Addressing the poor adaptability of traditional static vocabulary teaching models and the limitations of existing machine learning schemes such as fixed forgetting models and single feature dimensions, this paper proposes a vocabulary learning optimization algorithm (DF-MMFL algorithm) that integrates dynamic forgetting perception and multimodal feature fusion. This algorithm comprises three core modules: a multimodal data preprocessing module that constructs a 12-dimensional feature vector covering four categories of information, including vocabulary difficulty and answering speed, through improved Min-Max normalization and mutual information entropy filtering; a dynamic forgetting model module that, based on Ebbinghaus forgetting theory, designs a forgetting coefficient correction factor using real-time answering data and iteratively updates the forgetting coefficient through gradient descent; and a reinforcement learning recommendation module that optimizes the recommendation strategy using a short-term and long-term weighted composite reward function and an improved Q-learning algorithm. A controlled experiment was conducted with 500 first-year non-English major students, divided into four groups. Results showed that the experimental group achieved memory retention rates of 82.3% and 71.5% after one and two weeks, respectively, representing increases of 28.1 and 30.3 percentage points compared to the control group. The overall accuracy rate of vocabulary application was 68.9%, the recommendation accuracy rate was 90.1%, and the model converged after only 120 iterations. This study demonstrates that the DF-MMFL algorithm can effectively improve vocabulary memory retention and application capabilities, optimizing algorithm performance.

Huizhuo Zhang, Yajun Zhao, Wen Bai. Research on Algorithms for Improving English Vocabulary Teaching Effectiveness Using Machine Learning. Frontiers in Educational Research (2026), Vol. 9, Issue 8: 121-128. https://doi.org/10.25236/FER.2026.090816.

[1] Ye Junyao, Su Jingyong, Wang Yaowei, & Xu Yong. A long-term memory prediction model for language learning based on LSTM. Journal of Chinese Information Processing, vol. 36, pp. 133-138, December 2022.

[2] Wang Aiqing. Intelligent evaluation technology: data-driven analysis in personalized teaching. Artificial Intelligence Education Research, vol. 1, pp. 41-56, January 2025.

[3] Hao Yongbin, Zhou Lanjiang, & Liu Chang. An end-to-end multi-task Lao word segmentation method based on LSTM. Journal of Chinese Information Processing, vol. 35, pp. 75-81, September 2021.

[4] Yin Zhige, & Lu Jianfeng. Research on chemical named entity recognition with fusion of radical features. Computer and Digital Engineering, vol. 51, pp. 809-816, April 2023.

[5] Chen Enhong, Liu Qi, Wang Shijin, Huang Zhenya, Su Yu, Ding Peng, ... & Zhu Bo. Key technologies and applications of adaptive learning for intelligent education. Journal of Intelligent Systems, vol. 16, pp. 886-898, May 2021.

[6] Huang Lihe, Qu Huiyu, & Yang Jingjing. Automatic text analysis of elderly discourse by computers: progress and prospects. Language Strategy Research, vol. 7, pp. 88-96, March 2022.

[7] Huang Zhenhua, Yang Shunzhi, Lin Wei, Ni Juan, Sun Shengli, Chen Yunwen, & Tang Yong. A review of knowledge distillation research. Chinese Journal of Computers, vol. 45, pp. 624-653, March 2022.

[8] Wu Yajuan, Niu Jiakui, Xie Hongtao, & Ma Ning. Named entity recognition in well control domain based on dictionary and character vector fusion. Journal of Hainan University (Natural Science Edition, Chinese and English), vol. 40, pp. 125-133, February 2022.

[9] He Xuejian, Chen Anqi, Guo Zhiqiang, Wang Zhiru, & Chen Qun. A Chinese Question Matching Method Based on Progressive Machine Learning. Journal of Engineering Science, vol. 47, pp. 79-90, January 2025.

[10] Liu Ying, Guo Yingying, Fang Jie, Fan Jiulun, Hao Yu, & Liu Jiming. A Review of Deep Learning Cross-Modal Image and Text Retrieval Research. Computer Science and Exploration, vol. 16, pp. 489-511, March 2022.