A multi-dimensional analysis of interpreted and non-interpreted English discourses at Chinese and American government press conferences

The paper examines whether and how interpreted English discourse at Chinese government press conferences differs from non-interpreted original English discourse at American government press conferences in terms of communicative functions. Motivated by limitations of prior work that focused on isolated linguistic indicators and translation universals, the study adopts Biber’s multidimensional analysis (MDA) to capture co-occurrence patterns of linguistic features that reflect underlying communicative functions. It situates the inquiry in political press conference settings where consecutive interpreting is used and where governments communicate policy positions to domestic and international audiences. The study explicitly addresses three research questions: (1) Is interpreted speech at government press conferences different from non-interpreted speech from a communicative perspective? (2) What linguistic features distinguish interpreted from non-interpreted speech in terms of communicative dimensions? (3) What reasons account for these differences?

The review traces a shift from the search for translation universals (e.g., simplification, explicitation, normalization) to interpreting universals, noting inconsistent findings when isolated features are used. Studies such as Shlesinger and Ordan (2012), Kajzer-Wietrzny (2018), and others question the applicability of translation universals to interpreting, suggesting tendencies rather than laws and highlighting the role of communicative situations. Consequently, scholars have turned to Biber’s MDA to examine clusters of co-occurring features that index communicative functions. Prior MDA-informed work on interpreting (Li, 2014; Xiao, 2015; Xu, 2021; Zou and Wang, 2021) shows MDA’s value but often varies in feature selection or corpus design, limiting comparability. Research on Chinese government press conference interpreting has explored norms, interpreter roles, and estilistic tendencies, including target-orientation trends and gatekeeping. The present study addresses gaps by adhering rigorously to Biber’s (1988) six-dimension, 67-feature model and by comparing interpreted and non-interpreted discourses within the same communicative setting of government press conferences.

Design: A corpus-driven MDA following Biber (1988) with inferential statistics. Corpus: Two comparable subcorpora from 1998–2017. (1) IE: English consecutive interpretations of Chinese Premier’s press conferences during NPC/CPPCC sessions; (2) OE: original English transcripts of U.S. government press conferences (sourced from https://www.presidency.ucsb.edu). Token counts: IE 217,498; OE 317,117. The Chinese set includes interpretations of Premiers and Foreign Ministers; the U.S. set includes Presidents and Spokespersons. Analysis: Employed the Multidimensional Analysis Tagger (MAT; Nini, 2019) to annotate and compute z-scores for the 67 linguistic features and to derive dimension scores on Biber’s six dimensions: (1) involved vs. informational production; (2) narrative vs. non-narrative concerns; (3) explicit vs. situation-dependent reference; (4) overt expression of persuasion; (5) abstract vs. non-abstract information; (6) online information elaboration. MAT also reports closest text types. Dimension scores and feature z-scores for both subcorpora were exported to SPSS 20.0 for independent-samples t tests to assess differences between subcorpora. Four of the 67 features were not computed by MAT due to loading values < 0.35 (per MAT constraints).

- Text type classification: MAT classified OE as involved persuasion and IE as general narrative exposition, indicating a fundamental register difference. - Dimension-level differences: Significant differences on 5 of 6 dimensions (all p < 0.001) except Dimension 4 (p = 0.552). - Dimension 1 (Informational vs. involved): OE is highly involved (score 10.35) while IE is highly informational (score −7.37), showing IE’s higher information density and proximity to literate style. - Dimension 2 (Narrative vs. non-narrative): Both are non-narrative (negative scores), with IE more strongly non-narrative (−2.44 vs. OE −1.64). OE uses more past tense; IE favors present perfect for narrative events. - Dimension 3 (Explicit vs. situation-dependent reference): IE is more explicit/context-independent (IE 7.32; OE 1.86), indicating higher referential clarity and integration. - Dimension 4 (Overt expression of persuasion): No significant difference (OE 3.36; IE 3.14; p = 0.552), suggesting comparable persuasiveness, though achieved via different feature patterns. - Dimension 5 (Abstract vs. non-abstract): IE slightly more abstract (0.35) than OE (−0.85), though overall abstraction remains modest. - Dimension 6 (Online informational elaboration): Both show informational elaboration; OE slightly higher (2.25 vs. IE 1.04). IE prefers phrasal coordination (integration) over dependent clause elaboration. - Co-occurrence patterns: OE overuses features linked to involvement and spoken style (personal pronouns, private verbs, contractions, modals, discourse particles). IE overuses features linked to information density and integration (nouns, longer words, prepositional phrases, attributive/predicative adjectives), greater explicitness (nominalizations, phrasal coordination), and modestly higher abstraction (by-passives, WHIZ deletions, conjuncts).

Findings address RQ1 by confirming that interpreted and non-interpreted press conference discourses differ communicatively despite similar settings, with IE more informational, explicit, and abstract, and OE more involved and colloquial. For RQ2, specific linguistic feature clusters on Biber’s dimensions account for these differences: IE’s noun-heavy, adjective-rich, coordination-based integration versus OE’s pronoun- and clause-rich involvement and elaboration. For RQ3, three explanatory perspectives are advanced: (a) communicative context and cultural tenor: Chinese government press conferences, as formal institutional events aiming at trustworthy information dissemination, favor informative, explicit, and abstract styles; (b) norms shaping interpreter choices: adherence to adequacy and acceptability leads to directness via necessity modals, clearer cohesion (conjuncts), and efficient, integrated structures; (c) interpreters’ habitus and modes: consecutive interpreting in live-broadcast, high-stakes settings fosters risk avoidance, referential clarity, and compressed, well-planned integration. Together, these factors explain why IE exhibits higher information density, greater referential explicitness, more non-narrative content, slightly higher abstraction, and slightly lower online elaboration while remaining as persuasive as OE.

The study demonstrates that communicative functions, captured through Biber’s MDA, more effectively distinguish interpreted from non-interpreted political discourse than isolated linguistic indicators. Interpreted English at Chinese government press conferences, despite being orally delivered, aligns more with literate, information-dense, explicit, and modestly abstract styles, with strong non-narrative focus and integrated sentence planning, and achieves persuasion comparably to original English political discourse. MDA proves effective for revealing stylistic and functional differences in similar communicative settings through co-occurring feature patterns and enhances cross-study comparability. Future research may examine diachronic changes in political discourse, interpreter idiosyncrasies, and cross-language comparisons of interpreted English.

The article does not present a dedicated limitations section. Notes relevant to interpretation include: MAT did not compute 4 of 67 features due to loading values < 0.35; the corpus is restricted to government press conferences (1998–2017) and to English interpretations of Chinese press conferences versus original English from U.S. press conferences, which may limit generalizability beyond this setting. Suggested future work on temporal change, interpreter idiosyncrasy, and cross-language comparisons implies potential limitations in temporal coverage, individual interpreter effects, and language-pair generalization.