Machine translation (MT) is the use of software to translate text from one language to another. It first started to take shape in the 1950s, with classic results such as ‘Out of Sight, Out of Mind’ going into Russian and coming back out as ‘Invisible Idiot’.
Sowing the seeds of perfection
On a basic level, MT performs simple word-for word substitution from the source language to the target language, but that alone usually cannot produce a good translation of a text because recognition of whole phrases and their closest counterparts in the target language is needed, so an editor or proof-reader is required to polish the machine translation.
Most current machine translation software allows customisation, which means you can ‘train’ the machine in a specific field to produce a better result, but the input required to customise the basic dictionaries is intense and only valid where very large amounts of translation is required in a clearly identified sphere and group of languages.
In terms of day-to-day application, Machine Translation has its place in the translation world, most commonly in the form of MT plus post-editing. It can be used to save time for large quantities of translation material, but it is only a valid option if you can employ the services of strong post-editing team who understand subject matter, source material and required quality output levels in depth. It is not a human replacement option unless you are in a position to carry out your own editing, or the need for accuracy of translation is so significantly reduced as to make it viable as a stand-alone requirement (e.g. weather forecasts or website page translations for basic information purposes only).
Google Translate™ is probably the best base machine translation engine we’ve come across, simply because it is undergoing continuous update (training) by everyone who uses it, but again, its usage should be closely monitored and reviewed: there are many occasions where we have seen it apply a negative translation to a positive statement, or produced nonsense when the source text is perfectly intelligible. There is no doubt that Machine Translation has its place in the modern translation world. As with all technological developments, it is to be embraced, nurtured and fine-tuned as a tool to improve the translation process. It is important to understand how it may best fit into your requirements and enhance final translated output for you.
At RP Translate, we use Machine Translation combined with an editor (post-editing) when working with large numbers of verbatim. Significantly cheaper and usually quicker than Human Translation, our Machine translation/post-editing option is designed to give you translations from which to code or analyse, and whilst it is still essentially accurate, it does not give you a polished client-facing finish.
Caveat: Machine Translation is not applicable to all languages