Background: Argumentative writing represents a core dimension of academic literacy within higher education; however, research concerning “argument,” “argumentation,” and “argumentative writing” remains dispersed across distinct disciplinary paradigms and commonly draws upon non-equivalent conceptual definitions and analytical methodologies. This fragmentation has practical consequences for teaching and assessment, particularly as technology-enhanced writing environments and AI-mediated support expand the range of tools used to scaffold and evaluate argumentation. Purpose: To map the conceptual and methodological approaches to studying argument in academic writing in higher education, including definitions of argumentative writing, the argument models employed, the ways argument quality is operationalized, the dominant research directions, and the role of digital and AI-mediated environments in the teaching and assessment of argumentation. Materials and Methods: The review followed PRISMA-ScR guidance and used a PCC framework to define the sources eligibility. A structured search was conducted in Scopus on 27 September 2025 using predefined Boolean queries covering argumentative writing, academic writing, and higher education; backward reference searching was also applied to identify additional relevant studies. Records were screened against inclusion and exclusion criteria, and data were charted in a standardised extraction form. The synthesis combined (a) bibliometric keyword co-occurrence mapping using VOSviewer to identify major thematic blocks and (b) expert thematic coding to interpret conceptualisations, models, and methodological patterns across the literature under analysis. Inter-coder agreement was established during iterative coding. Results: Ninety-five sources were included in this review. Publication output increased after 2018, with the largest share of studies appearing around 2019–2020, and the evidence base was geographically concentrated in Asia and the Americas. Bibliometric mapping and expert synthesis converged on several recurring blocks: theoretical and definitional work on argument/argumentation in academic writing; pedagogical studies on teaching argumentative writing and related scaffolding; assessment-oriented research (rubrics, indicators of quality, and technology-supported evaluation); sociolinguistic and interlinguistic perspectives, especially in EFL/L2 contexts; and an emergent strand focused on digital writing environments, automated feedback, and AI-enabled support. Across the corpus, Classical, Toulmin-based, and Rogerian traditions function as influential modelling frameworks, but they are applied inconsistently, and operationalisations of argument quality vary substantially: most commonly privileging detectable structural elements over comparably stable measures of reasoning strength or epistemic integration. Conclusion: The review shows that the development of research on argumentative writing in higher education is constrained not by a lack of studies, but by the absence of conceptual and methodological coherence. Differences in the definitions of argument, the models used for its analysis, and the approaches to assessing its quality limit the comparability of findings and complicate the translation of research insights into teaching and assessment practice. Under these conditions, the integration of argumentation theories with more robust and substantively oriented approaches to argument assessment becomes particularly important, especially against the backdrop of the growing automation of the structural aspects of writing in digital and AI-mediated environments.