Two decades of digital humanities and artificial intelligence in language and literature studies: A bibliometric review of Scopus and Web of Science research (2004–2025)

The encounter between computational methods and the study of language and literature has moved from a specialist pursuit to a mainstream concern over the past two decades, an acceleration made abrupt by the public release of large language models (LLMs) in late 2022. However, the bibliometric evidence documenting this transformation is fragmented across incompatible databases, disciplinary labels, and short observation windows, leaving scholars without an integrated map of the field. This article synthesizes the quantitative findings of the principal Scopus- and Web of Science–indexed bibliometric studies published between 2004 and 2024, complemented by seminal works that define the intellectual tradition of Digital Humanities (DH) and computational literary studies. Following the PRISMA 2020 guidance, twenty-four core studies and foundational works were selected for a structured performance and science-mapping synthesis. The evidence shows a rapidly expanding but structurally uneven field: corpus sizes reported by prior analyses range from 185 documents in applied linguistics to more than 8,000 across the broad DH landscape; publication output in the artificial intelligence and language stream rose by roughly 190% between 2017 and 2022; and ChatGPT-focused educational research grew from a single indexed document in 2022 to 211 in 2023. Geographically, leadership is contested between the United States and China, while the field’s most cited outlet remains the former Literary and Linguistic Computing (now Digital Scholarship in the Humanities). Six thematic clusters emerged: natural language processing and intelligent tutoring, computer-assisted language learning, automated writing evaluation, computational stylistics, AI authorship and academic integrity, and speech-and multimodal interfaces. The review argues that the field’s central tension—between quantitative ambition and interpretive validation—remains unresolved and is intensified rather than settled by the generative AI. The implications for language and literature scholarship and an agenda for methodologically transparent computational inquiry are outlined.