Friday, June 29, 2007

On the Origin of Languages

Re-reading the intriguing book by Merritt Ruhlen: On the Origin of Languages: Studies in Linguistic Taxonomy. The basic premise is that contrary to widespread belief, the tools of comparative linguistics can be applied beyond the Indo-European family to reconstruct the roots of a world proto-language. Though it sounds far-fetched at first, Ruhlen makes a compelling case for his methodolgy, and gives some world-etymologies that I can't resist from reproducing below. It gives me particular satisfaction to see these ancient roots manifest themselves in the languages I read, write and speak: Russian, English, Hebrew, Latin. The asterisk indicates an unattested form.

KAMA 'hold (in the hand)'. Ruhlen gives the Proto-Afro-Asiatic root *km, from which we get the Arabic kamasa 'seize, grasp'. Could the Hebrew חמש 'five' be related (as in, the five fingers of the hand)? The Indo-European cognate is *gemo, which appears in Russian as жму 'I press'.

My favorite: MANA 'to stay (in a place)'. Proto-Afro-Asiatic: *mn, which manifests in Hebrew as אמן. This root means 'true, enduring', and is borrowed by many Indo-European languages as amen. Of course, the I-E family has this root occurring natively as well -- as the Proto-IE *men. The Latin manere will be more familiar to English speakers as remain.

Another favorite: MENA 'to think (about)'. Hebrew: מנה 'to count'; English: mind, mental; Russian: мнить.

I'll close with the colorful PUTI 'vulva'. Hebrew speakers will immediately recognize this as פות, via the Proto-Afro-Asiatic *pwt 'hole, anus, vulva'. Speakers of Romance languages might be pleased to learn that when they curse a woman as puta(-na), they're invoking an ancient root, dating back tens of thousands of years!

A final note, and an appeal to my more professional linguist readers. Ruhlen writes: "the Indo-European family has been established beyond doubt," and this has been my belief ever since I began to amateurishly dabble in comparative linguistics. However, I recently had an argument with a computational linguist/computer scientist/mathematician who claims that the IE-family is "merely" a hypothesis, and a rather controversial one at that. Does anyone know of a reputable linguist who doubts the common origin of the so-called Indo-European languages, and questions the basic structure of reconstructed Proto-IE?

Update: More world etymologies are available online.

Update II: To be fair and balanced, I'm linking to a harsh critique of Ruhlen and his methods, with a hat tip to Cosma.

Thursday, June 14, 2007

Shameless self-promotion

Some good news on the publications front. My paper with Kavita Ramanan has been accepted (with minor revisions) to the Annals of Probability. If you just want the main idea (actually, a much simpler proof of a more general form of the main result), this is the paper to read.

Another recent acceptance is my Universal Regular kernel extended abstract, to appear in MLG'07. It's a short 4-page writeup, and if you can resolve the issue of computing K_n, I guarantee you fame and fortune.

I'm hanging out at the FCRC conference (Mehryar is presenting our Rational Kernels paper with Corinna at COLT). So if you're around, find me and say hi, and most definitely come to Mehryar's talk on Thurs. at 2:40. I'll try and blog about the conference a bit later, but we always have Scott to count on.

Monday, June 11, 2007

Thesis online

After abusing our department coordinator's patience beyond all common decency, I've stopped making revisions on my thesis. I know, I know -- a week from now I'll casually glance at it and see something I'll want to change. But major OCD notwithstanding, enough is enough. I'm putting it online for public perusal; don't all rush in to download it all at once now. As always, questions and comments are more than welcome.

Saturday, June 9, 2007

Idea for Sci-Fi story

How can you tell if the reality you experience is "real" or is just a giant computer simulation? Philosophers realized long ago that of course you cannot; Hofstadter and Deutsch make the a posteriori obvious point that the question itself is meaningless. Any physical process may be viewed as a computation and therefore a "simulation".

But what if we allow the possibility of simulator malfunctions? Address errors, memory leaks, unknown-error-must-shut-down type things. What would it feel like to be in a simulation that suddenly displayed such artifacts? Parts of your universe are working fine as before, but you might locally observe very strange discontinuities and irregularities.

There are sophisticated theories of spacetime defects that I lack the mathematical apparatus to understand (any quantum gravitists want to help out?). Might any of these defects be explicable as computer bugs in the universal simulator? Could one at least get a decent sci-fi story out of this? I'm sure this vein has been already explored -- can anyone point me to a good story? Anyone up the the task of writing one?