Automated Arabic to English translation

An electronic text of the world history by Bar Hebraeus, known as the Book of the dynasties, has come into my hands.  This led me to wonder whether Google translate does Arabic.  And it does!

So I pasted in the first few paragraphs, to see what it made of it.  Well… it was largely gibberish.  But if you knew — say from Pococke’s Latin — vaguely what the subject was, you could see that it was giving you this, albeit in a primitive form.

I was actually quite impressed.  Worth a try, anyway.

One consequence of this is that I can do a word count on the book.  It’s around 84,000 words, which is quite a lot!  It’s divided into ten “dynasties” — really ten groups of who was the dominant power at the time.  Indeed they might be called “books”, I  think.  The 10th dynasty is described in the start as “from the kings of the Moslems to the Mongols”, for instance.  The end of the 9th dynasty is the Moslem takeover.

It would certainly be interesting to get a few pages of this translated. When I next have an income, I must look at this.

4 thoughts on “Automated Arabic to English translation

  1. Google’s translation from English to Arabic is a step forward but it can not be used for scientific purposes such as research. Many mistakes are encountered, mainly due to the difficult nature of Arabic where a word has so many different meanings, or one object has so many different words for it. For example, I have used a few lines from the Live of Abba Khael, the 56th Coptic Patriarch (http://sites.google.com/site/demontortoise2000/hist6-htm ) and used Google to translate it: here is the result:

    The English text:

    When Shenouti, the patriarch who built the sanctuary of my lord Mark at Alexandria, died—this sanctuary remained for one hundred and fifteen years—Abba Khael was consecrated after him. He had three beautiful qualities which resembled gold molten in the fire, on account of the trials which he suffered and underwent. There was an evil bishop of the see of Sakha, and there was among (the villages] of his see the village known as Dinusir.

    The Google Arabic translation:

    عندما Shenouti ، البطريرك الذي بنى حرم ربي مارك في الاسكندرية ، وتوفي — هذا الملاذ لبقيت 115 سنة — كرس أبا Khael بعده. وكان قد ثلاثة الصفات الجميلة التي تشبه الذهب المنصهر في النار ، وعلى حساب من المحاكمات التي تعرض وخضع. كان هناك أسقف شر انظر من سخا ، وكان هناك بين (القرى) له رؤية قرية تعرف باسم Dinusir.

    Now, if we try to translate this Google Arabic translation into English (not using Google), it gives us the following:

    “When Shenouti, the patriarch who built the sanctuary of my lord Marc at Alexandria, and he died — this refuge remained 115 years — Abba Kahel was consecrated after him. And he three the beautiful qualities that look like molten gold in fire, and in account of the trials that he exposed and subjected. There was a bishop of evil observe from Sakha, and there was between the villages he had a vision a village known by the name of Dinusir.”

    See, e.g., the confusion about the word “see”, which is used in the English text to mean parish. In the Google translation it is rendered “observe/look/see” once and “vision” on another occasion. As you can see the Arabic translation distorts the meaning very badly.

  2. True — but better than 0% which is what you would get if you couldn’t even read Arabic letters.

    But yes, these things have some way to go.

    How did you find translating Bar Hebraeus? Hard? Easy? What was his text like?

  3. The translation and the footnotes took me about two hours to do. I didn’t find it particularly difficult but you know Arabic is not an easy language to translate because a lot of it is about beatifying the sound of the text rather than achieving clarity and scientific precision of the words used.

Leave a Reply