Friday, July 2, 2010

Robo-Linguists At Last!!

(image from evasee)

Finally, the tedious job of linguists has been replaced by robots! The site proudly trumpets the triumph of algorithm over the comparative method with the post title: Computer program deciphers a dead language that mystified linguists. proudly proclaims the following:
The lost language of Ugaritic was last spoken 3,500 years ago. It survives on just a few tablets, and linguists could only translate it with years of hard work and plenty of luck. A computer deciphered it in hours.

However, just a brief scan of the original article (pdf) suggests that there's less here than meets the eye. The abstract begins thusly:

In this paper we propose a method for the automatic decipherment of lost languages. Given a non-parallel  corpus in a known re- lated language, our model produces both alphabetic mappings and translations of words into their corresponding cognates.

Producing an alphabetic mapping and a cognate set is nice, but "deciphering a dead language" it ain't.

HT: Sérgio Bernardino vi Twitter #linguistics)


Jason M. Adams said...

Yeah, typical pop science (mis)coverage. The technique is a tool to bootstrap linguists rather than a tool to replace them. Also, the fact that you need a known related language pretty much limits this technique to Ugaritic and other languages that have already been deciphered. According to Andrew Robinson (author of Lost Languages, but whose authority I can't really establish): "[having a known, related language] is almost certainly not the case with any of the important remaining undeciphered scripts."

Chris said...

I know some linguists I'd like to boot-strap, ifyaknowwahaddamean...

thewordguy said...

Reminds me a little of the old notion that computers would create a paperless society. Try telling that to the trees. Given an infinite number of monkeys writing code, I suspect we could develop software that mimics the human brain, but I doubt this paper heralds the arrival of unemployed linguists cluttering the street corners with their "Will parse for food" signs.

Chris said...

As most linguists already count as underemployed, they didn't need this competition hehe, but you'r right, nothing to worry about.

NLPers: How would you characterize your linguistics background?

That was the poll question my hero Professor Emily Bender posed on Twitter March 30th. 573 tweets later, a truly epic thread had been cre...