languagehat.com

Nicholas Andresen’s recent post annoys me somewhat because (like almost everyone else) he writes as if “AI” were some kind of “conscious” “agent” that does things with purpose (it starts “In September 2025, researchers released the internal monologue of OpenAI’s GPT-o3 as it decided to lie about scientific data. Here’s what it was thinking…”), but never mind, it’s got a lot of interesting material. He starts with the discovery that “a model’s own output could serve as scratch paper” (this is called “Chain-of-Thought”), and continues:

But something strange is happening to Chain-of-Thought reasoning.

Remember that screenshot we started with? “Glean disclaim disclaim.” “Synergy customizing illusions.” Online, people have started calling this kind of thing Thinkish. There’s a whole emerging vocabulary: “watchers” apparently means human overseers, “fudge” means sabotage, “cunning” means circumventing constraints. Some of the other words – ”overshadows,” “illusions” – seem to mean different things in different contexts, and some combinations resist interpretation entirely.

Weirdly, Thinkish reminds me of home.

I grew up near Gibraltar – a tiny British territory dangling off the southern tip of Spain – where a Spanish-English blend called “Llanito” is spoken. Here’s an example:

“Llévate el brolly que it’s raining cats and dogs.”

To a Llanito speaker, this is completely normal: “take the umbrella, it’s pouring”. To anyone else, it might take a minute to parse – there’s a Spanish verb, borrowed British slang, and an idiom that makes no literal sense in any language.

And Llanito feels great to speak. You’re never stuck – you can always grab the best-fitting piece from each language and move on. It’s just less bueno for everyone else.

That’s Thinkish. The model seems to grab whichever token fits best, optimizing for its own convenience. The problem is that we’re the outsiders now.

He goes on to Alfred the Great’s translation of The Consolation of Philosophy and how languages change, and returns to AI:
[Read more…]

I’ve praised Tessa Hadley many times at LH (e.g., last year), and I’m going to do so again; her latest New Yorker story, “The Quiet House” (archived), is every bit as good as I expected, and I thought I’d bring a couple of excerpted paragraphs here in the hope of enticing readers:

During the time of their youthful adventures, and although they proclaimed themselves feminists, they still more or less thought all those things about the inauthenticity of women. They didn’t so much think them with their conscious minds: the sensation of secondariness was built into the very texture of their imagination and their desires. They supplied to every adventure some invisible observer, male, to fulfill it and make it real. And yet the girls also took for granted, with contemptuous confidence, their right to travel alone and wear shorts and sleeveless tops if they wanted to, while girls their age in Italy and Greece were kept chastely at home. They learned how to say foul things in other languages, in order to put off the boys and men in those countries who followed them and propositioned them, pleading with them so insistently and cravenly—“like dogs,” Jane said. They saw the recoil and disgust on the boys’ faces, at hearing those words from a girl’s mouth.
[…]

The book group had degenerated somewhat, Jane and Geraldine both thought, into a kind of dining club, each member feeling obligated to put on a spread of delicious Ottolenghi-type dishes when it was their turn to host. Discussion of the books was too perfunctory; the two friends’ ideal would have been more like a seminar. They brought their books marked up and bristling with torn slips of paper, and were disappointed when they were hardly opened. Both of them devoured fiction: Jane, a history graduate, was susceptible to a serious theme and anything in translation, whereas Geraldine, who’d done English literature, insisted she cared only about the sentences. Life was hard, she said. Thank God for sentences.

And speaking of sentences, here’s the story’s first: “Geraldine woke out of busy dreams into the calms and shallows of old age.” I memorized it even as I read it. Now, that’s writing.

If you want to blow your wig, cast your pies on the Hepcats Jive Talk Dictionary (1945):

In the old days the people greeted a new ruler with the cry: “The King is dead, long live the King!” Today there’s a strange shout: “Webster is dead, long live the Hepster!” […]

The Editors feel that JIVE TALK is here to stay and believe that this book is the most complete dictionary of Jive words and phrases ever presented. It may be of no assistance to an Abercrombie but we’re sure it will help most people a great deal.

Zowie! And don’t miss the JEOGRAPHICAL JIVE section, which goes from KEEN from ABERDEEN to CONQUERS from YONKERS. Take a bow, Jay!

I was reading Mary M. Wiles’ Interview with Jacques Rivette (yes, I’m still on my Rivette kick; I recently saw his late sort-of-thriller Secret défense and posted a rant about the folly of calling it “Secret Defense” in English) and was struck by this passage:

mw: Why did you feel the need to return to the theater after La bande des quatre?

jr: After La bande des quatre, I was so pleased with the work we did with the actresses that I wanted to continue. And since they also wanted to do some theater, we got together a small group like that to work with each other on some classics; Corneille, Racine, and Marivaux, is what we were working on. And then at the end of a few months, they wanted to do a real performance, and that’s when we dropped Marivaux, which was too hard. But we kept—and we should have just kept Corneille. Because, it was already too much, Corneille and Racine. It was just that we wanted to continue to work on those classics … on Corneille, which was the driving force. And besides, the work that I did with them and with the guys who had joined them was much more interesting to me, the play by Corneille, that is, Tite et Bérénice [1670], than working on Racine’s plays, on Bajazet [1672]. Bajazet is fabulous, I can’t speak ill of Racine, but you grasp it all in the very first reading. That’s it, we said, we understand it all, but then, afterwards, what were we going to do with it? Whereas Corneille is hard, even for the French, he’s hard.

mw: You can imagine that for us …

jr: It’s like Latin for us. It isn’t French. All of those authors were fluent readers of Latin and Greek, but Corneille couldn’t read Greek, and that’s a big difference between them. Racine read Greek, and Corneille read Latin. And he read so much Latin that it’s almost Mallarmé, it’s so dense. Corneille’s Bérénice, it’s true that it’s a hard play, it’s overloaded, each verse says three things, and its characters are infinitely rich. Infinitely more things are happening between Titus and Bérénice, infinitely more things are happening in Corneille’s play than in Racine’s, where nothing happens. Bérénice isn’t the Corneille play that I like the best though, even if it’s got fabulous language. It’s fabulous as a poem, but it’s like Shakespeare for you. As a matter of fact, I gather it’s hard to translate, like Goethe is untranslatable, like Pushkin is untranslatable, like Dante is untranslatable, well, like all the major poets. I don’t know, but, in any case, it was difficult to stage.

mw: I’ve read a few of Corneille’s plays but have never seen them performed on stage.

jr: In France, I’m not the only one, once you get hooked on Corneille, you’re lost. It’s very deep. He’s an author I find very dense, so full of history, of thought. He’s a very rich author.

It makes me wish I’d kept up the French literature I studied almost sixty years ago with the imperious Mme Ruegg; we read Corneille and Racine and I loved them both, but I can no longer remember them well enough to know whether I agree with what Rivette is saying. It doesn’t matter, though; when somebody expatiates like that about literature they love, it’s the loving attention that matters, I don’t really give a damn whether they’re “right.” (I often have that reaction to Dmitry Bykov’s essays about Russian lit.) Anyway, I should give Bérénice a try.

Once, long ago, when Languagehat was a pup, Blogovia flourished, and friendly links o’erspread the world like a canopy. Now, like a region abandoned by industry, it is full of ghost towns, and all the young pups flock to the dread silos of Social Media. Every once in a while, though, one comes back to life, and such is the case of the blog of bulbul (aka Slavomír Čéplö), the much-loved bulbulistan, which has been missing in action for over a decade apart from a brief pondering of the passage of time in 2019. Now there are three posts in rapid succession: mentalist (a dive into the history and semantic range of Bulgarian пич, which reminded him of “a Slovak vulgar term for feminine genitals”), depowedlajo (an exploration of Targumic used for humorous effect in a Purim play published in Vienna in 1878 titled Reb Simmel Andrichau — I expect this will appeal to rozele among others), and now work, which begins:

So, anyway, been a while, right? How have y’all been the last *checks notes* few years? Yeah, I know, interesting times… How about instead of focusing on that shit, I show you what I have been up to since 2015 or so. Let’s start with some of the projects I have been working on that you might find interesting.

He talks about his work on HunaynNET (a project that collected all texts of classical science that were translated into both Syriac and Arabic: “The translations were then re-edited and aligned on the level of semantic and syntactic units […] The text is also tokenized and links to dictionaries and corpora are provided; and in some cases, we also provided aligned text of translations into modern languages”), Simtho (“an electronic corpus – the only one worthy of the name – of the Syriac language”), and the Zoroastrian Middle-Persian Corpus and Dictionary:

This DFG-supported ongoing project seeks to collect, annotate and analyze all available Zoroastrian texts written in Middle Persian to create a searchable corpus (in transcription) and finally an updated dictionary of Middle Persian. I was largely responsible for data processing, conversion and import, so none of what you see online is my work. The web application is still very much a work in progress, but once finished, it will be a one-stop shop for all your Zoroastrian Middle Persian needs, including manuscript images and comprehensive lexical resources.

Good stuff; congrats, bulbul, and don’t be a stranger!

For almost a decade I’ve had a copy of Jacek Hugo-Bader’s White Fever: A Journey to the Frozen Heart of Siberia hiding in the depths of the to-be-read pile to the left of my desk; it recently rose to the top, and I thought “I should read that,” so that’s what I’m doing, with great pleasure. Hugo-Bader hangs out with old hippies and the like, and has a whole section called “a small and impractical Russian-English dictionary of hippy slang,” most of which is probably long out of use (he visited Russia in late 2007). At one point he mentions “blatna music, in other words criminal, bandit, jail or prison camp music” (the book is translated from Polish, hence Polish forms like blatna rather than the Russian blatnaya), and goes on to write:

The word blatny probably comes from Yiddish, into which it passed either from the German word ‘Blut’, meaning blood, or ‘Blatt’ meaning a page, a sheet of paper, because whenever the bandits in Odessa came to rob the stores of an old Jew or a German, they stuck a revolver barrel in his face and said it was their ‘Blatt’, in other words their receipt or goods delivery document. Thus bandits came to be called blatniye in Russian.

I figured that was probably all nonsense, but it seems the Blatt idea is taken seriously, though via Yiddish rather than German; Wiktionary has “From Polish blat or Yiddish בלאַט (blat).” But the Russian version has different suggestions — they reference Yiddish blat ‘trusted,’ originally ‘illegal,’ “possibly stemming from an Old Hebrew etymon — pāliṭ “fugitive,” pālaṭ “he fled”), cf. German slang Platt ‘new member of a gang of thieves,’ platt ‘one of us, trustworthy.’” I have no idea how seriously to take any of that.

I wasn’t sure whether to post Naomi Kanakia’s The New Yorker offered him a deal, because it’s very long and wouldn’t be of much interest to someone who didn’t care about John Cheever or the strange phenomenon known as the “New Yorker story,” but if you do care about those things it’s fascinating — she goes into the whole history of the magazine’s stories and why they work (and why they’ve always been criticized), and why the magazine’s demands determined Cheever’s career. And I have a deep respect and affection for people who do a deep dive into a corpus they’re interested in so they can report on the results with authority rather than making the usual facile assumptions and moving on: “All told, I’ve probably read five hundred New Yorker stories over the last three months.” (Compare my appreciation of Vera Dunham’s In Stalin’s Time, for which she “waded through mountains of elephant shit,” and see this 2010 post linking to Slawkenbergius’ “thoughtful take on John Cheever.”) She starts off:

Two months ago, I read a seven-hundred-page collection of short stories by John Cheever. But somehow that wasn’t enough. I went on to read seven-hundred-page retrospective collections from Mavis Gallant, Alice Adams, and John O’Hara. And I still wanted more!

Normally when I get halfway through a story collection I think, “Okay…I’m done now”, but with these authors, it wasn’t like that. I wanted more. Not more of these particular writers, but more work that was like their work in some weird, indefinable way. […]

Not only were these stories similar to each other, but they also seemed quite different from other literary stories. These stories were mostly marked by their extreme restraint. They didn’t just eschew plot, they also eschewed lyricism, symbolism, surrealism, or any other devices that would call attention to themselves. Their plotlessness made them seem highbrow, but their unadorned style made them highly accessible. And I wondered how The New Yorker could’ve arrived at this unique-seeming combination of elements.

And she proceeds to the history of how the early “casuals” became short stories, and how the tastes of the first editor, Harold Ross (who “was never totally sold on the idea of publishing stories”), determined the kind of story that would become the hallmark of the magazine. If any of this sounds intriguing, give it a try and you may find yourself reading the whole thing (and perhaps developing a new respect for Cheever, who for so long was a punching bag for critics of all descriptions and who was persistently underpaid by the magazine).

The video How Far Back Can You Understand Northern English? nominally lasts twenty minutes, but it will take longer if, like me, you keep pausing it to read the footnotes. It was sent me by rozele, who says:

it’s a dialect coach called simon roper’s retrogression through cumbrian english from circa 2000 to circa 1200, followed by some overall comments, and then a subtitled (both IPA and a standard u.k. english rendering) repetition, with enough on-screen notes on his reconstruction to make me wish youtube had 10-second jump controls. i don’t know a lot of northern english dialectology, so can’t check his work, but i was quite impressed, on a bunch of levels.

one bigger-picture tidbit from the notes that was news to me, though i’m sure it’s familiar stuff to many of the hatters:

“In the north, from the Middle English period onwards, verbs agreeing with ‘thou’ tended to take the ending ‘-s’ (‘thou does’), unlike in the south where they took the ending ‘-(e)st’ (‘thou dost’). This remained true into the 20th century, and as far as I know is still true in northern dialects that retain ‘thou’.”

(which has me wondering whether the stereotyped u.s. quaker use of “does” in 2sg, which i’d always assumed was generalized from the 3rd person, is just a northernism. and it seems somehow relevant to the current singular “they are”, too.)

I learned a great deal from it, including the word gled ‘(red) kite’ (OED: “the Old Germanic form was probably *gliđon- and with o- umlaut gleđon-, < glið- weak grade of the root of *glîđan to glide v.”). At one point he discusses ingressive speech, which we covered in 2014. I was surprised how far back I could mostly understand what was being said (I think I started losing the trail around the fifteenth century), but my immersion in British cop shows has given me a head start in northern dialects — thanks, Vera, and thanks, rozele!

I was shamed by David Eddyshaw’s recent comment (“I highly recommend The English Understand Wool to the three Hatters who have not already read it”) — he, of course, had no intention of shaming me, as he could not possibly have known that I had failed to read the latest fiction by one of my favorite authors — and I have accordingly remedied my inexcusable omission. I have no intention of telling you “what it’s about”; the text itself will do that. I will quote a couple of early bits to convince you that you need to read the whole thing (it’s a short book and will take only a few hours). Here’s the opening of the novella:

—The English understand wool.

My mother sat on a small sofa in our suite at Claridge’s, from which the television had been removed at her request. She held in her lap a bolt of very beautiful handloomed tweed which she had brought back from the Outer Hebrides. She had in fact required only a few metres for a new suit.

I use the word “suit” because I am writing in English, but the French tailleur—she would naturally think of clothes in French—makes intelligible that one would travel from Marrakech to the Outer Hebrides to examine the work of a number of weavers, perhaps to establish a relationship with a weaver of real gifts. It makes intelligible that one would bring one’s daughter, so that she might develop an eye for excellence in the fabric, know the marks of workmanship of real quality, observe how one develops an understanding with a craftsman of talent. The word “suit,” I think, makes this look quite mad.

She had needed only a few metres, but she had bought the entire bolt to prevent it from falling into ignoble hands. We had stayed overnight on the way north in Inverness, where the shops were full of distinguished tweeds put to debased uses.

And here’s the start of ch. 4:

Maman was exigeante—there is no English word—in matters of protocol. Lunch, tea and dinner were served formally. English was spoken if my father was present, French if we were alone. It is important for the servants to become accustomed to the correct manner of serving; if the President of the Republic comes to dine, they must not be anxiously casting their minds back to the last important dinner. It is an advantage to them to speak both French and English flawlessly. (When they made mistakes, they were corrected.) If an opportunity arises in a great hotel, they will not be unprepared.

Maman spoke French with a pure Parisian accent. She used this in the normal management of the household; it was better for them to accustom themselves. She spoke the standard Arabic, the Arabic of television, of high-level functionaries, of international businessmen, on formal occasions where French was inappropriate. She spoke Darija, the Moroccan form of Arabic, when the servants were ill or had family problems. This was the hardest to learn because the language schools did not like to teach it, and private instructors felt they would lose face if they did not teach what was taught in schools. What she set out to do she did.

I will add that I am entirely in accord with DE’s “At several points I had to suppress an embarrassing desire to cheer out loud for the remarkable heroine, an eminently worthy spiritual sibling to Ludo.” Nobody writes like DeWitt.

I know “odd British pronunciations” is a hoary old trope, and we’ve had posts about it before, but I was struck when looking something up in my trusty BBC Pronouncing Dictionary of British Names by the density of names with unpredictable pronunciations on the spread pp. 70-71. Many of them, of course, are easy enough, e.g. Heriot [ˈherɪət] (hérriot), but around half seemed worth reproducing here:

Helwick Shoals and lightship [ˈhelɪk] (héllick)
Hely, f.n. [ˈhilɪ] (heeli)
Helyer, f.n. [ˈhelɪər] (hélli-er)
Heman, f.n. [ˈhimən] (heeman)
Heming, f.n. [ˈhemɪŋ] (hémming)
Hemingbrough [ˈhemɪŋbrʌf] (hémming-bruff)
Hemmerde, f.n. [ˈhemərdɪ] (hémmerdi)
Hene [ˈhinɪ] (heeni)
Heneage, f.n. [ˈhenɪdʒ] (hénnij) Appropriate also for the Barony of ~.
Henebery [ˈhenɪbərɪ] (hénneberi)
Heneghan, f.n. [ˈhenɪgən] (hénnegan)
Heneglwys [henˈegluɪs] (henégloo-iss)
Heneker, fm. [ˈhenɪkər] (hénneker)
Henig, f.n. [ˈhenɪg] (hénnig)
Henlere, f.n. [ˈhenlɪər] (hénleer)
Henriques, f.n. (henˈrikɪz] (henreekez)
Hepburn, f.n. [ˈheb3rn] (hébburn); [ˈhebərn] (hébbŭrn)
Hepburn [ˈheb3rn] (hébbŭrn)
Heppell, f.n. [ˈhepl] (heppl)
Hereford [ˈherɪfərd] (hérreferd) Appropriate also for Viscount ~.
Hergest Ridge [ˈhargɪst] (haargest)
Herklots, f.n. [ˈh3rklɒts] (hérklots)
Herkness, f.n. [ˈharknɪs] (haarkness)
Hermges, f.n. [ˈh3rmdʒiz] (hérmjeez)
Herries, Baron [ˈherɪs] (hérriss)
Herstmonceux, also spelt Hurstmonceux, Hurstmonceaux [ˌh3rstmənˈsju] (herstmo6n-séw); [ˌh3rstmənˈsu] (herst-mon-soo)
Hertford [ˈharfərd] (haarford) Appropriate also for the Marquess of ~.
Hertingfordbury [ˈhartɪŋfərdberɪ] (haartingfordberri)
Hervey, f.n. [ˈharvɪ] (haarvi)
Herwald, f.n. [ˈh3rwəld] (hérwald)
Heseltine, f.n. [ˈhesltain] (héssltin) Also the pronunciation of Peter Warlock, composer, for his nom-de-plume of Philip ~.
Heselton, f.n. [ˈhesltən] (hésslton)
Hesilrige, f.n. [ˈhezɪlrɪdʒ] (hézzilrij)
Hesleden [ˈhesldən] (hésslden)
Hesmondhalgh, f.n. [ˈhezməndhælʃ] (hézmənd-halsh) ; [ˈhezməndhɔ] (hézmond-haw)
Hespe, f.n. [hesp] (hessp)
Hessé, f.n. [ˈhesɪ] (héssi)
Hessle [ˈhezl] (hezzl)
Hethel [ˈhiθl] (heethl) ; [ˈheθl] (hethl)
Heugh, f.n. [hju] (hew)
Heugh, Northumberland [hjuf] (hewf)
Hever [ˈhivər] (heever)
Hewardine, f.n. [ˈhjuərdin] (héw-ardeen)
Heyrod [ˈherəd] (hérred)
Heysham [ˈhiʃəm] (hee-sham)
Heyshott [ˈheɪʃɒt] (hay-shot)
Heythrop [ˈhiθrəp] (heethrop) Appropriate also for the ~ Hunt.
Hibaldstow [ˈhɪblstoʊ] (hibblsto)
High Legh [ˈhaɪ ˈli] (hi lee)
High Wych [ˈhaɪ ˈwaɪtʃ] (hi witch)
High Wycombe [ˈhaɪ ˈwɪkəm] (hi wickem)
Higham, f.n. [ˈhaɪəm] (hi-am)
Higham, East Suffolk, West Suffolk [ˈhaɪəm] (hi-em); [ˈhɪgəm] (higgam)
Higham, Yorks. [ˈhaɪəm] (hi-em); [ˈhɪkəm] (hickam)
Hinchingbrooke, Viscountcy of [ˈhɪnʃɪŋ-brʊk] (hinshing-brook)
Hindolveston, also spelt Hindolvestone [ˈhɪndlˈvestən] (hindlvéston); [ˈhilvistən]
(hilvéston)
Hindsley, f.n. [ˈhaɪndzlɪ] (hindzli)
Hinwick [ˈhɪnɪk] (hinnick)
Hiorns, f.n. [ˈhaɪərnz] (hi-ornz)
Hippisley, f.n. [ˈhɪpslɪ] (hipsli)
Hiron, f.n. [ˈhaɪərɒn] (hiron)
Hirwaun, also spelt Hirwain [ˈhɪərwaɪn] (heerwin); [ˈh3rwɪn] (hirwin)
Hoathly, East and West [hoʊθˈlaɪ] (hoth-li)
Hodder & Stoughton, publishers [ˈhɒdər ənd ˈstautən] (hodder and stowton)
Hodghton, f.n. [ˈhɒdʒtən] (hojton)
Hoenes, f.n. [ˈhoʊnes] (honess)
Hogarth, fm. [ˈhoʊgarθ] (hogaarth); [ˈhɒgərt] (hoggart) The first is traditional for William ~, painter and engraver. The second is usual in Cumberland and Westmorland.
Hoggan, f.n. [ˈhɒgən] (hoggan)
Hoggard, f.n. [ˈhɒgard] (hoggaard)
Hoggarth, f.n. [ˈhɒgərt] (hoggart)
Hogh, f.n. [hoʊ] (ho)
Hoghton, f.n. [ˈhɔtən] (hawton)

Boy, that was a lot more work than I expected — I think I’ve got the bracketed pronunciations right, but the respelled ones in parens are catch-as-catch-can: I haven’t tried to reproduce the breves and what have you. I trust you’ll get the idea.