Tag: Machine translation

Playing with a 1905 Russian book, Finereader 12, and Google Translate

Posted on by Roger Pearse

This morning I decided to see what I could find out about a 1905 Russian edition of the works of Methodius of Olympus (d.311 AD), which I obtained in PDF form from a library in Chicago a year or so back.

Now I don’t know any Russian … not even the alphabet.  But I have tools at my disposal to help me!

First of all, we have Google Translate.  This will give us something, and we can even enter corrections as we go along.  We also have the ability to enter unicode using Charmap.  Finally modern software like Abbyy Finereader 12 does a remarkable job.

I started with the title page.  I didn’t actually get much from this, except for a reference to “complete” and “Greek”.  So it’s probably an edition of the complete works, translated from Greek or something.

On the reverse of the title page, I got this:

Отъ С.-Петербургскаго Духовнаго Цензурнаго Комитета печатать дозволяется. С.-Петербургъ, 27-го іюля 1904 г.

Цензоръ, Іеромонахъ Александръ.

Сиб. Типолитографія М. II. Фроловой. І’алерная, 6.

Erm, yes.

So I popped it into Google Translate.  I got this:

Ot St. Peterburgskago of spiritual Tsenzurnago Committee is permitted to print. St. Peterburg, 27th іyulya 1904
Tsenzor, Іeromonah Alexandre.
Sib. Tipolitografіya M. II. Frolova. І’alernaya <5.

OK…  But hang on…

Surely “Ieromonah” is “Hieromonk”?  And I wonder, O I wonder, what “Tsenzor” could mean?  It must be pronounced “Censor”!  Which means that “Tsenzurnago”, combined with “Committee of spiritual Tsenzurnago” is probably “Committe of spiritual censorship”!

So the notice must mean that this is permission to print, issued by the St Petersburg committee for spiritual censorship, signed by the Hieromonk Alexander, Censor.

The text has not scanned perfectly.  “I'” should actually be Г, and “<5” is actually “6”.  That makes the last line:

Sib. Tipolitografіya M. II. Frolova. Galernaya 6.

Which is probably something to do with the address of the printer.

Now this is a little thing, in a way: except consider what it means.  All we need to make progress is some industry.  If I started looking words up, and learning a bit about the language, I would soon learn even more.  All I need is time and industry.

And … candidly … it’s quite fun!

Share

Machine translating unknown texts – the copiale manuscript

Posted on by Roger Pearse

Via Dyspepsia Generation I find this story at Wired:

They Cracked This 250-Year-Old Code, and Found a Secret Society Inside

It was actually an accident that brought to light the symbolic “sight-restoring” ritual. The decoding effort started as a sort of game between two friends that eventually engulfed a team of experts in disciplines ranging from machine translation to intellectual history. Its significance goes far beyond the contents of a single cipher. Hidden within coded manuscripts like these is a secret history of how esoteric, often radical notions of science, politics, and religion spread underground. At least that’s what experts believe. The only way to know for sure is to break the codes.

In this case, as it happens, the cracking began in a restaurant in Germany.

The story has wide application:

On October 25, 2011,The New York Times published a story about the Copiale, focusing on Knight’s code-cracking techniques. A flood of media attention followed—along with hundreds of emails from people who claimed to have ancient ciphers of their own. In December, when I visited Knight, he had just received a picture from Yemen. Some Bedouins had found a stone with an unknown, squarish script. Perhaps Knight could tell them what it said?

Fascinating.

Share

Ve haf vays of meking you translate … not you Google, sit down!

Posted on by Roger Pearse

Michael Gilleland has been trying out German on Google Translate, with mixed results.

I used to tell students, “This passage makes sense in Latin, and your English translation has to make sense, too. It isn’t nonsense in Latin, and it can’t be nonsense in English.” Google Translate’s version of Wilamowitz’s German seems to fail the “intelligibility” test.

Does Microsoft’s Bing Translator do any better?

I wasn’t aware of the existence of Bing Translator (which for some reason makes me think of defunct US sit-com Friends) , but the more the better, in my experience —  you can sometimes get part of the meaning from one, and part from another.

German IS a problem in Google translate.  Part of the reason for this is the involuted word order, for which, I believe we have Martin Luther to thank.  Part of the reason is the lengthy sentences that German literature written by scholars tends to favour.

What I have found is that often you can get enlightenment by breaking down a sentence into bits.  If you put each clause on a separate line, and do the same with what looks like the main verb at the end, it helps.  You’ll often do better to translate one such sentence into more than one English sentence anyway.

But in the end, you will still need some knowledge of the language.  These toys do not do all the job for you.  But they help considerably.

I’m having difficulties with a (paid) translator at the moment.  The following neatly sums up the problem:

This passage makes sense in Latin, and your English translation has to make sense, too. It isn’t nonsense in Latin, and it can’t be nonsense in English.

Well said, Mr. G.

Share

Using Google translate on Manuel Paleologus, and contributing as you do

Posted on by Roger Pearse

I’m trying to finish up various little tasks that I started ages ago.  One of these was a translation of book 7 of the Dialogue with a Persian by Manuel Paleologus, which got Pope Benedict into such hot water with the Moslems awhile back.  I’ve been looking at the French version of this.

Here’s a tip.  Take a single sentence, and run it through Google translate.  You’ll get a box with the French; and opposite it, the English, more or less good.  But… at the bottom right, you’ll get “Contribute a better translation.”  Click this box, and edit the machine translation there into proper English.  Then hit the button and submit it to Google.

Firstly, when you copy back your edited version, it’s in a sensible font rather than Courier (which is what you get otherwise).  Secondly, since Google Translate works by using existing translations of texts, you’re actually increasing their database and making it more likely that the result will work for you next time.

The results, from French, are really very good; better than Systran, which I have used so far.  I need to see how good it is on Italian!

Share