English Learning with Duolingo: A case study of three EFL learners

BACKGROUND: Duolingo, as a popular language learning application, have approximately 200 million active users worldwide (Loewen et al., 2019). With artificial intelligence (AI) support, it can model and recommend courses for learners according to their performance in the placement test. This personalized learning experience makes learning relevant and lead to effective learning (Zou, Wang, Xie, Cheng, Wang, & Lee, 2020). Several studies compared Duolingo with regular classes and reported favorable learning outcomes (e.g., Rachels & Rockinson-Szapkiw, 2018). However, little research has been conducted to investigate the learning experiences of learners with different language proficiency levels.
METHODS: The current study conducted a case study on three EFL learners’ English learning with Duolingo. The three participants were of low, middle, and high proficiency levels respectively. Methods of think aloud, learning logs, and interviews were applied to investigate their learning experience. The participants learned with Duolingo for one month and reported their learning experiences by conducting thinking aloud protocols and writing learning logs twice a week. One month later, they were interviewed to further reflect on their learning experiences.
RESULTS: The results showed that the learners generally enjoyed learning with Duolingo, as it enabled them to learn, practice, and review language knowledge at their own paces. The AI algorithim embedded in it also assisted them in reviewing what they have learned previously at scientifically timed intervals. Take vocabulary learning as an example, if learners remembered target words correctly in consecutive two days, the system would prolong the review interval till the 4th day. Four days later, if they remembered it again, the review interval would be extended to 8th day, etc. If learners answered incorrectly, the words would be demoted to a shorter review interval. The learner with high proficiency also considered the gamified learning mode useful for grammar learning. The learners with low and middle proficiency reported that they benefited greatly from the personalized arrangements of the recommended courses (e.g., vocabulary, speaking, grammar).
CONCLUSIONS: Duolingo is conducive to successful language learning as it provides learners with personalized learning experiences and makes good use of spacing effect through AI algorithm. Copyright © 2020 The Education University of Hong Kong (EdUHK).