📚 Duolingo English Test: How Technological Innovation Reshapes Language Proficiency Assessment | 多邻国英语测试：技术创新重塑语言能力评估
The Duolingo English Test (DET) has emerged as a disruptive force in the world of high-stakes language assessment. Leveraging cutting‑edge technologies such as computer‑adaptive testing, artificial intelligence, and remote proctoring, the DET offers a fully digital alternative to traditional exams like TOEFL and IELTS. Its rapid adoption by thousands of universities and institutions worldwide signals a paradigm shift in how we measure English proficiency—one that prioritises accessibility, efficiency, and scientific rigour.
多邻国英语测试（DET）已成为高风险语言评估领域的一股颠覆性力量。它借助计算机自适应测试、人工智能和远程监考等尖端技术，为托福和雅思等传统考试提供了一种全数字化的替代方案。全球数千所大学和机构迅速采纳该测试，标志着英语能力衡量方式的一次范式转变——更加注重可及性、效率和科学严谨性。
1. The Rise of the Digital‑Native Language Test | 数字原生语言测试的崛起
The Duolingo English Test was launched in 2016 by the company behind the world’s most popular language‑learning app. Its creators recognised that existing proficiency exams were expensive, time‑consuming, and inaccessible to many learners, particularly in developing countries. By building a test that can be taken anytime, anywhere on a computer with a camera, they addressed a massive unmet need for affordable, on‑demand certification.
多邻国英语测试于 2016 年由全球最受欢迎的语言学习应用的公司推出。其创建者意识到，现有的能力考试费用高昂、耗时且对许多学习者——尤其是发展中国家的学习者——难以企及。通过打造一个可以随时随地通过带摄像头的电脑参加的测试，他们满足了人们对经济实惠、按需认证的巨大未满足需求。
In less than a decade, the DET has been accepted by over 5,000 institutions, including Yale, Stanford, MIT, Imperial College London, and many government immigration bodies. Its growth reflects not only the convenience of remote testing but also a growing confidence in technology‑driven evaluation. The test has demonstrated that a short, adaptive experience can produce scores that correlate strongly with longer, established tests, while adding innovative measures of productive skills via recorded video and writing samples.
在不到十年的时间里，多邻国英语测试已被耶鲁、斯坦福、麻省理工学院、伦敦帝国理工学院等 5000 多所机构以及许多政府移民部门所接受。其增长不仅反映了远程测试的便利性，也体现了人们对技术驱动评估日益增长的信心。该测试表明，一次简短的自适应体验可以产生与传统较长测试高度相关的分数，同时通过录制的视频和写作样本增加了对产出技能的全新衡量维度。
2. Computer‑Adaptive Testing: The Core Engine | 计算机自适应测试：核心引擎
At the heart of the DET lies a sophisticated computer‑adaptive testing (CAT) algorithm. Unlike linear exams where every test‑taker sees the same questions, the DET continuously tailors the difficulty of items in real time based on the candidate’s responses. If a question is answered correctly, the next item becomes slightly harder; an incorrect answer results in an easier question. This dynamic process hones in on the test‑taker’s true ability level within a shorter time and with greater precision.
DET 的核心是一个复杂的计算机自适应测试算法。与每个考生看到相同题目的线性考试不同，DET 会根据考生的回答实时不断调整题目难度。如果一个回答正确，下一题的难度会略有提升；错误回答则导致下一个题目变简单。这一动态过程能在更短的时间内以更高的精度锁定考生的真实能力水平。
The CAT model is built on item response theory (IRT), which estimates a candidate’s ability (θ) while accounting for item parameters such as difficulty, discrimination, and guessing probability. The DET’s algorithm uses a pool of thousands of calibrated items, ensuring that no two test sessions are identical. This not only enhances test security but also reduces test‑taker fatigue and anxiety, as individuals rarely encounter items that are far too easy or impossibly hard for their level.
CAT 模型建立在项目反应理论（IRT）之上，该理论在评估考生能力（θ）的同时考虑了题目参数，如难度、区分度和猜测概率。DET 的算法使用了一个包含数千个经过校准的题目的题库，确保没有两次测试会话完全相同。这不仅增强了考试安全性，还减少了考生的疲劳和焦虑，因为个人很少会遇到远低于或远高于其水平的题目。
3. AI‑Powered Scoring: Beyond Human Raters | 人工智能评分：超越人工评分员
One of the most groundbreaking aspects of the DET is its use of artificial intelligence to evaluate open‑ended responses. The test’s scoring engines are trained on thousands of human‑rated samples for tasks such as “Read then Speak,” “Listen then Speak,” and “Write about the Photo.” These machine‑learning models assess not just linguistic accuracy but also fluency, pronunciation, lexical sophistication, grammatical range, and relevance of content.
DET 最具突破性的一个方面是其使用人工智能评估开放式回答。该测试的评分引擎针对“阅读后说话”、“听后说话”和“描述照片”等任务，在数千个人工评分样本上进行了训练。这些机器学习模型不仅评估语言准确性，还评估流利度、发音、词汇复杂度、语法范围和内容相关性。
s. For the video interview and writing sample sent to institutions, a separate layer of AI‑ s that the automated scoring correlates with human scores at levels above r = 0.9, rivalling the reliability of expert human assessors while eliminating biases related to accent, appearance, or fatigue.
人工智能与分布式人工评分员网络的结合确保了评分的一致性和公平性。对于发送给机构的视频面试和写作样本，另一层人工智能辅助审核会检查异常模式，但最终解释权仍交予招生官员。多邻国发表的研究表明，自动评分与人工评分的相关系数高于 0.9，可与专家人工评估的可靠性相媲美，同时消除了与口音、外貌或疲劳相关的偏见。
4. Integrated Skills Assessment in One Flow | 一体化技能评估流程
Traditional proficiency tests often separate skills into distinct, lengthy sections. The DET interweaves reading, writing, listening, and speaking tasks in a seamless sequence that mirrors real‑world language use. A typical test section might ask a candidate to read a prompt, listen to an audio clip, and then speak a response—a truly integrated task requiring multiple cognitive processes simultaneously.
传统的能力考试通常将各项技能分割成独立而冗长的部分。DET 将阅读、写作、听力和口语任务无缝交织在一起，模拟真实的语言使用。一个典型的测试部分可能会要求考生阅读提示、听取音频片段，然后说出回应——这是一项需要同时动用多重认知过程的真正综合性任务。
This design is made possible by the test’s digital nature. Interactive question types such as “Interactive Reading” require candidates to complete missing words in a passage and then answer comprehension questions that build upon the reconstructed text. The “Interactive Listening” scenario simulates a conversation where the test‑taker first listens, then responds by selecting appropriate replies, effectively blending comprehension and pragmatic competence. The result is a richer, more authentic performance sample in just under one hour.
这种设计得益于测试的数字化本质。“互动阅读”等互动题型要求考生补全文章中缺失的单词，然后回答基于重构文本的理解题。“互动听力”场景模拟对话，考生先听，然后选择合适的回答，有效地融合了理解能力和语用能力。最终，在不到一个小时的时间里，便获取了更丰富、更真实的表现样本。
5. The Video Interview and Writing Sample: Unscripted Production | 视频面试与写作样本：无脚本产出
Beyond the scored section, the DET includes an unscored but critically important 10‑minute video interview and a writing sample. The candidate responds to a spoken prompt by recording a short speech, and writes an extended response to a written topic. These unscripted tasks are not graded by the AI but are shared directly with institutions as a verifiable sample of spontaneous communication.
在计分部分之外，DET 还包含一个不计分但至关重要的 10 分钟视频面试和一个写作样本。考生对口语提示做出回应并录制一段简短的演讲，并就一个书面主题撰写延展性回答。这些无脚本任务不由人工智能评分，而是直接分享给机构，作为可验证的自发性沟通样本。
This dual‑pronged approach addresses a long‑standing criticism of standardised tests: that they fail to capture a learner’s ability to produce language under realistic, low‑preparation conditions. Admissions officers report that the video interview provides valuable insight into a student’s pronunciation, spontaneity, and interpersonal readiness far beyond what a numeric score can convey. It also acts as an integrity check, allowing institutions to compare a candidate’s on‑camera performance with their written personal statement.
这种双管齐下的方法回应了标准化考试长期以来的一个批评：它们未能捕捉学习者在真实、低准备条件下的语言产出能力。招生官员报告说，视频面试能提供远远超出数字分数所能传达的、有关学生发音、自发性和人际交往准备度的宝贵洞察。它还能作为一种诚信核查，让机构可以将考生在镜头前的表现与其书面个人陈述进行比较。
6. Test Security and Remote Proctoring: AI Vigilance | 测试安全与远程监考：人工智能的警戒
Delivering a high‑stakes test in a home environment demands uncompromising security. The DET uses a multi‑layered remote proctoring system powered by artificial intelligence. Before the test begins, the software verifies identity by matching a government‑issued ID with a live photo and video scan. Throughout the session, the camera, microphone, and screen are monitored for suspicious behaviours such as looking away from the screen, the presence of another person, or attempts to launch unauthorised applications.
在家庭环境中开展高风险测试需要绝不含糊的安全保障。DET 采用了一套由人工智能支持的多层次远程监考系统。测试开始前，软件会通过将政府签发的身份证件与实时照片和视频扫描进行比对来验证身份。整个测试过程中，摄像头、麦克风和屏幕都会受到监控，以检测可疑行为，例如视线偏离屏幕、有他人在场或试图启动未经授权的应用程序。
If a violation is detected, the system can flag the session for human review. A dedicated team of proctors then watches the flagged recording and decides whether to certify or invalidate the results. Cutting‑edge features such as gaze estimation, keystroke dynamics analysis, and background audio analysis make it extremely difficult for test‑takers to cheat. Duolingo reports that fewer than 0.5% of tests are invalidated for integrity reasons, a rate comparable to that of large‑scale in‑person administrations.
如果检测到违规行为，系统会将本次测试标记为需要人工审核。专门的监考团队随后观看被标记的录像，并决定是认证成绩还是使其作废。视线估计、击键动态分析和背景音频分析等尖端功能使得考生极难作弊。多邻国报告称，因诚信问题而作废的测试不到 0.5%，这一比例与大规模线下考试相当。
7. Accessibility and Convenience: Democratising Language Assessment | 可及性与便利性：语言评估的民主化
A single TOEFL iBT test can cost upwards of USD 250, and IELTS fees are similarly prohibitive for many families, especially when travel to a test centre is required. The Duolingo English Test costs just USD 59, with unlimited score sending to institutions at no extra charge. This radical affordability, combined with the ability to test from home, has eliminated two of the biggest barriers to international education: cost and geography.
一次托福网考的费用可能高达 250 美元以上，雅思考试的费用对许多家庭同样令人望而却步，尤其是在需要前往考点的情况下。多邻国英语测试的费用仅为 59 美元，并且可以无限制地免费将成绩发送给院校。这种巨大的价格优势，加之在家即可参加考试，消除了国际教育中两大障碍：成本和地理位置。
For students in rural areas, conflict zones, or low‑income communities, the DET represents a gateway that previously did not exist. The test requires only a computer with a stable internet connection and a quiet room. The entire experience, from registration to receiving scores, takes under 48 hours, compared with weeks for traditional tests. This rapid turnaround has proved vital during admissions crises, scholarship deadlines, and pandemic‑era disruptions.
对于农村地区、冲突地区或低收入社区的学生而言，DET 代表着一条此前不曾存在的通道。考试只需一台能稳定连接互联网的电脑和一个安静的房间。从注册到收到成绩的整个体验不到 48 小时，而传统考试则需要数周。这种快速周转在招生危机、奖学金截止日期和疫情造成的混乱时期被证明至关重要。
8. Validity and Reliability: Scientific Rigour | 效度与信度：科学严谨性
ic properties. The DET’s development team has published extensive validity studies in peer‑ alidity is supported through confirmatory factor analyses that align item types with the communicative language ability model.
一项新测试要获得顶尖院校的信任，必须展示出扎实的心理测量学特性。DET 的开发团队在同行评审期刊上发表了大量效度研究，显示出与托福网考（相关系数 r = 0.77 至 0.85）和雅思（r ≈ 0.80）的高度相关性。通过验证性因素分析将题型与交际语言能力模型对齐，支持了结构效度。
Reliability is ensured via the CAT algorithm, which yields standard errors of measurement consistently below 3 points on the 10–160 scale. Additionally, DET reports subscores for Literacy, Comprehension, Conversation, and Production, each derived from multiple item types to reduce construct‑irrelevant variance. Independent research commissioned by Duolingo has found no significant score differences across gender, first‑language background, or socioeconomic status, evidence that the test mitigates cultural and economic bias.
信度通过 CAT 算法得以保证，其在 10–160 分量表上的测量标准误始终低于 3 分。此外，DET 还报告读写、理解、交谈和产出四个单项分数，每个分数均源自多种题型，以减少与构念无关的变异。由多邻国委托的独立研究发现，在性别、母语背景或社会经济地位方面均未出现显著的分数差异，证明该测试减轻了文化和经济偏见。
9. Comparison with TOEFL and IELTS: Convergence and Divergence | 与托福和雅思的比较：趋同与差异
While the DET shares the same communicative foundation as TOEFL and IELTS, it differs markedly in format and philosophy. TOEFL iBT lasts about 3 hours and IELTS about 2 hours 45 minutes; the DET is typically completed in 1 hour. The traditional tests include lengthy academic lectures and essay writing, whereas the DET focuses on shorter, more varied tasks that adapt to the learner’s level.
尽管 DET 与托福和雅思拥有相同的交际语言基础，但在形式和理念上却存在显著差异。托福网考大约持续 3 小时，雅思大约 2 小时 45 分钟；DET 通常 1 小时内完成。传统考试包含冗长的学术讲座和论文写作，而 DET 侧重于更短小、更多样化且能适应学习者水平的任务。
The DET’s scoring scale (10–160) aligns with CEFR levels from A1 to C2, but it does not provide a direct score‑to‑CEFR mapping for every subscore. Recent concordance tables published by Duolingo help institutions interpret results: a DET score of 120 roughly corresponds to IELTS 6.5 and TOEFL 80–90. However, the DET’s emphasis on integrated production and its video sample provide unique evidence of communicative ability that static scores cannot replicate.
DET 的评分量表（10–160）与 CEFR 的 A1 至 C2 等级对齐，但并非每个单项分都直接映射到 CEFR。多邻国最近发布的分数对照表有助于院校解读成绩：DET 120 分大致相当于雅思 6.5 分和托福 80–90 分。然而，DET 对综合产出的强调及其视频样本，提供了静态分数无法复制的独特交际能力证据。
10. Impact on Global Education and Immigration | 对全球教育与移民的影响
The DET’s rapid acceptance has reshaped international student mobility pathways. During the COVID‑19 pandemic, when test centres worldwide shut down, the DET became a lifeline for applicants who otherwise would have deferred or cancelled their plans. Governments in Ireland, Canada, and Australia have begun accepting the DET for visa and citizenship applications, further cementing its status as a high‑stakes instrument.
DET 的迅速被接受重塑了国际学生流动的途径。在新冠疫情导致全球考试中心关闭期间，DET 成为了那些否则可能推迟或取消计划的申请者的生命线。爱尔兰、加拿大和澳大利亚的政府已开始接受 DET 用于签证和公民身份申请，进一步巩固了其作为高风险工具的地位。
By dramatically lowering the financial and logistical barriers to certification, the DET has widened the applicant pool for universities. Institutions now see applications from talented students in sub‑Saharan Africa, South Asia, and Latin America who previously could not afford standardised tests. This diversification enriches campuses and aligns with the global equity goals of higher education.
通过大幅降低认证的财务和后勤障碍，DET 拓宽了大学的申请者群体。院校现在收到来自撒哈拉以南非洲、南亚和拉丁美洲的有才华学生的申请，而他们此前无力承担标准化考试的费用。这种多元化丰富了校园，也与高等教育的全球公平目标保持一致。
11. Future Directions: Continuous Adaptive Learning and Gamification | 未来方向：持续自适应学习与游戏化
Duolingo has hinted at deeper integrations between the English Test and its broader learning ecosystem. One prospect is a continuous assessment model where learners’ progress in the Duolingo app contributes to a dynamic proficiency estimate, reducing the need for a single high‑stakes event. Such a system would resemble the evolving gradebook, constantly updated by micro‑interactions and practice sessions.
多邻国已经暗示将在英语测试与其更广泛的学习生态系统之间进行更深度的整合。一个前景是持续评估模式，学习者在多邻国应用中的进步有助于形成动态的能力评估，从而减少对单次高风险考试的需求。这样一个系统将类似于不断更新的成绩册，通过微交互和练习课持续刷新。
Gamification elements, already central to the Duolingo learning app, could find their way into assessment. Short, game‑like diagnostic tasks that measure vocabulary depth, grammatical sensitivity, or processing speed could provide formative feedback long before a candidate sits the official test. These innovations would further blur the line between learning and assessment, turning language evaluation into an iterative, low‑anxiety process aligned with modern pedagogical principles.
游戏化元素已是多邻国学习应用的核心，它们也可能进入评估领域。简短的类游戏诊断任务可以衡量词汇深度、语法敏感度或处理速度，在考生参加正式测试之前许久便提供形成性反馈。这些创新将更进一步模糊学习与评估之间的界限，将语言测评转变为与现 代教学原则相一致的迭代式、低焦虑过程。
12. Conclusion: A Blueprint for the Future of Testing | 结论：未来测试的蓝图
The Duolingo English Test is more than a convenient alternative to legacy exams—it represents a fundamental rethinking of what a language test can be. By weaving together adaptive algorithms, artificial intelligence scoring, and secure remote delivery, it has produced an instrument that is scientifically sound, globally accessible, and remarkably efficient. Its success challenges the testing industry to embrace innovation without sacrificing rigour.
多邻国英语测试不仅仅是传统考试的一个便捷替代方案——它代表了对语言测试本质的根本性反思。通过将自适应算法、人工智能评分和安全的远程交付交织在一起，它创造了一个科学性强、全球可及且异常高效的工具。它的成功挑战着考试行业，在拥抱创新的同时不能牺牲严谨性。
As the DET continues to evolve, it will likely drive further convergence between language learning and language assessment. The ultimate beneficiary is the learner, who gains a fairer, faster, and more affordable pathway to demonstrating their English proficiency to the world.
随着 DET 的不断发展，它很可能会推动语言学习与语言评估的进一步融合。最终的受益者是学习者，他们获得了一条更公平、更快捷和更实惠的途径，向世界证明自己的英语能力。
Published by TutorHao | Duolingo English Test Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图