Only logged in customers who have purchased this product may leave a review.
LEXICAL ERROR ANALYSIS IN ENGLISH-TO-ROMANIAN MACHINE TRANSLATION
- Autor: Angela DUMITRAN
- An aparitie: 2026
- Numar de pagini: 198
- Format: B5
- ISBN: 978-606-16-1761-6
- Limba textului: engleză
How reliable is Machine Translation when human health is on the line? This volume, based on the author’s doctoral research, investigates the lexical errors produced by Google Translate when rendering Covid-19-related content from English into Romanian. Drawing on a corpus of 230 texts – official documents from the World Health Organization and Gavi, the Vaccine Alliance, alongside patient information leaflets covering active ingredients, dosage, storage, and warnings – the study offers a detailed, data-driven map of where automatic translation still falls short in specialized, high-stakes domains.
The analysis identifies and classifies recurring error types, from collocation mismatches (the most frequent category, at nearly a third of all errors) to borrowings, false friends, stylistic inconsistencies, misread abbreviations, faulty prepositional usage, and semantic distortions. Rather than treating these as isolated glitches, the book traces their probable causes back to the structural differences between English and Romanian and to the particular complexity of Covid-19 terminology, asking not only what goes wrong but why – and what computational linguistics and machine learning might do about it.
Beyond its statistical findings, the volume speaks to a broader concern: as Machine Translation becomes ever more embedded in the circulation of medical and public-health information, the line between a fluent translation and an accurate one can carry real consequences for readers who rely on it. The book argues for continued human oversight in such domains, while pointing toward concrete strategies for improving the linguistic sensitivity of systems like Google Translate.





Reviews
There are no reviews yet.