Representative image: Isaac Chou/Unsplash.
Every now and again, when historical linguist Sarah “Sally” Thomason was teaching at the University of Pittsburgh, she would receive weird calls. From people who spoke what they believed were unknown languages in their sleep and wanted her to identify them. Or others who claimed they’d discovered that the world’s original language, the one from which all other languages were descended, was Turkish, say, or Japanese.
And then there was a local hypnotist who called wanting to verify that a subject talking under hypnosis was speaking a language from a past life.
“His people were just producing gibberish. Interesting gibberish, but still gibberish,” says Thomason, now at the University of Michigan.
Thomason’s scholarship usually focuses on what happens when languages meet, as well as the Native American language Montana Salish. But over the years she has encountered a surprising variety of fringe notions about language, which she and Canadian William Poser, a linguistic consultant on aboriginal languages in northern British Columbia, dubbed “ fantastic linguistics” in the 2020 issue of the Annual Review of Linguistics.
In a discussion with Knowable Magazine, Thomason dissects fantastic-linguistics claims and talks about what one can learn from this pseudoscience. This conversation has been edited for length and clarity.
A common theme in fantastic linguistics is the search for the world’s original language. You note there was an unusual candidate for this oldest language – Dutch.
That claim’s really interesting. The 16th century Dutch linguist-physician Johannes Goropius Becanus argued that the oldest language must be the simplest one, the simplest language was the one with the shortest words, and Dutch words are shorter than Latin, Greek or Hebrew words, so Dutch was therefore the earliest language.
From one perspective, it’s a pretty good idea. It does seem fairly likely that the world’s oldest language was relatively simple, with not so many words when it first got started, and not much in the way of variation in syntax, and that it developed and got more complex over time.
Where Becanus went wrong is that he didn’t reckon with language change. There’s no reason to believe that modern languages would retain the hypothesised simplicity of the original human language.
Also, just because Dutch has shorter words than other languages doesn’t mean that it’s simpler overall. It doesn’t, for example, mean that the sentence structure of Dutch is simpler than that of other languages.
What other candidates are there for the world’s original language?
A popular claim is that Hebrew has to be the oldest language, because the oldest parts of the Bible are mostly written in Hebrew. That is possibly the mistake that laymen make the most about language – that language is writing.
If you talk to people in mainstream America, written language is the main part of language. All this stuff about how language is going to pot, how people are messing up language, is about how people aren’t talking the way language is written. But nobody talks the way they write.
It’s a really widespread misconception that language is mainly written language. It goes with the idea that written language is better than spoken language, that it’s pure and grammatical, and if you don’t write in fully grammatical prose, it proves that you’re stupid.
It’s hard to explain to people why this idea is wrong, because literate people can’t remember when they couldn’t read, and reading is so much of what they do in adult life.
Your Fantastic Linguistics review is full of extraordinary claims, such as those made in 1983 by the Harvard marine biologist Howard “Barry” Fell, who argued that rock carvings found on Native American sites were made by Irish visitors, centuries before the arrival of Columbus.
Yes, the petroglyphs on caves in West Virginia. He thought they were Ogam, a form of Old Irish writing. Imagine a monument stone with vertical edges – Ogam was written in the form of lines etched horizontally along one side or the other of the vertical edge, with some lines going across the vertical edge.
And the West Virginian petroglyphs also consisted of straight lines etched on cave walls. But straight lines are common in Native American rock carvings, right? These West Virginian petroglyphs aren’t necessarily Ogam?
For these carved lines to resemble Ogam, Fell had to put in lines that are not there on the cave wall – to add a central line that the Ogam “letters” could fall across, or be on one side or the other. He also had to introduce gaps between groups of lines to make them look like separate letters. And he added lines where needed and made up other linguistic stuff, too. He didn’t really know the language.
It could be that Fell was a charlatan who knew perfectly well he was lying, but it’s possible that he was engaging in self-deception. I’d guess the latter. Self-deception is a powerful thing. He wanted to find evidence of the Old World in the New World without training in language. Wishful thinking.
How did Fell account for the apparent errors in Ogam in the carvings?
He said they were made by boy druids who were still just learning, who just got some things wrong. It’s really great.
Many fantastic linguistic claims involve the paranormal. For example, in xenoglossy, people apparently speak languages they had no way of learning, such as the cases described by University of Virginia psychiatrist Ian Stevenson, who believed in and studied reincarnation. When his patients were hypnotised and made to “age-regress” – to remember memories from earlier in their lives – they would supposedly remember details from past lives.
And they would supposedly speak languages such as Swedish or German.
And to investigate these hypnotised people, Stevenson brought in speakers of the languages supposedly spoken during xenoglossy?
In the case of the hypnotised, so-called Swedish speaker, Stevenson would get a Swedish interviewer to ask a question in Swedish. If the hypnotised guy didn’t understand, he was asked the question again in English. So he heard Swedish with an immediate translation into English, and so had a chance to learn words in Swedish.
And if you look at the transcripts, the hypnotised person knew only a handful of words of Swedish. The replies given to the interviewer were only one- or two-word answers, and often nonsensical – for example, when a person was asked what he would pay for some item at the market, he said, “My wife.”
And the interviewer was a believer in past lives. When they got an answer that did not sound quite Swedish, they might say it sounded Norwegian, or a Swedish dialect they didn’t know. The believer made it fit.
You also analysed transcripts from the interviews with a so-called German speaker.
I know German. In the German case, who was a woman, the woman’s husband was very interested in reincarnation, and really wanted this xenoglossy to happen.
She was asked many yes-no questions, and the thing in both English and German is that with yes-no questions, the intonation rises at the end, whereas with content questions such as, “What did you eat?” you have a falling intonation. So with the yes-no questions, she only needed to know the words for “yes” and “no” in order to answer. And they questioned her about her life in Germany around the time of Martin Luther, but nobody there knew if the answers were right or wrong – they had no way of checking the answers for accuracy.
She did know some German words. Deploying what little German she knew, at one point they asked, “What is there after sleeping?” And she clearly heard the word “Schlafen,” which means “sleep,” and she said, “Bettzimmer,” or “bed-room.” She knew the German words for “bed” and “room,” and she responded with a calque, a word formed by a literal word-for-word translation of a word from another language, from the English word “bedroom.” But the German word for bedroom is “Schlafzimmer,” or “sleep-room,” not “bed-room.” So she knew some German words, but she didn’t know German.
Stevenson does make some effort to account for this badness in language. He didn’t do it the way I would, by taking into account the changes that happen in language after several hundred years, or the difficulty of discussions through the ether. He reasoned that she was clearly an uneducated servant girl who did not have much language. That’s naïve – a servant girl would have plenty of language, just not the same as a professor’s.
I never met Stevenson, but I do believe he was genuine, not cheating. In each of his books, he puts a lot of effort into detecting fraud. I think that shows his good faith. But he didn’t analyse the speech enough – he wasn’t doing what he thought he was doing and, not knowing about language, made his mistakes.
Why do you think linguistics attracts these pseudoscientific ideas?
Because everyone speaks a language, right? So people think they know all about language. The notion that there may be some things about language that might not be accessible to one’s contemplations is not something that occurs to most people. So language becomes an easy target for these kinds of speculations.
And it’s exciting for people to see and make connections across continents and millennia. For instance, if you have a word that’s identical across two languages, they must be related, right? It just so happens that the word for “bad” in Persian is also “bad.” But that’s totally irrelevant – those words appear fairly similar, but those English and Persian words are not related, they’re not etymologically connected. If you take any bilingual dictionary, you’ll find words that look and sound similar and kind of mean the same thing. It’s coincidence.
What about words that might be similar across languages because of onomatopoeia — because they imitate the sound they describe?
Surprisingly, onomatopoetic words very often differ significantly from language to language. Admittedly, rooster-crow words with K’s in them are pretty similar in many languages – English cock-a-doodle-do, French cocorico, German kikeriki, Gujarati kuk-de-kuk, for instance – but they aren’t identical words. “English-speaking” dogs say bow-wow or woof-woof; Chinese dogs say wang-wang; Estonian dogs say auh-auh; Greek dogs say ghav-ghav; Burmese dogs say woke-woke; Balinese dogs say kong-kong. Some similarities, but certainly not identical.
A more striking set of examples would be nursery words for “mama” and “papa” or “dada.” The ones with bilabial consonants – sounds formed by the closure or near-closure of the lips – at the beginning, like “mama” and “papa,” do recur in many languages, but sometimes “mama” means “daddy,” not “mommy.” One theory that’s pretty popular among linguists is that infants’ babbling produces such sounds, and meaning gets attached to repeated syllables like “mama” later, via positive reinforcement by parents eager to believe that their baby is naming them.
But all these sound-symbolic words, even when they’re similar in several or many languages, comprise a very small part of any language’s vocabulary. When I gave the English and Persian “bad” example, I was talking about truly accidental similarities.
So these pseudoscientific beliefs stem from a lack of scientific rigour?
Stevenson did try and address xenoglossy scientifically – he was a psychiatrist; he really did try. Where he and most others went wrong is that he didn’t know there was stuff he needed to know about language and was ignorant of.
And once you make that error, it’s easy to make others.
Right.
Why, as a linguist, do you think we should learn about fantastic linguistics?
You mean, why would anyone else pay attention to fantastic linguistics other than for amusement? Because it’s educational.
How might fantastic linguistics be educational for linguists?
Linguistics, like most sciences, is divided into a number of quite distinct subfields, and specialists in one subfield are more or less guaranteed to be mostly ignorant about some other subfields. For instance, several of the pseudoscientific claims that Poser and I reported on in our article focused on language history, especially language origins and language relationships. These topics are studied in the subfield of historical linguistics. I’d guess that a rather small minority of currently active professional linguists have taken even a single course in historical linguistics during their student years; job ads targeting historical linguists are extremely rare.
So, amusing linguists who haven’t studied historical linguistics with wacky claims while explaining the methodological pitfalls of the claims can help them learn about historical linguistics. For instance, one could compare claims about what the world’s original language is to the way historical linguists demonstrate that two or more languages are related.
And how might fantastic linguistics be educational for non-linguists?
I can easily imagine a class on linguistic pseudoscience. You have these really far-out claims, and here’s what people say is the evidence for those claims. Can you test this evidence? Is there anything good in there, solid in there? You can teach science by looking at pseudoscience. How is an example not scientific? What makes this pseudoscience?
How do you prove these claims are pseudoscience to their believers?
You might not be able to. People are really good at not believing in things they don’t want to, like climate change. But at least you can prove it to your own satisfaction.
Charles Q. Choi is a science journalist who has written for the New York Times, the Washington Post, Science, Nature and Scientific American, among other publications. In his spare time, he has traveled to every continent. @cqchoi, cqchoi.com.
This article originally appeared in Knowable Magazine, an independent journalistic endeavour from Annual Reviews.