Google translate FTF!

Posted: September 2, 2009 in languages, polish
Tags: , ,

Somehow, when we use Google Translate to work out translations for articles from Norwegian newspapers, the results are almost perfect. However, this is not the case with Polish. Not at all.

Is it just because nobody cares about developing a better Polish/English translator? This is a possibility!

Also, Polish is difficult to automatically translate from/to, because:

1. the word order in a sentence is basically, almost random. You can put anything almost anywhere, depending on what information is most important to you in any given sentence. There are some rules, like, usually adjectives come before nouns, but if the text was written in an “olde Polish”-ish style, this would not be the case at all, and  a “Latin” word order would be preserved as much as possible, etc, etc, etc. This sort of style is often used nowadays for parodies and making ironic statements, so it’s not like it would be difficult to spot or anything.

2. many words are totally (also grammatically) ambiguous. Many are not, but then, their English counterparts totally (also, grammatically) are.

3. there are declensions, conjugations and grammatical gender. You can write an algorithm that doesn’t ignore them, but the results could easily be disastrous. Also, you’d need an awesome tagging tool to tag your dictionary, anyway.

Which is how we get brilliant results like that (used random text from my chat log* :P):

1. Będę miała permanentne wgniecenie od tego ciągłego walenia się po twarzy <— should be translated as something akin to “I will get a permanent dent on my face from all this face-palming”, but instead got us this:

I have a permanent dent on the continuous pounding her face.

This is, basically, incomprehensible. It’s so wrong, I actually have no idea what could have gone wrong to make it so absolutely wrong.
2. No bo yukaty były. więc musiałam chodzić za ludźmi i robić im zdjęcia <— should end up something like “‘Cause there were yukatas, and I just had to follow people and take photos of them”. However, it ended up looking like that:

what else were yukaty. So the people have to walk and to do their pictures

The first short sentence is, well, short, so it’s somehow understandable that it wouldn’t be translated correctly. However, in the second sentence, weirdness happened. First, the translator ignored the verb conjugation that suggested that the subject was 1. singular, 2. had feminine grammatical gender, which would suggest that “people” (plural; and strictly speaking Polish  has no feminine is not the subject of the sentence. Secondly, it seemed that the translator was unaware that Polish rarely uses personal pronouns together with verbs (like the English “I walked for 5 minutes” and so on)’ in fact, pronouns are only used in sentences like “*I* will do this” or *It was *you* who did it” for emphasis. The second sentence omits the personal pronouns “I” (female I), which Google Translate interprets thus: no noun with “walk” –> search for the closest noun in the same sentence: “people” –> “people” = noun for “walk” –> “people” = subject –> SUCCESS! And, of course, “people”, preceded by an adverb, is not the nominative form, which makes it impossible for “people” to be the subject at all.

Whatever the cause, (I suspect a sucking tagging tool), Google translate can’t find subject in a Polish sentence.

3. Slyszałaś o ich prysznicach? Bo ja tak, ze szczegółami, od zgnębionych, mieszkających tam tajskich studentów should be translated as “Have you heard about their showers? Because I did, in detail, from the poor Thai students living there”, but:

You heard about the shower? Because I am so in detail, from the oppressed, Thai students living there.

So: Google Translate ignored 1. a possessive pronoun and translated it as an article, um, why? b. ignored the plural in “showers”, 3. didn’t recognise “Because I did” for what it is, but that part must have been totally difficult, so.

I’d say, a sucky tagging tool,  still.

4. Let’s finally take a look at how Google Translate will work with a bit larger chunk of text. This is my e-mail to a friend:

Nos spalił Ci się na slońcu, ponieważ zaniedbałaś pokrycia go grubą warstwą mleczka do opalania z filtrem SPF +150. Przed opalaniem się, a nawet przed wyjsciem na zewnątrz w słoneczny dzień nidgy nie powinnaś zapominać o zastosowaniu grubej warstwy mleczka SPF +150, zwłaszcza jesli Twój nos wystaje wysoko ponad powierzchnie płaskie.
Na wszystkie inne pytania dotyczące nosow, mleczka SPF +150 i ich zastosowania w ciekawych aczkolwiek atletycznych i ryzykownych dla stawów aktach seksualnych odpowie Ci Dan Savage.

Now you can figure out what it was all about.

Burned down your nose in the sun, because it zaniedbałaś cover a thick layer of milk for sunscreen SPF 150. Before firing in August, even before the output on the outside on a sunny day nidgy you should not forget to apply a thick layer of milk SPF-150, especially if your nose protrudes above the flat surfaces.
On all other questions concerning nasal, milks the SPF 150 and their use in athletic and exciting but risky for joints sexual acts you respond by Dan Savage.

Now, the plus side: IT’S STILL FUNNY! If for all the wrong reasons. However, one thing eludes me: WHY would anyone want to translate “to sunbathe, get a tan” as “firing in August”??? Also, “going out” is “output” now, LOL <3

* The fail part is: I couldn’t really find much usable Polish (stuff that wasn’t LOL MORONS) in my chat log *facepalm*.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s