That just moves the question to “Why did Latin have genders.”
And part of the answer, I believe, will have to do with the very flexible word order.
Flexible word order goes nicely hand in hand with having different noun cases, but I don’t see how it relates to gender.
Well, the OP asked “Why did so many languages evolve this way?” (emphasis added).
And part of the answer is that many of the foreign languages that an English-speaking person is most likely to learn are romance languages, and are derived from a language that had three genders.
As to why Latin had genders, maybe it “just did”. Languages are bound to have some fairly arbitrary (and perhaps inefficient) features, we shouldn’t assume everything has utility.
The familiar Indo-European two- or three-gender system is only one way of doing noun classes. The site says Niger-Congo languages can have “ten or more” noun classes, and Caucasian languages from two to eight.
Germanic languages (including Old English) have genders, too. Slavic languages have genders, too. (Polish has three or five, depending on how you count them.) So, it’s not just the languages deriving from Latin that have them.OK, so now we can move the bar back to Indo-European languages. But, as stated above, there are some other language groups that have genders (in terms of noun classifications), too.
Mijin, but even if Latin had genders, there is also the fact that the languages evolved differently, for many words in those languages, although they mean the same, and are even spelled almost the same, take up different genders. See the above with the -gem/-je grammar convention, also “la nariz/o nariz/the nose” and “la leche/o leite/the milk”.
I suspect it all comes down to how a language deals with basic classification of at least some nouns. For instance, in Indo-European languages, male, female and sometimes neither. Using German as the example I’m familiar with, it seems natural that once you start distinguishing between “der Mann”, “die Frau”, and “das Kind” (along with their associated pronouns), you lock yourself into using the der/die/das distinction for all nouns. How and why any given noun picks up the gender that it does is often arbitrary, but I think that’s the basis of it – an impulse toward consistency with the nouns that a language originally distinguishes.
I don’t think that disambiguation can have much to do with it, as there are so many languages that get along just fine without any grammatical gender for nouns at all, or even for personal pronouns in many cases.
That’s not how it originated in the parent language, though. It really has nothing to do with sex. And just look at English, which does have he / she / it but puts almost all nouns into the “it” category.
Ur Indo-European had three genders (and about 7 cases). It presumably had quite free word order, just as Latin does. If an adjective has to agree with the modified noun in gender, number, and case, then quite likely the adjective can appear anywhere in the sentence, with presumably slightly differing connotations. English, for largely unknown reasons, moved to a fixed (or nearly so) word order and so the Old English declensions ceased to have any value in disambiguation. A couple centuries later, most of the declensions disappeared and have not returned. Actually, English uses declensions in just four places, three of them involving adding an -s.
Now the real question is why languages developed inflections at all. In the case of endings, the best guess is that they came from postpositions (just like prepositions except they follow the noun–AFAIK, the only postposition in English is “ago”) that moved closer and closer to the noun and eventually joined it. That leaves open the question of why Latin, say, has five different declensions. Maybe they started out similar, but under phonetic influences became different. Not very satisfying, but there it is.
What is known about how genders originated in Indo-European languages? (I do know that many other languages have a plethora of genders/noun groupings that are divorced from male/female entirely, but that’s not my point – it’s just that once some nouns get put into grammatical groups, you’d expect the tendency would be to categorize the rest of them as well.)
“Ago” is more usually counted as an adverb, not a pre/postposition. If you allow it in the latter category then there are all sorts of other words used in much the same way that you could also count as postpositions: “away”, “hence”, “before”, etc. Many other words which are unequivocally prepositions can optionally be used postpositionally: consider for example “through the whole night” vs. “the whole night through”.
What is known and what I know are two different things. I remember learning a theory that the three-gender system arose later in the Proto-Indo-European period, creating a new category of feminines out of some neuter plurals (both end in -a). I don’t remember why, but one of the suggestions was that was why masculine nouns outnumbered feminines by such a large margin.
Looking around online, I see some references to the theory, where the “collective and abstract suffix in -h[sub]2[/sub]” led to the formation of the feminine. The most recent information I can find suggests that this is no longer held: oh well.
Or extremely differing connotation, in some cases. One of the famous prophecies of the Oracle had a prince asking what would happen if he went to war, and the answer could be parsed as either “You will go, you will return, you will never die”, or as “You will go, you will never return, you will die”. OK, technically that’s an adverb, not an adjective, but the same principle applies.
It’s because they were neuter in Latin (e.g., viaticum), so Portuguese and Spanish could have gone either way (masculine or feminine), since both lack neuter. Why one picked one way and the other picked the other, I’m not sure, and might not be knowable. Often, it’s because of some existing form to serve as a model – e.g, since the Spanish forms ended up ending in “e”, and that the default gender is usually masculine unless the word ends in “a”, they were perceived as masculine. Is there some model for Portuguese -m as feminine?
ETA – turns out some of these “-je (Span), -gem (Port)” words came from French “-age”…but what I said before still applies, since the French ending is ultimately from that same Latin neuter “-aticum”.