Escrita e autopós-edição de resumos acadêmicos em língua adicional: uma análise de elementos retóricos em corpus comparável

Abstract

English, as the dominant language of scientific communication, poses significant challenges for non-Anglophone researchers, who are required to read, conduct research, and publish in an additional language. In this context, machine translation has emerged as an accessible alternative to support academic writing, particularly when combined with self-post-editing, here understood as the revision carried out by the author after the machine-generated translation. Building on Nunes’ 2021 research project, this study investigates the rhetorical organization in a comparable corpus composed of academic abstracts written in English as an additional language (L2) and of abstracts self-post-edited in the same language from machine translations of texts originally written in Brazilian Portuguese (L1). Grounded in the expanded CARS (Create a Research Space) model, the study examines the distribution of rhetorical moves and steps in a comparable corpus comprising 36 abstracts produced by doctoral students from the Federal University of Minas Gerais in the fields of Exact and Earth Sciences, Humanities, and Applied Social Sciences. The texts were produced in an empiricalexperimental context and analyzed through both manual and automatic classification using the INCEpTION annotation tool. The results reveal a consistent predominance of Move 3, associated with the presentation of objectives, methods and results, across all modes of textual production, whereas Moves 1 and 2, related to topic contextualization and niche construction, occur less frequently and show irregular distribution across disciplinary areas. The comparison between texts written directly in L2 and machine-translated and self-post-edited texts in this language indicates that the latter primarily functions as a mechanism for textual refinement, enhancing fluency and informational clarity without substantially altering the overall rhetorical organization of abstracts. These findings highlight the importance of training for rhetorical organization within text production and of genre-based literacy so as to contribute to a broader understanding of machine translation as a practice within contemporary academic writing.