I am excited to announce that I shall soon be publishing a Portuguese translation of my website.
Last night I posted a question to a Portuguese translators' group in which I asked whether I should use the rules of spelling as detailed in the new Orthographic Agreement, which is not popular among linguists, for good linguistic reason on a number of fronts. It has therefore met with much resistance from translators and much confusion from the public at large.
Here is a screenshot of my question:
This morning a translator friend in England who also translates from Portuguese sent me a message wondering why she could not figure out what I was trying to say until she realised that she was reading a Bing translation. All became clear when she clicked on "View original". She sent me a screenshot of the reason for her confusion:
As soon as I laid eyes on the above, the compulsive editor built into my human brain immediately winced seven times in quick succession:
It seems that even Microsoft's grammar checker in Word objected about the start of the second sentence in the above image. I do like to, actually, but that is not the subject of this blog.
If you are not a translator, then I need to tell you that what you are looking at is 22 English words (23 in the Portuguese) translated by an automatic translation program called Bing, which has rather small search engine capabilities compared to those of Google Translate - another "machine".
Those 22 words contain 7 translation errors of which 3 are, in my view, serious ones. If the text were 100 times longer and comprised 2,200 words, it is not entirely unreasonable to assume that Bing would make in the region of 700 errors of which 300 could be considered serious. Even if it is unreasonable to make that assumption, I would wager that the incidence of error would be unacceptably high for all except the most frivolous of purposes, such as trying to figure out what your foreign friends of your foreign friends are saying on Facebook.
What is an unacceptably high incidence of error in a translation?
Most translators use the rule of thumb that one error is one error too many. In practice these days, I suppose an "acceptable" number of minor typographic errors a proofreader could expect to pick up in a translation would be in the region of between 2 and 5 in a text 2,200 words long. Having said that, I revised a 214,000-word biography last summer. The translator made only 4 typographic errors in the entire book, which I found remarkable. It was, however, a revision assignment, and my revisions of her translation were many.
Neither Bing nor Google Translate seem to make many typographic errors at all. I would like to know what is the point of perfect typing when the perfectly typed translation is, in fact, rubbish?
Google Translate did a slightly better job of "translating" my little Facebook post than Bing:
Where Google Translate has the advantage over Bing as it is presented to readers in Facebook is that it offers alternatives. If these are chosen judiciously, the result is closer to the original meaning of the source text:
Close, but not close enough. It seems not all abbreviations are recognised by this machine. And, despite my simple sentence structure in Portuguese, I unwittingly produced a text from which you could not tell the gender of the translator. This makes translation of the second sentence quite tricky.
As it happened, the sentence was conceived in Portuguese. I gave no thought to how it might be translated. That is not unusual. That is how most people write. Unless one works for an international organisation which has rules about using simplified language, one is under no obligation to make what one writes "easier" to translate into any given language.
The translation difficulties encountered in the above 23 Portuguese words would be very different if they were being translated into a language other than English. This does not mean that one language is more difficult than another. It means that the process and the method of conveying the meaning in different languages is different because the deep structure (or grammar) of every language is different from every other.
So how would I translate my own words in Portuguese into my own mother tongue, English?
I will save you the trouble of scrolling up and repeat the text here:
Alguém está a traduzir o meu site para português. Surgiu a pergunta sobre o Acordo Ortográfico.
If I were to base my translation solely on the information in the text itself, then I suppose this would have to do:
Someone is translating my website into Portuguese. The question arose about using the new Orthographic Agreement.
You will notice from my translation that English requires "expansion" of the phrase "about using the new Orthographic Agreement" which is not necessary in Portuguese (although the opportunities for me to be more eloquent in my use of Portuguese abound).
Suppose, however, that the gender of the "someone" who was translating the text had been revealed as female - or, in this case, that I know that fact from extra-textual information. That tiny piece of the puzzle, known in the trade as "context", possibly helps to makes a more natural-sounding translation:
Someone is translating my website into Portuguese. She asked me whether she should use the new Orthographic Agreement.
You will have to wait and see what my decision has been after considering all the arguments presented by my learned colleagues.
In the meantime, I hope that I have convinced you that human involvement in the translation process is by far the better approach if your aim is to communicate effectively.
In closing, I should perhaps tell you that I would consider either of my two translations above as acceptable equivalents of the Portuguese source text. It would depend, as translators are fond of saying, on the context.
©2016 Allison Wright
This blog will focus solely on aspects related to translation.