This study explores the impact of human-AI collaboration on the quality of AI-translated legal documents. Focusing on Chinese-English legal translations produced by a neural machine translation (NMT) system and a Large Language Model (LLM), the study compares four translation workflows with varying degrees of human-AI collaboration: (1) no human input: NMT translation alone; (2) low human input: LLM translation with minimal human instruction; (3) medium human input: LLM translation with context-specific instructions; and (4) high human input: LLM translation with context-specific instructions and a human-verified glossary. The translations were evaluated using automatic metrics (BLEU and chrF++) and manual error analysis. Results show that automated legal translation quality improves with increased levels of human input, with the best quality achieved from a detailed prompt carrying context-specific instructions and a bilingual glossary. However, despite the quality improvements due to increased human input, all the automated translations still fell short of professional human translations, especially in handling specialized legal terminology and complex syntax. These results suggest that full automation is not yet feasible in high-stakes domains such as legal translation, underscoring the irreplaceability of human translators and the importance of human-AI collaboration in guaranteeing adequate translation quality in the legal context.
Recent advances in Neural Machine Translation (NMT) have reshaped the language services industry, with the number of language service providers offering machine translation services nearly doubling between 2018 and 2026 (EUATC, 2026. ELIS 2026 results & report. https://euatc.org/eu-relations/elis-2026-report/). Human post-editing has become standard practice, raising questions about how NMT is influencing translators' workloads and workflows. While its application in high-stakes domains, such as medical and legal translation, has grown, concerns remain regarding accuracy and ethics, where human oversight is essential. This paper focuses on the experiences of legal translators in the UK through a small-scale survey. It explores three dimensions: (a) translators' use of NMT in their professional practice, (b) participation in available training opportunities, and (c) shifting client expectations regarding accuracy and remuneration in specialized domains. Results show that NMT adoption has been more common among the less experienced, post-2010 cohort of legal translators. However, uptake of official training opportunities was low among participants. Several respondents reported concerns about workflow changes, rising quality expectations, and pressure to meet unrealistic client demands.
This research utilizes framing theory to analyze how the Iranian Students News Agency (ISNA) reshapes news reports originally produced by US and UK sources regarding the Palestinian-Israeli conflict that commenced on October 7, 2023. The study examines a parallel corpus of 81 English-language news articles from Axios and The Guardian, and their Persian translations published by ISNA. After examining the corpus, four major frames were identified: political, responsibility, conflict, and moral/ humanitarian. ISNA intensified the depiction of Israeli aggression, downplayed critical portrayals of Hamas, and foregrounded the United States' culpability in the conflict through lexical shifts, omissions, and additions. Furthermore, ISNA's headlines recast policy critique as a moral-blame narrative. This framing constructed a narrative where Iran's geopolitical stance was aligned with advocacy for the Palestinian cause; simultaneously, the moral standing of the U.S. and its allies was undermined. This study demonstrates that news translators are not passive conduits but active agents who reshape source narratives to align with domestic ideological imperatives by showing how translation operates as an active framing tool.
Audiovisual translation has gained traction in recent years, being one of the most researched areas within translation studies. While contemporary practices have been discussed from a variety of vantage points, historical research on AVT is still at an early stage of development. Most studies into the history of AVT have been devoted to intertitling as well as the genesis of either subtitling or dubbing and exemplified by isolated case studies with no systematic approach. This paper presents the results of an archival study on Polish subtitling done in the 1930s, demonstrating the gradual development of the modality, starting from intertitles typical of the 1920s, moving through the transition year of 1930 to subtitling in the 30s that resembled contemporary practices. It offers a systematic overview of the interwar film translation practice in Poland with special reference to subtitling and indicates subsequent improvements that led to the standardization of early subtitling.
Dependency distance, the linear distance between two syntactically related words, has shown promise in measuring syntactic complexity and processing load in interlanguage research, psycholinguistics, translation studies, and into-A interpreting. Drawing on a large-scale learner corpus of Chinese - English consecutive interpreting, the present study explored the potential of dependency distance as an indicator for cognitive load in learners' into-B interpreting. A negative binomial regression was conducted to examine the relationship between cognitive load (measured by the frequency of disfluencies) and both dependency-based and traditional linguistic complexity indices (i.e. source-text mean dependency distance, target-text mean dependency distance, source-text mean clause length, target-text mean clause length, source-text lexical density and percentage of numerals). The results showed that: (1) longer target-text mean dependency distance significantly increased disfluency frequency, whereas source-text mean dependency distance did not; (2) mean clause length in both source and target texts had no significant effect; and (3) percentage of numerals was reaffirmed as significant predictors of disfluencies. These findings provide preliminary evidence that dependency distance is a more sensitive indicator for cognitive load in learners' into-B interpreting than traditional syntactic complexity measures, aligning with decay - and interference-based accounts of dependency distance rather than expectation-based facilitation.
Suspense, as a textual element, plays a central role in audiovisual storytelling, yet its subtitling has received limited attention from a multimodal perspective. While previous research has focused primarily on verbal strategies in literary texts, suspense in audiovisual texts emerges from the interaction of verbal, paraverbal, and non-verbal modes. This article proposes a framework for analyzing suspense as a multimodally constructed translation problem in AVT, integrating insights from narratology, film studies, multimodality, and descriptive translation studies, including a problem-solution approach. The framework is applied in the analysis of three examples that show how semiotic modes such as dialogue, onscreen writing, sound design, music, and visual composition interact to create suspense, and how subtitling solutions may reproduce, modify, or disrupt these textual elements. The study underscores the need for greater multimodal awareness in subtitling practice and highlights implications for future research and for the development of AI-assisted subtitling tools.
This study investigated whether and how subtitles on educational videos influence attentional and cognitive processing by examining mind wandering (MW), comprehension, and cognitive load (CL) - bridging translation studies with constructs and indicators from educational psychology and cognitive science. English L2 university students (n = 102) were assigned to one of three experimental conditions: watching a lecture in English without subtitles (NS, control), translated Mandarin Chinese subtitles (TLS, L1), or same-language English subtitles (SLS, L2). Using experience sampling, content learning and psychometric indicators, we measured MW rates, perceived CL, and comprehension scores, while controlling for working memory capacity and English proficiency. Our results indicated that TLS reduced MW (vs NS and SLS) and improved comprehension (vs NS), while SLS reduced CL (vs NS and TLS) but did not enhance comprehension. Our findings support the previously reported subtitle principle and suggest a dynamic between MW, CL, and comprehension.
This paper examines the paratextual discourse that shaped English translations of Homer from Chapman to Butler, treating prefaces as theoretical interventions rather than ancillary materials. Although Homeric translation has been extensively studied, the prefatory spaces that codify principles, justify departures from literalism, and anticipate reception remain understudied. To address this gap, the study analyzes major translators' prefaces: Chapman's defense of rhetorical freedom, the Dacier-de La Motte debate, Pope's domestication through heroic couplets, Newman's archaism against Arnold's grand style, and Butler's advocacy of prose clarity. The analysis draws on Genette, Skopos theory, Nord's functionalism, and Chesterman to show how translators balance fidelity, function, and ethics. Prefaces reveal how translators define their methods and present Homer to new cultural contexts. Ultimately, translation history emerges as inseparable from its paratexts: prefaces reveal how translators articulate their methods and recast Homer as a continually reinvented cultural palimpsest.
