Simulating differences between Spanish dialects from Europe and Latin America
Translation is a difficult process and even more so if dialects of a language are involved. The differences in word choice and frequency between several dialects can easily lead to translation errors. One language where this problem is in play is Spanish, for which a different dialect can be found in every Spanish speaking country. With the help of a survey, information about these differences was gathered and analysed using a machine learning model called Multilink. A subjective frequency of word usage score was at the core of this analysis. This statistic was processed and used as input for the model. The results showed interesting differences, not only between the dialect of Spain and those of Latin America, but also between the dialects within Latin America. These findings show that, even though generally the issue is acknowledged already, more attention needs to be paid to translation processes involving dialects. Translators such as Google Translate for example, should find a way to incorporate this dialect information into their translations and make distinctions in translation based on this.
Faculteit der Sociale Wetenschappen