Current Trends, Innovations, and Challenges in AI-Powered Language Teaching: A Systematic Literature Review

In this study, we successfully identified 808 articles from various databases: 623 from Scopus, 73 from Web of Science, and 112 from ProQuest, with no entries from the register. Before screening, 364 duplicate records were removed, along with 46 records deemed ineligible by the automation tool, leaving 398 records to proceed to the screening stage. During the screening stage, 260 records were excluded because they did not meet the inclusion criteria, leaving 138 reports for retrieval. Of these, 50 reports were not retrieved, reducing the number of reports assessed for eligibility to 88. Of the 88 reports assessed, 50 articles were excluded because the article title and abstract were irrelevant. In comparison, 260 reports were excluded because the substance of the article did not match the study criteria. This comprehensive review resulted in 88 new studies that were included in the final analysis. The selected articles were then reviewed by 2 independent reviewers and the results were used to calculate the kappa value.本研究从多个数据库中共检索得到808篇文献，其中斯高帕斯数据库（Scopus）623篇、Web of Science数据库（Web of Science）73篇、ProQuest数据库（ProQuest）112篇，未从登记库中获取到相关文献。正式筛选前，研究团队剔除了364条重复记录，另有46条记录被自动化工具判定为不符合要求，最终剩余398条记录进入筛选环节。在筛选阶段，260条记录因未满足纳入标准被排除，仅剩138篇文献待获取全文。其中50篇未能检索到有效资源，因此进入资格评估环节的文献数量缩减至88篇。在对这88篇文献进行资格评估时，50篇因标题与摘要内容无关被排除；另有260篇文献因研究内容不符合本次研究标准被排除。本次全面系统的文献综述最终纳入88篇新研究用于最终分析。随后，由2名独立评审员对入选文献开展审校，并基于审校结果计算kappa系数。