Grammatical gender, a thread.

This has been going on a while, so I thought it worth articulating what I think and why. I respond to some of the earlier discussions, but I won't link anyone in. They are free to engage, or not, as they wish.
This is *a* linguist's view on the matter, neither original nor imbued with particular authority beyond what training and interest have given me, nor asserted on behalf of 'linguistics' as a whole.
It then outlines various different ways that noun (NB) gender has been characterised by ancient writers about language. First, let us be clear: these grammarians might have been native speakers of Greek or Latin, but we are not obligated to accept their analyses.
It's worth noting that in Romanian, there is a long-standing argument about how many genders there are (two sets of endings in sg and pl, but three overall patterns). So this is not a matter of 'ancient languages'; this is a matter of difficulty in understanding what 'gender' is.
In other words, the idea that ἵππος, νῆσος and λόγος represent three genders is not surprising, or odd, or indefensible.
What it does not, however, mean, is that grammatical gender is 'shifting'. It also doesn't make Graeco-Roman grammar particularly complex.
The reason I say this is that a noun 'having' a gender is only half the story. If all we did was speak in individual nouns, nominal gender wouldn't matter. The reason gender *does* matter is that it predicts things about the syntax of sentences.
For example, grammatical gender predicts the shape of adjectives; 'of black male horses' and of black female horses' only differ in adjective suffix (NB not ending!) in Greek: μελάνων/μελαινῶν ἵππων. It will predict pronoun forms, participles in compound tenses etc.
The notorious 'schema atticum', according to which neuter plural subjects take singular verbs, is another lovely -hideous feature of Greek grammar determined by grammatical gender.
(We could talk about animacy and split ergativity in Hittite in this context too)
The crucial point is that *none* of this behaviour is affected by an analysis that accords five not three genders to nouns. Even if ἵππος, νῆσος, and λόγος are said to be three different genders ('common', 'feminine', 'masculine'), only two agreement patterns apply to them.
Concealing the alleged 'gender diversity' of ancient grammatical thought is thus no conspiracy (as the Medium article linked above suggests): it reflects a real fact about Greek and Latin syntax, in which there are in total three possible agreement patterns (per number).
Agreement patterns are the only way to determine a noun's gender in Greek and Latin, incidentally.

Now, the article's stated goal is to 'foster an inclusive classroom environment', and about the desirability of this goal I have no doubt at all - I shouldn't think any of us do.
The argument, then, is that teaching a pluralistic view of gender in ancient nouns fosters a pluralistic attitude towards gender in humans. The problem is that the two things are importantly different.
"How can they be? They are after all designated by the same word." Indeed, and that fact has provoked lots of fanciful, not to say dangerous speculation about gendered features of language - 'masculine, adult' vs. 'feminine, childish' languages for example...
...or 'female vs. male concepts' being at the root of gender in language (French 'pied', 'foot' is masculine because it dominates, while 'main', 'hand' is feminine because it is recipient). Plainly this tells us more about the analyst (and his society) than about the language!
Gender systems in language were even held responsible for European savagery in colonial contexts (see Kilarski, Nominal Classification, 188-9, also used by the author of the Medium piece linked up-thread).
The problem is that many languages across the world have a 'gender' system. Sometimes, as in Bantu langs, gender has consequences not only for nouns, but for every component of a sentence, which take appropriate class marking depending on the class of the subject.
In many languages, esp. those without a native grammatical tradition, these classes are unnamed. Greek and Latin grammarians, however, did label their classes, and they called them masculine, feminine and neuter. Why?
The reason was that one group of nouns contained those which referred to females, another to males. What is interesting about that is that the 'other' group contained nouns referring to both, but was labelled 'neither' (in Latin, 'neuter'!)
That's already an important clue that this was never about 'gender-in-the-world'; if it was, the name of the 'neuter' (Grk. οὐδέτερον) makes little sense.
And this brings us to another important problem, the non-linguistic concept of 'gender'.
We distinguish sex - broadly speaking, biologically marked differences between males and females - from gender - broadly speaking, socially marked differences between masculine and feminine people. The thing is, grammatical gender kind of targets both.
(Hence you will read linguists referring to 'natural gender', which is rather a contradiction in terms on the modern definition of gender; what it captures is that a distinction of 'grammatical gender' can refer both to gender and to sex).
The point is that even if nouns designating masculine people and male animals behave similarly syntactically, this isn't unique to them. Automatically this is a crucial divide between the grammatical and 'natural' gender.
Even if it wasn't, what would this prove? Hittite's noun classes are based on (in)animacy, not gender. This doesn't mean there was no gender in Hittite society. Again, this shows the autonomy of the two uses of the word, which is all I seek to demonstrate.
English likewise has no grammatical gender, even if it has gendered pronouns; no *syntactic* consequences result from changing the gender of the subject of a sentence. Georgian has neither grammatical gender, nor gendered pronouns. But Britain and Georgia has gendered people.
(*have, urgh)
My worry, then, is that relating grammatical gender to social gender inevitably privileges languages where that connection works; teaching the two together, even in strongly contrastive terms, reinforces the elision between the two.
Finally, using the fact that a noun like "hippos" is sometimes masculine, sometimes feminine, as a point of comparison for 'fluidity' in modern gender expressions (as per the Medium piece) doesn't capture what grammatical gender actually is.
It therefore rather misleads students about how gender is expressed syntactically in Latin and Greek. My own practice, for all it is worth, is to teach, in a language class, what linguists know, think and say about grammatical gender; social gender can be taught elsewhere.
And a final thing - all this has been about 'noun gender'. There was once a practice to talk about 'verbal gender' too. Annoyingly I can't find a reference in the grammars I've rifled through (and Google, unsurprisingly, is not precise enough to search for this).
What is interesting is that the term 'verbal gender' has been displaced by 'voice' or 'diathesis'. Now this does argue that we do have some aversion to seeing 'gender' "etymologically" as being a loan of Latin 'genus' (in context, roughly 'category').
But that's only natural; and I'm last person to wish to pursue an etymological fallacy. Nonetheless, the usage of linguists has retained more of the Latin term than the modern (highly political term). And that again is only natural for a technical term.
Linguistics does have its own language sometimes, and I don't think arguing that it is important to use it precisely, even against what we might call 'natural use', is unreasonable, particularly in didactic contexts.
(I remember defending my use of the word "argument", as in "argument of a predicate", to an examiner during my 'upgrade'/'confirmation' viva).
As I say - this is just me, trying to sort things out. Happy to listen to corrections, criticisms, enthusiastic applause. /end
You can follow @benjcartlidge.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: