I don’t always reblog stuff from elsewhere, but when I do, it’s for a good reason. This article from Language Log (‘Reference to humans with this and that’ by Geoffrey K. Pullum) deserves to be reblogged because it neatly illustrates three interrelated points about grammar which, in my opinion, every self-respecting linguist should either agree with or be able to argue coherently against.
Point 1: Grammar is inaccessible to introspection. In other words, just because you are able to pass judgement on whether some sentence or other is or isn’t grammatically acceptable does not imply that you are able to explain why. As an English speaker, I may be able to tell that this is my friend John is an OK sentence in English but that *this wants to know where the police station is is not because it “feels wrong”. I know this because my brain makes that conclusion from what it knows about English grammar. Just because I am able to make that conclusion does not mean I know what rules or principles my brain followed to make it: they are encoded in such a way that they are not accessible to introspection. To be able to formulate those rules, I need to observe authentic language in use and deduce patterns from samples, like one would with any other natural phenomenon. It would be much easier if I could just “read” what’s “written” inside my brain but sadly, it doesn’t work like that.
Point 2: Non-native speakers can have native-speaker intuitions. I am not a native speaker of English and sometimes, when I come across an aberrant sentence, I am able to conclude it is grammatically incorrect because I can recall the facts and rules about English I once memorized at school. But that’s not what’s happening with *this wants to know where the police station is. This sentence appears to be breaking some rule that my brain knows about whether and when the pronoun this can be used on its own to refer to humans. My brain is telling me that this sentence does not feel right, but I’m pretty sure I never learnt such a rule in school: I’m certainly not explicitly aware of its existence or its content now. I must have acquired it through exposure and osmosis, just like native speakers do, even though I’m not one. Therefore, it seems that even adult non-native language learners can acquire (portions of) their non-native languages in the same way that native speakers do and have the same kind of intuitions (which are inaccessible to introspection, see Point 1).
Point 3: Grammar is not required to be elegant. When attempting to formulate the grammatical rules that govern a particular phenomenon, linguists normally search for “elegant” general patterns that explain a lot with very few rules. Sometimes, that search is in vain and you only end up with a bulleted list of special cases riddled with exceptions and exceptions to exceptions. That’s what happened to Pullum when he attempted to describe in detail whether, when and how pronouns like this and that can be used to refer to humans without an accompanying noun. It appears there is no elegant way to describe this phenomenon in English, and there is no shame in admitting this. While languages do sometimes display grand elegant patterns (which are teachable in school, see Point 2; think inflection classes, gender and number agreement), even these often degenerate into a maze of wanton complexity and irregularity when you attempt to explore them in detail. Human languages are always more complex and more irregular than they appear to need to be (except when they have been deliberately designed or regulated to overcome this). There probably isn’t, and certainly isn’t required to be, any grand scheme to explain the motivation behind the rules of language’s grammar, apart from a high-level human desire to communicate well.
The article that inspired these three points was mainly about English, but the conclusions are valid for every language. And because you’ve been so good and got as far as here, here’s a final bonus point about English in particular.
Bonus Point: English grammar is not simple. It is often stated that English has a simple grammar due to its extremely bare nominal and verbal inflection, its almost complete lack of case and gender, its more or less fixed word order and so on. Yes, these features do make English quite approachable for second-language learners and, incidentally, for computational applications. But English grammar has its fair share of complexity too, as illustrated in the example with this and that referring to humans. A slightly less specialized example is the use of definite and indefinite articles in English, which is notoriously hellishly complicated. If you attempt to describe these phenomena fully and explicitly in a grammar book or in a computational grammar, you will either fail or end up with a conundrum of “exceptions to exceptions” (see Point 3). Either way, an intuition acquired by osmosis will produce better results (see Point 2). So don’t let anyone tell you that English has a simple grammar: no language does. I have yet to see a naturally evolved human language where an apparent simplicity in one area of grammar is not compensated by an explosion of face-shredding complexity in another.