Item — HN Reader

The same way in that a CS PHD is going to be able to pickup programming languages better and more quickly than someone who has learned C++ and python, they understand the underlying mechanics of language, what makes it work. But a toddler is something different, they primarily mimic and play with the language gauging function on result, how do people react to what I say. Learning languages from a young age is why the OP study shows no cognitive advantage and why people with only a single language tend not to have a good understanding of their language, it is a part of them.

But I was mostly speaking of peers, the sort who would be posting on HN, don't think we have many toddlers on the site. Learning multiple languages from a very early age seems to result in the languages being a single language from the perspective of the speaker, they have different situations in which they speak in different ways much like the average monolingual will speak more formally in a job interview compared to when they are out at the bar. Second and third, and etc languages are very different things between those that learned them from the start and those that learn them later in life.

4 comments

> The same way in that a CS PHD is going to be able to pickup programming languages better and more quickly than someone who has learned C++ and python, they understand the underlying mechanics of language, what makes it work.

... have you ever actually met anyone who got a PhD in CS? Because first off, very few people will actually study the "mechanics of language" in doing a CS PhD. If you're studying compilers or formal methods or maybe software engineering, sure, you'll study up on programming language semantics; but if you're learning AI or operating systems or computer architecture, you're very unlikely to touch those semantics. But even assuming you have someone who has actually taken those graduate-level courses (and I have!), it's not actually helpful for learning new languages. Learning semantics is essentially learning how to give meaning to the expression "a + b" (there are many ways to do this, and they have different tradeoffs), but if you want to learn a language, all you really want to learn is that "a + b" adds two numbers together, let's move on.

There's a similar gap between learning languages and being a linguist. As I understand it, most linguists are not polyglots (actually, I think a majority of them may be monolingual!), and most polyglots are not linguists. The linguist is the person who will tell you that English doesn't have a future tense: instead, you indicate the future by using specific helping verbs in the present tense. Anyone instructing English as a second language will slap the linguist on the head for being an idiot and tell you that the future tense in English is the "will X" construction.

> The same way in that a CS PHD is going to be able to pickup programming languages better and more quickly than someone who has learned C++ and python

Huh? This is about as convincing to me as the claim that learning linguistics makes it easier to learn a human language. It's also sort of a non-sequitor.

Someone who understands python and C++ would be trivially able to understand a massive swath of modern programming languages. I see little that having some deeper theoretical background would help.

And in human languages, even more so than programming, theory is almost irrelevant to regular practice. As the upstream comment says, we learn by practice, not by memorizing and applying rules.

>Someone who understands python and C++ would be trivially able to understand a massive swath of modern programming languages. I see little that having some deeper theoretical background would help.

Yes, they will be able to quickly pickup related languages with relative ease just as someone who knows a romance language will be able to pickup other romance languages without much difficulty. But that person who knows C++ and Pynthon with nothing of the underlying workings will have a more difficult time learning assembly, Forth or Lisp than a related language just as the person who knows a romance language will have a more difficult time learning Swahili than another romance language. Not a non-sequitor at all. When you look at CS or linguistics and how they relate to the languages they deal with what they really are are the study of what those languages have in common and understanding that will decrease what you need to learn when learning new languages.

Understanding a language and using a language are very different things and this is why toddlers can pickup languages with more ease than others, they do not care so much about understanding, and this is the primary limitation of our first language(s) and why being bilingual offers little cognitive advantage for most people.

I don't really know much about linguistics or linguists (though I would certainly believe that linguists find picking up new languages easier), but I don't think the analogy holds for a CS graduate.

PhD graduates don't have magic in-depth knowledge about programming languages. PhD graduates (in any field, really, not just CS), choose to dive very deeply into a very specific topic in their field. They do not just "go to more CS classes". They spend time becoming an expert on a very narrow topic. And for a CS PhD, that very likely would not be "general principles of programming language design".

I would not expect your average CS PhD to be any better at picking up new programming languages than a CS undergraduate.

Also, you asserted downthread:

> Understanding a language and using a language are very different things and this is why toddlers can pickup languages with more ease than others

No, toddlers pick up languages faster than adults because their level of brain development is more receptive to language acquisition. As people age, their brain's change to optimizes learning skills for other tasks, and language acquisition skills usually suffer.

>PhD graduates don't have magic in-depth knowledge about programming languages.

No magic involved, they generally have a good understanding of how a computer functions which means they can relate a language directly 1:1 to what the computer is doing instead of relating it to what another language they already know does or by simple trial and error.

>No, toddlers pick up languages faster than adults because their level of brain development is more receptive to language acquisition.

Toddlers are in the 12-36 month range, they primarily mimic and use language on simple cause and effect terms, if I say this than this is the result. They do not understand the concept of language or languages and if you teach them with multiple languages they do not see them as separate languages they see them as a sound and a result. We develop those distinctions and understanding of meaning during childhood and adolescence.

Natural language acquisition uses specialised circuitry in brain, and doesn't really run on general logic / intelligence, like programming languages do.

That's why analogies between the two are of limited use at best.

The rest of your comment isn't really wrong. I even agree with most of it. It's just not really relevant to your point.

> Learning multiple languages from a very early age seems to result in the languages being a single language from the perspective of the speaker, they have different situations in which they speak in different ways much like the average monolingual will speak more formally in a job interview compared to when they are out at the bar.

Last I checked the research was still open on that one. What you are saying might work as a gross simplification, though.

> Second and third, and etc languages are very different things between those that learned them from the start and those that learn them later in life.

Yes. And exactly when you learn them and in what context also seems to make some context.