Machine translation (MT) is becoming more and more common, but does it work for audiovisual translation (AVT)? In this article, we will discuss why a human audiovisual translator can be your best choice, depending on the type of translation.
Use of machine translation
In machine translation, unlike in computer-assisted translation (CAT), the translation is done entirely by the computational systems. As a result, MT outputs are usually not very good, so human post-editing is indispensable.
Depending on the type of content, MT can be very helpful. For example, repetitive materials such as FAQs, tickets, e-mails, or other texts using a very specific language.
However, audiovisual content is very challenging for machine translation. Films, series, and TV programmes are highly creative and often use humour and colloquial language. In addition, even in the case of less creative videos, subtitling involves sound and image, and MT does not understand context.
A few months ago, the Association of Audiovisual Translation and Adaptation of Spain (ATRAE) protested against the use of post-editing for subtitling as they found out that the series “Squid Game” was subtitled in Spanish using post-edited machine translation (MTPE). They urged the major audiovisual platforms to stop using MTPE and, instead, to trust the talent of human translators.
Limitations of machine translation
In MT, it is common to occur errors of ambiguity. Again, the machine does not understand context and how a word can mean something in one sense and something else in another. For example, “bank” can appear with “river” in geography and with “savings and loan” in finance.
Idiomatic expressions are also very challenging for MT, and they are present in all languages and all types of communication. Expressions need to be adapted for the target culture, taking into consideration the way the language is used by the speakers and not only the meanings in dictionaries or synonyms. Contrarily to the capability of a human, MT simply replicates existing translations.
Another issue for MT is the different languages and language varieties. Translating is not easy, and it is not about finding exact equivalents. While in American English it is said “set the table” and “make a decision”, in British English, the corresponding expressions are “lay the table” and “take a decision”. Are MT databases extensive enough to include all language pairs and language varieties?
From a personal experience, I have worked with a company in which they used MT with English as a pivot language, and I was asked to translate to Brazilian Portuguese. The quality of the translation into English was poor and because of that, it affected the translation into Brazilian Portuguese.
Human audiovisual translator
Subtitles are dynamic content. Translating them requires specific strategies, such as the reduction of the text so that the audience can read the subtitles and watch the video. Machine translation does not translate concisely taking the reading speed into consideration as a human audiovisual translator does.
The professional will adapt your material to the target audience, following the technical aspects of subtitling to guarantee that the audience will have time to read the content. Above all, they will compress the information while keeping the meaning of the original content.
In conclusion
Understanding the problems of machine translation in audiovisual translation is important to improve the quality. Post-editing machine translation is not the best approach for highly creative content. Sometimes it is better to start from scratch rather than to fix a very bad translation done by a machine. Therefore, machine translation cannot replace human AV translators.
If you are interested in getting your videos subtitled by a professional human audiovisual translator, get in touch.
Pingback: How to Get Quality Subtitles for Your Video Content | Rocha Translations
Pingback: Translation and Subtitling Rates | Rocha Translations