One thing I've never seen discussed on this topic (possible I just missed it, I only read popular accounts) is whether speaking multiple languages is a proxy for higher sociability / stronger social ties. That's a known dimension that improves health and aging and I wonder if just being able or interested in speaking with a broader swath of people is what helps more than the cognitive demands of switching.
Raised multilingual here (as in father spoke one, mother spoke another, living abroad and US back and forth). I don't know about the stronger social ties but I have found that thinking in a different language helps me get to sleep easier. There are times when I'm spinning around in webs in English (work, life etc) at night, and when I switch over to Spanish thoughts I fall asleep easier.
Maybe just stuff like that is enough to make a difference.
Switching into another language also helps if you are stuck in an environment where you do not want to pay attention (e.g. on a bus with blaring ads that you can't mute or with a rude neighbor yapping on their phone).
I grew up speaking multiple languages. My monolingual friend once asked in which language I dream, and I said - "not sure, I think in all three different ones I'm currently fluent". "But how do you know?", he kept pressing. "Ah, well, they come with subtitles", I joked. But since then I often wonder, how do we recognize a language in our dreams?
I think my dreams use semantic vectors that come with a tag saying they came from sensory input even though they didn't.
I am mostly monolingual but I would expect them to come with a language tag too. I believe a dream about Bob talking about cats in Spanish would directly activate your neural network like V_cats + V_spanish + V_bob_is_talking without activating the parts that would decode auditory input into that semantic vector.
In my world, most multilingual speakers are children of immigrant families, and that transcends socio-economic boundaries. Plenty of immigrants were already wealthy before coming to the SF Bay Area, where we import highly educated, specialized workers. On the other hand, we also import physical laborers who are also generally multilingual but not in the same social class.
Some of these immigrants are very well supported with a strong social network, while others struggle with isolation.
> One thing I've never seen discussed on this topic (possible I just missed it, I only read popular accounts) is whether speaking multiple languages is a proxy for higher sociability / stronger social ties.
Yes. This is exactly what you should be asking with this kind of churnalism. The research is hopelessly confounded by social status traits that correlate with wealth.
Sadly, this beneficial activity doesn’t look promising in the longterm. Real-time interpretation of foreign languages through earbuds is already available in its nascent phase, and China at least has begun cutting foreign-language programs because such AI translation is seen as the way of future. Once this tool becomes adopted enough societally, the learning of foreign languages is going to become a very niche hobby. It’s already becoming a niche hobby when many developed-country Gen Z are content with traveling and working abroad with only a knowledge of English alongside their native language.
I think any multilingual person will be skeptical of your analysis. Translation is not a substitute for understanding the original. It's good that we have translations, but it isn't the same. As you get into consuming art, literature, poetry, commentary, this is relevant.
I'll give you a very simple example I saw recently. I saw a dad joke in Spanish on social media and it gave me a chuckle. It was a comment to a lawyer's video and it translates to "can I study law if I'm a leftie?" It relies on the fact that law is also a word for "right".
AI won't be able to succinctly translate that joke and have it hit the same way. As an experiment I just fed that into ChatGPT, and it did explain the pun in 6 paragraphs with quotations and a bulleted list, but that kills the simplicity of the humor.
My point is that the AI will have that info understood as part of "being a translator," and react accordingly. The phenomenon you're describing is a basic fact of linguistics. Many things don't have direct translations, but they are still translated by fluent speakers of both languages.
1) But for many people and in many situations, translation---including MT---is good enough. And that overcomes any perceived need to learn another language (which learning may be in opposition to learning something else, something perceived as more important).
2) Most people don't "consume" literature or poetry, and it's unclear to me that non-performance art requires language. (I appreciate Rennaissance painting, but apart from the occasional Latin writing I don't need language to understand it.)
Disclaimer: Besides English, I'm reasonably fluent in Spanish (more so 35 years ago when I last lived in a Spanish speaking country), and used to be reasonably fluent in German, French, Tzeltal and Shuar (and a bit of Italian and written classical Greek). So I have every reason to hope this study is correct. But I have my doubts.
Translation is good enough, but quite often no translation exists. MTL is at this point good enough for simple texts and for everyday living like restaurant menus and shopping, but any even mild complexity in a text, like long-running context (necessary for consistent translation of terminology), puns, purposeful ambiguity, etc., MTL is completely unable to cope with. It just doesn't work and it will likely never work.
This has been said for ages and, as a person in a very multilingual environment, I still cannot see it happening, no matter how good the translations are. I may be wrong, but talking through translation earbuds seems dystopic and uncomfortable.
there's also no substitute for the reduced inferential distance when two people speak the same language. the literal meaning of words encodes just a subset of the communication
> Once this tool becomes adopted enough societally, the learning of foreign languages is going to become a very niche hobby.
That could a long time or it could be a very different implementation than the one you describe. Half of the world knowing 2 or more languages and that has been growing over time. I don’t see the evidence that technology will soon close the gap of speaking and understanding another language when in comes to communicating with family, music, business, participating in community events, volunteering, or even intimacy.
> China at least has begun cutting foreign-language programs because such AI translation is seen as the way of future. Once this tool becomes adopted enough societally, the learning of foreign languages is going to become a very niche hobby.
If this happens, professional or reliable translations will only be accessible to those who can afford/pay them, leaving everyone else stuck with the errors produced by LLMs.
To use machine translation, one have to know the language to review the output; otherwise, you're doomed to mistakes. Whatever you do with LLMs, the same thing will happen.
I would name all the marketing surrounding the A"I" as the LLM's blindness virus, or something similar.
Even if the translation is perfect and "real-time", it comes with a significant latency that will make any conversation less natural. Good for some situations while traveling, not something you would want to rely on for everyday life.
With AI there will always be people who learn more, faster using AI. And there will be people who use AI to avoid learning. When moving abroad, I hope folks will do the former (and the latter as tourists).
When I move back to Japan I'll wear something like Even Realities glasses, and when unknown vocabulary gets used, display that. Personally I think this will help me learn better than before. But let's wait and see!
I'm very hopeful about learning-acceleration tools. Just think what will be possible with Neuralink.
That seems like saying weightlifting is pointless because we have robots for lifting and moving things around. But lifting weights can improve health and lifespan. It may also be that thinking in another language is different from merely translating because of the idioms and cultural norms and references involved.
Mass market GPS navigation devices haven't made learning how to navigate a city like a local obsolete. If you've ever seen a gig economy driver get lost despite having step by step instructions...
> Dr Amoruso said: “In simple terms, people who spoke more languages tended to have brains that looked younger than expected for their chronological age. The effect was not only related to the number of languages spoken. Higher language proficiency and earlier acquisition of a second language were also associated with more delayed brain ageing. This suggests that multilingual experience matters as a gradient: it is not simply about being bilingual or not, but about the depth and duration of language experience.”
Some peer criticism from: Vanhove, J. (2026). Does multilingualism really protect against accelerated ageing? Critical comments on Amoruso et al. (2025). Journal of Multilingual and Multicultural Development, 1–10.
"As the authors correctly point out in their discussion, their observational study does not allow them to establish any causal links between the degree of multilingualism in a country and the extent to which its inhabitants show signs of accelerated ageing. Unfortunately, they do not seem to have kept this insight in mind when they came up with the title (‘Multilingualism protects …’) and the abstract (‘These results underscore the protective role of multilingualism …’) that grabbed the media's attention."
"The authors use country-level data on multilingualism, namely the estimated percentage of monolinguals in the country (the Mono variable in BAG_OR_cross.csv), the estimated percentage of people in the country who know exactly one additional language (One), those who know exactly two additional languages (Two), and those who know at least three additional languages (Three), always at the time of data collection."
"According to a simple OLS regression model at the country level with the monolingualism percentage as its sole predictor, a 10-percentage-point difference in monolingualism is associated with a difference in the average GAP value of about 0.36 years (some 133 days; 95% CI: (58,207) days). If log-per-capita GDP is controlled for, a 10-percentage-point difference in monolingualism is associated with a difference in the average GAP value of about 0.29 years (some 105 days; 95% CI: (22,188) days)."
It is also known that programming languages are treated by your brain similarly to human languages. But it is not known whether this means make your brain younger.
41 comments
[ 3.1 ms ] story [ 49.3 ms ] threadMaybe just stuff like that is enough to make a difference.
I grew up speaking multiple languages. My monolingual friend once asked in which language I dream, and I said - "not sure, I think in all three different ones I'm currently fluent". "But how do you know?", he kept pressing. "Ah, well, they come with subtitles", I joked. But since then I often wonder, how do we recognize a language in our dreams?
I am mostly monolingual but I would expect them to come with a language tag too. I believe a dream about Bob talking about cats in Spanish would directly activate your neural network like V_cats + V_spanish + V_bob_is_talking without activating the parts that would decode auditory input into that semantic vector.
Some of these immigrants are very well supported with a strong social network, while others struggle with isolation.
Yes. This is exactly what you should be asking with this kind of churnalism. The research is hopelessly confounded by social status traits that correlate with wealth.
AI won't be able to succinctly translate that joke and have it hit the same way. As an experiment I just fed that into ChatGPT, and it did explain the pun in 6 paragraphs with quotations and a bulleted list, but that kills the simplicity of the humor.
If I changed the meaning of the sentence, I could say, "can I study civil rights if I'm a lefty?"
If it were text, I might give you the translation I already gave yesterday and add a footnote explaining the pun.
Neither of these are ideal. One changes the meaning, the other is hard to understand.
What's the problem with an LLM-translatoin bot doing it? It would just do that, give a possible translation, and explain the issue.
That said, there are plenty of jokes in English. Stop me if you've heard this one...
2) Most people don't "consume" literature or poetry, and it's unclear to me that non-performance art requires language. (I appreciate Rennaissance painting, but apart from the occasional Latin writing I don't need language to understand it.)
Disclaimer: Besides English, I'm reasonably fluent in Spanish (more so 35 years ago when I last lived in a Spanish speaking country), and used to be reasonably fluent in German, French, Tzeltal and Shuar (and a bit of Italian and written classical Greek). So I have every reason to hope this study is correct. But I have my doubts.
That could a long time or it could be a very different implementation than the one you describe. Half of the world knowing 2 or more languages and that has been growing over time. I don’t see the evidence that technology will soon close the gap of speaking and understanding another language when in comes to communicating with family, music, business, participating in community events, volunteering, or even intimacy.
If this happens, professional or reliable translations will only be accessible to those who can afford/pay them, leaving everyone else stuck with the errors produced by LLMs.
To use machine translation, one have to know the language to review the output; otherwise, you're doomed to mistakes. Whatever you do with LLMs, the same thing will happen.
I would name all the marketing surrounding the A"I" as the LLM's blindness virus, or something similar.
(When half of words are wrong it is not understandable.)
When I move back to Japan I'll wear something like Even Realities glasses, and when unknown vocabulary gets used, display that. Personally I think this will help me learn better than before. But let's wait and see!
I'm very hopeful about learning-acceleration tools. Just think what will be possible with Neuralink.
https://www.fens.org/news-activities/news/speaking-another-l...
"As the authors correctly point out in their discussion, their observational study does not allow them to establish any causal links between the degree of multilingualism in a country and the extent to which its inhabitants show signs of accelerated ageing. Unfortunately, they do not seem to have kept this insight in mind when they came up with the title (‘Multilingualism protects …’) and the abstract (‘These results underscore the protective role of multilingualism …’) that grabbed the media's attention."
"The authors use country-level data on multilingualism, namely the estimated percentage of monolinguals in the country (the Mono variable in BAG_OR_cross.csv), the estimated percentage of people in the country who know exactly one additional language (One), those who know exactly two additional languages (Two), and those who know at least three additional languages (Three), always at the time of data collection."
"According to a simple OLS regression model at the country level with the monolingualism percentage as its sole predictor, a 10-percentage-point difference in monolingualism is associated with a difference in the average GAP value of about 0.36 years (some 133 days; 95% CI: (58,207) days). If log-per-capita GDP is controlled for, a 10-percentage-point difference in monolingualism is associated with a difference in the average GAP value of about 0.29 years (some 105 days; 95% CI: (22,188) days)."
It's 1,790.16 KB (1,833,122 bytes), which is ~800,000 bytes larger than the code space of the main microprocessor of the product I'm working on.