In the current translation context, understood as professional practice, everybody is aware that the expert’s task does not only consist in sitting in front of a text that has been written in one language and render it in the target language. Even though interlanguage reproduction remains the fundamental objective of translation services, workflow dynamics have evolved by leaps and bounds. From a chronological perspective, we went from translating by hand on paper with a physical source at hand to preparing and digitalizing the texts by using a computer and preparing their translations using editing software. Gradually, the latter made way for different translation software suites, which offered the possibility of integrating the source text and the target text on the same screen in a visual manner while providing specific tools that proved very useful to harness greater control over the text editing process while conducting more comprehensive quality tests. Finally, the last consolidated revolution in the translation sector was sparked by the incorporation of the automatic translation, which enabled instant translations to be generated using text comparison with a corpora of source text pairs and thus translating them in a translation engine. This new resource was of transcendental importance, since writing and translation times could be extraordinarily streamlined, so that progressively, it gained popularity among translation agencies and clients, as the new resource also implied cost reductions. However, this automatic translation also had a series of drawbacks that entailed a real problem when trying to guarantee work quality, one of them being that the translation results provided by the translation engine were conditioned due to segmentation. In other words, when text was introduced into an automatic translation engine and this text was segmented by commas, the tool would only recognize a sentence up to a comma and provide a translation that was entirely independent of the context and which, occasionally, was totally unrelated to the phrase in the next segment. These errors, which did not only take place at the comma level, but rather the same errors could show up between entire sentences, represented a great drawback for the quality of a translated product, as it generated terminology inconsistencies, ungrammatical sentences and a forced and stilted style. Hence, after some time, the industry ended up concluding that, for the time being, automatic translation could not exist on its own, since it could not be entrusted with the task of faithfully and consistently translating a text, something that neither the client’s needs nor the translation industry’s standards could allow to happen.
It is for this reason that the post-editing step came into play. This step, which professionals are well acquainted with today, is that which is responsible for receiving an automatic translation in the aforementioned conditions, and correcting each and every one of the errors concerning translation, terminology consistency, punctuation, grammar and style that may have been generated, so that the final product is as close as possible to an old-school translation. In principle, the usefulness of the entire process, despite the necessary involvement of the post-editor, lies in the fact that there are significant time savings thanks to all those segments that the translation engine has generated correctly and which, during post-editing, only require manual confirmation. Check Post-editing pros and cons (from the translator’s perspective) and Post-editing pros and cons (from the client’s perspective) where we analyze these issues in detail.
However, in addition to the step known as post-editing, there is another step, perhaps not so visible in language service agencies, known as “pre-editing”. Essentially, the objective of this process is similar to that of post-editing, since it works with a text that is going to be in contact with automatic translation at some point. In other words, pre-editing takes into account that a text with certain characteristics may incur into a greater error ratio in the automatic translation, and for this reason, its mission is to make the appropriate modifications before giving the text the translation engine treatment. Pre-editing takes a text and analyzes parameters such as sentence length. If sentences are too long or have extensive punctuation, there will more likely be broken during segmentation, which could lead to incoherent translations. For this reason, pre-editing is responsible for transforming long and complex sentences into simpler ones to be fed to the automatic translation engine. Likewise, it analyzes terminology to ensure that it is consistent throughout and that, as far as possible, synonyms are not being used that could lead to inconsistency issues in the translation. Finally, it is responsible for listing the information in a logical order that is easy to interpret for machines. In this way, much simpler texts are achieved that, when subjected to automatic translation, will obtain better results, so that the post-editor’s involvement will be reduced.
As we can see, automatic translation is always linked to the participation of the human factor, whether during pre-editing (where we prepare and modify a text to be fed to the translation engine) or during post-editing (where we correct the generated errors). Indeed, the most popular variety is the latter, since pre-editing entails modifying the source text, a task that should only be assigned to those who generate said document, since a translator is not allowed to modify any source material at will. Quite the opposite, what a language professional can indeed do is assess the different clients with regard to drafting keys that ensure that the texts thus produced have a reduced risk of generating errors during automatic translation.
Image reference: https://www.zacharyer.xyz/products.aspx?cname=cat+glasses+computer&cid=43
