The Efficacy, Integration, and Pedagogical Implications of Generative AI in Second Language Acquisition

Abstract

The advent of sophisticated Generative Artificial Intelligence (GenAI), particularly Large Language Models (LLMs) such as ChatGPT and Gemini, has precipitated a significant paradigm shift in educational technology. This paper presents a systematic literature review (SLR) of empirical studies published between late 2022 and 2025 to synthesize the current understanding of GenAI's role in language teaching and learning. The objective is to move beyond speculative discourse and consolidate evidence-based findings regarding the efficacy, integration strategies, and perceived challenges of these tools in Second Language Acquisition (SLA). Employing the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, this review analyzed 42 empirical articles sourced from major academic databases, focusing on studies with measurable linguistic outcomes or in-depth qualitative feedback. The results reveal three predominant themes: (1) GenAI demonstrates a significant positive impact on productive language skills, particularly writing fluency, grammatical accuracy, and lexical diversity, often functioning as an advanced automated writing evaluation (AWE) tool. (2) Student and educator perceptions are marked by a dichotomy; students report high levels of motivation, engagement, and reduced language anxiety, while educators express profound concerns regarding academic integrity, algorithmic bias, and student over-reliance. (3) The primary challenges identified are not purely technological (e.g., 'hallucinations' or inaccuracies) but are deeply pedagogical, centering on the lack of established frameworks for effective integration. The discussion posits that GenAI functions as a 'pedagogical catalyst,' demanding a shift from product-oriented assessment to process-oriented learning. This review concludes that GenAI's potential is maximized not as a replacement for instructors, but as a supplementary tool that requires a new, critical "AI literacy" for both learners and teachers.