MAPPING CONSISTENT STYLISTIC PATTERNS OF UNDERGRADUATE LEARNERS ACROSS WRITING TEXTS

Abstract

This study investigates the consistency of writing style (or stylistic biometric identity) of undergraduate students engaged in different writing tasks. With the growing use of computer writing and computer-supported writing tools, author verification and support for academic integrity are increasingly important. Here, this study adopts a quantitative approach to computational stylometry. The samples were purposively selected 100 undergraduate students and constructed a corpus of students' work, including several writing samples for each essay, report, and reflective task. Stylometric features of interest included vocabulary, syntax, sentence length, word frequency, and readability index—descriptive statistics indicating low variability in the selected features and stylistic consistency among students. Set out to confirm these observations by means of inferential statistics. There were no significant differences in most writing tasks (p > 0.05) as reported by the paired-samples t-test, reaffirming stylistic consistency. However, there were differences (p < 0.05) between the high- and low-stylistically consistent groups, as assessed by an independent-samples t-test, for lexical diversity and syntactic complexity. Also found a high positive correlation between lexical diversity and stylistic consistency (r = 0.58); readability and consistency (r = 0.49), and a moderately negative correlation between syntactic complexity and consistency (r = -0.52), according to the correlation test. The accuracy of predicted stylistic consistency for all the students is 82%. Findings show that students typically maintain a consistent writing style across different writing tasks, with minor variations due to the writing context. Research demonstrates the use of stylometric analysis for individualized learning, author recognition, and academic plagiarism detection. Findings can be improved by training the system with more sophisticated natural language processing approaches and larger datasets to enhance generalization.

Keywords: Stylometry, writing style, undergraduate learners, computational Linguistics, Academic integrity, ESL writing, text analysis.