductory remarks about interpreting). (2) Units of analysis: segmented 98 sequences, each comprising a speaker’s reference and the interpreter’s potential response, ending when interpreter and speaker roles realign (no longer distinguishable as separate speaker positions). Each unit tagged with video time codes. (3) Annotation/transcription: Imported videos into ELAN (Wittenburg et al., 2006). Annotated multi-tier verbal/nonverbal cues (director and, when visible, interpreter), focusing on gaze, facial expression, hand gestures, head movements, and body orientation. Exported to MS Word and transcribed with Jefferson (1983) and Mondada (2007) conventions to synchronize multimodal semiotics with utterances. (4) Topic identification: Conducted micro-analytic review to assign a single overarching topic to each sequence, and quantified topic frequencies across the corpus. Ethical/data notes: Public-domain AIT recordings; excerpts included; full transcripts currently unavailable due to ongoing research.

Quantitative: Across 9 h 24 min, speakers referenced interpreters or their output 98 times (~1 mention every 5.8 min), far more frequent than European Parliament debates reported by Bartłomiejczyk (2017). Six topics emerged with frequencies: Repair 31.6% (n=31); Confirmation 25.5% (n=25); Humor 21.4% (n=21); Care 8.2% (n=8); Reminder 7.1% (n=7); Appreciation 6.1% (n=6). Qualitative: - Appreciation: Speakers explicitly thanked interpreters, sometimes by name, accompanied by gaze and touch (e.g., shoulder pat), rendering interpreters salient; interpreters produced interactional non-renditions (e.g., declining to translate praise), demonstrating footing shifts and text ownership. - Reminder: Directors reminded journalists to shorten questions or pause for interpreting, sometimes humorously, turning to the interpreter and acknowledging her presence; interpreters responded with first-person non-renditions requesting cooperation. - Confirmation: Bilingual directors monitored completeness (e.g., adding 'and other targets of opportunity'), using gaze shifts to the interpreter and script checks; interpreters acknowledged and supplied omitted items, becoming visible through overlap and floor-holding markers. - Repair: Directors corrected slips (e.g., 'Ireland' vs. 'Iran'); interpreters apologized and issued corrective non-renditions, assuming ownership to ensure accuracy; nonverbal monitoring (head/gaze) directed audience attention to the interpreter. - Humor: Directors joked and sometimes said 'don't translate this' while laughing/gesturing; interpreters engaged with creative non-renditions (e.g., Photoshop quip), prompting playful regulatory gestures from the director and audience laughter. - Care: Directors expressed consideration (e.g., 'drink some water'), with thumbs-up and passing water, publicly highlighting the interpreter. Overall, nonverbal resources (gaze, gestures, touch, body orientation) worked with verbal references to co-construct interpreter visibility and a relaxed communicative environment while supporting accuracy.

Addressing RQ1, interpreters at AIT became visible frequently and across six recurrent, context-dependent topics, with repair and confirmation most common, reflecting directors’ bilingual monitoring in a diplomatic press setting with consecutive interpreting and side-by-side seating. Humor, care, reminders, and appreciation further contributed to visibility, often amplifying a collegial, relaxed atmosphere. Addressing RQ2, sequential speaker–interpreter interactions, including interpreters’ non-renditions and footing shifts, showed interpreters as active co-participants who can serve as ice-breakers and rectifiers, enhancing both rapport and informational accuracy. Nonverbal semiotics were integral to these processes, often directing audience attention to the interpreter and indexing stance or coordination without explicit verbalization. The findings nuance interpreter role conceptualizations by emphasizing contextual variability and collaboration with speakers, and they demonstrate the analytical value of multimodal CA for capturing interactional mechanisms in conference/press settings traditionally assumed to limit interpreter agency.

The study demonstrates, via multimodal conversation analysis of AIT press conferences, that speakers regularly bring interpreters into view through varied references, and interpreters respond as active co-constructors of discourse. The sequential organization and layering of verbal and nonverbal resources contribute to interpreter visibility, interpretive accuracy, and a relaxed communicative climate. The work challenges expectations of political/diplomatic interpreter invisibility and underscores the need for context-sensitive analysis of interpreter roles. Future research should broaden datasets beyond a single predominant speaker, triangulate with perception data from interpreters and users, and further extend multimodal CA to diverse conference contexts to compare visibility patterns and role enactments.

Examples are limited and, due to data availability, findings on speaker references are largely drawn from one director (Stephen Young) across six press conferences, raising concerns about idiosyncrasy and generalizability to the broader AIT context. Full annotated transcripts are not yet publicly available. A fuller account of interpreter visibility would benefit from additional datasets and perception-based evidence from interpreters and interpreting users.