Despite all the jokes about its inaccuracy, Google Translate is still on the top of our minds when it comes to free and automatic translation as it is a free machine translation software that, until 2016, was running on a statistical-based engine, but has now converted into a neural-based engine.
As long as it used for personal purposes, all is fine and well: Google Translate does an amazing job at helping you understand the user manual for the piece of electronics you just bought or helping you with your homework. But how good is Google Translate for Italian when you want to translate a marketing content? Maybe you’ll never think to use it to translate an entire catalogue or your carefully drafted presentation to launch a new product but what about something innocent like the company’s Facebook post? Can it really be employed blindly for official communication, even if just for a quick Instagram caption?
Table of Content1. When it doesn’t work 2. Google Translate vs colloquialisms 3. When it does work 4. So… is Google Translate for Italian good or not?
Let’s say you have received some great feedback for your products and you want to post the best comments on your social pages in all languages. Do you really need a translator for seven lines of text about a lipliner review?
I tested Google Translate for Italian on a real product review for a Revlon lipliner and well…
Google’s translation, which is called raw output, is full of errors but it also makes you wonder if this lipliner wants to murder you.
In the sentence “Prevents bleeding of your lip color” the term “bleeding” has been translated literally into the word “sanguinamento” which describes a loss of blood when here it obviously means “sbavare” as in going over the lips’ borders. And then, towards the end, a common word like “sharpener”, which even a middle schooler would correctly translate into “temperino”, has become “affilatoio”, which is a knife or similar kind of blade sharpener.
The aforementioned are all example of lexical ambiguity which is still one of the main problems of machine translation. “In fact,” writes the linguist Dijako in his dissertation, “ambiguity resolution has been one of the trickiest linguistic aspects of MT since research began in this area.”
Even if you manage to get past the images of blood and knife, Google’s output still has all the common errors that we can find in machine translation such as a style that doesn’t reflect spoken ad casual language and syntactic ambiguity (when a word may be assigned more than one grammatical or syntactic category – e.g. noun, verb or adjective). In the original, “constant” refers to “perfect” (like saying always perfect) but in Italian it got referred to “point”.
Let’s try Instagram now. Again, in order to get as real as possible, I wanted to test Google Translate on a real Instagram caption and the first company that popped into my mind was Coca-Cola.
It’s a good example because their language is easy and not technical at all. Just everyday words right? Wrong. Look at the translation for the caption of Coca-Cola’s Valentine’s Day post.
We can forgive Google for having translated Valentine’s Day and Valentines’ Year the same way, as it needs a good deal of transcreation, but what about that “like”?
I am actually not surprised, as this is one of the most common and standard mistakes for machine translation: a computer doesn’t know grammar and therefore can’t digest any lexical and grammatical ambiguity.
Everybody deserves a second chance, so here’s another Coca-Cola’s Instagram post that doesn’t contain any grammar ambiguity nor idioms nor colloquialisms.
This output works pretty well! It is grammatically correct and no term has been misunderstood but it sounds a bit unnatural. The original English post is rather earnest, but the Italian translation is too stiff due to the repetition of “new ways”: the Italian language tends to avoid using the same word twice not only in the same sentence but also in the same paragraph. This is particularly true the more colloquial the tone, whereas repetitions are accepted in essays and in more official text. So, in order to achieve the correct tone, I would personally suggest striking off the second “nuovi modi”.
To sum up, Google Translator provides good translation for more technical type of texts which have the following characteristics:
- The sentences are neither too long nor too short and there is no need to rework the structure of the original when translating into the target text.
- The language is free of ambiguities and the general meaning is conveyed clearly and plainly.
- There are no idioms, colloquialism or wordplay to be conveyed (so everything related to marketing is a real challenge).
So, in some cases, Google Translate can provide good enough translations, but you never know which one those cases are. Are you willing to play Russian roulette with your content? If the answer is no, maybe you are in need of somebody who will look over the machine-translated output and improves it. I can help you with that! Check out my translation services or write to me to ask for a quote on your Machine-Translation Post-Editing project.