[英語教學資源中心] 生成式AI襄助英語教學　生成式字典與EMI 語料庫

國立政治大學英語教學資源中心主辦，人工智慧跨域研究中心、雙語及多元文化推動辦公室協辦，名為「Jason老師的生成式AI雜記」的講座於5月24日上午在商學院玉山國際廳舉行。本次講座邀請國立清華大學資訊工程學系教授張俊盛主講，探討人工智慧（AI）科技如何有效地幫助閱讀與寫作，以及如何應用於英語授課（EMI）課程。本講座吸引各領域教授、大學及研究生、中等學校師生，以及社會人士的參與。此外，數十名其他大學的參與者，也透過網路直播的方式參與本次講座。本場活動由英語教學資源中心主任鍾曉芳開場，她提及講者張俊盛是最投入英語教學的資訊工程教授之一，並衷心感謝他一直以來在這領域的貢獻。

張俊盛以Nils J. Nilsson著作《The Quest for Artificial Intelligence》中的名言開場：「AI的探索始於夢想——就如其他的探索一樣」（The quest for AI begins with dreams – as all quests do）。講者表示今日生成式AI達到發展的頂峰，是基於過往不少人士的無私、慷慨與利他主義，這些使各式的資訊能夠串連在一起。接著，講者以科普的形式，闡述過去數十年以來不同重要網路工具的發明（例如：全球資訊網、維基百科與變換器），分享人們對AI觀感上的轉變（樂觀與悲觀），更介紹了ChatGPT及大型語言模型（LLMs）如何產生以及其受人重視的原因。

接著講者講述生成式AI幫助英語教學與學習的幾個方法：生成式字典、EMI語料庫（corpus）與EMI語料索引（concordance）。例如，利用語言學習工具Linggle，當中的生成式字典可以透過字詞搜索，解釋該字詞在不同語境中的意義。學習者再也不用在紙本或網路字典查閱不懂的字詞，但最終得到的僅是脫離語境的解釋。利用該工具，學習者能大幅減少在閱讀期間不必要的干擾、降低認知上的負擔，還能延長知識在記憶裡保留的時間。事實上，Linggle還有其他不同的功能，例如在寫作時建議適合的用字或字詞搭配、語法修改，更提供了不同語法主題的教學影片，以及個人化的文章評分等服務。

講者深信生成式AI有助於EMI的進行。例如，美國不少大學的開放式課程（OCW）就是EMI語料庫典型的材料，對想進行EMI課程的教師而言，十分具有參考價值。耶魯大學的40門課程，總計1,073個課堂的影片及教學內容，均有逐字稿上傳至網路（oyc.yale.edu）。若將所有的逐字稿放到資料夾裡，並透過Linggle工具內置字典的協助，在英語授課課堂裡常見的用語，就會一目了然。教師還能利用Google或其他方式進行語料索引（同時在搜尋器裡輸入「oyc.yale.edu」），就能進一步了解這些用語會在何種課堂情況中被使用。另外，KWIC搜尋也可以在AntConc中進行：選取課堂內容逐字稿的檔案後，相關的常用詞串（lexical bundle）就能被呈現出來。張俊盛更示範了如何透過Unix grep的N-Gram，以獲取相關用語更精細的搜尋結果。

在問答環節裡，講者與校內外的參與者熱烈互動，帶領參與者們看見生成式AI應用在教學中的益處，並引發其共鳴及高度讚賞。本中心再次對張教授帶來生成式AI在EMI的創新教學啟示，表達由衷的感謝。 

中心官網：https://www.eminccu.com/ 

中心FB：https://www.facebook.com/nccu.emi/ 

中心IG：https://www.instagram.com/nccu.emi/ 

On the morning of May 24th, 2024 (Friday), Professor Jason S. Chang from the Department of Computer Science at National Tsing Hua University was cordially invited to give a lecture entitled “Jason’s Gen AI Diary of Generative Dictionary and EMI.” This event, organized by the EMI Resource Center at National Chengchi University and co-hosted by the Interdisciplinary Artificial Intelligence Center and the Bilingual Education and Multicultural Promotion Office at National Chengchi University, took place in the E. Sun Hall at the College of Commerce, NCCU. The lecture provided an insightful introduction to how AI technology can be used to effectively assist in reading, writing, and teaching in EMI. It attracted a wide range of audience members, including professors, graduates, undergraduates, language teachers, high school students, and the general public. Additionally, dozens of participants from other universities joined the lecture online. The event was opened by Professor Siaw-Fong Chung, the Center’s director, who regarded Professor Chang as one of the computer science professors most dedicated to English language teaching and expressed her sincere gratitude for his contributions to the field.

Prof. Chang began his talk with a quote from the book “The Quest for Artificial Intelligence,” written by Nils J. Nilsson and published by Cambridge University Press in 2009: “The quest for AI begins with dreams – as all quests do.” He used this quote to emphasize that the current advancements in Generative AI are largely due to the unselfishness, generosity, and altruism of numerous individuals who have pieced together various bits of information. He then proceeded to introduce important tools, such as ChatGPT and large language models (LLMs), from the perspective of popular science. He narrated the differences in perceptions of AI, both optimistic and pessimistic, and discussed the emergence of significant inventions like the World Wide Web, Wikipedia, and the Transformer model over the past few decades.

When it comes to how Generative AI benefits English teaching and learning, the speaker proposed several means: Generative Dictionary, EMI Corpus, and EMI Concordance. Instead of toilsome reading for understanding unknown words in dictionaries, whether online or print, which explain words out of context, Prof. Chang introduced a language learning toolkit called Linggle. Linggle’s generative dictionary can explain vocabulary items in context with just a few words, significantly minimizing disruptions to the flow of reading, reducing learners’ cognitive load, and enhancing their retention. Additionally, Linggle offers functions such as suggestions for suitable collocations in writing, videos teaching a spectrum of language items, error correction for writing, and personalized essay scoring.

The speaker was deeply convinced that Generative AI is conducive to EMI. For example, open courseware (OCW) from many universities in the US is a typical source of EMI corpus, serving as a useful reference for the language used in teaching English. Notably, Yale University provides YouTube videos and transcripts of 1,073 lessons from 40 courses in diverse disciplines online (oyc.yale.edu). When all transcripts are compiled into a folder, the Linggle Toolkit with Built-in Dictionaries can identify phrases commonly used in EMI lessons. Teachers can find the contexts in which these phrases are used through various means such as Google (by keying in the link “oyc.yale.edu” in the search engine) or performing a KWIC (key word in context) search in AntConc, selecting transcript files of courses to list such lexical bundles. More refined search results of utterances commonly adopted in EMI lessons, as demonstrated by Prof. Chang, can also be achieved through N-Gram generations using Unix grep.

The enthusiastic interaction between the speaker and participants from various backgrounds, especially in-service language teachers, during the question-and-answer session highlighted the widely recognized and highly acclaimed importance and advantages of utilizing Generative AI in teaching and learning. Therefore, we express our greatest gratitude to Prof. Chang for his inspirational enlightenment on the innovative pedagogical implications of Generative AI in EMI.

For more information, please check our website：

www.eminccu.com

FB：https://www.facebook.com/nccu.emi

IG：https://www.instagram.com/nccu.emi