Language Log

Locally consistent hallucinations, anyhow… Zoë Hannah, "We pushed this ChatGPT game to the limits, but playing it the right way is more fun", Polygon 7/30/2024:

Apparently, we all like playing god, and we all like doing it badly. I bet none of us thought that removing the ladder from our Sims’ pools was such a universal experience until it became a pretty popular meme, and it’s no secret that lots of mods are centered on adding, uh, explicit elements to games. So, naturally, when I started playing around with DeepGame, Utile Labs’ ChatGPT-based choose-your-own-text-adventure game, I put my best sicko foot forward.

The game, which runs on ChatGPT and is available to anyone with an account, generates stories in a variety of genres. You start off with a command like “Play a romantasy story” or “Surprise me” and let the GPT do its thing — and despite my desire to break the game, I found it much more enjoyable when I took it just a little more seriously.

Like me, you may have been puzzled by Donald Trump's repeated references to Hannibal Lecter in his rally speeches. Given the contexts, I figured it was a connection between "political asylum" and "insane asylum" — and Miles Klee has the receipts ("Why Is Trump So Obsessed With Hannibal Lecter?: A Complete Timeline", Rolling Stone 7/30/2024):

How an off-script moment from early in the election cycle led to a bizarre MAGA ritual celebrating a fictional cannibal

[…] How did Trump end up name-checking Lecter as part of his pitch to the MAGA base? Responding to a request for comment on the matter, campaign communications director Steven Cheung replied, “President Trump is an inspiring and gifted storyteller and referencing pop culture is one of many reasons why he can successfully connect with the audience and voters. Whereas, Kamala [Harris] is as relatable as a worn-out couch.”

Absent any further explanation, a forensic review of the former president’s speeches over the past year is in order. What’s clear is that this all began with a simple misunderstanding — or several.

By now, we've had dozens of posts about the Great Firewall, VPNs, internet censorship, and so forth, but they're all from the vantage of the outside trying to look in.  Of course, that gives us a skewed picture of what the situation is really like with regard to the internet inside and outside of the PRC.  This is not a healthy situation, for nearly one fifth of the world's population (17.72%) live inside the borders of China.  To be ignorant of how they are living is dangerous, for we may make erroneous assumptions about what one fifth of humanity is doing and thinking.

Fortunately, at last I have found an American expat who has been living and working in the PRC for more than a decade at a remote location and is well connected with many Chinese colleagues.  He is an active scholar and very well informed about the internet, AI, databases, and so forth, both inside and outside of the PRC.  I should note that he does not live among expats.  In fact, he is the only Westerner where he is located, quite far from major metropolitan areas, so he truly understands what Chinese of all walks of life do on a day-to-day basis.

Economics buzzword.  From the Wall Street Journal's China newsletter:

Boone Ashworth, "Wear This AI Friend Around Your Neck", Wired 7/30/2024:

The latest attempt at an AI-powered wearable is an always-listening pendant. But it doesn’t help you be more productive, it just keeps you company.

AVI SCHIFFMANN SHOWS up to the WIRED office with a Friend hanging around his neck. It dangles there like a pendant on a necklace. It’s about the size and shape of an AirTag—a soft, round little puck that rests right next to Schiffmann’s heart, just atop the Dark Side of the Moon logo on the shirt behind it.

The Friend, to be clear, is an AI wearable. It’s a pal, a buddy, but mostly an AI chatbot that lives inside the pendant. It always has an opinion to share about what’s going on around it, which it communicates using text messages and push notifications on the phone it’s paired to.

Schiffmann and his Friend (this one’s name is Emily) have come to WIRED’s San Francisco office to meet with me and my colleague Reece Rogers to talk publicly about this new AI wearable for the first time. Before we get started, I tell Schiffmann I’d like to record our chat and ask if he’s cool with that. This is considered a good journalistic practice, sure, but also it’s a legal requirement in California, which requires two-party consent before taping a private interaction. So I ask permission to turn on a tape recorder and Schiffmann just laughs.

“I am the last person who would mind that,” he says.

That makes sense. After all, the pendant around his neck has already been listening to us this entire time.

I've eaten a lot of muxu beef / pork / chicken / shrimp in my day, and I love the combination of meat strips, black "wood ear" fungus, scrambled eggs, daylily, and cucumber served wrapped in a thin, soft pancake.  Usually I'm compulsive about knowing the meaning of the names of dishes that I eat, but muxu has always defeated me.  I'm not even sure how to pronounce the name (it's also transcribed as moo shu, mushu, and mooshi) nor how to write it in characters (variants include completely unrelated 木须, 木樨, etc.).

When I first encountered the dish decades ago, I spent a fair amount of time trying to unravel the jumbled meanings, pronunciations, and written forms of the name.  However, since I was getting nowhere fast, I soon gave up on those investigations (in the days before the internet and search engines, things were much harder to figure out).  Then I spent so many years wandering around overseas, and I simply didn't encounter muxu for a long time.

In last Friday's post ("Annals of intervocalic coronal reduction"), I presented a case from 2015 where Donald Trump pronounced "president" as if it were "prison". This provoked a lot of interesting commentary about the nature and prevalence of various reduced pronunciations of that word, and so I thought I'd add a bit more evidence to the discussion. As I noted a few months ago ("'There's no T in Scranton'", 3/10/2024)

Shuang Li's INTERVIEW: NPR Media Dialog Transcripts dataset […] contains 3,199,859 transcribed turns from 105,817 NPR podcasts, comprising more than 10,648 hours. That dataset is just the transcripts, but some years ago, Jiahong Yuan and I downloaded the audio and aligned it with the texts. And I wrote a simple search script […]

Running that script to search for the word string "president of the united states" turns up 2,443 phrasal clips, from which I selected 12 (literally) at random.

[I chose a consistent context because the often-extreme across-word co-articulation in spontaneous speech means that the last syllable or two of "president" may overlap with what follows, wherefore I've included "…of the united" in the audio clips…]

"Taigi a political question of identity", By Hugo Tseng, Taipei Times (7/27/24)

The issue of whether to call the language spoken in Taiwan “Minnanese” (閩南語) or “Taigi” (台語, taiyu, also called Hoklo or Taiwanese) has long been a subject of debate. On the surface, it seems to be a simple question about language, but in essence it is a political question of identity.

Perhaps we could gain some inspiration from the duality of English as a language. English was, at its earliest, the language of the Angles — the Germanic people from the German-Danish border who invaded and settled in what is now known as England, whose name meant the “Land of the Angles.”

Through colonization and the spread of the language across the world, English — even as it melded with and adopted local characteristics and traits from other languages — remained essentially the same. In the US, Australia and other Anglophone countries, English is the name of the language, but the name is appended with a qualifier — the name of the country where it is used — such as American English or Australian English.

[Serendipitously, right while we are in the midst of energetic discussions over the classification of and terminology for the languages of Taiwan, I received a communication from the international body that is charged with such matters for all the languages of the world, namely, an arm of the ISO.

The following (after the page break) is a guest post by Janell Nordmoe, Registrar of ISO 639-3 Language Coding Agency.  For those who are not familiar with it, "ISO 639 is a standard by the International Organization for Standardization (ISO) concerned with representation of languages and language groups." (source)

There have been significant changes with the publication of 639:2023, including that the decision on CRs rests with the Maintenance Agency, not SIL as Language Coding Agency for 639-3.

This link describes the four sets within ISO 639, the Maintenance Agency.

——————————————————-

What word do you hear in this clip?

If you heard "prison", you agree with me, and with Google's speech-to-text algorithm.

[This is a guest post by Nathan Hopson]

I recently received the following delightful question from Hilary Smith (University of Denver) about the origins of the term for protein in Chinese (dànbáizhì) and Japanese (tanpakushitsu). Thanks to her for pointing me down this lovely rabbit hole!

The hanzi/kanji used are identical (蛋白質), though in written Japanese the term is often タンパク質 or たんぱく質 because the 蛋 character is not one of the “regular use” kanji (常用漢字 jōyō kanji) selected by the officially announced by the Japanese education ministry for mastery during compulsory education.

From a Facebook page with Army background in Taiwan:

Facebook page for Voice of Han Broadcasting Network
(漢聲廣播電台 hànshēng guǎngbō diàntái)
from Taiwan's Ministry of National Defense

"Not Lost In Translation: How Barbarian Books Laid the Foundation for Japan’s Industrial Revolution", by Alex Tabarrok, Marginal Revolution (July 22, 2024)

I am grateful to Alex Tabarrok and his colleague Tyler Cowen at Marginal Revolution University of George Mason University's Mercatus Center for introducing me to what is one of the most mind-boggling/blowing papers I have read in the last decade.

First, here is Tabarrok's introduction, and that will be followed by selections from the revolutionary paper to which I am referring.

Japan’s growth miracle after World War II is well known but that was Japan’s second miracle. The first was perhaps even more miraculous. At the end of the 19th century, under the Meiji Restoration, Japan transformed itself almost overnight from a peasant economy to an industrial powerhouse.

After centuries of resisting economic and social change, Japan transformed from a relatively poor, predominantly agricultural economy specialized in the exports of unprocessed, primary products to an economy specialized in the export of manufactures in under fifteen years.