The practice of linguists has been very mixed especially since 1951, tending toward heavy reliance on axioms vs. minimal reliance on axioms (historically, trending in that direction). I disagree with the position that "that has to depend on what you mean by grammatical gender", as admitting a fundamental methodological error, namely the arbitrariness of concepts and their names. This error is related to the error of improper axiomatization. An axiom is not an arbitrary stipulation that defines what words and relations you allow in your particular theory, it is a self-evident truth. Allowing the meaning of terms to be arbitrarily stipulated on an individual basis means that rational discussion and knowledge itself is impossible, instead we just have vague feelings that so-and-so might be right. Because we can never talk about reality without invoking some words – "gender", "noun", "person", "subject", "agree", "dog", "eat" – and I have no reason to believe that when you say "person", you're referring to what I'm referring to when I say "dog".

The logical error that I see here (in these pages), is that the putative ontological status of grammatical gender is utterly in flux. Does it refer to a causal principle of linguistic theory, in the same way that Merge is claimed to be a causal principle of linguistic theory (specifically, about the syntactic component). Or does it refer to an abstract class of phenomena? The same question comes up with "agree(ment)" – is this a fundamental operation in the theory, or is this a carving-out of observed data patterns? How does "agree" have any relationship to "grammatical gender"? Is there even a difference between linguistics as data-fitting and linguistics as real science seeking causal principles? I suppose you can tell where I stand on the matter.

The discussion is about the data-fitting approach. How do we validate terminological choices when we are only talking about the nature of communicative behavior, and not about the nature of the underlying grammatical computations? Let's take German and French as examples of data patterns. The gender-related facts of French are not the same as those of German – does that mean we have to construct separate terms for German grammatical gender versus French grammatical gender? How can we talk about the unity of these systems, or should we even try?

Immediately above, I may have appeared to advocated a Kantian disintegration of the concept of "gender" into myriad special cases – that is exactly the opposite of what I am advocating. Science operates by integrating observations, and not by an infinite regress of hair-splitting. However, in the context of a discussion of communicative behavior as opposed to grammatical theory, there is an important distinction between gender and grammatical gender. Social theorists have all sorts of classifications, but in ordinary usage the term refers to male versus female. Grammatical gender, a terms of linguistic theory, has nothing to do with genitalia and everything to do with agreement governed by a property of a noun. That is how it is used by the overwhelming majority of linguists. An individual can declare that they hereby define "grammatical gender" to be any means of distinguishing males from females (having words for "man" and "woman" for example), or animate from inanimate (any language with a distinction between "dead" and "alive"). But this is not how linguists talk about grammatical gender.

Again, "grammatical gender" is a data-classification term employed in aid of constructing grammatical theory, it is not the same as "gender" as used by social theorists. It's not a question of "how you define grammatical gender", it is a question of what the term actually means in the context where it is used. It refers to the partitioning of nouns into two or more subsets, according to some criterion (including lexical marking), where other elements in the sentence "agree" with that noun, analogous to agreement in person, number and case. Lushootseed has a marker -s- that can be added to determiners to indicate "female", and it has absolutely no grammatical gender or agreement. Niger-Congo languages often have complex noun classifications and agreement patterns which are partially semantic and partially lexical – but have zero correlation with biological gender. It is impossible to distinguish "he fell" and "she fell", at best you can say "(singular human) fell".

What we have here is a case of equivocation. The problem is the very word "gender". In linguistics in the phrase "grammatical gender" it means little more than "class", but because outside linguistics it has connotations of "male/female" and because grammatical genders have names like "masculine" and "feminine" notions of natural gender interfere. The fact that (at least in the languages I know) most things male may have masculine gender and most things female may have feminine gender does not help, but should not be allowed to get in the way. If instead of "grammatical gender" we referred to "agreement requiring noun classes" and feminine gender was called "agreement requiring noun class 1" and masculine "agreement requiring noun class 2" we would not be having this discussion.

Grammatical agreement is no more than saying that the form of a word depends on other words related to it. A language may require agreement for any one or more grammatical category such as number, case, person or gender or indeed none. If English is studied you will not find anything indicating the possibility of grammatical gender except in third person personal pronouns. Since that is the only instance you can reasonably ask: "Does this justify me asserting that English has grammatical gender?" The answer is "no" because there is another explanation. Which pronoun you use is determined entirely by the nature of the thing referred to. In "My wife is out - she will be back soon" using "she" depends on knowing that "wife" refers to a female and not on it belonging to any noun class other than nouns which refer to things female or some class such as "beginning with the letter w".

If you want to know whether a given language has grammatical agreement, and if it has if it has grammatical gender, all you have to do is study it. You do not need axioms or a detailed methodology.

I started off by saying "that has to depend on what you mean by grammatical gender" and offered two possible definitions. I could have taken a different tack which I do now:

Rock100

You take the view that English has grammatical gender. It is my view, based on wide reading over many years, that there is a consensus that English does not have grammatical gender. What is you understanding of what grammatical gender involves which leads you to conclude that English has grammatical gender?

> Grammatical agreement is no more than saying that the form of a word
> depends on other words related to it.
If it were like this, it would be enough to claim that English has a grammatical gender. There is the requirement to agree some nouns with different pronounces. And this requirement is conditioned by something other than number, case, etc. You see, you must justify such agreement with something different from the existing features of the language. And if you name such an additional feature I will rename it to the grammatical gender and vu a la. Technically speaking you must either justify the feature with existing features (or axioms if you will) or add a new one. The adding a new one is the disaster because it leads us to the grammatical gender. Though you may argue about it does not look like grammatical gender in other languages I will call the grammatical gender because it looks enough close for me. You have no choice but to prove that your new feature has an equivalent in another language and it is not the grammatical gender in that language.
So, there must be a helper for your position somewhere else. Most likely this would be sociology. It looks even less science than linguistics, so it may help. And it really does and I do feel there might be something in it – you may introduce different kinds of gender, different classifications (humans, other animates, robots, artificial intelligences, etc.), and so on and so forth. All you (the followers of the absence of grammatical gender in English) need is to provide some good enough grounds. Good enough means that total majority of, say, Russian textbooks stops claiming that English has the grammatical gender at least for pronounces and you may not take into account my personal opinion.
So far your arguments are not good enough. Let us consider your sociological example
> Which pronoun you use is determined entirely by the nature of the thing referred to.
> In "My wife is out - she will be back soon" using "she" depends on knowing that "wife"
> refers to a female and not on it belonging to any noun class other than nouns which refer
> to things female or some class such as "beginning with the letter w".
The Putin’s propaganda tells us that many of American wives have penises nowadays. In my opinion such propaganda ruins your argument completely. You claim that if you see the word wife you imply “she”. But I bet (I have no physical possibility to check it now) if you come closer and see the wife’s penis you will refer to her as he. Like “This is my wife. He is cute.” Believe it or not but the same thing will happen in Russian – he will be cute but the feeling of the situation will be different as if the wife is Pamela Anderson (I am not sure if my subjunctive mood is correct but it is supposed to be it). There is sociologically something that makes us to expect the agreement of wives with “she”. And there is no doubt that you either shall eliminate every single one who thinks that way (the propaganda says you have already started that) or as in case with technical point of view you have to provide some solid explanation why would we like to associate the word “wife” with “she”. So far your arguments does not look very convincing (at least for me and many authors of English textbooks in Russia).
And I do believe that it might be technically possible to provide technical and sociological (and any other if there is a need) justifications (just to make Russians to rewrite their English textbooks). Apple Inc. costs nowadays more than all Russian assets, including Gazprom, Sberbank and my microwave. So, there are enough resources why do not give it a try?..
> What is you understanding of what grammatical gender involves which leads you
> to conclude that English has grammatical gender?
1. There is a grammatical need to agree some nouns with other words in the text.
2. The need (1) cannot be satisfied with existing grammatical feature of the language such as number, case, person, etc.
3. There are non-grammatical (sociological, etc.) needs to agree some nouns with other words in the text.
4. The features (1) and (3) can be combined and the resulting feature looks very like the grammatical gender in other languages.

> panini
> The problem here is really simple. In the science of linguistics, "agree" and "gender"
> have a specific meaning. Rock100 is using words in a completely different way. "She fell"
> is not an example of agreement, by linguistic standards. If you want to have a rational
> discussion of linguistic matters, you either have to declare your theory of what technical
> terms mean, or you have to learn what those words mean in the context of linguistic
> science. As I said. The linguistic theory of "gender" has very little relation to social theories
> of "gender".
You are right. But I do not think linguistics has a full featured unambiguous theory beneath. So far you, linguists, are as the ancient Greeks talking about atoms – you may be right or you may be wrong. The good thing is that this forum is not a private party.

> /папа (ends in -a), the word is MASCULINE but -a is a nominative case ending
> for FEMININE. So папа gets all feminine grammatical case endings./
Technically speaking the grammatical case endings are controlled by declension and not by grammatical gender. Father and mother are the nouns of the first declension so they change the same way. There are not many such masculine common words indeed but they do exist. Plus nouns that are Proper Names. Plus common gender nouns.
But you might probably double-check the info above -- I can easily be wrong in details (I just hope it is acceptable for the lounge section).

> There is no mistake/error. There is politics.
You have left me no choice but to check it. This book www.ozon.ru/context/detail/id/30048120/ (probably the most reputable one if you are interested) reads on the page 188 that nouns declension numeration is relative and the first and the second declensions are rearranged in school program.

> "There is no swifter route to the corruption of thought than through the
> corruption of language."
> George Orwell
>
> For the last 100 years (give or take) some language reforms have taken place
> in Russian. Not always for the best.
Yeah, probably that Orwell was a little bit Russian too. Russians indeed like to change something in schoolbooks to make parents (or schools) to buy new ones. For example, it is possible in math to reorder the theorems so that the resulting whole sets will be equivalent but the look and idea of derivations will be completely different. That is how, for example, the new geometry (> 1000 years old knowledge) school textbooks appear in Soviet union/Russia.

It seems simpler to restrict the term 'grammatical gender' to its established meaning, i.e. (a), and to use some other term for (b).

Agreed, though I would expand (a) to include verbs, articles and pronouns, though not necessarily all of them in every language which has grammatical gender.

There is another term for (b): "pronominal gender system". See here: https://en.wikipedia.org/wiki/Gender_neutrality_in_genderless_languages#English where it says:

"English lacks grammatical gender, but can be considered to have a pronominal gender system with semantic gender represented in the pronouns. This system of gender is quite minimal compared to languages with grammatical gender."

without changing their form (they look the same). This is simplification.

That has to be moot.

It is all too easy to equate complexity with morphological complexity. However, the word "on" having so many different meanings in English has to be a complexity. Native English speakers do not regard it as a complexity, just normal. It must though be a nightmare for those learning English as a second language - it would certainly seem so if all the information on the Wiki page were presented to them at once.

English is an analytic language with isolating tendencies and words can have multiple meanings and functions. Sometimes it is difficult to say immediately what the function of a word is.