Research on translated language highlights its distinct characteristics compared to both source and target languages. Most studies focus on translation universals across language pairs, examining lexical, syntactic, semantic, and discursive levels. Interpreted language, or "interpretese," has received less attention, particularly in political contexts. Previous research has attempted to identify universal features using linguistic indicators, but more recent empirical work employs multivariate analysis of comparable corpora of interpreted and non-interpreted texts. Multi-dimensional analysis (MDA), based on Biber's model, is increasingly used in interpreting studies because it identifies differences based on communicative dimensions rather than pre-selected linguistic features. This study uses MDA to distinguish interpreted and non-interpreted discourse in a political setting, exploring the factors contributing to these differences.

Studies on translation universals, advanced by corpus-based translation studies, have identified linguistic features overused or underused in translated texts compared to originals. However, the features (simplification, explicitation, normalization, leveling) vary across studies, suggesting interpreting universals may be tendencies, not laws. Some research contradicts the assumption of simplified interpreted language, indicating that interpreting universals are context-dependent. This leads to the adoption of Biber's MDA, which analyzes co-occurrence patterns of linguistic features to understand communicative functions of interpreted discourse in specific settings. Previous MDA studies on interpreted discourse show mixed results depending on factors like language pair, interpreting mode, and corpus size. This study leverages Biber's MDA to analyze interpreted and non-interpreted political discourse, focusing on government press conferences to improve comparability and control for context.

This study employs a corpus-driven approach using Biber's MDA. The comparable corpus includes two subcorpora: (1) transcripts of Chinese-English consecutive interpretations from Chinese Premier's press conferences (1998-2017) and (2) original English transcripts of American government press conferences from the same period. The Multi-dimensional Analysis Tagger (MAT) software was used to analyze 67 linguistic features across Biber's six dimensions. MAT generated dimension scores based on z-scores of feature frequencies. These scores were then imported into SPSS 20.0 for Student's t-tests to compare the two subcorpora. The use of MAT ensures comparability and replicability with previous studies. The selection of American press conferences as a comparable corpus reflects the similarity in communicative setting, speaker identity, spontaneous speech delivery, and time frame.

MAT identified "involved persuasion" as the closest text type for the original English (OE) subcorpus and "general narrative exposition" for the interpreted English (IE) subcorpus. This unexpected difference in text type suggests different communicative functions. Student's t-tests revealed significant differences between the two subcorpora on five of six dimensions (p < 0.001): * **Dimension 1 (Informational vs. Involved Production):** OE showed high involvement, while IE was highly informational, indicating a shift towards written register characteristics in interpreted discourse. * **Dimension 2 (Narrative vs. Non-Narrative Concerns):** Both showed non-narrative focus, but IE had significantly more non-narrative content. * **Dimension 3 (Explicit vs. Situation-Dependent Reference):** IE demonstrated more context-independent reference than OE. * **Dimension 5 (Abstract vs. Non-Abstract Information):** IE was more abstract than OE, although the difference was less pronounced than other dimensions. * **Dimension 6 (On-line Informational Elaboration):** OE showed slightly higher information elaboration than IE. Dimension 4 (Overt Expression of Persuasion) showed no significant difference, suggesting both discourses effectively conveyed persuasion, albeit through different linguistic means.

The findings challenge the notion of interpreted discourse as simply a translated version of the source text. The differences in text type and communicative dimensions highlight the influence of various factors. The formal nature of government press conferences as institutional discourse contributes to the abstract, informative style of interpreted speech. Interpreting norms, particularly the emphasis on adequacy and acceptability, affect the interpreters' choices. Furthermore, the interpreters' habitus, shaped by live broadcast modes and risk aversion, leads to a preference for clarity and cohesion in interpreted discourse. The study reveals that seemingly similar communicative contexts can yield drastically different discourse patterns due to interplay between linguistic and non-linguistic factors.

This study demonstrates that communicative functions, not pre-defined linguistic features, are key to understanding interpreted discourse. The unique style of interpreted discourse in government press conferences is distinguished by its higher information density and integration, clearer reference, and more formal style compared to non-interpreted counterparts. These features stem from the nature of the discourse, interpreting norms, interpreters' habitus, and the working/broadcasting mode. MDA proves highly effective in distinguishing these differences, offering a more nuanced understanding of interpreted discourse. Future research could explore the impact of temporal change in political discourse, interpreter idiosyncrasies, and cross-linguistic comparisons of interpreted English.

The study's focus on Chinese and American government press conferences limits generalizability to other contexts. The corpus size, while substantial, may not fully capture all variations in interpreted discourse. The reliance on Biber's 1988 model might benefit from incorporating more recent developments in MDA. Subjectivity in qualitative analysis of examples should be acknowledged. Investigating the role of individual interpreter styles could provide further insight.