How AI Is Transforming Vocabulary Learning in Taiwan
As artificial intelligence (AI) becomes increasingly integrated into everyday life, educators are exploring whether it can also serve as a powerful tool for language learning. Professor Ting-Chia Hsu, Distinguished Professor in the Department of Technology Application and Human Resource Development at National Taiwan Normal University, led a research team investigating how AI-based image recognition technology can be applied to English vocabulary acquisition. The findings indicate that combining AI with self-directed learning strategies not only improves elementary school students’ vocabulary learning outcomes, but also strengthens their learning behaviors and motivation.
The origins of the project can be traced back to the COVID-19 pandemic in 2021. As schools adopted online and hybrid learning under the government’s policy of “suspending classes without suspending learning,” students’ capacity for self-directed learning emerged as a critical factor influencing educational outcomes. Hsu regards that period as a turning point in both technological development and learning environments. Her team sought to provide students with a new learning approach that would enable them to build personalized vocabulary databases by using AI to recognize real-world objects.
To achieve this goal, the research team developed a mobile application that integrates AI image recognition with Self-Regulated Learning (SRL) strategies. Students can photograph everyday objects, such as toothbrushes, newspapers, or schoolbags, and receive instant identification through the system, along with bilingual Chinese-English vocabulary, text-to-speech functionality, and personalized dictionary records. This design allows learners to connect vocabulary with real-world experiences, creating a dynamic and repeatable learning cycle.
In the study, students were divided into two groups: one using AI image recognition (AI-IR) and the other using optical character recognition (AI-OCR). The results revealed that students who used AI-IR in conjunction with self-regulated learning strategies achieved significantly better vocabulary acquisition outcomes than those using AI-OCR. According to Hsu, the key advantage lies in AI-IR’s ability to recognize concrete objects rather than text alone, enabling learners to connect language learning with authentic life contexts and thereby creating a more meaningful learning experience.
A closer examination of student behavior showed that the AI-IR group engaged in more active learning practices, including peer discussion, self-monitoring, repeated verification of answers, and observation of classmates’ learning processes. By contrast, students in the AI-OCR group exhibited less interaction and a narrower range of learning behaviors. These findings suggest that AI image recognition not only improves learning performance but also promotes student engagement and participation.
.jpg)
Figure caption: The research team developed a mobile application integrating AI image recognition, enabling students to learn vocabulary by photographing real-world objects.
With regard to learning attitudes, statistical measures showed no significant difference in the two groups’ learning anxiety. However, classroom observations revealed that students in the AI-IR group were more willing to participate in learning activities, displayed a more relaxed learning atmosphere, and demonstrated stronger exploratory tendencies. Hsu believes that learning vocabulary through photographing real objects combines enjoyment with authenticity, reducing anxiety associated with rote memorization, while making learning more vivid and engaging.
The project also faced several implementation challenges. Hsu noted that elementary school students were initially unfamiliar with AI-based tools. Poor image quality or insufficient lighting could result in recognition errors, while students’ curiosity about surrounding objects occasionally distracted them from the learning task. To address these issues, the team provided brief operational training before the activities began, successfully increasing recognition accuracy to over 98 percent. Teachers also used worksheets and instructional guidance to help students remain focused on target vocabulary items.
Discussing the practical implications of the research, Hsu explained that teachers can readily incorporate AI image-recognition applications into language classrooms. By photographing real-world objects, students receive immediate vocabulary items, pronunciation support, and translations, all of which can be stored in a personalized dictionary. This approach enables learners to understand and remember vocabulary within authentic contexts. By linking textbook content to everyday objects, learning moves beyond abstract words on a page and becomes a tangible, interactive experience.
The AI-supported learning activities were also closely integrated with self-regulated learning strategies. The research team found that students maintained clear objectives throughout the planning, implementation, and reflection stages of learning. Teachers can further cultivate autonomous learning abilities through task sheets, collaborative group work, and self-assessment activities. Working in pairs, students discussed answers, verified recognition results, and checked spelling and pronunciation, thereby strengthening both vocabulary acquisition and collaborative skills.
Looking ahead, Hsu believes this project represents only the beginning of AI applications in education. Future developments could extend across languages and disciplines, integrating speaking, writing, reading comprehension, science, and STEM education to serve a broader range of learners. AI systems may also provide sentence-construction support and contextual suggestions, expanding learning from isolated vocabulary items to practical language use, and creating a more comprehensive language-learning experience. Beyond this, AI technologies may increasingly function as learning companions and instructional assistants, supporting students’ self-monitoring and strategic learning processes.
From the watershed moment of the pandemic to the widespread adoption of AI, education is entering a new era of possibility. AI image recognition offers a multisensory learning experience that combines visual, auditory, textual, and hands-on elements. Such multimodal stimulation can strengthen cognitive connections, improve memory retention, and enhance language comprehension. By encouraging students to explore the world around them through photography, the technology stimulates curiosity, increases engagement, and reduces learning anxiety, ultimately fostering deeper and more meaningful learning experiences.
Source: Hsu, T. C., Chang, C., & Jen, T. H. (2024). Artificial Intelligence image recognition using self-regulation learning strategies: effects on vocabulary acquisition, learning anxiety, and learning behaviours of English language learners. Interactive Learning Environments, 32(6), 3060-3078. https://doi.org/10.1080/10494820.2023.2165508
Ting-Chia Hsu Distinguished Professor | Department of Technology Application and Human Resource Development
Professor Ting-Chia Hsu is a Distinguished Professor in the Department of Technology Application and Human Resource Development at National Taiwan Normal University. Her expertise spans computational thinking education, AI literacy, and adaptive digital learning. She has long been engaged in the development of educational technologies, instructional materials, and learning tools, producing work with substantial educational impact that has also reached international markets. She currently serves as chief editor of the Kang Hsuan Information Technology textbook series for Taiwan’s junior high school curriculum guidelines, and as chief editor of the 2026 Taipei City White Paper on AI Education Policy. She also founded the Taiwan Association for Learning Analytics and Artificial Intelligence Literacy Development (TALA-AILD). Her honors include the Wu Ta-You Memorial Award from the National Science and Technology Council and the Outstanding Researcher Award from the Asia-Pacific Society for Computers in Education. She has also been repeatedly recognized among Stanford University’s World’s Top 2% Scientists.