“Grammatical gender” is just a term. You can define it in various ways, and it doesn’t really matter, as long as you make your definition clear. Even linguists don’t have perfect agreement about how they use this term.
It is clear that linguists don’t mean “having different words to refer to males and females”. So the existence of words with different forms for male and female, like “waiter” and “waitress” or “hen” and “rooster”, isn’t considered to be an example of grammatical gender. (For comparison, we also have the words “child” and “adult”, and “puppy” and “dog”, but this doesn’t mean English has a category of “grammatical age” or “grammatical maturity”.)
According to Anna Kibort and Greville G. Corbett, the “definitional characteristic of gender” is agreement in gender of other words with the noun. This is where it gets tricky to classify English. Unlike most languages with grammatical gender, English doesn’t exhibit gender agreement on articles, adjectives or verbs. The only place we could point to to justify a category of “grammatical gender” in English is the pronouns.
So, it comes down to whether you think it’s correct to say that pronouns “agree” with their antecedents in gender, in the same sense that adjectives or articles agree with their antecedents in languages like Spanish or German.
Apparently, linguists disagree about this. Kibort and Corbett say
Barlow (1992:134-152) discusses the issue of the scope of agreement and concludes that there are no good grounds for distinguishing between agreement and antecedent-anaphor relations. If antecedent-anaphor relations are accepted as agreement, languages in which gender distinctions are absent from noun phrase modifiers and from predicates and in which free pronouns present the only evidence for gender, can be counted as having a (pronominal) gender system. Such languages are rare, and the best known example is English, which is typologically unusual in this respect (Corbett 2005:126; see also §6 below); another such language is Defaka (Niger-Congo, Nigeria; Jenewari 1983:103-106).
So even if we say English should be considered to have grammatical gender, it’s not a typical example.
An idealized "typical" example of grammatical gender
Typically, in a language with grammatical gender, every noun in the language is associated with a single specific gender, which governs the agreement of other words that refer back to that noun, usually including any articles or adjectives modifying the noun. Different languages assign gender to nouns in different ways. Some languages use strictly semantic criteria, like “male animate” vs. “female animate” vs. “inanimate” or “male human” vs. “female human” vs. “other”. There are also more complicated semantic assignment systems like “male human, domesticated animal, or concrete noun” vs. “female human, wild animal, or abstract noun”. Other languages use a mix of semantic criteria and formal criteria relating to the shape and structure of the word, like what sound it ends with or whether it has certain affixes. In languages with formal criteria, the gender of a noun doesn't necessarily relate in any obvious way to whether it is "male", "female", or "neither". (Corbett 2013)
In real languages, things are often more messy. Even in languages with generally robust and obvious gender systems, some nouns may be "epicene" or "common" and have more than one pattern of agreement, and some nouns might have an uncertain gender classification that varies from speaker to speaker.
An example from French
A clear example of grammatical gender from French: the phrase for "a mouse" is une souris, which is grammatically feminine, marked by the use of the feminine indefinite article une. Even if it is a male mouse, we cannot use the masculine article and say un souris -- that’s absolutely impossible. If you wanted to indicate that it is male, you would have to use a longer phrase like une souris de sexe masculine. (example taken from the following thread on Wordreference)
How pronoun "agreement" is unusual
In many languages, pronoun agreement is based less on the identity of the noun, and more on the identity of the referent, than adjective or article agreement. (We can see this in English for the category of plurality: a grammatically plural pronoun like “they” can refer back to a grammatically singular noun like “the group”.)
Taking our French example of a male mouse, une souris de sexe masculin, the mouse might be referred to with the masculine pronoun il, since it is male, although the grammatical gender of the word souris is still unambiguously feminine.
J'ai une souris qui s'appelle Gérald. Il est très gentil, c'est une bonne souris.
“I have a mouse called Gerald. He is very nice; he
is a good mouse.”
(Sentence taken from pointvirgule's post on Wordreference)
So we see in languages with grammatical gender, the gender of a pronoun isn’t guaranteed to be the same as the gender of the word it refers to.
Similarly, in English, the choice of pronoun is rarely based on the specific word we're using. It generally has more to do conceptualization of the actual thing that we’re referring to.
Depending on the circumstances, we can say any of the following:
- I took my cat to the veterinarian because it was sick”
- “I took my cat to the veterinarian because she was sick”
- “I took my cat to the veterinarian because he was sick”
using the exact same noun in each case.
So the agreement of pronouns -- the only place where English regularly makes a gender distinction -- is almost never based on the specific noun used, but on semantic criteria pertaining to the referent. (This holds even in freak cases such as using “she” for ships: the pronoun isn’t specifically tied to the word “ship”, but is used to refer to things that are ships. You can use it in a sentence like “The vessel set sail on her maiden voyage this evening” where the word “ship” is never even used.)
If we do say English has grammatical gender, it may not be a three-way system
It's generally accurate to say the "target" of agreement (the pronoun) can take three values based on gender when referring to a singular noun: he/him/his..., she/her/hers..., it/its... (although in modern times, with greater recognition of additional gender categories for humans, the they/them/theirs... set of pronouns has also been added to this group by some people).
However, this is only half of the gender system. The other half is the gender classification of "controllers" (the nouns that are being referred to). The classification of controller gender seems to be generally more complicated (there are examples like Italian nouns that behave like masculines in the singular, and feminines in the plural: should this be called a third "neuter" gender?) and English is no exception.
Kibort and Corbett say English nouns have been analyzed as belonging to anywhere from three to nine "gender classes" based on possible agreement patterns with third-person pronouns and interrogative and relative pronouns.
Summary: English either doesn't have gender, or it has a "pronominal gender system"
In sum, the phenomenon of grammatical gender is characterized by agreement.
Most familiar European languages with grammatical gender exhibit the following features:
- every word has to have a grammatical gender
- gender is marked by the agreement of other words in the noun phrase like adjectives or articles
- a word generally only has a single grammatical gender, which takes precedence over the natural gender of the thing that the word is referring to for purposes of agreement
- gender is generally reflected in the morphology of nouns
English doesn't exhibit these features. The only place gender shows up in English is in pronouns, so English is either not considered to have grammatical gender, or it is said to have a "pronominal gender system". In addition, English pronouns are almost always selected according to semantic criteria, "natural gender". For language teachers, it is probably useful to emphasize the differences between the natural gender system of English and the grammatical gender system of a language like French, rather than using terminology that emphasizes the similarities.
Kibort, Anna & Greville G. Corbett. "Gender." Grammatical Features. 7 January 2008. http://www.grammaticalfeatures.net/features/gender.html
Greville G. Corbett. 2013. Systems of Gender Assignment.
In: Dryer, Matthew S. & Haspelmath, Martin (eds.)
The World Atlas of Language Structures Online.
Leipzig: Max Planck Institute for Evolutionary Anthropology.
(Available online at http://wals.info/chapter/32, Accessed on 2016-12-12.)