Have you ever wanted to understand the content of a document written in a foreign language quickly and consulted an automatic translation programme like Google Translate? Did you find that some of the “translation” just did not make sense? For those of you who are not familiar with the translation industry. you may wonder why using machine translation is not sufficient. Allow us to explain why with our guide to the processes that a translator usually adopts when translating a document.
Step 1: Reading the source text
The translator receives the document and begins by reading it twice. The first perusal is at reading speed, to get the gist and an initial idea of the content. The second time, the translator carefully analyses the text to identify potential translation problems and specific terminology. They also scan for characteristics regarding style and register, level of specialization, syntax, cultural references, etc.
Step 2: Preparation and research
After the translator has completed Step 1, they would usually familiarize themselves with the subject area (unless they are already specialists in this field). They normally read information about the subject area as well as parallel texts (similar documents to the one to be translated, both in form and style) in both the source and target languages.
This research is crucial as it is here that the translator can verify the typical usage of certain terms in a given context. It also enables the translator to become familiar with the register and style that they are aiming to produce.
If the translator is unable to identify any problematic terminology at this stage, then they will consult other sources such as specialist dictionaries, encyclopaedias etc. The translator will always double check their terminological choices by verifying with more than one reliable resource.
Step 3: Drafting the first version
At this stage, the translator would normally draft their first version on a word processor and make a backup version to cover all eventualities. The last thing they want is to miss a client’s deadline due to IT faults. Once the translator has finished the first draft, they normally put it to one side for a while before they start to revise it, so that they can come back to it with a fresh pair of eyes.
Step 4: Editing and revising
The translator will then read over their first version and re-draft any sections that sound unclear, or that follow the language patterns of the source language rather than the target language. The aim is to produce a translation of the document which conveys the meaning of the source text accurately and reads naturally in the target language. Once the translator is happy with the revised version, they will usually put it to one side again, so that they can come back to it with a fresh perspective to carry out proofreading and quality assurance checks.
Step 5: Proofreading and Quality Assurance
At this stage, the translator will check the punctuation, grammar, spelling, and any numerical figures etc. In order to do this effectively, the translator might print out a version of the document and carry out their own comprehensive checklist to ensure that the translation is accurate and error-free. Upon completion of the proofreading and Quality Assurance checks, the translation is complete and ready to send to the client.
We hope that this guide has clarified the stages involved in the translation process that a computer simply cannot perform, such as verifying that a lexical item has been used in the correct context. Machine translation can be useful to get a rough idea of the meaning of a document, but in order to ensure accuracy and quality it is always a good idea to consult the services of a fully-qualified translator.
Computer says ‘no’… to computer translations
Without a doubt, GoogleTranslate is one of the technological marvels of our times. I myself, a professional translator, have used it to translate pages from Eastern languages where I’m interested in the content, if not necessarily the way it’s said. I’m not about to get on my high horse and say that it’s an absolute abomination of the modern era and is destroying the translation industry, because it most certainly is not.
That said, I think a demonstration is in order to show GoogleTranslate’s shortcomings. There’s no doubt about Translate’s ability to interpret individual words and it’s even useful for advanced high-tech or sci-tech terminology that you can’t find in your bog-standard dual language dictionary. But Translate’s real shortcoming is not a shortcoming in code: its shortcoming is the fact that it is not human, and like humans languages change and evolve and sometimes say one thing to mean the exact opposite (cue Michael Jackson’s ‘Bad’).
So let’s have a little play around with GoogleTranslate and see what comes up.
Ehrm... Not quite.
The above is an expression you’d most likely hear on the streets of Madrid—two businessmen discussing how much something has cost: Doscientos pavos por la cara, m’a dicho, logo trescientos más. The first problem with this expression is that it’s vaguely slangy—pavo literally means ‘turkey’. Although I was anticipating this to catch Translate out, the machine gets at least a half-point for the American ‘bucks’. However, as I accessed the site from google.co.uk would ‘quid’ have been too much to ask?
Secondly, Translate doesn’t take into account regional dialects. The m’a dicho and logo above are both Madrilene dialect corruptions of me ha dicho (he/she/it [has] told me) and luego (lit. ‘later’, in this case ‘and then’). This catches the machine out as it’s not ‘proper’ Spanish. Mind, if we do use proper Spanish (see below), it’s not much better, and the translation of luego as ‘after’ is actually the opposite of what is meant in Spanish!
Ehrm... Still don't get it.
Lastly, Translate can’t manage colloquial expressions. Por la cara (lit. ‘by the face’, ie ‘up front’) is almost as common as pavo and yet, because GoogleTranslate works primarily on a one-word basis, ‘up front’ becomes ‘for the face’.
I’m not trying to denigrate Translate and I think it does a great job for what it is. But I wouldn’t want it translating my website, or the Articles of Association for my new start-up or even an e-mail from a friend. Language is a particularly human trait. And although computers superficially use a language that looks and reads like our own, would you ever want to replace our spoken English with programmer’s code English? No, me neither!
So what would the above example look like translated by a human? I personally would say: He told me 200 quid up front and then 300 more at a later date. And that’s the difference between GoogleTranslate and the human mind.
So if you ever consider having a translation done and are weighing up the options, take the human translator every time. Because you can communicate with another human much more easily than you can with a computer, and the whole purpose of a translation is effective communication.