PEUMO: computational tool to support academic writing in engineering

Abstract

The production of academic and scientific genres is a cognitively demanding task central to university learning and to students’ integration into disciplinary and professional communities. In engineering programs, written communication has become increasingly relevant, as graduates are expected not only to master technical processes but also to articulate results clearly. However, writing instruction in engineering is often limited by overloaded curricula, persistent views of writing as peripheral, and scarce access to specialized, theoretically grounded support. In the Spanish-speaking context, this situation is compounded by the lack of digital tools that integrate corpus-based evidence, genre pedagogy, and automated feedback for disciplinary writing. Responding to this gap, the present study introduces PEUMO, a free web-based platform that supports the writing of technical reports in engineering.

PEUMO is grounded in Corpus Linguistics (CL) and Genre-Based Pedagogy (GBP). It draws on corpus-informed descriptions of engineering genres, rhetorical modeling of the final degree project, and evidence of common linguistic difficulties identified in previous studies. The platform provides automated feedback across four levels—lexico-grammatical, formal, stylistic, and discursive—and includes instructional capsules, concordancing tools, and a specialized engineering corpus. Technically, PEUMO combines a front-end text editor with a back-end analysis server (Redilegra/PACTE), using rule-based scripts, natural language processing (SpaCy, Connexor), readability metrics (INFLESZ), and AI models such as BETO for purpose classification.

Beyond corrective feedback, PEUMO incorporates a process module based on Graham’s (2018) Writer(s)-within-Community model. It operationalizes sub-processes of conceptualization, ideation, transcription/translation, and reconceptualization through corpus-driven functions such as thesis indices, length estimates, keyword extraction, conceptual clouds, previous findings, contribution spaces, rhetorical patterns, similarity searches, and a phrasebank organized by sections of the genre. This design enables students to consult authentic disciplinary exemplars and supports metacognitive engagement with their writing.

The study reports a validation focused on usability and perceived usefulness, conducted within a quasi-experimental intervention with engineering students at a Chilean university. A 16-item Likert-scale perception survey, adapted from validated instruments, was administered to 22 students who used PEUMO during a semester-long intervention grounded in CL and GBP. Results show a predominantly positive evaluation, with a significantly higher proportion of favorable responses. Students valued feedback on paragraph and sentence length, first- and second-person pronouns, connectors, passive voice, and gerunds. At the process level, they most appreciated keyword identification, the phrasebank, and rhetorical-pattern examples. Less frequent use was reported for functions related to conceptualization, ideation, syntactic complexity, readability, and communicative-purpose identification.

Overall, the findings confirm the potential of PEUMO as a disciplinary academic literacy tool that integrates CL and GBP to provide meaningful and actionable feedback. Students perceived the platform as useful, clear, and relevant to their writing tasks, reinforcing the value of accessing corpus-based textual evidence during composition. The study also identifies areas for improvement—such as usability refinements, updates to underlying NLP components, and broader technological enhancements—and outlines future developments, including the integration of explainable generative AI to strengthen automated feedback and support for writing sub-processes. PEUMO thus contributes to students’ autonomy, metacognitive reflection, and rhetorical awareness in engineering writing.