Análisis de calidad de traducción de dos traductores automáticos utilizados en un estudio de traducción en Lima, 2023
Loading...
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Femenina del Sagrado Corazón
Abstract
Esta investigación se centró en evaluar la calidad de la traducción automática mediante el análisis de dos traductores automáticos, utilizando la métrica de evaluación SAE J2450, previamente establecida. Se optó por un enfoque cuantitativo, de nivel exploratorio y se llevó a cabo mediante un diseño de estudio transeccional exploratorio en Lima, Perú. La muestra consistió en dos motores de traducción automática, DeepL y Yandex, a los cuales se les sometió un fragmento de un documento técnico del campo de la automoción. Luego de evaluar ambos traductores automáticos, se determinó que la calidad promedio de sus traducciones es medianamente baja, con un puntaje de 74.50 %. Esto se atribuyó en gran medida a la presencia significativa de errores terminológicos y sintácticos. Asimismo, el resultado de calidad de DeepL fue de un 72.39 %, mientras que Yandex alcanzó un 76.62 %. Ambos casos revelaron una calidad de traducción medianamente baja, con errores de terminología como los más frecuentes. Por último, para mejorar la calidad en las traducciones se recomienda la intervención de un traductor profesional que pueda corregir y optimizar los posibles errores identificados durante el proceso de traducción automática.
This study focused on evaluating the quality of machine translation by analyzing two machine translators using the previously established SAE J2450 evaluation metric. The study chose a quantitative, exploratory approach and was performed using an exploratory cross-sectional study design in Lima, Peru. The sample included two machine translation programs, DeepL and Yandex, and a fragment of a technical document from the automotive field was submitted on these platforms. After the evaluation of both machine translators, the average quality of their translations was found to be moderately low, with a score of 74.50 %. This was largely due to the significant presence of terminological and syntactic errors. Likewise, the quality score for DeepL was 72.39 %, while Yandex reached 76.62 %. In both cases, the translation quality was moderately low, with terminology errors being the most frequent. Finally, to improve the quality in translations, the intervention of a professional translator who can correct and enhance possible errors identified during the machine translation process is recommended.
This study focused on evaluating the quality of machine translation by analyzing two machine translators using the previously established SAE J2450 evaluation metric. The study chose a quantitative, exploratory approach and was performed using an exploratory cross-sectional study design in Lima, Peru. The sample included two machine translation programs, DeepL and Yandex, and a fragment of a technical document from the automotive field was submitted on these platforms. After the evaluation of both machine translators, the average quality of their translations was found to be moderately low, with a score of 74.50 %. This was largely due to the significant presence of terminological and syntactic errors. Likewise, the quality score for DeepL was 72.39 %, while Yandex reached 76.62 %. In both cases, the translation quality was moderately low, with terminology errors being the most frequent. Finally, to improve the quality in translations, the intervention of a professional translator who can correct and enhance possible errors identified during the machine translation process is recommended.
Description
Keywords
Traducción--Tesis, Traducción automática, Traductores (Programas para computadoras), Control de calidad