Language Log

[This is a guest post by Scott Mauldin]

I recently visited Marrakesh and was fascinated by the signs that I submit in the attached photographs. Ostensibly these were originally a kind of business sign that artisans and professionals could hang on their businesses or homes to advertise their profession, but they have evolved into something slightly different for touristic consumption as they now sometimes feature the faces of celebrities or even items.

They're interesting in themselves as a cultural item, but if you look closely at the photos the truly fascinating bit are the "errors" and deviations from standard French spelling. These signs are often made by artisans without a formal education in French and sometimes are phonetic renderings that encode Maghrebi French pronunciations.

Definition

In artificial intelligence (AI), a hallucination or artificial hallucination (also occasionally called delusion) is a confident response by an AI that does not seem to be justified by its training data.

(source)

I had mentioned such AI hallucinating in a previous post once or twice (see "Selected readings"), so it's good to have a concrete example.

Is the account below an instance of ChatGPT "hallucinating"?  Its explanation of gato-por-liebre (cat-for-hare) in Spanish would seem so.

[The following is a guest post by Conal Boyce.]

From Taiwan News (3/25/23), by Keoni Everington:

"Taiwanese 'Hello Kitty' English-Chinese dictionary has 70 'egregious errors'

Publisher ACME Cultural Enterprise Co has admitted errors but not recalled dictionaries"

Cover of dictionary, example of misspelling. (Eryk Smith photo)

From the recent meeting between Putin and Wang Yi (Director of the Office of the Central Foreign Affairs Commission of the Chinese Communist Party):

Next time you hear or use the expression "scot-free", don't think that it has anything to do with Scots language or Scot people.  I have always avoided using this expression because I didn't want to disparage a whole people.  But "scot-free" is such a useful phrase that I wished I could use it with a good conscience.  So finally I looked it up and found that it has a completely different derivation from that of the name of the language and the people.

(colloquial) Without consequences or penalties, to go free without payment.

• • Synonym: beat the rap

    to get off scot-free

  • (archaic) Free of scot, free of tax.

From Middle English scotfre, from Old English scotfrēo (“scot-free; exempt from royal tax or imposts”), equivalent to scot (“payment; contribution; fine”) +‎ -free.

(Wiktionary)

(AH Dict. 2016 5th ed.)

At this most important moment of his career, when he is about to be crowned emperor for life of the CCP / PRC, Xi Dada commits a whole slew of bloopers and blunders, gaffes and goofs, and the camera has caught him in flagrante delicto:

【辱包素材】习近平二十大开幕报告口误合集
接踵(zhòng)而至
打铁(xuě)必须自身硬
解决(放)人类面临(yín)的共同问题
妇女儿童(téng)
着(zháo)力
供(gòng)给
塑(shuò)造
痼疾(jì)
步骤(zòu)
生动活泼(生吞活剥🐻)
道德水准(平😈)
斗争取得(斗争争🤨)
物质文明(墓志铭🪦) pic.twitter.com/ahPJ1hoLoK

— Dope Pig_习蜜归来 🇻🇳 🇰🇵 看反贼正在土崩瓦解，保皇派出头之日已经到来 (@Dope__Pig) October 18, 2022

Eric P. Smith writes:

Is there a name for a typographical error like the following?  If not, perhaps we should call it a “Firebug”.

Since 2021, Truss has served as the Secretary of State for Fireugb Cinnibweakth and Development affairs.

Liz Truss, who may well be the UK’s next prime minister, was Secretary of State not for some obscure Scottish Gaelic department with an indecipherable name, but for “Foreign Commonwealth and Development affairs”.  The typist’s right hand has strayed one quantum to the left, so that O has become I, M has become N, and so on.  The hands will have physically collided with the left index finger on the T of “Commonwealth” and the right index finger above the G next to the H, and the collision must have jogged the right hand back onto the straight and narrow, apparently without the typist even noticing.

From John Rohsenow:

In this age of typing on computers and other digital devices, when we daily input thousands upon thousands of words, we are often amazed at the number and types of mistakes we make.  Many of them are simple and straightforward, as when our fingers stumblingly hit the wrong keys by sheer accident.  People who type on phones warn their correspondents about the likelihood that their messages are prone to contain such errors because they include some such warning at the bottom: 

Please forgive spelling / grammatical errors; typed on glass // sent from my phone.

M. Paul Shore called my attention to a highly useful Latin expression that, in his opinion, is much needed in various scholarly communities, but that few people are aware of, much less use.

Paul writes:

For the last four-and-a-half decades of my life, from late teens to early sixties, I've had the nagging feeling that there ought to be a Latin scholarly expression that one could use when presenting the correction of an erroneous word or words in quoted material alongside the error itself. But in all my tens of thousands of pages of reading of scholarly works in the social sciences and humanities (which is not to be compared, of course, with the hundreds of thousands of pages, or more, that you must've read), I never ran across such an expression until last night, when I saw it in independent scholar Nigel Simeone’s meticulously annotated book of selected correspondence of Leonard Bernstein, published by Yale University Press. There it was, in black and white: recte! Meaning, of course, “correctly”, as in “Victor Mare [recte Mair]”, or “Edwin Pullyblank [recte Pulleyblank]”. It’s so exciting to discover this, after all these decades of desiring it, that I almost feel like applying to a graduate program at my somewhat advanced age, choosing a thesis or dissertation topic that requires the use of lots of defective sources, just so that I can splash “recte“ on as many pages of my work as possible.

[This is a guest post by Penglin Wang]

The Chinese transcription of foreign words has made a unique and valuable contribution to our understanding of linguistic situations in early Inner Asia, but it was sometimes inevitably fraught with logographic confusion and scribal errors. Even given quite advanced word-processing and printing in modern times, one can hardly prevent miswriting or misspelling from happening. In ancient China, presumably, it was historians and other authors who heard foreign words spoken and jotted them down, and then further changes developed through the involvement of scribes, typographers, and printers, with each possibly committing their own miswritings and infelicities. It is therefore necessary to reinterpret certain transcriptions on the basis of the known philological and linguistic relevance of what came to be written down.

I wrote to a colleague who helped me edit a paper that it had been accepted for publication.  She wrote back, "I’m glad it is excepted".

Some may look upon such a typo as "garden variety", but I believe that it tells us something profoundly significant about the primacy of sound over shape, an issue that we have often debated on Language Log, including how to regard typographical errors in general, but also how to read old Chinese texts (e.g., copyists' mistakes, deterioration of texts over centuries of editorial transmission, etc.).

Often, when you read a Chinese text and parts of it just don't make any sense, if you ignore the superficial semantic signification of the characters with which it is written, but focus more on the sound, suddenly the meaning of the text will become crystal clear.  In point of fact, much of the commentarial tradition throughout Chinese history consists of this kind of detective work — sorting out which morphemes were really intended by a given string of characters.

[This is a guest post by Bernhard Riedel]

I stumbled across what was probably a mis-MT in the context of the Olympic Games.  (article in Korean)

"During a foot kick on the way to the gold medal, some hangul became visible. But…"

On the black belt of the athlete from Spain, one can see "기차 하드, 꿈 큰" which is wonderful gibberish. Netizens in Korea were puzzled but also quick to guess an erroneous machine translation.

기차(汽車): (railway) train (definitely *not* related to "to train")
하드: (en:hard, transliterated)
꿈: dream (noun built from the verb 꾸다(to dream) with the nominalizer ㅁ/음)
큰: big (from the verb 크다) in the form used when modifying a noun that follows