Modeling listeners' perceptions of quality in consecutive interpreting: a case study of a technology interpreting event

Interpreting quality is shaped by multiple stakeholders, but listeners are central because interpreting output is ultimately for them. Prior work has largely emphasized product-oriented assessment of interpreting outputs and paralinguistic features, leaving a gap regarding listeners' own subjective factors in shaping perceived quality. This study targets that gap in consecutive interpreting (CI), where the interpreter is visible and impressions may be especially influential. Building on literature, the authors propose a model with six listener variables—quality expectations, perceived characteristics of the interpreter, experiences with CI, domain knowledge, perceived dependence on CI, and perceived communicative effect—to predict perceived CI quality and examine interrelations among these factors. The study asks: which subjective factors influence CI quality perceptions (RQ1), what causal relations exist among them (RQ2), and what implications arise for practice and training (RQ3). Eleven hypotheses posit direct effects on perceived CI quality and interplays among the variables, including expected negative effects of quality expectations and experiences (via mismatch or stricter standards) and a negative effect of domain knowledge, alongside positive effects of perceived interpreter characteristics, communicative effect, and perceived dependence.

Interpreting quality research comprises two broad lines. Product-oriented work treats quality as an idealized output and defines/assesses criteria such as content, language, and delivery, using methods including error analysis, checklists, rubric-based rating scales, multi-method scoring, comparative judgment, and emerging automatic assessment. These criteria and measures are applied across professional and training contexts and from different stakeholder perspectives. Interaction-oriented research, less developed, views quality as subjectively constructed in communicative interaction, especially by listeners. Two main strands there examine (a) listeners' quality expectations and (b) effects of paralinguistic features (accent, fluency, prosody, pauses, intonation, hesitation) on perceived quality. However, studies have focused more on output features than on listener variables. The present work emphasizes listener subjectivity, proposing six key variables—quality expectations, experiences with CI, domain knowledge, perceived dependence on CI, perceived interpreter characteristics, and perceived communicative effect—based on theory and prior empirical findings, and formulates a model to test their predictive roles and interrelations in CI.

Design: Case study with a questionnaire-based survey during a simulated CI-mediated technical conference, testing a structural model with PLS-SEM. Participants: 115 computer science students recruited; 107 valid responses (85 male, 22 female), ages 21–25, from late undergraduate to master’s level at Central South University, China. They varied in domain knowledge, CI experience, and English proficiency (hence dependence on CI). Materials: A 12-minute English presentation segment on “AI and Logistics,” adapted from the 2023 World Computing Conference, interpreted consecutively into Mandarin. The English speaker (Pakistani female, US-educated, fluent in English) delivered from a prepared transcript with slides. A professional male conference interpreter (Mandarin A, English B, 5 years’ experience in technological conferences) provided CI, seated beside the speaker in a lecture hall setup to simulate a real event. Measures: Two-part questionnaire. Demographics; 24 items covering seven constructs on 7-point Likert scales: Experiences with CI (3 items), Quality expectations (4), Domain knowledge (3), Perceived dependence on CI (3), Perceived characteristics of the interpreter (4), Perceived communicative effect (3), Perceived quality of CI (4). Items adapted from prior studies (e.g., Bühler, Kurz, Pöchhacker, Cattaruza & Mack; Amini et al.; Cheung; García Becerra; Chiaro; Kurz; Wang & Mu). Translation/back-translation used; expert review; pilot with 20 students showed Cronbach’s alpha 0.833–0.945. Procedure: Pre-event online survey one week prior measured Experiences with CI, Quality expectations, and Domain knowledge. On Dec 25, 2023, participants attended the live simulated talk with CI in a university lecture hall. After the talk, they completed a target-language comprehension test (10 minutes), then a post-event online survey measuring Perceived dependence on CI, Perceived characteristics of the interpreter, Perceived communicative effect, and Perceived quality of CI (~15 minutes). Incentives were offered for correct comprehension and survey completion. Data analysis: Descriptive statistics (SPSS 22) followed by PLS-SEM (SmartPLS 3). Measurement model assessed via factor loadings, Cronbach’s alpha, composite reliability (CR), average variance extracted (AVE), and discriminant validity (HTMT). Structural model fit via GOF, R², Q²; hypotheses tested with bootstrapping (5,000 resamples) to obtain path coefficients (β), t, and p values.

Descriptive results (7-point scales): Experiences with CI had the lowest mean (M=3.184, SD=1.519), indicating limited prior exposure. Highest means were Perceived communicative effect (M=5.835, SD=0.976) and Perceived dependence on CI (M=5.782, SD=1.342). Other constructs’ means exceeded 5, indicating high expectations, substantial domain knowledge, positive interpreter impressions, and favorable perceived CI quality. Measurement model: All item loadings >0.70; Cronbach’s alpha 0.833–0.945; CR >0.70; AVE >0.50; HTMT <0.85, supporting reliability, convergent, and discriminant validity. Model fit and predictability: GOF=0.563 (large). R²: Perceived quality of CI (PQC)=0.781; Perceived communicative effect (PCE)=0.416; Perceived characteristics of the interpreter (PCI)=0.232; Quality expectations (QE)=0.203. Q² values >0 (PQC=0.223; PCE=0.077; PCI=0.171; QE=0.147) indicate predictive relevance. Hypotheses (β, p):

• H4 PCI→PQC supported: β=0.320, p<0.001 (positive).
• H5 PCI→PCE supported: β=0.447, p=0.008 (positive).
• H6 EC→QE supported: β=0.189, p=0.010 (positive).
• H8 DK→QE supported: β=0.391, p<0.001 (positive).
• H9 DK→PQC supported: β=−0.232, p=0.001 (negative).
• H10 PDC→PQC supported: β=0.207, p=0.010 (positive).
• H11 PCE→PQC supported: β=0.247, p=0.012 (positive). Unexpected directions (not as hypothesized but significant positive effects):
• QE→PQC: β=0.448, p<0.001 (contrary to predicted negative H1).
• QE→PCI: β=0.482, p<0.001 (contrary to predicted negative H2).
• QE→PCE: β=0.297, p=0.025 (contrary to predicted negative H3).
• EC→PQC: β=0.109, p=0.031 (contrary to predicted negative H7). Overall, domain knowledge negatively predicts perceived CI quality; the other factors positively predict it. The strongest predictors of PQC are Quality expectations (β=0.448), Perceived characteristics of the interpreter (β=0.320), Perceived communicative effect (β=0.247), and Perceived dependence on CI (β=0.207). QE is itself positively predicted by EC and DK; PCI positively predicts PCE.

Findings affirm that greater domain knowledge leads to stricter evaluations and lower perceived CI quality, consistent with prior work. In contrast, five listener variables—quality expectations, perceived interpreter characteristics, perceived communicative effect, perceived dependence on CI, and experiences with CI—positively relate to perceived CI quality. Contrary to expectations that higher QE and more EC would heighten mismatch and dissatisfaction, positive paths suggest that, when interpreter performance is strong, higher expectations and greater experience can align with or be exceeded by performance, yielding higher perceived quality. QE also enhances PCI and PCE, indicating direct and indirect pathways by which expectations shape perceived CI quality. Interpreter visibility in CI amplifies the effect of first impressions (appearance, voice, delivery), which in turn drives perceived communicative effect. PCE’s positive link to PQC underscores that listeners judge quality by communicative success, not only linguistic detail. Perceived dependence on CI modestly elevates PQC, though its influence is contingent on other variables. Together, results depict a mutually reinforcing cycle among QE, PCI, PCE, and PQC, clarifying how subjective listener factors co-construct CI quality perceptions and informing strategies for professional practice and training (e.g., managing expectations, projecting a positive interpreter image, and optimizing communicative effectiveness).

The study proposes and tests a listener-centered model of CI quality perceptions incorporating six subjective variables. Results show: (1) Domain knowledge negatively predicts perceived CI quality, while quality expectations, perceived interpreter characteristics, perceived communicative effect, perceived dependence on CI, and experiences with CI positively predict it, with quality expectations the strongest predictor. (2) Quality expectations also positively predict perceived interpreter characteristics and perceived communicative effect, underscoring their foundational role and indirect contribution to perceived quality. (3) Domain knowledge and experiences with CI significantly raise quality expectations, and perceived interpreter characteristics drive perceived communicative effect. The work advances interaction-oriented perspectives on interpreting quality and offers practical guidance for interpreters and trainers to manage expectations, cultivate favorable first impressions, and enhance communicative effectiveness. Future research should broaden settings and modes, diversify listener populations, combine qualitative methods, study natural conference contexts, and integrate additional listener variables to enrich the model.

• Context and sample: Single simulated technical CI event with computer science students; generalizability to other interpreting modes (e.g., SI), settings (e.g., community/court), and broader listener populations is limited.
• Self-report bias: Heavy reliance on self-reported questionnaires; future studies should incorporate qualitative interviews/observations and triangulation.
• Simulated environment: Controlled simulation may not fully capture dynamics of natural conferences; field studies are needed.
• Scope of variables: Only six listener factors were examined; additional variables (e.g., cultural background, motivation, cognitive load) could provide a more comprehensive account.