Definition of language

[Aedl Foxe]

I know this is a simplistic question, but it's one that's been niggling at me lately. How is the term "language" to be defined?

A friend of mine parroted off the standard answer, "Language is a system that is used to communicate information", but I don't think that's what I'm looking for; it's a definition of function, not of form.

My working definition is "Any formal system that is isomorphic to cognitive structures", but I'm not sure if that's either broad or specific enough to be correct.

Thoughts?

[Aiwendil]

My working definition is "Any formal system that is isomorphic to cognitive structures", but I'm not sure if that's either broad or specific enough to be correct.

Well, for one thing that definition would appear to be contingent upon a particular resolution of the philosophical debate over the "language of thought"; i.e., there are people who claim that thought does not constitute a language, and if one holds this view, one is bound to find your definition untenable. More generally, one could worry about your use of the term "cognitive structure" being ill-defined.

Aside from such fundamental considerations, I'd suggest that your isomorphism criterion is too strong. I wouldn't expect a perfect one-to-one mapping to exist between any natural language and cognitive structures. Assuming that when you say isomorphism, what you have in mind is a bijection between sentences of a language and ideas (or what used to be called "representations") in a mind, then trivial counterexamples exist such as "Today I saw a square with five sides" or "Buddhism weighs nine grams", which are sentences but cannot be mapped to representations.

[Aedl Foxe]

Well, for one thing that definition would appear to be contingent upon a particular resolution of the philosophical debate over the "language of thought"; i.e., there are people who claim that thought does not constitute a language, and if one holds this view, one is bound to find your definition untenable. More generally, one could worry about your use of the term "cognitive structure" being ill-defined.

Aside from such fundamental considerations, I'd suggest that your isomorphism criterion is too strong. I wouldn't expect a perfect one-to-one mapping to exist between any natural language and cognitive structures. Assuming that when you say isomorphism, what you have in mind is a bijection between sentences of a language and ideas (or what used to be called "representations") in a mind, then trivial counterexamples exist such as "Today I saw a square with five sides" or "Buddhism weighs nine grams", which are sentences but cannot be mapped to representations.

Your counterexamples help to refine what I'm trying to say. Although the sentence "Today I saw a square with five sides" is not mappable to one representation, each element in the sentence has a legitimate cognitive referent. I would therefore argue that the sentence is mappable to "cognitive structures", just not mappable to any observable or imaginable (self-consistent?) reality. But thank you for posting those; they helped me realize what it was I was trying to get across.

As far as fundamental considerations go, that's one of my goals in exploring this question, is to better refine my understanding of the underlying problems. The argument that "thought does not constitute a language" (is this connected to the Sapir-Whorf hypothesis? An anti-corollary?) would depend itself upon a functional definition of language, which is what I'm seeking in the first place. :3 The trouble about "cognitive structures" is troublesome, too; I've been thinking of it in terms of the "classes" you find in object-oriented programming languages, but it bears further investigation.

Further thoughts? Thank you for your commentary.

[goofy]

features of human language

[Aedl Foxe]

features of human language

Thank you for that, but 1) it's not a concise definition of human language (though admittedly I'm not sure if one is possible, so thanks!) and 2) I'm trying not to focus on just human language. How about computer "languages"? Is there a way to draw a correlation between the two "languages" that's meaningful, that is, the bits of a computer are analogous to the neurons (or, in fact, these magical "cognitive structures") and the languages are analogous to each other?

[gmalivuk]

It doesn't solve the initial bootstrapping problem, but a rough-and-ready criterion might be "A language is a system which can communicate the same information as a given other human language". This, plus one uncontroversial example of a human language, could get you pretty far.

[Aedl Foxe]

"A language is a system which can communicate the same information as a given other human language"

I hate to be pedantic (I don't really, cuz that's kind of the point of this conversation), but what is supposed to be meant by "information"? And why is human language the benchmark?

[gmalivuk]

And why is human language the benchmark?

Do you have any non-human communications in mind that everyone uncontroversially considers to be languages? Because if not, then I don't think you'll be very successful in capturing what people actually mean when they talk about languages.

As for what information means, pick your favorite definition from the various technical ones, making sure that it matches what we already think of as the boundaries of "language", however vague those boundaries may be.

Like I said, it's not perfect, but it seems to work as a rough-and-ready rule. It's been used more than once in this very forum to argue that something is not, in fact, a language, because it is incapable of expressing a proposition that is quite simple to convey in, for example, English.

[TheGrammarBolshevik]

A language is a dialect with an army and a navy.

[Aedl Foxe]

A language is a dialect with an army and a navy.

Hehe. Ahh, Linguistics 110, good times. :3 But that refers only to human language.

What if I maintain that a computer language has all the capabilities that a human language does for communicating information?

[gmalivuk]

Then I guess we'd be forced to admit that TGB's post wasn't meant to be taken completely seriously.

[TheGrammarBolshevik]

Linguistics 110

…do you go to Harvard?

[gmalivuk]

No one else ever used 110 for an intro class before...

[TheGrammarBolshevik]

Heh, true. Course numbers here are sometimes so weird, though, that I forget that it's normal for intro classes to fall in the lower 100s.

[Aedl Foxe]

o_o I'm confused... I only ever went to BYU, and Intro to Linguistics was LING 110. Is that unorthodox?

[TheGrammarBolshevik]

Nope. I just hadn't considered that the name of Harvard's Linguistics 110 is probably not unique.

[gmalivuk]

Indeed.

[The Mighty Thesaurus]

How is a computer language not a human language?

Do you have any non-human communications in mind that everyone uncontroversially considers to be languages? Because if not, then I don't think you'll be very successful in capturing what people actually mean when they talk about languages.

But what is the point in defining it in terms of human language? If one doesn't recognise an alternative, the term is tautological. There's also the problem of using a word in its definition.

[Aedl Foxe]

But what is the point in defining it in terms of human language? If one doesn't recognise an alternative, the term is tautological. There's also the problem of using a word in its definition.

Thank you! That was what I was trying to ask, but I kept stumbling over myself. >>;

[gmalivuk]

But what is the point in defining it in terms of human language? If one doesn't recognise an alternative, the term is tautological. There's also the problem of using a word in its definition.

Thank you! That was what I was trying to ask, but I kept stumbling over myself. >>;

Do you people understand the meaning of the word "rough"? I admitted in my first post that it didn't solve the bootstrapping problem of getting at least one language from first principles.

However, I don't believe it's tautological just because there's a human language used as a benchmark. Plenty of non-human languages might be able to convey the same propositional information as a given human language. Incomplete, sure. But not tautological.

[Aedl Foxe]

Do you people understand the meaning of the word "rough"? I admitted in my first post that it didn't solve the bootstrapping problem of getting at least one language from first principles.

However, I don't believe it's tautological just because there's a human language used as a benchmark. Plenty of non-human languages might be able to convey the same propositional information as a given human language. Incomplete, sure. But not tautological.

I take your meaning of the word "rough". :3 Please don't think my dissatisfaction with your definition is meant to be an attack on you. But mine is the type of mind that isn't satisfied with rough. So I will continue to nitpick, by your leave, until I am satisfied. Yours is a good launching-off point.

No, I agree that it's not tautological. It's just unhelpful. It's as if we ask "what is a fruit", and someone points to an apple and says "That is a fruit, and all things like unto it". I, personally, am not satisfied with definitions of a class that draw upon examples of that class; it hits too close to Shaped Like Itself for me.