人机协同视域下英语口译课程学习评价生态的重构

Reconstructing the Learning Assessment Ecology of English Interpreting Courses from a Human-AI Collaborative Perspective

摘要: 在以大语言模型(Large Language Model, LLM)和自动语音识别(Automatic Speech Recognition, ASR)为代表的新一代人工智能技术迅速发展的背景下，翻译行业生态与口译人才能力结构正发生深刻变化，高校英语口译课程的学习评价体系面临前所未有的重构压力。现有研究多聚焦于智能技术赋能教学模式与课堂活动的设计，对于人工智能条件下学习评价范式如何整体转型、评价目标–主体–证据关系如何在“人机协同”视域下重构关注不足。本文在系统梳理教育评价范式演进与学习导向评价理论的基础上，结合人工智能介入教育的最新发展，揭示当前高校英语口译课程学习评价在目标与标准偏狭、主体结构单一、证据生成碎片化等方面的结构性困境。在此基础上，引入教育生态学与人机协同学习理论，提出以“学习证据链”为关键中介的英语口译课程学习评价生态重构框架，细化学习证据链的构成要素、数据来源、分析方法与呈现路径，并构建基于LLM与ASR的技术实现流程。进一步，以某高校英语专业《英汉交替传译》课程为例，构建基于学习证据链的人机协同评价教学情境，展示师–生–机多主体协同的互动方式，讨论其在课堂实践中的可能成效与潜在挑战。最后，从技术使用偏离教育初衷的风险、数据隐私与学术诚信、制度规范与素养培育等维度，探讨人机协同学习评价的伦理边界与治理路径。研究旨在为人工智能时代高校英语口译课程学习评价的理论探索与实践改革提供一个兼具规范性与操作性的分析框架。

Abstract: With the rapid advancement of a new generation of artificial intelligence technologies represented by Large Language Models (LLMs) and Automatic Speech Recognition (ASR), the ecology of the translation industry and the competence structure of interpreters are undergoing profound changes, placing unprecedented pressure on the reconstruction of learning assessment systems in university English interpreting courses. Existing studies primarily focus on how intelligent technologies empower teaching models and classroom activities, while paying insufficient attention to how assessment paradigms should be transformed as a whole under AI conditions, and how the relationships among assessment goals, agents and evidence can be reconfigured from a human-AI collaborative perspective. Drawing on a systematic review of the evolution of educational evaluation paradigms and theories of learning-oriented assessment, and incorporating recent developments of AI in education, this paper identifies several structural dilemmas in current learning assessment for English interpreting courses, including narrow assessment goals and standards, a single and teacher-dominated agent structure, and fragmented generation of assessment evidence. Building on educational ecology and human-AI collaborative learning theories, it then proposes a reconstruction framework for the learning-assessment ecology of English interpreting courses with the “learning evidence chain” as a key mediator, and further specifies its components, data sources, analytical methods and modes of presentation, together with a technical pipeline based on LLMs and ASR. Taking a university English interpreting course (Chinese-English consecutive interpreting) as an illustrative case, the paper designs a human-AI collaborative assessment scenario grounded in the learning evidence chain, and discusses the potential benefits and challenges of multi-agent collaboration among teachers, students and machines in classroom practice. Finally, from the perspectives of risks associated with the misuse and overreliance on technology, data privacy and academic integrity, as well as institutional regulation and literacy cultivation, it explores the ethical boundaries and governance pathways of human-AI collaborative learning assessment. The study seeks to provide a normative yet operational analytical framework for theoretical inquiry and practical reform of learning assessment in English interpreting courses in the age of artificial intelligence.