Language Log

I just thought of a (not so) funny phenomenon that amounts to a linguistic paradox.  Namely, as languages die out, one after another, so do they arise at a steady rate. This has been a verity throughout human history.  So inexorable are these trajectories that a clever mathematician might be able to work out an equation to account for them.

The cycle of language saṃsāra संसार is ceaseless.

We know well enough how languages disappear — usually forever (see "Archive for Language extinction"; see also here).  We have also witnessed the birth of languages, which happens for a variety of reasons (social, political, linguistic, etc.).  Increasingly, however, artificial languages are being invented in astonishing numbers.

We have seen ChatGPT tell stories (and variants of the stories it tells), fancify Coleridge's famous poem on Xanadu, pose a serious challenge to the Great Firewall of China, mimic VHM, write Haiku, and perform all manner of amazing feats.  In a forthcoming post, we will witness its efforts to translate Chinese poetry.  Today, we will watch ChatGPT make a credible foray into Akkadiology.

Translating old clay tablet by using chatGPT

Jan Romme, Jan's Stuff (5/15/23)

The author commences:

You might have heard how I asked chatGPT to pose as a Jehovah’s Witness, write a “witnessing letter” with 2 or 3 bible scriptures in it, and then translate that letter into an English rap song, Eminem style.  Or you might have missed that news. My point is, I like to play with AI’s.

I’m increasingly stupefied by how much AI models like OpenAI’s chatGPT, Google’s BARD, and Facebooks LLaMMa and others are capable of.

Bill Benzon has been our most prolific humanistic commentator about GPTs, almost as prolific as GPTs themselves.  Here he introduces his latest creation in / on the genre:

"From 'Kubla Khan' through GPT and beyond", 3 Quarks Daily (3/27/23)

In a covering note to me, Bill writes:

A story about how I came to be interested in GPTs. It’s also implicitly a critique of the large language model business. You have a bunch of very smart and clever people creating engines that pump out language by the bucketful, but who seem to have little interest in or knowledge about language itself, much less linguistics, psycholinguistics, or the various cognitive sciences. It’s crazy. But the machines they’re producing are marvelous and fascinating.

Less than half a year ago, we were treated to yet another among countless claims for the decipherment of the mysterious Voynich manuscript (henceforth "Vm"):  "Voynich code cracked?" (5/16/19).  I was skeptical then and am even more skeptical now after having read this article:

Peter Bakker, "The Voynich manuscript: the decipherment of ms. 408", Lingoblog (9/10/19)

Urgent bipartite action alert for The Economist: First, note that my copy of the July 18 issue did not arrive on my doormat as it should have done on Saturday morning, so I did not have my favorite magazine to read over the weekend; please investigate. And second, the guerilla actions of the person on your staff who enforces the no-split-infinitives rule (you know perfectly well who it is) have gone too far and are making you a laughing stock. Look at this sentence, from an article about Iran (page 21; thanks to Robert Ayers for pointing it out; the underlining is mine):

Nor do such hardliners believe compliance will offer much of a safeguard: Muammar Qaddafi's decision entirely to dismantle Libya's nuclear programme did not stop Western countries from helping his foes to overthrow and kill him.

Matt Treyvaud forwarded this from the Lojban mailing list:

"Lojban changes to hanzi writing system" (4/1/15)

Some people complained that although the spelling in Lojban is very easy to grasp the grammar is not. So the committee for the development of Lojban (BPFK) decided to fix this issue and to make the spelling hard as well.  Especially for those people who are not familiar with hanzi (Chinese characters).