This is a followup question of this question that I asked 3 and a half year ago. So based on what I could gather there, "descriptive" grammar comes after a language, hoping the rules are best describing the language itself.

In that sense, how could people say someone's grammar was "wrong" (or grammatically incorrect)? Because that someone was definitely speaking a language and whenever a language exists, grammar comes with it. Language "is" grammar. Saying someone's grammar was "wrong" is saying someone was not speaking any kind of language? Then what defines a language?

We are talking about most of the natural languages here, not those artificial languages where grammar rules were defined first.

  • There is such a thing as a wrong grammar, namely a grammar that does not accurately describe the facts of a language. Thousands of wrong grammars have been posited; thousands of fixed have been proposed that make the grammars less wrong. Generative grammar equivocates between the "mental object" and "scientific theory" uses of the term grammar.
    – user6726
    Aug 23, 2018 at 16:39

2 Answers 2


Linguists frequently describe utterances as "ungrammatical", and in the context of linguistics, this is a fairly meaningful, objective and well-defined statement (although what exactly it means depends on the linguist). I think most linguists would disagree with the statement that "Language 'is' grammar" if we define "language" as all utterances of any kind that anybody makes, for whatever reasons.

Linguists have various opinions about grammar, but typically, "descriptive" grammarians think that there are rules (in the sense of "scientific laws", not in the sense "things that you'll be punished for breaking") underlying language. They don't think of grammar as just being an after-the-fact summary of the different utterances that people have made, but as some kind of system, underlying speech, that can be studied by linguists.

The point of descriptivism is that descriptive linguists expect any grammatical rules that are proposed to be useful for explaining linguistic data. Other criteria may also be applied (e.g. restrictions on the kind of rules that are allowed to be in the grammar) as long as they are motivated by a theory of how language works, rather than by a theory of how language "should be".

Many theories of how language work distinguish a grammatical component from some other components that contribute to the phenomenon of language; e.g. the limitations of human memory.

Aside from there being limitations on language that aren't based in grammar, it's also the case that people can draw on resources other than the naturally acquired grammatical system of their native language to form utterances. For example, an utterance like "Dost thou thinketh that I beeth a fool?" isn't a grammatical modern English sentence, but the ungrammatical parts ("thinketh" etc.) may be intentionally used to produce a certain effect. This isn't a matter of idiolects, either: as far as I know, nobody has an idiolect where "Dost thou thinketh that I beeth a fool?" is actually grammatical.

Obviously words like "wrong" and "incorrect" look like value judgements, which is why I've mainly used the word "ungrammatical" in the preceding parts of this post. But in the context of descriptive linguistics, "grammatically incorrect" can be used as a synonym of "ungrammatical". Saying that an utterance is "grammatically incorrect" doesn't mean that it's "wrong" from a moral or rational standpoint. It just means that it's not consistent with—or not a valid product of—whatever grammatical system you are talking about.

Linguists and non-linguists don't mean the same thing by "grammatically incorrect"

When non-linguists say that someone's grammar is "wrong", they usually mean something different, and more complicated to define. (Since clothing is frequently used an an analogy to the social aspects of language, compare the concept of "wrong" fashion.) The meaning of "incorrect grammar" as used by a non-linguist isn't entirely meaningless, but it's typically a value-based judgement where part or all of the meaning being expressed is "I don't like this usage" or "I think this usage is socially inferior". If that's who you're talking about when you say "how could people say someone's grammar was "wrong" (or grammatically incorrect)?", then you're right to say that these people are making an essentially arbitrary and subjective judgement in regarding the rules of non-prestige dialects as "wrong".

  • Thanks. But then if any language carries its own grammar (something like idiolect), then by definition all grammar rules applicable to that language are right (or grammatically correct). Those that may deemed "wrong", are just those that are not applicable to that language. So there is no "wrong" grammar, only a different set of grammar is needed, no? Aug 23, 2018 at 3:32
  • @user1589188: The idea of "a language" is an abstraction. One way to think of it is as in Draconis's answer, some kind of set of rules that are "shared" between speakers. If you use a social definition like that, then not everything is grammatical in that language: only things that are part of the shared rule system are grammatical. It's irrelevant whether things might be grammatical in a different rule system. Aug 23, 2018 at 3:38
  • @user1589188: The other way to think about "grammar" is as a specific system that is a part of each individual speaker's capacity to speak a language (or to speak language in general, if you get more abstract). In that case, the speaker is physically capable of producing utterances that aren't a product of that system; we just wouldn't expect the speaker to judge "ungrammatical" utterances as acceptable unless the situation is complicated by some other factor. Aug 23, 2018 at 3:39
  • What you said would be correct if a grammar rule system is a fixed/dead system. A live natural language can evolve and rules be added or outdated. And since it is language that causes rules, language is the source, how can we say the source is wrong and then use a byproduct of the source to judge the source. If the source was wrong, then its byproduct couldn't be right (some form of fallacy)? Aug 23, 2018 at 4:26
  • @user1589188: It's a bit complicated to explain how grammatical systems evolve. Many popularizations stress the fact that evolution can occur as a result of children acquiring different grammatical systems from adults. From what I remember, speech does in fact change over the course of a person's adult lifetime as well, but I don't know to what extent this is considered to involve changes in the grammar that a person acquired as a child. Also, even if the changes that occur during a person's lifetime might be very noticeable from a social perspective, they'll be pretty minor relative to... Aug 23, 2018 at 4:47

Descriptivism certainly allows for a sentence to be ungrammatical!

The key to (natural) language is a shared underlying set of rules, that a group of speakers can use to communicate with each other. If someone deviates from these rules, we say that their utterance was ungrammatical.

For example, I maybe move can around words the in sentence my (maybe I can move the words around in my sentence) and say that. But any English-speaker would say the syntax there is entirely wrong. I've deviated from the mental rules that English-speakers share, and thus that sentence is not valid English.

The key difference between prescriptivism and descriptivism is that prescriptivism tries to define the rules according to some universal standard of correctness, whereas descriptivism tries to define them according to how the native speakers use them. Any English speaker can say that my example sentence was utterly wrong; even if I suddenly claim it's right, the consensus among millions and millions is that it's wrong. So a descriptivist would agree that it's wrong.

  • Thanks. I agree you can say a spoken language does not conform to English grammar (if there really is a standard English grammar set of rules), but not the language is ungrammatical. Unless we do not allow language spoken by some minority to exists at all, any language carries its own grammar yet to be defined by linguists (e.g. Yoda language). Aug 23, 2018 at 3:40
  • @user1589188 That just depends on your definition of "ungrammatical". When a linguist says something is ungrammatical, they mean that it's not correct according to the rules of the language currently being spoken. "Arma virumque cano Trojae qui primus ab oris…" might be perfectly grammatical Latin, but it's still ungrammatical in an English context.
    – Draconis
    Aug 23, 2018 at 3:42
  • Exactly my point. What rules or whose rules? My language has my own grammar rules and how does it make any sense when someone use a different set of rules to claim my grammar is wrong? Even within English rules, old or middle or modern or Internet English we are talking about? UK or US? West coast or East coast? What state? Which city? Age group? Social class? At the end we may just be using a wrong set of rules to apply on a different language, which obviously is "wrong" by itself. Aug 23, 2018 at 3:57
  • 1
    @user1589188 In the definition I'm using for this answer, language is defined by consensus. Even if I say that suddenly we need invert to infinitives, that doesn't make it a rule of English, because the worldwide consensus of English-speakers doesn't allow it. In particular, you'd consider it incorrect, because your mental model of English doesn't allow that inversion. Sumelic would too, or my brother, because that inversion goes against the mental rules we share. Language can change over time, but only as the consensus changes, which is a slow process.
    – Draconis
    Aug 23, 2018 at 4:32
  • "Descriptivism" is a straw man; there is no such thing. There are linguists and there are peevers and they really are not two sides in a debate; linguists have an analytic job to do, while peevers have their individual esthetic tastes to enforce. It's rather like comparing biologists' and fashion consultants' opinions on breeding flowers.
    – jlawler
    Aug 23, 2018 at 15:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.