Or so Viorica Marian says in this Scientific American piece:
Psycholinguistics is a field at the intersection of psychology and linguistics, and one if its recent discoveries is that the languages we speak influence our eye movements. For example, English speakers who hear candle often look at a candy because the two words share their first syllable. Research with speakers of different languages revealed that bilingual speakers not only look at words that share sounds in one language but also at words that share sounds across their two languages. When Russian-English bilinguals hear the English word marker, they also look at a stamp, because the Russian word for stamp is marka.
Even more stunning, speakers of different languages differ in their patterns of eye movements when no language is used at all. In a simple visual search task in which people had to find a previously seen object among other objects, their eyes moved differently depending on what languages they knew. For example, when looking for a clock, English speakers also looked at a cloud. Spanish speakers, on the other hand, when looking for the same clock, looked at a present, because the Spanish names for clock and present—reloj and regalo—overlap at their onset.
The story doesn’t end there. Not only do the words we hear activate other, similar-sounding words—and not only do we look at objects whose names share sounds or letters even when no language is heard—but the translations of those names in other languages become activated as well in speakers of more than one language. For example, when Spanish-English bilinguals hear the word duck in English, they also look at a shovel, because the translations of duck and shovel—pato and pala, respectively—overlap in Spanish.
She goes on to describe similar findings for American Sign Language and finishes with suggested implications (“Not only is the language system thoroughly interactive with a high degree of co-activation across words and concepts, but it also impacts our processing in other domains such as vision, attention and cognitive control”). It’s all very cute, but I find it hard to believe; I can easily conceive that researchers get the results they’re looking for in such experiments. On the other hand, I am a known curmudgeon, and far too lazy to actually click through to the studies and evaluate them for myself, so I’m putting it out there for others to chew over. (Thanks, Trevor!)
Why are you finding it hard to believe? From my personal experience (as well known, the most scientific research tool) it is absolutely clear that words from one language actuate similar-sounding words in another language. A cute thing would be to see what happens with people who learned different languages in the early childhood. The usual story is that languages learned early do not interact or at least not as much as with languages learned later.
Another nice thing would be to see whether switching from single words to sentences changes the effect. Does the linguistic environment narrow the focus?
I think it’s easier for me to imagine the mechanism they’re talking about because it seems similar to a computer’s predictive text suggestions.
I wonder how much every era’s current technology controls our metaphors of how the human mind works. Like the (assumedly prehistoric) “sparking” ideas, or the “blank slate”, or later Edisonian light bulbs switching on…. now I guess our brains have algorithms running in the background that are constantly spitting up questionably relevant search results?
Why are you finding it hard to believe?
Because of this bit of unreconstructed Whorfianism: it is safe to say that the language you speak influences how you see the world not only figuratively … The “not only” there presuming everyone would agree.
Because the article is clearly marked ‘Opinion’/Blog, not science. Because ‘Scientific American’ has long since ceased to be a serious journal.
Because someone hearing “cand…” and glancing at candy is not evidence of a high degree of co-activation across words and concepts,. It could easily be explained as a strategy to disambiguate phonetically unclear utterances by reference to context of the dialogue — the visual context in that case. And observing that attention-focussing even after hearing the “…le” might be explained merely as delayed signal-processing from phonetic to phonological to focus.
What if the conversation is over the phone? Do listeners still focus on candy in their own visual field, even though the speaker couldn’t possibly be referring to that?
Because this (from the abstract to one of Marian’s own papers) is not saying much at all: We conclude that the varying linguistic information available to speakers of different languages affects visual perception, leading to differences in how the visual world is processed. “affects”, “differences” does not substantiate “across words and concepts”.
Because I too can easily conceive that researchers get the results they’re looking for in such experiments.. Redo the experiments asking the participants to look for ‘tallow’ or ‘timepieces’. There must be gazillions of Spanish words starting “re…”; what if there are many such objects in the visual field at the same time?
I honestly don’t understand what you are objecting to. The only thing that piece says is that linguistic priming works across languages. It might be true or not true (or true in some cases) or the effect size can be small or something else. But one way or the other, I cannot understand what is so exciting about all that.
I think it would be rather surprising if there were not some sort of effect along these lines. By analogy with the much-ballyhooed business about colour words and colour perception, I would also anticipate that real-world effects might be small.
As AntC says, you’d need to take a lot of care with your experiment design if you wanted to demonstrate anything rigorously, and even then you might not have proven quite what you thought you had proven.
It reminds me of the cod-psychology (or real psychology, in its place, I guess) of word-association tests: there, the opposite problem presumably arises of how you could ever be sure that your respondent is free-associating on the basis of the concept expressed by your probe word, rather than the huge aura of purely linguistic associations every content word evokes in any natural language.
This may vitiate the Voight-Kampff test in certain circumstances. I’ll point that out to the Tyrell Corporation if I can figure out how to do so safely.
Why are you finding it hard to believe?
What AntC said. I’m not saying I find it hard to conceive that such an effect could exist, just that proving it does exist would require a rigor in experimentation and a constant awareness of the problem of finding what one went looking for that I seriously doubt the studies cited exhibited. Industrious researchers have “proved” everything from phrenology to telepathy with “scientific” experiments.
@ AntC
Because ‘Scientific American’ has long since ceased to be a serious journal.
Scientific American has never been a “serious journal”, if by “serious journal” you mean a scientific journal publishing original research. It’s always been a popular science publication. Well, OK, it’s apparently true that the emphasis shifted from its original “engineering and inventions and patents” focus to “science in general” sometime during the early/mid-20th Century. And it’s no longer warning us about things like the dangers of chess, as it did back in 1859. But I don’t think the latter is necessarily a characteristic of a “serious journal”.
I suspect AntC means that although SciAm has, of course, always been a popular science publication, you used to be able to take what it said pretty seriously.
Gerard Piel, thou shouldst be living at this hour.
Amen (or ameen or even omain, if you prefer). I used to say I got all the news worth learning about from S.A., and have even been known to describe myself as a scientific American. But nowadays, fuggeddaboudit.
For example, when looking for a clock, English speakers also looked at a cloud.
Under what conditions?
I subscribed mainly for Martin Gardner’s Mathematical Games column, and then Metamagical Themas. But that was a log time ago.
Under what conditions?
Cricket.
https://lords-stg.azureedge.net/mediafiles/lords/media/images/father-time-storm.jpg
(The clock is just out of the shot.)
Liberals may sneer, but we ignore the moral dangers of chess (by common consent a game of foreign, indeed oriental origin) at our peril. For example, my wife is a much stronger chess player than I am. I need hardly spell out the pernicious effect such things are liable to have on Christian marriage, the bedrock of our entire society.
Play shove-ha’penny! Vote Conservative for a chess-free future!
Thank you. Yes what everybody said I said.
As a symptom of what Scientific American has become these days: Marian’s own papers make modest claims somewhere between the unremarkable (as David E points out) and the unfalsifiable. Or at least the unfalsifiable by this methodology: sample size for example “Twenty monolingual English speakers and twenty Spanish-English bilinguals were recruited from Northwestern University ” for the clock/cloud/Spanish “re…” study. Judging by their mean age, they were students; could it be they were studying Psychology/Linguistics/Psycholinguistics? Would _any_ of their linguistic behaviour count as any sort of evidence?
But of course unremarkable claims don’t sell newspapers, whereas Whorfianism is a sure-fire money-earner in a popular science publication. How much of the piece is authentic Marian/how much got ‘sexed up’ by a journalist?
I’ve no objection to what Scientific American used to be/you could “take what it said pretty seriously”; and I used to read Martin Gardner myself (still got some Hexaflexagons somewhere). Nowadays it’s at best a gateway to the original research.
It’s often sarcastically remarked that the subject pool in most modern cognitive science experiments (at least in America) is drawn overwhelmingly from college students. Depending on the institution, they may be paid (typically in the neighborhood of $10–20* these days) for their time, or they may be required to participate, as a course requirements for psychology classes they are themselves enrolled in. In any case, these students may not represent typical members of the population.
* I once tried to get into a study that paid much better (hundreds of dollars, because it involved multiple MRI scans). I participated in a screening interview (for which I was paid $40), but they said I did not qualify for the main study, because I was nor paranoid enough.
It’s not just cognitive science: the whole of the behavioral sciences, including experimental human psychology and experimental economics, heavily depend on WEIRDos (members of Western Educated Industrialized Rich and Democratic societies) as subjects, and mostly students at that — who are on the extreme end of a great many social scales. The article is very much worth reading.
@John Cowan: Cognitive science was a neologism developed in the 1970s to distinguish researchers and departments that were interested in the scientific study of the mind, without any direct interest in clinical treatment of individuals with psychological issues. The term came into use because the field denoted by the name psychology was becoming dominated by therapeutic practitioners, and some scientists in the United States wanted to distinguish their field from that of clinical psychologists.
In principle, cognitive science could be construed include essentially anything to do with the scientific study of animal thought processes. In practice, it may be useful to distinguish it from experimental economics, for example. However, mention of “experimental branches of psychology [and] cognitive science” is pleonastic. (It reminds me of the phrase “alcohol and substance abuse”—which is used in spite of the fact that substance abuse was coined specifically to encompass alcohol, illegal drugs, and prescription drugs.)
Looking past this peeve of mine, the real content of that paper (which I was already familiar with, prior to this) is carefully thought out and well expressed. The authors, coming from a “psychology” department outside the United States, may just not have appreciated the meaning for which cognitive science was coined.
Well, historical origin is not always current use. Parts of linguistics, AI, neurology, and even anthropology fit into the cog sci umbrella, or bubble, or whatever.
they said I did not qualify for the main study, because I was nor paranoid enough
You have to wonder what their real reason was.
As I once actually said to Mariette DiChristina, Michael Shermer was to Martin Gardner as Starship was to Jefferson Airplane. And now DiChristina and Shermer are both gone, and the magazine even more resembles the old National Lampoon parody Scienterrific American.
It’s sad. I was given my subscription in 1959, when I was eleven. The above came out of an email conversation after I wrote her on the fiftieth anniversary of that. (“Its finally happened, I saw the cover of my first issue on the 50 Years Ago page.”) I was mentioned in their podcast. On consideration, I might have quietly abandoned the magazine after that.
The bilingualism is a total red herring here, the same effect would exist for English-only speakers if you structured the search space to contain words that are similar or begin similarly (as she clearly did for Russian/English and Spanish/English). And honestly, this just doesn’t jibe with my own experience of fluency in multiple languages. And not just fluency – I’m barely an intermediate speaker of Portuguese after 2 years but it’s not like você invokes EN “voice” or BCS “voće” as unrelated but similar sounding words.
Finally, the decision to design the experiment as a visual search for the named object seems particularly misguided because it doesn’t rely on language at all, but cognition; what your eyes are searching for is a visual match to some abstract notion of the object called up by your brain not a word matching exercise in which the influence of a different langauge might lead you astray.
Yes, a good point.
Silly me. I’ve thought of a plausible everyday scenario.
The researchers take a blank wall on which they have hung a clock and painted a large cloud. The researcher says:
“Look at the cl-”
Instinctively my eyes go to the picture of the cloud, which seems to be a strange thing to have painted on the wall. Then I hear the end:
“-ock”.
In a split second, my eyes switch from the cloud to the clock. But it’s too late. I’ve been discovered looking at the cloud.
For the female participants they also have a large picture of a cock. To the eternal embarrassment of the more staid members of the group, they are all caught taking furtive glances at the cock. Fortunately, the researchers can put this down to ‘clock’ and ‘cock’ starting with the same consonant, and rhyming.
The bilingualism is a total red herring here, the same effect would exist for English-only speakers
There were a lot of mistakes, sloppiness, and outright fraud perpetrated in psychology research and other disciplines. As well as less obvious stuff like drawer effect. But I am amazed that well-meaning people in general think that these researches and their community are unreformed idiots. Of course, there is linguistic priming within a language. The claim is that it doesn’t disappear across languages.
Thanks for the WEIRD paper, JC; I’ve seen it before, but it certainly bears rereading.
It occurred to me while reading it that a currently rather prevalent school of linguistics could without great unfairness be characterised as: intensively studying WEIRD languages in order to generalise about Human Language.
See also: neurology, the application of neurobiology – not just in the US.
Bathrobe – but cock doesn’t start with “cl”. A really well-designed and incontrovertible experiment would show you a picture of a clock, a cloud, and a clitoris and see which one you glance at. That’s science! (or a new discipline, cognitive cunnilinguistics?)
The whole point of these studies would be to test how much language does influence visual search tasks which you might otherwise expect to rely only on “some abstract notion of the object called up by your brain”. The experiments might not test that as well as they intended, the sample may be different from the wider population, or the results may appear blown up to more than they are by articles starting with the claim that languages “influence our eye movements” rather than that influence on visual search tasks is seen in eye movements, but I don’t see why testing for such as influence could possibly be misguided in itself.
No, of course it’s not, it’s a perfectly legitimate thing to study — it’s just that it’s so easy for such studies to be badly done and so probable that they will (given the attractions of both confirming one’s own biases and producing exciting results that will get media attention) that it’s hard to have high expectations of them.