Abstract This paper presents a three-part methodology for identifying special-purposes words to teach in data-driven learning (DDL) vocabulary activities. Previous methods have focused on either identifying important words for an English for Specific Purposes (ESP) context or identifying learner vocabulary gaps—and little or no research has addressed how to determine whether specific words are well suited to teaching through DDL rather than through a more traditional, deductive approach. The system in this study used a corpus-based approach to identify words that are (1) important to the ESP context, (2) difficult for students, and (3) well suited to teaching through DDL. This study applied the system to the context of civil engineering and found that it was overall effective in identifying 18 words that are prevalent in civil engineering writing, that were problematic for the students whose writing was examined, and that showed indications of being well suited to DDL. This paper discusses a drawback of the system—the time required to apply it—and discusses two valuable strengths: revealing how the words functioned in civil engineering discourse and identifying words not overtly connected to engineering (e.g., existing or using) that could be easily overlooked by an instructor.

中文翻译：

---

选择专业词汇来教授数据驱动学习：土木工程的一个例子

摘要 本文提出了一种由三部分组成的方法，用于识别在数据驱动学习 (DDL) 词汇活动中教授的特殊用途单词。以前的方法侧重于识别特定用途英语 (ESP) 上下文的重要单词或识别学习者的词汇差距——很少或没有研究解决如何确定特定单词是否非常适合通过 DDL 而不是通过更多传统的演绎方法。本研究中的系统使用基于语料库的方法来识别 (1) 对 ESP 上下文很重要、(2) 对学生来说困难以及 (3) 非常适合通过 DDL 进行教学的单词。本研究将该系统应用于土木工程的上下文中，发现它在识别土木工程写作中普遍存在的 18 个单词方面总体上是有效的，这对于接受写作考试的学生来说是有问题的，这表明他们非常适合 DDL。本文讨论了该系统的一个缺点——应用它所需的时间——并讨论了两个有价值的优势：揭示了土木工程话语中的词如何发挥作用，以及识别与工程（例如，现有或使用）没有明显联系的词，这些词可以很容易地被导师忽视。