Rethinking Higher Education/Chapter 5/en-zh

Language: EN · ZH · EN-ZH · ← Book

📌 Hinweis (Stand 8.5.2026): Diese Seite wurde strukturell überarbeitet, damit jeder Absatz seinen eigenen Tabellen-Row hat. Bisherige chinesische Übersetzungen wurden automatisch zugeordnet — die Zuordnung ist nicht in jedem Fall korrekt. Bitte prüfen Sie die rechte Spalte und verschieben/korrigieren Sie die ZH-Übersetzungen, falls sie nicht zum DE-Absatz passen. Bei nicht übersetzten Absätzen steht (zu übersetzen).

English (Source)	中文 (Übersetzung)
	(zu übersetzen)
Language: EN · ZH · EN-ZH · ← Book	(zu übersetzen)
	(zu übersetzen)
Learning a Foreign Language with and without AI: An Empirical Comparative Study	(zu übersetzen)《使用与不使用人工智能的外语学习：一项实证比较研究》
Martin Woesler	(zu übersetzen)吴漠汀
'Abstract'	(zu übersetzen)摘要
This study compares the self-reported learning outcomes, motivations, and attitudes of 133 Chinese university students learning a foreign language — 85 in an AI-assisted group and 48 in a traditional human-teacher group — over a period of approximately one month. Drawing on a comprehensive survey instrument with 126 variables covering demographics, learning methods, sensory modality preferences, attitudes toward AI in education, and self-assessed improvement across ten language skill areas, the study finds a complex picture that challenges both techno-optimistic and techno-pessimistic narratives. The human-teacher group reported higher overall improvement (63.2% vs. 51.9%), yet the AI group reported greater gains in speaking and listening — precisely the interactive skills that AI chatbots are designed to practise. Both groups expressed strong preference for human teachers, but the AI group simultaneously valued AI’s availability, speed, and pressure-free environment. Attitudes toward AI autonomy were cautious in both groups: over 70% agreed that AI needs ethical control, and fewer than 20% endorsed AI dominance over humans. These findings contribute to the growing literature on AI in language education and are discussed in relation to the qualitative findings of Fang Lu (this volume) and the philosophical framework of Ole Döring (this volume).	(zu übersetzen)本研究通过一项为期约一个月的追踪调查，对比了133名中国大学生在外语学习中的自陈学习成果、动机与态度。其中85名学生属于人工智能辅助组，48名学生属于传统人类教师组。研究采用包含126个变量的综合调查问卷，涵盖人口统计学特征、学习方法、感官模态偏好、对教育中人工智能的态度，以及在十个语言技能领域的自评提升情况。研究发现呈现出复杂的图景，对技术乐观主义与技术悲观主义两种叙事均构成挑战。 数据显示，人类教师组报告的总体提升比例更高（63.2% vs. 51.9%），但人工智能组在口语和听力方面的进步更为显著——而这恰恰是人工智能聊天机器人被设计用来训练的交互技能。两组受试者均强烈倾向于选择人类教师，但人工智能组同时认可了AI的随时可用性、响应速度及无压力环境。两组对人工智能自主性的态度均持谨慎立场：超过70%的受试者认为需要对AI实施伦理管控，仅不足20%支持AI对人类的主导地位。这些发现丰富了人工智能语言教育领域的研究成果，并与方璐（本卷）的质性研究结果及Ole Döring（本卷）的哲学框架进行了对话式探讨。
Keywords: AI-assisted language learning, comparative study, foreign language education, human-AI interaction, digital education, sensory modalities, student attitudes, China, European Union, complementarity thesis	(zu übersetzen)关键词：​ 人工智能辅助语言学习；比较研究；外语教育；人机交互；数字教育；感官模态；学生态度；中国；欧盟；互补性论题
'1. Introduction'	(zu übersetzen)1.介绍
The integration of artificial intelligence into language education has moved from speculative futurism to daily practice with remarkable speed. Chinese university students in 2025 routinely use AI chatbots — ChatGPT, Kimi, DeepSeek, Doubao — as conversation partners, pronunciation coaches, grammar checkers, and vocabulary tutors. Yet the empirical evidence for whether AI-assisted language learning produces better outcomes than traditional human instruction remains surprisingly thin. Most existing studies are small-scale, focus on a single AI tool, or measure outcomes over very short periods. What is missing is a comparative study that examines not only learning outcomes but also the motivational, attitudinal, and perceptual dimensions of AI-assisted versus human-taught language learning.	(zu übersetzen)人工智能融入语言教育的速度之快，已使其从充满想象的未来图景迅速转变为日常实践。至2025年，中国大学生已普遍将ChatGPT、Kimi、DeepSeek、豆包等AI聊天机器人用作会话伙伴、发音教练、语法校对员和词汇导师。然而，关于人工智能辅助语言学习是否优于传统人工教学，其实证依据仍出人意料地匮乏。现有研究大多规模有限，或聚焦于单一AI工具，或仅在极短时间内测量学习效果。当前尤为缺失的是一项比较研究：它不仅应考察学习成果，还需深入探究人工智能辅助与传统课堂教学在学习动机、学习态度及感知体验等维度的差异。
This study addresses that gap. We surveyed 133 Chinese university students — 85 who chose or were assigned to learn a foreign language with AI assistance, and 48 who learned with human teachers — after approximately one month of study. The survey instrument, comprising 126 variables, captures demographics, prior language knowledge, daily study time, reasons for group choice, AI usage methods, feedback quality perceptions, self-assessed improvement across ten specific skill areas, the importance of twelve sensory and social modalities in learning, and attitudes toward fourteen aspects of AI in education and society.	(zu übersetzen)本研究旨在填补这一空白。我们对133名中国大学生展开了调查——其中85人自主选择或通过分配进入人工智能辅助外语学习组，其余48人则在人类教师指导下学习——调研时段覆盖约一个月的学习周期。所使用的调查问卷包含126个变量，系统采集了受试者的 demographic 信息、既有语言基础、日均学习时长、组别选择动因、AI使用方式、对反馈质量的感知、在十个具体技能领域的自评提升程度、学习中十二种感官与社会模态的重要性权重，以及对教育与社会中人工智能十四个维度的态度倾向。
Our findings are situated within a growing body of work on digital education in China and Europe, including the qualitative case studies of Fang Lu (this volume), who examined AI’s effects on critical thinking in Chinese language courses at Boston College, and the philosophical analysis of Ole Döring (this volume), who interrogates the conceptual foundations of „artificial intelligence“ in pedagogical contexts. Where Fang Lu provides depth through individual cases and Döring provides philosophical breadth, we contribute breadth through quantitative comparison across a substantial participant pool.	(zu übersetzen)本研究立足于中欧数字教育领域日益丰富的学术脉络之中。我们既呼应了方璐（本卷）的定性案例研究——她深入考察了人工智能对波士顿学院中文课程中批判性思维的影响；也关联了Ole Döring（本卷）的哲学分析——他对教学语境下“人工智能”的概念基础进行了深刻审视。方璐通过个案提供了研究深度，Döring通过哲学提供了思想广度，而我们则通过对大规模受试者群体的定量比较，为该领域贡献了实证广度。
'2. Literature Review'	(zu übersetzen)2.文献综述
'2.1 AI in Language Education: The State of the Art'	(zu übersetzen)2.1 语言教育中的人工智能：研究现状
The application of technology to language learning has a long history, from language laboratories in the 1960s through Computer-Assisted Language Learning (CALL) in the 1990s to the current generation of AI-powered tools. Chapelle (2001) provided an early framework for evaluating technology in second language acquisition, emphasising the importance of language learning potential, learner fit, and practical considerations. Golonka et al. (2014) reviewed 350 studies on technology types in foreign language learning and found that while technology shows promise for vocabulary acquisition and reading comprehension, evidence for speaking and writing gains was limited.	技术在外语学习中的应用历史悠久，从20世纪60年代的语言实验室，到20世纪90年代的计算机辅助语言学习（CALL），再到当前这一代由人工智能驱动的工具。Chapelle（2001）较早提出了二语习得技术评估框架，强调了语言学习潜力、学习者适配性及实际考量因素的重要性。Golonka等人（2014）对350项关于外语学习技术类型的研究进行了综述，发现尽管技术在词汇习得与阅读理解方面展现出潜力，但在提升口语与写作能力方面的证据则较为有限。
The emergence of large language models (LLMs) — ChatGPT, Claude, and their Chinese counterparts Kimi, DeepSeek, and Doubao — has fundamentally changed the landscape. Unlike earlier chatbots that relied on scripted dialogues and keyword matching, LLM-based chatbots can sustain open-ended, contextually appropriate conversations across virtually any topic. Huang, Hew, and Fryer (2022) conducted a systematic review of chatbot-supported language learning and found positive effects on vocabulary acquisition and speaking confidence, but noted that most studies suffered from small sample sizes, short durations, and lack of control groups.	(zu übersetzen)大语言模型（LLMs）——如ChatGPT、Claude及其中国对应产品Kimi、DeepSeek与豆包——的兴起，从根本上改变了这一格局。与早期依赖预设对话脚本与关键词匹配的聊天机器人不同，基于LLM的聊天机器人能够在几乎任何主题下维持开放式、语境适配的对话。Huang、Hew与Fryer（2022）对聊天机器人辅助语言学习进行了系统性综述，发现其对词汇习得与口语信心具有积极影响，但同时指出，多数研究存在样本量小、周期短及缺乏对照组等局限。
Jeon (2022) explored AI chatbot affordances with young Korean EFL learners and found that students appreciated the chatbot’s patience, availability, and non-judgmental nature — findings that our data strongly corroborate. Kim (2019) reported that AI chatbot interaction improved English grammar skills among Korean university students, a finding that our data only partially support (grammar improvement was actually lower in our AI group).	Jeon（2022）探讨了人工智能聊天机器人对韩国年轻英语作为外语（EFL）学习者的可供性，发现学生们赞赏聊天机器人的耐心、随时可用性及非评判性——这些发现在我们的数据中得到了有力佐证。Kim（2019）报告称，与人工智能聊天机器人的互动提升了韩国大学生的英语语法技能，而这一发现在我们的数据中仅得到部分支持（事实上，我们研究中人工智能组的语法提升幅度更低）。
'2.2 Foreign Language Anxiety'	(zu übersetzen)2.2外语焦虑
The psychological dimension of language learning has been extensively studied since Horwitz, Horwitz, and Cope (1986) developed the Foreign Language Classroom Anxiety Scale (FLCAS). MacIntyre and Gardner (1994) demonstrated that language anxiety has measurable effects on cognitive processing in the second language: anxious learners process information more slowly, recall less vocabulary, and produce less complex utterances. Krashen’s (1982) „affective filter“ hypothesis posits that negative emotional states — anxiety, self-doubt, boredom — create a mental barrier that impedes language acquisition.	自Horwitz、Horwitz与Cope（1986）编制《外语课堂焦虑量表》（FLCAS）以来，语言学习的心理维度已得到广泛研究。MacIntyre与Gardner（1994）证实，语言焦虑会对二语认知加工产生可测量的影响：焦虑的学习者信息处理速度更慢，词汇回忆量更少，且产出的话语复杂度更低。Krashen（1982）的“情感过滤”假说提出，焦虑、自我怀疑、厌倦等负面情绪状态会形成阻碍语言习得的心智屏障。
The relevance for AI-assisted learning is direct. If AI chatbots can lower the affective filter by providing a judgment-free practice environment, they may enable learners to process and produce language more effectively than they would in the anxiety-producing context of a human classroom. Our data suggest that this mechanism is operative: the AI group’s most highly rated advantage was „no fear of making mistakes“ (76.6%), and the AI group reported greater improvement in precisely those skills — speaking, listening, communicative confidence — that are most inhibited by anxiety.	(zu übersetzen)这对人工智能辅助学习具有直接意义。如果人工智能聊天机器人能够通过提供无评判的练习环境来降低情感过滤（affective filter），那么相较于令人焦虑的人类课堂环境，它们或许能使学习者更有效地进行语言输入与输出。我们的数据表明这一机制确实发挥了作用：人工智能组评价最高的优势正是“无需担心犯错”（76.6%），且该组报告进步最为显著的技能——口语、听力、交际信心——恰恰是最易受焦虑抑制的领域。
'2.3 The Chinese Context'	(zu übersetzen)2.3中国语境
China‘s educational AI landscape is distinctive. The Chinese government’s „New Generation Artificial Intelligence Development Plan“ (2017) and „Education Modernization 2035“ plan both identify AI as a strategic priority for educational reform. Chinese students have access to a range of domestically developed AI tools — including Kimi (Moonshot AI), DeepSeek, Doubao (ByteDance), and Ernie (Baidu) — in addition to international tools like ChatGPT (accessible via VPN). The cultural context is also relevant: Chinese classroom culture traditionally emphasises teacher authority, student deference, and face-saving behaviours that can inhibit oral participation — precisely the conditions under which AI’s judgment-free environment may offer the greatest benefit.	(zu übersetzen)中国的人工智能教育生态具有鲜明的独特性。中国政府发布的《新一代人工智能发展规划》（2017）与《中国教育现代化2035》均将人工智能确立为教育改革的战略重点。除ChatGPT（可通过VPN访问）等国际工具外，中国学生还能使用一系列本土研发的AI工具，包括Kimi（月之暗面）、DeepSeek（深度求索）、Doubao（字节跳动）及Ernie（百度）。文化背景同样至关重要：中国传统课堂文化历来强调教师权威、学生顺从及“面子”观念，这些因素往往会抑制口语参与——而正是这样的环境，使得AI提供的无评判环境可能发挥出最大的效益。
'3. Study Design and Methodology'	(zu übersetzen)3.研究设计与方法论
'2.1 Participants'	(zu übersetzen)2.1参与者
A total of 133 Chinese university students participated in the study. The AI group comprised 85 participants (74% female, 26% male; mean age 23.8 years, range 19–38). The human-teacher group comprised 48 participants (89% female, 11% male; mean age 23.1 years, range 20–32). All participants were enrolled at Chinese universities, predominantly studying English (AI: 38%, Human: 29%) or German (AI: 16%, Human: 25%) as their foreign language major. The gender imbalance — more pronounced in the human group — reflects the general demographics of foreign language departments at Chinese universities.	(zu übersetzen)本研究共有133名中国大学生参与。其中，人工智能组含85人（女性74%，男性26%；平均年龄23.8岁，年龄范围19–38岁）；人类教师组含48人（女性89%，男性11%；平均年龄23.1岁，年龄范围20–32岁）。所有参与者均为中国高校在读生，所学专业以外语为主，其中人工智能组修读英语者占38%、德语者占16%，人类教师组修读英语者占29%、德语者占25%。人类教师组的性别失衡现象更为显著，这反映了中国高校外语院系的总体性别分布特征。
Participants were not randomly assigned. Some chose their group; others were assigned (44.7% of the human group reported passive assignment). This self-selection introduces a potential confound: students who chose the AI group may have been more technologically curious or more dissatisfied with traditional instruction. We address this limitation in Section 5.	(zu übersetzen)参与者并未被随机分配。部分学生自主选择组别，另一部分则为被动分配（人类教师组中44.7%的参与者报告为被动分配）。这种自我选择引入了一个潜在的混淆变量：选择人工智能组的学生可能技术好奇心更强，或对传统教学更不满意。我们将在第5节探讨这一局限性。
'2.2 Survey Instrument'	(zu übersetzen)2.2调查问卷
The survey was administered in Chinese via an online questionnaire platform (问卷星) on 28 March 2025. It comprised the following sections:	调查于2025年3月28日通过在线问卷平台（问卷星）以中文进行。包括以下部分：
(a) Demographics: name (anonymised before analysis), date of birth, gender (5 items). (b) Prior language proficiency: self-assessed CEFR levels for Chinese, English, German, French, Japanese, Korean, and up to three additional languages (9 items). (c) Study language and starting level: same structure as (b) but for the language being studied in the experiment (9 items). (d) Study habits: daily study time in minutes, group assignment, daily AI usage time in minutes (3 items). (e) Reasons for group choice: 5–6 reasons rated by relative importance (percentage, totalling approximately 100%) (6–10 items depending on group). (f) AI learning methods (AI group only): chatting with AI, task completion, VR classroom, AI teacher — each rated by usage share (5 items). (g) Reasons for interest in current learning method: 9–10 reasons rated by importance (10 items). (h) AI feedback quality and handling (AI group only): categorical rating and yes/no response (2 items). (i) Self-reported overall improvement: percentage estimate (1 item). (j) Sensory modality importance: 21 items covering visual, auditory, textual, gestural, spatial, tactile, olfactory, gustatory, social (3 sub-items), emotional (2 sub-items), VR immersion (2 sub-items), and AI immersion (2 sub-items), each rated 0–100%. (k) Sensory modality ability: same 21 items, rated for personal capacity (0–100%). (l) Group satisfaction and willingness to switch (4 items). (m) Attitudes toward AI: 14 statements rated 0–100% agreement. (n) Improvement areas: 10 language skill areas rated by relative improvement (percentage, totalling approximately 100%) (11 items).	(zu übersetzen)(a) 人口统计学信息：姓名（分析前已匿名化处理）、出生日期、性别（5项）。 (b) 既往语言能力：对汉语、英语、德语、法语、日语、韩语及最多三种其他语言的欧洲语言共同参考框架（CEFR）自评等级（9项）。 (c) 所学语言及起始水平：结构与(b)相同，但针对实验中正在学习的语言（9项）。 (d) 学习习惯：每日学习时长（分钟）、组别分配情况、每日使用AI时长（分钟）（3项）。 (e) 组别选择原因：按相对重要性评分的5–6项原因（百分比，总和约100%）（视组别而定，6–10项）。 (f) AI学习方法（仅限AI组）：与AI聊天、任务完成、虚拟现实课堂、AI教师——各项按使用占比评分（5项）。 (g) 对当前学习方法感兴趣的原因：按重要性评分的9–10项原因（10项）。 (h) AI反馈质量与处理方式（仅限AI组）：分类评级与是否回答（2项）。 (i) 自陈总体提升程度：百分比估算（1项）。 (j) 感官模态重要性：涵盖视觉、听觉、文本、手势、空间、触觉、嗅觉、味觉、社交（3个子项）、情感（2个子项）、虚拟现实沉浸感（2个子项）及AI沉浸感（2个子项）共21项，每项按0–100%评分。 (k) 感官模态能力：同上21项，按个人能力评分（0–100%）。 (l) 组别满意度与转换意愿（4项）。 (m) 对AI的态度：14项陈述，按同意程度以0–100%评分。 (n) 提升领域：10个语言技能领域，按相对提升程度评分（百分比，总和约100%）（11项）。
'2.3 Data Processing'	(zu übersetzen)2.3数据处理
Responses were recorded on a 0–100% scale, with 0% indicating „not at all“ and 100% „completely“ or „exclusively.“ For items requiring percentage allocation across multiple options (e.g., reasons for group choice, improvement areas), respondents were instructed that their ratings should sum to approximately 100%. Not all respondents achieved exact summation; we report the raw percentages without normalisation. Missing values were excluded pairwise. All statistical analyses were conducted using Python (descriptive statistics, no inferential tests given the exploratory nature and self-selection design).	(zu übersetzen)回复采用0–100%的量表记录，0%表示“完全不”，100%表示“完全”或“ exclusively（仅此一项）”。对于需要在多个选项中分配百分比的题目（如组别选择原因、提升领域），要求受访者的评分总和约为100%。并非所有受访者均能达到精确的总和；我们报告原始百分比，未进行归一化处理。缺失值采用两两剔除的方式处理。所有统计分析均使用Python完成（仅含描述性统计，鉴于研究的探索性特征及自我选择设计，未进行推断性检验）。
'3. Results'	(zu übersetzen)3.结果
'3.1 Daily Study Time and AI Usage'	(zu übersetzen)3.1每日学习时长与AI使用情况
Both groups reported similar daily study times: AI group mean 106 minutes (median 60, SD 103), human group mean 96 minutes (median 60, SD 90). The high standard deviations reflect wide variation: some students studied 10 minutes daily, others 360 minutes. Within the AI group, mean daily AI usage was 32 minutes (median 15), suggesting that AI constituted roughly 30% of total study time, with the remainder spent on textbooks, exercises, or other non-AI methods.	(zu übersetzen)两组报告的每日学习时间相近：人工智能组平均为106分钟（中位数60，标准差103），人类教师组平均为96分钟（中位数60，标准差90）。较大的标准差反映出个体差异显著：部分学生每日学习10分钟，另一些则达360分钟。在人工智能组内，每日使用AI的平均时长为32分钟（中位数15），这表明AI约占其总学习时长的30%，剩余时间用于课本、练习或其他非AI方式。
'3.2 Self-Reported Overall Improvement'	(zu übersetzen)3.2自陈总体提升程度
The human-teacher group reported higher overall improvement after one month: mean 63.2% (median 70%, SD 27.5%, n=42) versus the AI group’s mean 51.9% (median 50%, SD 18.1%, n=82). This finding is notable: despite similar study times, students learning with human teachers perceived greater progress. However, the human group’s higher standard deviation (27.5% vs. 18.1%) indicates more heterogeneous experiences — some human-group students reported very high improvement (up to 100%), while others reported as low as 5%.	(zu übersetzen)人类教师组报告的一个月后总体提升程度更高：平均值为63.2%（中位数70%，标准差27.5%，n=42），而人工智能组的平均值为51.9%（中位数50%，标准差18.1%，n=82）。这一发现值得注意：尽管两组学习时间相近，但接受人类教师指导的学生感知到的进步更大。然而，人类教师组更高的标准差（27.5% vs. 18.1%）表明其体验异质性更强——部分人类教师组学生报告了极高的提升（高达100%），而另一些则低至5%。
'3.3 AI Feedback Quality'	(zu übersetzen)3.3AI反馈质量
Among AI-group participants, perceptions of AI feedback quality were generally positive: 38% rated it as „very pertinent“ (75–100 points), 54% as „okay“ (50–74 points), and only 4% as „average“ (25–49 points). None rated it as poor. Three-quarters (76%) reported handling AI feedback promptly, while 18% did not.	(zu übersetzen)在人工智能小组中，参与者对AI反馈质量的感知总体较为积极：38%将其评为“非常切题”（75–100分），54%评为“尚可”（50–74分），仅4%评为“一般”（25–49分）。无人给出差评。四分之三（76%）的受访者表示会及时处理AI反馈，18%则表示不会。
'3.4 AI Learning Methods'	(zu übersetzen)3.4人工智能学习方法
The most popular AI learning methods were chatting with AI software (mean usage share 68.6%) and asking AI to complete tasks (66.3%). AI teacher functionality received moderate use (51.3%), while VR classroom was the least used (31.9%). This pattern suggests that conversational AI — the free-form chatbot interaction — dominates current AI-assisted language learning, with structured pedagogical AI tools playing a secondary role.	(zu übersetzen)最受欢迎的人工智能学习方法依次为：与AI软件聊天（平均使用占比68.6%）和请AI完成任务（66.3%）。AI教师功能的使用程度中等（51.3%），而VR课堂使用最少（31.9%）。这一模式表明，对话式人工智能——即自由形式的聊天机器人互动——主导了当前的人工智能辅助语言学习，而结构化的教学型AI工具则处于次要地位。
'3.5 Motivations'	(zu übersetzen)3.5动机
Reasons for choosing the AI group (rated by importance):	(zu übersetzen)选择人工智能组的原因（按重要性评分）：
'1. Novelty / trying new things: 75.4%'	(zu übersetzen)1.新颖性/尝试新事物：75.4%
'2. Learn anytime, anywhere: 72.5%'	(zu übersetzen)2.随时随地学习：72.5%
'3. Immersive learning experience: 66.9%'	(zu übersetzen)3.沉浸式学习体验：66.9%
'4. Bored with traditional methods: 60.8%'	(zu übersetzen)4.对传统方法感到厌烦：60.8%
'5. Cheaper than human teachers: 59.9%'	(zu übersetzen)5.比人类教师便宜：59.9%
The top two motivations — novelty and flexibility — suggest that early AI adopters are driven more by curiosity and convenience than by dissatisfaction with traditional teaching.	(zu übersetzen)排名最高的两大动机分别是新颖性与灵活性，这表明早期人工智能使用者主要受好奇心和便利性驱动，而非源于对传统教学的不满。
What makes AI learning attractive (rated by importance):	(zu übersetzen)让AI学习具有吸引力的因素（按重要性评级）：
'1. No fear of making mistakes / reduced pressure: 76.6%'	(zu übersetzen)1.不怕犯错/减少压力：76.6%
'2. Large knowledge base / diverse topics: 74.7%'	(zu übersetzen)2.庞大的知识库/涵盖话题广泛：74.7%
'3. Learn anytime, anywhere: 71.9%'	(zu übersetzen)3.随时随地学习：71.9%
'4. Fast response speed: 70.4%'	(zu übersetzen)4.响应速度快：70.4%
'5. Adaptive difficulty matching: 67.8%'	(zu übersetzen)5.自适应难度匹配：67.8%
'6. Adjustable speed, volume, voice: 65.3%'	(zu übersetzen)6.可调节速度、音量、语音：65.3%
'7. More encouragement: 64.5%'	(zu übersetzen)7.更多鼓励：64.5%
'8. Much cheaper: 59.4%'	(zu übersetzen)8.成本低得多：59.4%
'9. More accurate pronunciation correction: 58.5%'	(zu übersetzen)9.更准确的发音纠正：58.5%
The highest-rated advantage — „no fear of making mistakes“ at 76.6% — aligns with a substantial body of research on foreign language anxiety. The AI chatbot creates what language educators call a „low-anxiety practice environment“ in which learners can experiment without social embarrassment.	(zu übersetzen)获评最高的优势——“不怕犯错”（76.6%）——与大量关于外语焦虑的研究相吻合。AI聊天机器人创造了语言教育者所称的“低焦虑练习环境”，学习者可以在其中大胆尝试，无需担心社交尴尬。
Reasons for choosing the human group:	(zu übersetzen)选择人类组的原因：
'1. Prefer learning with real people: 65.7%'	(zu übersetzen)1.偏好与真人一起学习：65.7%
'2. Stimulates deeper thinking: 63.8%'	(zu übersetzen)2.激发更深层次思考：63.8%
'3. Better at detecting learning problems: 63.6%'	(zu übersetzen)3.更善于发现学习问题：63.6%
'4. More precise level assessment: 61.2%'	(zu übersetzen)4.对学习者的水平评估更为精准 (61.2%)
'5. More diverse feedback methods: 60.5%'	(zu übersetzen)5.更多样化的反馈方式：60.5%
'6. Emotional communication in feedback: 58.2%'	(zu übersetzen)6.反馈中的情感交流：58.2%
'7. Trust traditional teaching: 52.4%'	(zu übersetzen)7.信任传统教学模式：52.4%​
'8. Don’t want to change methods: 52.3%'	(zu übersetzen)8.不想改变方法：52.3%
'9. AI not mature yet: 45.3%'	(zu übersetzen)9.AI尚未成熟：45.3%
'10. Passively assigned: 44.7%'	(zu übersetzen)10.被动分配：44.7%
The human group’s top reasons centre on relational and cognitive depth: human teachers offer personal connection, deeper thinking, and more nuanced assessment. This contrasts with the AI group’s emphasis on convenience and psychological comfort.	(zu übersetzen)人类组的首要原因集中于关系联结与认知深度：真人教师能提供人际互动、激发深层思考，并给予更具辨识度的评估。这与 AI 组对便利性与心理舒适感的侧重形成了鲜明对比。
'3.6 Improvement Areas'	(zu übersetzen)3.6改进方面
Students assessed their improvement across ten specific language skill areas. The results reveal a striking complementarity:	(zu übersetzen)学生评估了他们在十个具体语言技能领域的进步。结果显示出一种显著的互补性：
Areas where the AI group reported greater improvement: - Speaking: +12.6 percentage points (AI 58.4%, Human 45.8%) - Listening: +10.2 pp (AI 53.6%, Human 43.5%) - Confidence in communication: +8.3 pp (AI 55.2%, Human 46.9%) - Synonyms/varied expressions: +5.6 pp (AI 56.8%, Human 51.2%)	(zu übersetzen)AI组报告进步更显著的领域：口语：+12.6个百分点（AI 58.4%，人类 45.8%）；听力：+10.2个百分点（AI 53.6%，人类 43.5%）；沟通信心：+8.3个百分点（AI 55.2%，人类 46.9%）；同义词/多样化表达：+5.6个百分点（AI 56.8%，人类 51.2%）。
Areas where the human group reported greater improvement: - Reading: +14.0 pp (Human 63.7%, AI 49.8%) - Grammar: +10.1 pp (Human 57.0%, AI 46.9%) - Syntax: +9.3 pp (Human 57.1%, AI 47.8%) - Vocabulary: +5.2 pp (Human 60.7%, AI 55.5%) - Writing: +5.0 pp (Human 51.5%, AI 46.5%)	(zu übersetzen)人类组报告进步更显著的领域：阅读：+14.0个百分点（人类 63.7%，AI 49.8%）；语法：+10.1个百分点（人类 57.0%，AI 46.9%）；句法：+9.3个百分点（人类 57.1%，AI 47.8%）；词汇：+5.2个百分点（人类 60.7%，AI 55.5%）；写作：+5.0个百分点（人类 51.5%，AI 46.5%）。
The pattern is clear: AI-assisted learning appears to strengthen interactive, oral skills (speaking, listening, communicative confidence), while human-taught learning produces greater gains in structural, analytical skills (reading, grammar, syntax). This finding has direct pedagogical implications: AI and human instruction may be most effective not as substitutes but as complements, each addressing different aspects of language competence.	(zu übersetzen)这种模式十分清晰：AI辅助学习似乎更能强化互动性与口语技能（如口语、听力、沟通信心），而真人教学则在结构性及分析性技能上带来更大提升（如阅读、语法、句法）。这一发现具有直接的教学启示：AI教学与人类教学最有效的模式并非相互替代，而是互为补充，各自针对语言能力的不同维度发挥作用。
'3.7 Sensory and Social Modality Preferences'	(zu übersetzen)3.7感官偏好和社会形态偏好
Participants rated the importance of twelve sensory and social modalities for their language learning. Several large differences emerged between groups:	(zu übersetzen)参与者针对十二种感官及社会形态在其语言学习中的重要性进行了评分。各组之间出现了若干显著差异：
Modalities rated higher by the AI group: - Auditory perception: +40.7 pp (AI 79.6%, Human 38.9%) - Written text: +37.4 pp (AI 74.5%, Human 37.1%) - Intrinsic motivation: +35.1 pp (AI 77.5%, Human 42.4%) - Extrinsic motivation: +30.0 pp (AI 69.1%, Human 39.1%) - Visual perception: +29.3 pp (AI 74.6%, Human 45.2%) - Emotions/motivation: +29.0 pp (AI 72.6%, Human 43.6%) - Environmental immersion: +20.6 pp (AI 69.9%, Human 49.3%) - Group dynamics: +17.7 pp (AI 64.6%, Human 46.9%)	(zu übersetzen)AI组评分更高的模式：听觉感知：+40.7个百分点（AI 79.6%，人类 38.9%）；书面文本：+37.4个百分点（AI 74.5%，人类 37.1%）；内在动机：+35.1个百分点（AI 77.5%，人类 42.4%）；外在动机：+30.0个百分点（AI 69.1%，人类 39.1%）；视觉感知：+29.3个百分点（AI 74.6%，人类 45.2%）；情感/动机：+29.0个百分点（AI 72.6%，人类 43.6%）；环境沉浸：+20.6个百分点（AI 69.9%，人类 49.3%）；群体动力：+17.7个百分点（AI 64.6%，人类 46.9%）。
Modalities rated higher by the human group: - Taste: +32.1 pp (Human 76.3%, AI 44.2%) - AI teacher immersion: +31.7 pp (Human 83.9%, AI 52.2%) - VR immersion: +29.3 pp (Human 83.0%, AI 53.7%) - VR ethics: +29.3 pp (Human 81.3%, AI 52.0%) - AI chatbot immersion: +27.2 pp (Human 79.4%, AI 52.2%) - Social impressions: +21.5 pp (Human 81.5%, AI 59.9%) - Smell: +16.0 pp (Human 59.8%, AI 43.8%)	(zu übersetzen)人类组评分更高的模式：味觉：+32.1个百分点（人类 76.3%，AI 44.2%）；AI教师沉浸：+31.7个百分点（人类 83.9%，AI 52.2%）；VR沉浸：+29.3个百分点（人类 83.0%，AI 53.7%）；VR伦理：+29.3个百分点（人类 81.3%，AI 52.0%）；AI聊天机器人沉浸：+27.2个百分点（人类 79.4%，AI 52.2%）；社交印象：+21.5个百分点（人类 81.5%，AI 59.9%）；嗅觉：+16.0个百分点（人类 59.8%，AI 43.8%）。
These results require careful interpretation. The AI group placed significantly greater importance on the primary language-learning modalities — visual, auditory, and textual — as well as on motivational factors. The human group, paradoxically, rated AI and VR immersion as more important than the AI group did. One interpretation is that human-group students, not having experienced AI immersion directly, may idealise it, while AI-group students, having used AI tools daily, are more measured in their assessment.	(zu übersetzen)这些结果需要仔细解读。AI组认为主要的语言学习模式——即视觉、听觉和文本——以及动机因素显著更重要。矛盾的是，人类组对AI和VR沉浸的重视程度反而比AI组更高。一种解释是，未直接接触过AI沉浸体验的人类组学生可能将其理想化了，而每天使用AI工具的AI组学生在评估时则更为审慎客观。
The human group’s higher rating of social impressions (81.5% vs. 59.9%) is consistent with their stated preference for learning with real people and reflects the importance of social presence in language education — a factor that current AI tools, despite rapid advances, cannot fully replicate.	(zu übersetzen)人类组对“社交印象”（social impressions）给出了更高的评分（81.5% 对 59.9%），这与他们直言不讳的“偏爱与真人学习”的倾向不谋而合。这也折射出“社交存在感”（social presence）在语言教育中的举足轻重——尽管AI技术日新月异，但这一关键因素至今仍无法被完全复制。
'3.8 Attitudes toward AI in Education and Society'	(zu übersetzen)对教育与社会中人工智能的态度
Fourteen attitude statements were rated on a 0–100% agreement scale. The results reveal a nuanced picture:	(zu übersetzen)十四项态度陈述采用0–100%的认同度量表进行评分，结果呈现出一幅细致入微的画面：
Both groups strongly like human teachers: AI group 77.7%, Human group 83.6%. Even after a month of AI-assisted learning, AI-group students retain strong appreciation for human instruction.	(zu übersetzen)两组均强烈喜爱人类教师：AI组77.7%，人类组83.6%。即使经过一个月的AI辅助学习，AI组学生依然保持着对人类教学的强烈认可。
The AI group is more positive toward AI teaching: current AI teacher approval was 57.3% (vs. 38.2% in human group), and future advanced AI teacher approval was 66.4% (vs. 53.3%). However, even in the AI group, current AI teacher approval (57.3%) is substantially lower than human teacher approval (77.7%).	(zu übersetzen)两组对AI教学的态度存在明显差异：AI组对当前AI教师的认可度为57.3%（人类组为38.2%），对未来先进AI教师的认可度为66.4%（人类组为53.3%）。然而即使在AI组内，当前AI教师的认可度（57.3%）仍显著低于人类教师的认可度（77.7%）。
Both groups express fear of AI dependency: - „Fear AI replaces thinking ability“: AI 60.1%, Human 61.0% - „Fear knowledge/skills decline“: AI 60.6%, Human 66.5% - „Fear losing independence / AI addiction“: AI 59.6%, Human 71.6%	(zu übersetzen)两组均表达了对AI依赖的恐惧：担忧AI取代思考能力：AI组60.1%，人类组61.0%；担忧知识与技能退化：AI组60.6%，人类组66.5%；担忧丧失独立性/AI成瘾：AI组59.6%，人类组71.6%。
The human group consistently reports higher fear of AI dependency, with the largest gap on addiction (71.6% vs. 59.6%). The AI group, perhaps through direct experience, has developed a more moderate but still cautious view.	(zu übersetzen)人类组始终表现出对AI依赖更高的恐惧，其中成瘾方面的差距最大（71.6% 对 59.6%）。AI组或许因亲身实践，形成了更为温和但仍持谨慎的态度。
Both groups strongly endorse AI ethics: „Need to control AI with ethics“ received 72.8% (AI) and 68.7% (Human) agreement.	(zu übersetzen)两组均强烈支持AI伦理：“需用伦理规范约束AI”获得了72.8%（AI组）和68.7%（人类组）的认同率。
Both groups reject AI dominance: „Let AI control humans“ received only 14.4% (AI) and 21.5% (Human) agreement. „Only AI robots, no humans, is enough“ received 15.2% and 19.3%. These findings suggest that Chinese university students in 2025 maintain a firmly humanist orientation: they welcome AI as a tool but reject it as a master.	(zu übersetzen)两组均反对AI主导：“让AI控制人类”仅获14.4%（AI组）和21.5%（人类组）认同；“仅需AI机器人、无需人类”仅获15.2%和19.3%认同。这些发现表明，2025年的中国大学生仍秉持坚定的人文主义取向：他们欢迎AI作为工具，但拒绝让其成为主宰。
Romantic attachment to AI or teachers is minimal: „Fell in love with an AI“ averaged approximately 20% in both groups, and „fell in love with a human teacher“ averaged 20–33%. These low figures suggest that immersive AI interaction has not, for this cohort, produced the emotional dependency that some commentators have predicted. The Chinese cultural context may be relevant here: the pragmatic orientation toward AI as a tool rather than a companion, combined with clear social norms around human relationships, may provide a cultural buffer against the parasocial attachment that has been reported in some Western studies of human-AI interaction.	(zu übersetzen)两组对AI或教师产生浪漫依恋的比例极低：“爱上AI”在两组的均值均约为20%，“爱上人类教师”的均值在20–33%之间。这些低数值表明，沉浸式AI互动并未在该群体中引发部分评论者所预测的那种情感依赖。中国文化语境可能与此相关：将AI务实定位于工具而非伴侣，加之明确的人际关系社会规范，或许构成了抵御准社会关系依恋的文化缓冲——此类依恋现象在一些西方的人机互动研究中曾有报道。
The willingness to use AI as a labour-saving device was moderate (approximately 39% in both groups), suggesting that most students do not view AI primarily as a shortcut. Combined with the strong endorsement of ethical AI control, this pattern indicates a cohort that views AI as useful but limited — a sophisticated position that contradicts stereotypes of Chinese students as uncritical technology adopters.	(zu übersetzen)两组将AI用作省力工具的意愿处于中等水平（两组均约为39%），表明大多数学生并不主要将AI视为捷径。结合对AI伦理控制的强烈支持，这种模式显示出该群体视AI为有用但有限的工具——这是一种复杂的立场，与将中国学生视为不加批判的技术采用者的刻板印象相悖。
'3.9 Detailed Attitude Analysis'	(zu übersetzen)3.9详细态度分析
To understand the nuanced attitudes more clearly, we can group the fourteen attitude items into thematic clusters:	(zu übersetzen)为更清晰地把握这些微妙的态度，我们可以将这十四个态度项目归纳为若干主题集群：
Cluster A — Teaching preference: - „I like human teacher teaching me“: AI 77.7%, Human 83.6% - „I like current AI teacher teaching me“: AI 57.3%, Human 38.2% - „I’d like future advanced AI teacher“: AI 66.4%, Human 53.3%	(zu übersetzen)A组——教学偏好：“我喜欢人类教师授课”：AI组77.7%，人类组83.6%；“我喜欢目前的AI教师授课”：AI组57.3%，人类组38.2%；“我愿意接受未来先进的AI教师授课”：AI组66.4%，人类组53.3%。
Both groups prefer human teachers, but the AI group shows significantly greater openness to both current and future AI instruction. The 20-point gap between human teacher approval (77.7%) and current AI teacher approval (57.3%) in the AI group — after direct experience with AI tools — suggests that familiarity breeds qualified appreciation rather than enthusiasm.	(zu übersetzen)两组均偏好人类教师，但AI组对当前及未来的AI教学表现出显著更高的开放度。在亲身体验AI工具后，AI组对人类教师的认可度（77.7%）与对当前AI教师的认可度（57.3%）之间仍存在20个百分点的差距，这表明熟悉催生的是有限度的认可，而非热情。
Cluster B — Fear of AI: - „Fear: AI replaces thinking ability“: AI 60.1%, Human 61.0% - „Fear: knowledge/skills decline“: AI 60.6%, Human 66.5% - „Fear: lose independence, AI addiction“: AI 59.6%, Human 71.6% - „Not afraid: focus on other areas“: AI 55.7%, Human 53.4%	(zu übersetzen)B组——对AI的恐惧：“担心：AI取代思考能力”：AI组60.1%，人类组61.0%；“担心：知识/技能退化”：AI组60.6%，人类组66.5%；“担心：失去独立性、对AI上瘾”：AI组59.6%，人类组71.6%；“不担心：可专注于其他领域”：AI组55.7%，人类组53.4%。
Both groups harbour substantial anxiety about cognitive atrophy — a concern that Fang Lu’s qualitative data make vivid. The human group’s higher fear of addiction (71.6% vs. 59.6%) may reflect a less differentiated understanding of what AI interaction actually involves: the unknown is often more frightening than the known.	(zu übersetzen)两组均对认知萎缩怀有显著焦虑——这种担忧在方璐（Fang Lu）的质性数据中得到了生动体现。人类组对成瘾的更高恐惧（71.6% 对 59.6%）可能反映出他们对AI互动实际内涵的认知不够细化：未知往往比已知更令人恐惧。
Cluster C — AI governance: - „Need to control AI with ethics“: AI 72.8%, Human 68.7% - „Give AI freedom to develop next gen“: AI 47.5%, Human 50.0% - „Let AI control humans“: AI 14.4%, Human 21.5% - „Only AI robots, no humans, is enough“: AI 15.2%, Human 19.3%	(zu übersetzen)C组——AI治理：“需用伦理规范约束AI”：AI组72.8%，人类组68.7%；“应给予AI自由研发下一代”：AI组47.5%，人类组50.0%；“让AI控制人类”：AI组14.4%，人类组21.5%；“仅需AI机器人、无需人类”：AI组15.2%，人类组19.3%。
The governance attitudes reveal a clear hierarchy: strong endorsement of ethical control, ambivalence about AI autonomy, and firm rejection of AI supremacy. The consistency across both groups suggests that these attitudes reflect a broader generational consensus rather than group-specific effects.	(zu übersetzen)C组（AI治理）态度呈现出清晰的层级：对“需用伦理规范约束AI”的支持最为强烈（AI组72.8%、人类组68.7%），对“应给予AI自由研发下一代”的态度存在分歧（AI组47.5%、人类组50.0%），而对“让AI控制人类”（AI组14.4%、人类组21.5%）和“仅需AI机器人、无需人类”（AI组15.2%、人类组19.3%）则持明确否定态度。两组态度的一致性表明，这些观点反映的是更广泛的代际共识，而非特定实验组的独有特征。
'3.10 Group Satisfaction and Switching Willingness'	(zu übersetzen)3.10组别满意度与转换意愿
Both groups reported high satisfaction with their assignment: AI group 80.9% (median 80%), human group 76.7% (median 85%). However, willingness to switch groups tells a different story: 47% of the AI group and a remarkable 68% of the human group expressed willingness to switch. The human group’s high switching rate suggests that many human-group students are curious about AI-assisted learning even while satisfied with their current experience — consistent with the broader cultural moment in which AI is perceived as novel and attractive.	(zu übersetzen)两组对其所处实验安排的满意度均处于高位：AI组为80.9%（中位值80%），人类组为76.7%（中位值85%）。然而，组间转换意愿却揭示了另一番图景：AI组中有47%的学生表达了换组意愿，而人类组的比例更是高达68%。人类组极高的转换意愿表明，即便对当前体验感到满意，仍有大量学生渴望尝试AI辅助学习——这与当下将AI视为新奇且富有吸引力的宏观文化氛围不谋而合。
Among AI-group respondents who described their switching preference, the most common response was „AI group: convenient“ (便利), suggesting that those who would remain valued practical accessibility above all. Among human-group respondents, several articulated thoughtful positions: „AI is not yet mature“ (AI不完善), „human teaching methods are more suited to me“ (human组的教学方法比较适合我), and notably: „I prefer exploring on my own. Humans will never be replaced by AI“ (我更喜欢自己探索。人类永远不会被AI取代) — a statement that encapsulates the humanist position shared by the majority of respondents.	在阐明换组偏好的受访者中，AI组最普遍的留组理由是“AI组：便利”，表明这部分学生最看重的是实际的可及性与便捷性。而在人类组，不少学生给出了深思熟虑的理由：“AI尚不完善”、“人类的教学方法更适合我”，尤为值得注意的是这句：“我更喜欢自己探索。人类永远不会被AI取代”——这句话精准概括了大多数受访者所秉持的人文主义立场。
'4. Discussion'	(zu übersetzen)4.讨论
The results paint a nuanced picture that resists simple conclusions. We organise our discussion around five themes: the complementarity of AI and human instruction, dialogue with the companion essays in this volume, the anxiety-reduction mechanism, modality differences, and implications for European-Chinese comparative education.	(zu übersetzen)这些结果呈现出一幅拒绝简单定论的细腻图景。我们将围绕五大主题展开讨论：AI与人类教学的互补性、与本卷其他论文的学术对话、焦虑缓解机制、模态差异，以及中欧比较教育的启示。
'4.1 The Complementarity Thesis'	(zu übersetzen)互补性命题
Our central finding — that AI-assisted learning strengthens interactive oral skills while human teaching strengthens structural analytical skills — supports what we call the Complementarity Thesis: AI and human instruction are not substitutes but complements, each better suited to different dimensions of language competence. This finding challenges both the techno-optimist position (that AI will replace human teachers) and the techno-pessimist position (that AI cannot teach effectively).	(zu übersetzen)我们的核心发现是：AI辅助学习更能强化互动口语技能，而人类教学更能提升结构分析能力——这支持了我们所提出的“互补性命题”：AI与人类教学并非替代品，而是互补品，各自更适用于语言能力发展的不同维度。这一发现既挑战了“AI将取代人类教师”的技术乐观主义立场，也反驳了“AI无法有效开展教学”的技术悲观主义立场。
The mechanism is plausible and grounded in established SLA theory. AI chatbots provide unlimited, patient, judgment-free conversation practice — precisely the conditions that promote speaking fluency and listening comprehension. This aligns with Long’s (1996) Interaction Hypothesis, which posits that conversational interaction — including negotiation of meaning, recasts, and comprehension checks — drives language acquisition. AI chatbots provide abundant interaction, albeit without the human interactional moves that Long emphasised. Human teachers provide structured instruction, error analysis, and metalinguistic explanation — precisely the conditions that promote grammatical accuracy, reading comprehension, and syntactic awareness. This aligns with Swain’s (2000) Output Hypothesis, which argues that learners need not only comprehensible input but opportunities to produce language and receive corrective feedback that pushes them beyond their current competence.	这一机制合乎情理，且有成熟的二语习得理论为依据。AI聊天机器人能提供无限量、有耐心且无评判压力的对话练习——这正是提升口语流利度与听力理解的理想条件。这与朗（Long, 1996）的“互动假说”相契合，该假说主张，包括意义协商、重铸与理解核查在内的会话互动是推动语言习得的关键动力。AI聊天机器人虽能提供大量互动，却缺乏朗所强调的人类互动策略。人类教师则提供结构化教学、错误分析与元语言讲解——这正是提升语法准确性、阅读理解能力与句法意识的理想条件。这与斯温（Swain, 2000）的“输出假说”相契合，该假说认为，学习者不仅需要可理解输入，更需要产出语言并获得纠错反馈的机会，从而被推至现有能力水平之上。
The Complementarity Thesis has practical implications: rather than debating whether AI should replace human teachers (a question our data clearly answer: no), educators should ask how AI and human instruction can be orchestrated to serve different learning objectives within a unified curriculum.	(zu übersetzen)互补性命题具有实际指导意义：与其争论AI是否应取代人类教师（我们的数据已明确给出答案：不应取代），教育者更应思考如何在统一的课程体系内，统筹安排AI与人类教学，以服务于不同的学习目标。
'4.2 Dialogue with Fang Lu'	(zu übersetzen)4.2与方璐的对话
Fang Lu’s qualitative study (this volume) identifies a critical risk of AI-assisted language learning: the potential erosion of critical thinking, creativity, and independent judgment. Her case studies — an elementary student whose AI-assisted writing was structurally perfect but intellectually shallow, and an advanced student whose AI-assisted translation was fluent but lacked cultural nuance — illustrate the „pulling seedlings to help them grow“ (拔苗助长) phenomenon: AI accelerates surface-level performance while undermining deeper cognitive development.	方璐的定性研究（本卷）指出了AI辅助语言学习的一个关键风险：可能侵蚀批判性思维、创造力与独立判断。她的案例——一名小学生借助AI写出的文章结构完美但思想浅薄，一名高水平学生借助AI完成的译文流畅却缺失文化意蕴——阐释了“拔苗助长”现象：AI加速了表层表现，却损害了更深层的认知发展。
Our quantitative data both support and complicate Fang Lu’s findings. The human group’s greater improvement in grammar and syntax — skills requiring analytical reasoning rather than pattern reproduction — is consistent with her concern that AI may bypass rather than develop cognitive skills. However, the AI group’s greater improvement in communicative confidence suggests that AI serves a genuine and important function that human instruction often fails to provide: creating a psychologically safe space for oral practice.	(zu übersetzen)我们的定量数据既支持又使方璐的发现更为复杂。人类组在语法与句法（这些技能需要的是分析性推理而非模式复制）上的更大进步，印证了她关于AI可能绕过而非发展认知技能的担忧。然而，AI组在交际自信上的更大提升则表明，AI确实发挥着人类教学常难以企及的重要功能：为口语练习创造一个心理安全的空间。
The implication is not that AI should be avoided but that its role should be carefully defined. AI appears most beneficial for fluency development and anxiety reduction; human instruction appears most beneficial for accuracy development and analytical thinking. A well-designed curriculum would deploy both.	(zu übersetzen)其启示不在于应避免使用AI，而在于应审慎界定其作用。AI似乎最有益于提升流利度与降低焦虑；人类教学则最有益于提高准确度与发展分析性思维。一套设计完善的课程应将二者兼而用之。
'4.3 Dialogue with Ole Döring'	(zu übersetzen)4.3与Ole Döring的对话
Döring’s philosophical paper (this volume) challenges the very concept of „artificial intelligence“ as applied to teaching, arguing that the German philosophical tradition’s distinction between Vernunft (reason, judgment) and Verstand (understanding, calculation) reveals a fundamental category error in claims that machines can „teach.“ What machines do, Döring argues, is process — not understand, not judge, not care.	(zu übersetzen)Döring的哲学论文（本卷）对应用于教学领域的“人工智能”这一概念本身提出了挑战。他指出，德国哲学传统中对“理性”（Vernunft，即判断力）与“知性”（Verstand，即理解与计算）的区分，揭示了声称机器能够“教学”这一论断中存在根本性的范畴错误。Döring认为，机器所做的仅仅是处理——它们不理解、不判断、也不关心。
Our attitudinal data resonate with Döring’s analysis. When students say they „like“ human teachers at 78–84% but only „like“ AI teachers at 38–57%, they may be responding to precisely the distinction Döring identifies: the human teacher offers Vernunft — judgment, care, understanding of the individual learner — while the AI offers Verstand — calculation, pattern-matching, information retrieval. Both are useful, but they are not equivalent.	(zu übersetzen)我们的态度数据与Döring的分析形成了呼应。当78–84%的学生表示“喜欢”人类教师，而仅有38–57%的学生表示“喜欢”AI教师时，他们很可能正是在回应德林所指出的那种本质区别：人类教师提供的是“理性”（Vernunft）——即判断力、关怀与对个体学习者的理解；而AI提供的则是“知性”（Verstand）——即计算、模式匹配与信息检索。两者皆有其用，但绝不等同。
The students’ strong endorsement of ethical AI control (70%+) and strong rejection of AI dominance (<20%) further support Döring’s humanist position. These 133 Chinese university students, while enthusiastically using AI tools, maintain a clear conceptual boundary between human and machine agency.	(zu übersetzen)两组学生对“需用伦理规范约束AI”的高度认同（均超70%）以及对“AI主导人类”的明确反对（均低于20%），进一步印证了Döring的人文主义立场。这133名中国大学生在积极使用AI工具的同时，仍对人类与机器的能动性保持着清晰的概念边界。
'4.4 The Pressure-Free Environment'	(zu übersetzen)4.4无压力环境
The highest-rated advantage of AI learning — „no fear of making mistakes“ at 76.6% — deserves particular attention. Foreign language anxiety is one of the most extensively documented barriers to language acquisition. Traditional classroom settings, with their inherent social dynamics of performance, judgment, and face, create anxiety that inhibits practice, particularly oral practice. The AI chatbot circumvents this entirely: there is no audience, no judgment, no loss of face.	(zu übersetzen)AI学习中被评价最高的优势——“不怕犯错”（占比76.6%）——值得特别关注。外语焦虑是被文献记载最多的语言习得障碍之一。传统课堂环境因其固有的表现、评判与面子等社会动态而产生焦虑，这种焦虑会抑制练习，尤其是口语练习。AI聊天机器人则完全规避了这一点：没有观众，没有评判，也没有丢脸的风险。
This finding suggests that AI’s primary educational contribution may be not as a teacher but as a practice partner — a tireless, patient interlocutor who never judges, never loses patience, and never generates social anxiety. If this is correct, the optimal educational model is not „AI instead of human teachers“ but „AI as supplement to human teachers,“ specifically for the practice component of language learning where anxiety most inhibits performance.	(zu übersetzen)这一发现表明，AI在教育中的主要贡献或许并非作为教师，而是作为练习伙伴——一个永不疲倦、极富耐心且从不评判、从不失去耐心、也从不引发社交焦虑的交谈对象。若果真如此，最优的教育模式便不是“用AI取代人类教师”，而是“将AI作为人类教师的补充”，专门用于语言学习中因焦虑而最阻碍表现的练习环节。
'4.5 Modality Differences and Their Implications'	(zu übersetzen)4.5模态差异及其启示
The large differences in sensory modality preferences between groups — AI students valuing visual, auditory, and textual input more highly, human students valuing social impressions, VR immersion, and physical senses more highly — suggest that the two groups may have fundamentally different learning orientations. AI-group students appear to be cognitively oriented learners who prioritise information input channels. Human-group students appear to be socially and physically oriented learners who prioritise relational and embodied experience.	(zu übersetzen)这组学生在感官模态偏好上存在的巨大差异——AI组更看重视觉、听觉和文本输入，而人类组更看重社交印象、VR沉浸感和身体感官——表明这两组学生可能具有截然不同的学习导向。AI组的学生似乎属于认知导向型学习者，他们优先考虑信息的输入渠道；而人类组的学生则似乎属于社会与体验导向型学习者，他们更看重关系互动和具身化的学习体验。
Whether these differences are causes or consequences of group choice is unclear. Students who prefer cognitive input channels may have selected the AI group because AI tools deliver precisely those channels. Alternatively, a month of AI-assisted learning may have habituated students to valuing cognitive input over social experience. Longitudinal research would be needed to disentangle these possibilities.	(zu übersetzen)目前尚不清楚这些差异是学生选择组别的原因还是结果。倾向于认知输入渠道的学生可能正是因为AI工具恰好能提供这些渠道而选择了AI组。另一种可能是，经过一个月的AI辅助学习，学生们逐渐习惯于更看重认知输入而非社交体验。要理清这些可能性，还需要进行纵向追踪研究。
'5.6 Implications for European-Chinese Comparative Education'	(zu übersetzen)中欧教育比较的启示
Our findings have specific relevance for the European-Chinese educational dialogue that this volume addresses. European language education, shaped by the Common European Framework of Reference for Languages (CEFR) and the communicative approach, has traditionally emphasised oral competence, interaction, and task-based learning. Chinese language education, shaped by examination-driven culture and grammatical-translation pedagogy, has traditionally emphasised reading, writing, grammar, and vocabulary. The emergence of AI as a practice partner may help bridge this gap: Chinese students who lack opportunities for authentic oral practice with human speakers can use AI to develop the communicative skills that European pedagogical approaches prioritise.	(zu übersetzen)我们的研究结果对本卷所探讨的中欧教育对话具有特定意义。受《欧洲语言共同参考框架》（CEFR）及交际教学法影响的欧洲语言教育，历来强调口语能力、互动与任务型学习；而受应试文化与语法翻译法影响的中国语言教育，历来侧重阅读、写作、语法与词汇。AI作为练习伙伴的出现，或有助于弥合这一鸿沟：缺乏真人口语练习机会的中国学生，可利用AI发展欧洲教学法所看重的交际技能。
At the same time, the European emphasis on critical thinking, learner autonomy, and reflective practice — values articulated in the EU Digital Education Action Plan (2021-2027) — provides a necessary counterweight to the risk that AI practice may develop fluency without depth. Fang Lu’s case studies illustrate this risk concretely: the student whose AI-assisted writing was fluent but intellectually empty had developed surface competence without the deeper cognitive engagement that human interaction promotes.	与此同时，欧洲对批判性思维、学习者自主性与反思性实践的重视——这些正是《欧盟数字教育行动计划（2021-2027）》所倡导的价值——为AI练习可能带来的“流利而无深度”的风险提供了必要的制衡。方璐的案例研究具体阐释了这一风险：那名借助AI写出流畅文章却思想空洞的学生，仅发展了表层能力，而未获得人类互动所能促进的深层认知投入。
A European-Chinese model of AI-integrated language education might therefore combine Chinese students’ enthusiastic adoption of AI tools with European pedagogical frameworks that insist on critical thinking and reflective practice. The technology provides the medium; the pedagogy provides the purpose.	(zu übersetzen)因此，一种中欧融合的AI语言教育模式，或将中国学生对AI工具的热忱采纳，与欧洲坚持批判性思维与反思性实践的教学框架相结合。技术提供媒介，教学法赋予目标。
'5.7 Recommendations for Practice'	(zu übersetzen)5.7实践建议
Based on our findings, we offer four recommendations for educators considering the integration of AI into foreign language teaching:	(zu übersetzen)基于我们的研究结果，我们向考虑在外语教学中引入人工智能的教育工作者提出四点建议：
First, use AI for oral practice, not as a replacement for instruction. The data suggest that AI’s greatest contribution is in developing speaking fluency and communicative confidence through low-anxiety conversational practice. This function complements rather than replaces human instruction.	(zu übersetzen)第一，将AI用于口语练习，而非以此替代教学。数据显示，AI的最大贡献在于通过低焦虑的对话练习提升口语流利度与交际自信。这一功能是对人类教学的补充，而非替代。
Second, maintain human teaching for analytical skills. Grammar, syntax, reading comprehension, and writing — the skills that showed greater improvement in the human group — appear to benefit from the structured, explanatory, and corrective instruction that human teachers provide.	(zu übersetzen)第二，保留人工教学以培养分析能力。语法、句法、阅读理解与写作等技能（数据表明人类授课组在这些方面的提升更为显著）似乎更能从人类教师提供的结构化、具解释性与纠正性的指导中获益。
Third, address students’ AI anxiety proactively. Over 60% of students in both groups expressed fear that AI would replace their thinking ability or erode their skills. These concerns are legitimate and should be addressed through explicit discussion of AI’s limitations, ethical frameworks for AI use, and assignments that require independent critical thinking.	(zu übersetzen)第三，积极化解学生的AI焦虑。两组中均有逾60%的学生担忧AI会取代自身思维能力或削弱技能。此类顾虑合情合理，应通过明确探讨AI的局限性、制定AI使用的伦理规范，以及布置要求独立批判性思维的作业来予以疏导。
Fourth, design assessment that AI cannot shortcut. As Fang Lu’s cases illustrate, AI can produce polished output that masks shallow understanding. Assessments should include oral examinations, spontaneous responses, and tasks that require genuine analytical reasoning — areas where AI assistance is either unavailable or visibly artificial.	(zu übersetzen)第四，设计AI无法“走捷径”的评估方式。正如Fang Lu的案例所展示的，AI能够生成掩盖了浅层理解的精致输出。因此，评估环节应包含口试、即兴回答以及要求真实分析推理的任务——在这些领域中，AI辅助要么不可行，要么会暴露出其人为堆砌的痕迹。
'6. Limitations'	(zu übersetzen)6.研究局限性
Several limitations constrain the interpretation of these results:	(zu übersetzen)几项局限性制约了对这些结果的解读：
First, the study relies entirely on self-reported data. Students’ perceptions of their improvement may not correspond to their actual improvement as measured by standardised tests. A pre-post test design would provide more robust evidence.	(zu übersetzen)第一，本研究完全依赖自我报告数据。学生对自身进步的主观感知，未必与标准化测试所测得的实际进步相符。采用前后测设计将能提供更有力的证据。
Second, the non-random group assignment introduces self-selection bias. Students who chose the AI group may differ systematically from those who chose or were assigned to the human group — in technological literacy, learning motivation, personality, or other unmeasured variables. The AI group’s higher male percentage (26% vs. 11%) and broader age range suggest some demographic differences, though the practical significance of these differences for language learning outcomes is unclear.	(zu übersetzen)第二，非随机分组引入了自选择偏差。选择AI组的学生，可能在科技素养、学习动机、性格或其他未测量变量上，与选择或被分配至人类组的学生存在系统性差异。AI组中男性比例更高（26% vs. 11%）且年龄跨度更大，这表明存在一定的人口统计学差异，尽管这些差异对语言学习成果的实际意义尚不明朗。
Third, the one-month observation period is short. Language learning is a long-term process, and the relative advantages of AI versus human instruction may shift over longer periods. The AI group’s advantage in speaking may be an early-stage fluency gain that plateaus, while the human group’s advantage in grammar may compound over time.	(zu übersetzen)第三，为期一个月的观察周期较短。语言学习是一个长期过程，AI与人类教学的相对优势可能会随着时间推移而发生变化。AI组在口语上的优势可能仅是早期的流利度提升，随后趋于平稳；而人类组在语法上的优势则可能随时间不断累积。
Fourth, the sample is entirely Chinese university students, predominantly female, studying English or German. Generalisability to other cultural contexts, age groups, genders, or target languages is uncertain. The cultural specificity of our findings should be emphasised: Chinese classroom culture’s emphasis on face-saving and teacher authority may amplify the anxiety-reduction benefits of AI in ways that would be less pronounced in cultures with more informal teacher-student relationships.	(zu übersetzen)第四，样本完全由中国大学生构成，且以女性为主，其所学语种为英语或德语。研究结果能否推广至其他文化背景、年龄段、性别或目标语种尚不确定。应着重强调研究发现的文化特异性：中国课堂文化对面子与师道权威的重视，可能会放大AI在缓解焦虑方面的益处；而在师生关系更为随意的文化中，这种效应则不那么显著。
Fifth, all measurements are self-reported. The „improvement areas“ data (Section 4.6) represent students’ perceptions of where they improved, not objectively measured gains. Students may overestimate improvement in areas they practised most (confusing practice with progress) or underestimate improvement in areas where gains are less consciously perceived.	(zu übersetzen)第五，所有测量均为自我报告。第4.6节中的“进步领域”数据反映的是学生对其进步所在的主观感知，而非客观测量的进步。学生可能高估了其练习最频繁领域的进步（将练习误认为进步），或低估了那些进步不易被意识到的领域的进步。
Sixth, the survey was conducted at a single time point. Longitudinal data — tracking motivation, attitudes, and outcomes over a full semester or year — would provide a richer picture. A follow-up study with the same participants after six months or one year of continued study would be particularly valuable for testing whether the Complementarity Thesis holds over longer learning periods.	(zu übersetzen)第六，本次调查仅在单一时间点实施。若能获取追踪整个学期或学年的纵向数据——涵盖学习动机、态度变化及学习成果的动态记录——将能呈现更为丰富的图景。尤其有价值的是，针对同一批参与者在持续学习六个月或一年后开展后续追踪研究，这将有力检验“互补论题”在更长学习周期中是否成立。
Despite these limitations, the study offers one of the larger-sample comparative investigations of AI-assisted versus human-taught language learning available to date, and the breadth of the survey instrument — covering motivation, modality preferences, attitudes, and skill-specific improvement — provides a multidimensional picture that most existing studies lack.	(zu übersetzen)尽管存在上述局限性，本研究仍是迄今为止关于AI辅助教学与传统人工教学对比的规模较大的实证调查之一；且该调查问卷覆盖学习动机、模态偏好、态度及分技能提升等多个维度，勾勒出一幅大多数现有研究未能提供的多维图景。
'6. Conclusion'	(zu übersetzen)6.结论
This study of 133 Chinese university students learning foreign languages with AI assistance (n=85) and with human teachers (n=48) yields four principal findings:	(zu übersetzen)本研究以133名中国大学生为对象，其中85人接受人工智能辅助外语学习，48人接受人类教师面授，得出四项主要发现：
First, human-taught students reported higher overall improvement (63.2% vs. 51.9%), but the pattern is skill-specific: AI-assisted students improved more in speaking (+12.6 pp), listening (+10.2 pp), and communicative confidence (+8.3 pp), while human-taught students improved more in reading (+14.0 pp), grammar (+10.1 pp), and syntax (+9.3 pp). This supports a Complementarity Thesis: AI and human instruction serve different, complementary functions in language education.	(zu übersetzen)第一，接受人类授课的学生自评总体进步率更高（63.2% vs. 51.9%），但这一优势呈现显著的技能差异性：AI辅助组在口语（+12.6个百分点）、听力（+10.2个百分点）及交际信心（+8.3个百分点）上进步更为明显，而人类授课组则在阅读（+14.0个百分点）、语法（+10.1个百分点）及句法（+9.3个百分点）上提升更为突出。这支持了“互补论题”：AI与人类教学在外语教育中发挥着不同且互补的作用。
Second, the primary perceived advantage of AI learning is not informational but psychological: „no fear of making mistakes“ was rated highest at 76.6%. AI’s greatest contribution to language education may be creating a pressure-free environment for oral practice — addressing one of the most persistent barriers to language acquisition.	(zu übersetzen)第二，AI学习的主要感知优势并非信息层面，而是心理层面：“不怕犯错”以76.6%的比例被评为最高优势。AI对外语教育的最大贡献，或许在于为口语练习营造了一种无压力的环境——从而应对语言习得中最顽固的障碍之一。
Third, both groups maintain strongly humanist attitudes. Even after a month of AI-assisted learning, AI-group students rate human teachers higher than AI teachers (77.7% vs. 57.3%). Both groups endorse ethical AI control (>68%) and reject AI dominance over humans (<22%).	(zu übersetzen)第三，两组学生均保持着强烈的人文主义态度。即便在接受了一个月的AI辅助学习后，AI组学生仍对人类教师的评价高于AI教师（77.7% vs. 57.3%）。两组学生均支持对AI进行伦理管控（>68%），且反对由AI主导人类（<22%）。
Fourth, the human group’s paradoxically higher valuation of AI and VR immersion suggests curiosity about technologies they have not experienced, while the AI group’s more measured assessment reflects the moderating effect of actual use.	(zu übersetzen)第四，人类组对AI与虚拟现实沉浸体验的悖论性评价更高，这反映出他们对于尚未亲身尝试的技术抱有好奇；而AI组的评估更为审慎，则体现了实际使用带来的调节效应。
These findings carry clear implications for educational design. The evidence does not support replacing human teachers with AI, nor does it support excluding AI from language education. Instead, it points toward an integrated model in which AI serves as a complementary practice partner — providing the unlimited, judgment-free conversational practice that develops oral fluency and communicative confidence — while human teachers provide the structured instruction, analytical guidance, and social presence that develop grammatical competence, reading comprehension, and critical thinking. Such a model would honour both the technological possibilities documented in our data and the philosophical concerns articulated by Döring and the pedagogical warnings articulated by Fang Lu.	(zu übersetzen)这些发现对教学设计具有明确的指导意义。证据既不支持以AI取代人类教师，也不支持将AI排除在语言教育之外。相反，它指向一种融合模式：AI充当互补的练习伙伴——提供无限量、无评判的对话练习，以培养口语流利度与交际自信；而人类教师则提供结构化教学、分析性指导与社会临场感，以发展语法能力、阅读理解与批判性思维。这样的模式既能尊重我们数据中记载的技术可能性，也能回应Döring所阐发的哲学关切与方璐所提出的教学警示。
These findings carry clear implications for educational design. The evidence does not support replacing human teachers with AI, nor does it support excluding AI from language education. Instead, it points toward an integrated model that leverages the complementary strengths of both: AI for fluency development and anxiety reduction, human teachers for accuracy development and critical thinking. As AI capabilities continue to advance, the question will be not whether to use AI in language education but how to use it wisely — a question that requires continued empirical research, philosophical reflection, and pedagogical innovation.	(zu übersetzen)这些发现对教学设计具有明确的指导意义。证据既不支持以AI取代人类教师，也不支持将AI排除在语言教育之外。相反，它指向一种融合模式，旨在发挥两者的互补优势：利用AI培养流利度并缓解焦虑，依托人类教师提升准确性并发展批判性思维。随着AI能力的持续进阶，问题的关键将不再是要不要在语言教育中使用AI，而是如何明智地使用AI——这一命题要求我们持续开展实证研究、哲学反思与教学创新。
'Acknowledgments'	(zu übersetzen)致谢
Co-funded by the European Union. Views and opinions expressed are however those of the author only and do not necessarily reflect those of the European Union [101126782].	(zu übersetzen)本项目获欧盟共同资助。文中所表达的观点仅为作者个人立场，并不一定代表欧盟的官方观点 [101126782]。
We thank the student participants for their candid responses, and the colleagues who administered the survey.	(zu übersetzen)我们谨向如实作答的学生参与者，以及协助开展问卷调查的各位同仁致以诚挚谢意。
'References'	(zu übersetzen)参考文献
Chapelle, C. A. (2001). Computer Applications in Second Language Acquisition. Cambridge University Press.	(zu übersetzen)Chapelle,C.A.(2001).《第二语言习得中的计算机应用》.剑桥大学出版社.
Döring, O. (this volume). AI and pedagogy: Between artificial intelligence and human understanding.	(zu übersetzen)Döring,O.（本卷）.《人工智能与教学法：在人工智能与人类理解之间》.
Garrett, N. (2009). Computer-assisted language learning trends and issues revisited: Integrating innovation. The Modern Language Journal, 93(s1), 719–740.	(zu übersetzen)Garrett,N.(2009).《计算机辅助语言学习的发展趋势与议题再探：融合创新》.《现代语言杂志》,93（增刊1）,719–740.
Godwin-Jones, R. (2015). Contributing, creating, curating: Digital literacies for language learners. Language Learning & Technology, 19(3), 8–20.	(zu übersetzen)Godwin-Jones,R.(2015).《贡献、创造、策展：面向语言学习者的数字素养》.《语言学习与技术》,19(3),8–20.
Golonka, E. M., Bowles, A. R., Frank, V. M., Richardson, D. L., & Freynik, S. (2014). Technologies for foreign language learning: A review of technology types and their effectiveness. Computer Assisted Language Learning, 27(1), 70–105.	(zu übersetzen)Golonka,E.M., Bowles,A.R., Frank,V.M., Richardson,D.L., & Freynik,S.(2014).《外语学习技术：技术类型及其有效性综述》.《计算机辅助语言学习》,27(1),70–105.
Horwitz, E. K., Horwitz, M. B., & Cope, J. (1986). Foreign language classroom anxiety. The Modern Language Journal, 70(2), 125-132.	(zu übersetzen)Horwitz,E.K., Horwitz,M.B., & Cope,J.(1986).《外语课堂焦虑》.《现代语言杂志》,70(2),125-132.
Huang, W., Hew, K. F., & Fryer, L. K. (2022). Chatbots for language learning — Are they really useful? A systematic review of chatbot-supported language learning. Journal of Computer Assisted Learning, 38(1), 237–257.	(zu übersetzen)Huang,W., Hew,K.F., & Fryer,L.K.(2022).《聊天机器人用于语言学习——它们真的有用吗？一项关于聊天机器人辅助语言学习的系统性综述》.《计算机辅助学习杂志》,38（1）,237-257.
Jeon, J. (2022). Exploring AI chatbot affordances in the EFL classroom: Young learners’ experiences and perspectives. Computer Assisted Language Learning, 37(1-2), 1-26.	(zu übersetzen)Jeon,J.(2022).《探索外语课堂中AI聊天机器人的可供性：年轻学习者的经验与看法》.《计算机辅助语言学习》,37(1-2),1-26.
Kim, N. Y. (2019). A study on the use of artificial intelligence chatbots for improving English grammar skills. Journal of Digital Convergence, 17(8), 37–46.	(zu übersetzen)Kim,N.Y.(2019).《利用人工智能聊天机器人提升英语语法技能的研究》.《数字融合研究》,17(8),37-46.
Krashen, S. D. (1982). Principles and Practice in Second Language Acquisition. Pergamon Press.	(zu übersetzen)Krashen,S.D.(1982).《第二语言习得的原则与实践》.培格曼出版社.
Lai, C., & Zheng, D. (2018). Self-directed use of mobile devices for language learning beyond the classroom. ReCALL, 30(3), 299–318.	(zu übersetzen)Lai,C., & Zheng,D.(2018).《课外移动设备的自主性语言学习使用》.《ReCALL期刊》,30(3),299–318.
Long, M. H. (1996). The role of the linguistic environment in second language acquisition. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of Second Language Acquisition (pp. 413–468). Academic Press.	(zu übersetzen)Long,M.H.(1996).《语言环境在第二语言习得中的作用》.载于 W.C. Ritchie & T.K. Bhatia（编）,《第二语言习得手册》(第413-468页).学术出版社.
Lu, F. (this volume). AI in Chinese teaching: Opportunities and challenges from the perspective of critical thinking.	(zu übersetzen)方璐(本卷).《中文教学中的人工智能：批判性视角的机遇与挑战》.
MacIntyre, P. D., & Gardner, R. C. (1994). The subtle effects of language anxiety on cognitive processing in the second language. Language Learning, 44(2), 283–305.	(zu übersetzen)acIntyre,P.D., & Gardner,R.C.(1994).《语言焦虑对第二语言认知加工的微妙影响》.《语言学习》,44(2),283-305.
Swain, M. (2000). The output hypothesis and beyond: Mediating acquisition through collaborative dialogue. In J. P. Lantolf (Ed.), Sociocultural Theory and Second Language Learning (pp. 97–114). Oxford University Press.	(zu übersetzen)Swain,M.(2000).《输出假说及展望：通过协作对话中介习得》.载于 J.P. Lantolf（编）,《社会文化理论与第二语言学习》(第97-114页).牛津大学出版社.
World Economic Forum. (2025). The Future of Jobs Report 2025. Geneva: WEF.	(zu übersetzen)世界经济论坛.(2025).《2025年未来就业报告》.日内瓦：世界经济论坛.
	(zu übersetzen)参考文献