Language Log

« previous post | next post »

Over the past couple of days, I've continued to use Google's alpha Persian-English translation system as part of an attempt to keep track of what's happening in Iran.

On long passages, the results are still at the fever-dream stage of machine translation, where enough relevant words and phrase-fragments emerge to leave a sort of impressionistic residue of content, but without much overall coherence. For example, I tried it on a bulletin from Mehr News yesterday evening that claimed to be a statement from the Assembly of Experts announcing full support for Kahmenei's speech on Friday. This sentence

به گزارش خبرگزاری مهر ، در این بیانیه آمده است: مجلس خبرگان رهبری ضمن تشکر از حضور شکوهمند و حماسه‌ساز مردم در انتخابات ریاست جمهوری، حمایت قاطع خود را از بیانات روشنگرانه، وحدت‌بخش و داهیانه‌ مقام معظم رهبری در نماز جمعه تهران اعلام می‌دارد و با شکرگزاری به درگاه الهی نسبت به نعمت عظما و بی‌بدیل ولایت فقیه، این رکن رکین حدوث و تداوم انقلاب؛ همگان را به تبعیت از دستورات و رهنمودهای معظم‌‌له فرا می‌خواند.

comes out in the automatic translation as

Mehr News Agency reported, the statement states: the Assembly of Experts also thanked the glorious presence Hmas·hsaz and presidential elections, support their statements Rvshngranh decisive, and Vhdtbkhsh Dahyanh Ayatollah Khamenei Friday Prayers in Tehran and ready Thanksgiving Portal to the Divine favor Zma Bybdyl and velayat-e faqih, the pillars of the revolution and continuity Rkyn Hdvs; everyone to obey the orders and guidelines Mzmlh call girl.

(More recent reports, by the way, have claimed that the statement (or at least the bulletin about it) was essentially a fake, signed by one not-necessarily-representative member of the group and not by the Assembly as a whole.)

Although I've mostly used the MT system on texts for which no human translation was available, it's interesting to compare machine and human translations where both exist, for example in the case of Mir Hossein Mousavi's statement on Saturday. Here are a few comparisons of sentences in the machine- and human-translated versions:

For self-belief I see that what you put in between, and you say you learn, that mission Nbrym learn from history and shoulder the responsibility on the shoulders of the fate of generations and the evenings are empty Let us put.

It is my duty to share with you what I believe, and to learn from you, may we never forget our historical task and not give up on the duty we are given by the destiny of times and generations.

In this period, and especially in our time of life Imam illuminati capitals huge life and property, and prestige in Mubarak left foot building will consolidate the achievements and was a valuable result.

In those times, especially when our enlightened Imam [Khomeini] was alive, large amount of lives and  matters were invested to legitimize this foundation and many valuable achievements were attained.

This is clearly a need in the military and police presence in the street scene and will see some of them and sent them to hear the heart of each interested country's revolution and the pain brings, will not face.

It is clear that in this case, there won’t be a need for security forces on the streets, and we won’t have to face pictures and hear news that break the heart of anyone who loves the country and the revolution.

As I reported a couple of days ago, this tool in its current state seems to be much more useful in getting information out of the short phrases found in Persian-language tweets. In translating a message of 10 or 15 words, there's less scope for the sort of switching-stations-in-marginal-reception sort of thing that tends to happen in statistical phrase-based translation:

نمایندگان مجلس خواستار دیدار با بازداشت شگان روزهای اخیر شدند

Parliament demanded to meet with detained in recent days were Shgan

کشته شدن 13 نفر در درگیریهای روز گذشته تایید شد اما برخی خبرها از کشته شدن 19 نفر حکایت دارد

13 people being killed in clashes yesterday confirmed news, but some 19 people being killed is a story

For comparison, I've given below  the first paragraph of output from the (much more mature) Google Arabic-English translation, applied to an Al-Arabiya story about recent events:

Iranian sources confirmed that many of the members of the Council of Experts, on the appointment or removal of all the guardian jurist, who are your terms of reference of the senior, who recently met the President of the Council Hashemi Rafsanjani, are prone to the formation of the collective leadership of running the country instead of one guardian jurist, in the light of what they say bias Wali Faqih Ayatollah Ali Khamenei, for the outgoing President Ahmadinejad, and he has the leadership that some of the conditions stipulated by the Constitution, as a special report for "Arabic. Net."

This is far from perfect, but it's good enough for me to believe that I can understand quite precisely what this interesting rumor is, and not just what sorts of things it's probably in general about.

Given this technology's sustained progress over recent years, we can look forward to the time when the Persian translation is as good as the Arabic translation now is — or in fact when both will be better.

June 21, 2009 @ 6:51 am · Filed by Mark Liberman under Computational linguistics

Permalink