Scholarlink

<p>The rise of Generative AI (GenAI) tools, such as ChatGPT, is transforming translators’ information-seeking behavior (ISB), traditionally centered on web search. This study compares student translators’ ISB in web-based and GenAI-driven contexts using a literature-informed ISB analytical framework, developed from a systematic review of existing ISB theories and models, with a focus on time-related, query/prompt-related, and process-related aspects. To compare these two conditions, twenty-four student translators completed two tasks under each condition. Their on-screen activities were recorded and analyzed using a comparative approach, which involved evaluating differences in the three aforementioned aspects. The data were collected through screen recording, coded using NVivo, and analyzed using mixed-effects regression models to assess behavioral patterns and determine how these aspects vary between web-based and GenAI-based ISB. Findings reveal distinct patterns: 1) Time-related: GenAI-based ISB takes longer, with extended information-seeking durations, whereas web-based ISB is quicker and more efficient; 2) Query/prompt-related: Both focus on source comprehension, but GenAI-based ISB addresses complex, segment-level tasks with broader objectives, whereas web-based ISB handles immediate, word-level issues with narrower goals; 3) Process-related: GenAI-based ISB is dynamic, involving frequent switching with less depth, while web-based ISB is more linear and structured, supporting deeper exploration within online resources. Overall, GenAI-based ISB is dynamic and interactive, allowing broader exploration of translation tasks, yet it may sacrifice depth and lead to increased reliance. In contrast, web-based ISB is more structured and precise, well-suited for word-level tasks, but lacks the flexibility for more complex translation challenges. Theoretically, this study extends ISB frameworks by demonstrating GenAI’s dynamic yet reliance-prone model, while pedagogically, it highlights the need for balancing GenAI and web-based tools and fostering critical evaluation through multi-step tasks, error analysis, and cross-verification.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Student translators’ web-based vs. GenAI-based information-seeking behavior in translation process: A comparative study

  • 期刊论文

  • 发表日期：

    2025年4月5日

查看全文
  • Yuxing Cai,
  • Sha Tian

摘要

The rise of Generative AI (GenAI) tools, such as ChatGPT, is transforming translators’ information-seeking behavior (ISB), traditionally centered on web search. This study compares student translators’ ISB in web-based and GenAI-driven contexts using a literature-informed ISB analytical framework, developed from a systematic review of existing ISB theories and models, with a focus on time-related, query/prompt-related, and process-related aspects. To compare these two conditions, twenty-four student translators completed two tasks under each condition. Their on-screen activities were recorded and analyzed using a comparative approach, which involved evaluating differences in the three aforementioned aspects. The data were collected through screen recording, coded using NVivo, and analyzed using mixed-effects regression models to assess behavioral patterns and determine how these aspects vary between web-based and GenAI-based ISB. Findings reveal distinct patterns: 1) Time-related: GenAI-based ISB takes longer, with extended information-seeking durations, whereas web-based ISB is quicker and more efficient; 2) Query/prompt-related: Both focus on source comprehension, but GenAI-based ISB addresses complex, segment-level tasks with broader objectives, whereas web-based ISB handles immediate, word-level issues with narrower goals; 3) Process-related: GenAI-based ISB is dynamic, involving frequent switching with less depth, while web-based ISB is more linear and structured, supporting deeper exploration within online resources. Overall, GenAI-based ISB is dynamic and interactive, allowing broader exploration of translation tasks, yet it may sacrifice depth and lead to increased reliance. In contrast, web-based ISB is more structured and precise, well-suited for word-level tasks, but lacks the flexibility for more complex translation challenges. Theoretically, this study extends ISB frameworks by demonstrating GenAI’s dynamic yet reliance-prone model, while pedagogically, it highlights the need for balancing GenAI and web-based tools and fostering critical evaluation through multi-step tasks, error analysis, and cross-verification.