Prompt engineering competence and learning outcomes in Legal ESP: Evidence from Vietnamese law undergraduates
Multidisciplinary Reviews · 17.07.2026 · 10 (1) · 2027032
DOI: 10.31893/multirev.2027032 ↗Kurz gefasst
Fachartikel in der Zeitschrift Multidisciplinary Reviews, online erschienen am 17.07.2026. Verfasst von Vo Thi My Hanh. Die Arbeit nennt 46 Literaturangaben.
Abstract (Originalsprache)
Currently, generative AI is being increasingly used by university students for learning and research purposes. However, the effectiveness of using generative AI among student groups can vary greatly for a number of reasons. In particular, within the context of Legal English for Specific Purposes (Legal ESP), students must work with specialised legal terminology and contexts. The present study investigated whether prompt engineering competence (PEC) was associated with two academic outcomes among Vietnamese law undergraduates: legal terminology mastery and performance on an AI-supported Legal ESP task. The participants were 107 law students who were completing the final lessons of the Legal ESP course. A 24-item PEC scale, a 30-item legal terminology test, an AI-assisted task evaluated by two independent raters, and prompt logs coded for five recurring error types were used as the instruments for the data collection. The findings of the study indicate that the participants were at a moderate level of PEC overall. The students with stronger PEC tended to obtain better outcome results. The Spearman analyses revealed significant associations between the overall PEC and the legal terminology scores (ρ = 0.73, p < 0.01). The overall PEC was also positively associated with AI-assisted task performance (ρ = 0.72, p < 0.01). Regression results further identified PEC as a significant predictor of legal terminology performance (R² = 0.561, Adj. R² = 0.557) and AI-assisted task scores (R² = 0.539, Adj. R² = 0.534). Two models were statistically significant at p < 0.001. Prompt-related problems were widespread. A total of 81.3% of the students made at least one error. In addition, 49.5% made three or more error types. The most detrimental patterns were linked to limited verification and weak revision practices. These results reveal that AI use in Legal ESP should be supported through explicit instruction in prompt design, checking strategies, and iterative improvement.
Zitieren
APA
Hanh, V. T. M. (2026). Prompt engineering competence and learning outcomes in Legal ESP: Evidence from Vietnamese law undergraduates. Multidisciplinary Reviews, 10(1), 2027032. https://doi.org/10.31893/multirev.2027032
Vancouver
Hanh VTM. Prompt engineering competence and learning outcomes in Legal ESP: Evidence from Vietnamese law undergraduates. Multidisciplinary Reviews. 2026;10(1):2027032. doi:10.31893/multirev.2027032
BibTeX
@article{Hanh2026,
author = {Hanh, Vo Thi My},
title = {Prompt engineering competence and learning outcomes in Legal ESP: Evidence from Vietnamese law undergraduates},
journal = {Multidisciplinary Reviews},
year = {2026},
volume = {10},
number = {1},
pages = {2027032},
doi = {10.31893/multirev.2027032}
}
EUVIMED