Genitive Marker to Subject Marker.

[GBR]

If I'm not mistaken, Japanese が ga shifted at some point from marking genitive case to marking subjects. How on earth did this happen, and why? It's still used in place names in the old way, occasionally, e.g. 幡ヶ谷 hata ga ya, 'valley of flags' (n.b. the second character is an alternative way of writing が used only in this context), but most native speakers I've spoken to don't know the old meaning.
Also, what other shifts have occurred in grammatical markers in Japanese and beyond?

[Drydic]

Don't ask me to remember how but genitives sometimes become ergatives.

[Hakaku]

In Old Japanese, the particles ga and no had overlapping functions as both genitive and nominative markers, and were ultimately distinguished by their degree of politeness. Such a contrast still exists in many Kyushu dialects and Ryukyuan languages, though the exact usage varies. Often, ga is less polite or more familiar and occurs more with self-referential pronouns (hence why we still have waga in standard Japanese), whereas no is more polite.

As for why modern Japanese differs, there's no one answer to that, but dissimulation would likely be the driving factor, with both particles now occupying more distinctive roles.

I also don't know what particles were specifically used in Old Japanese, but significant shifts and variations have occurred in many dialects and the Ryukyuan languages as well. Consider the following differences in topicalization:

Code: Select all

        Japanese   Kagoshima    Okinawan
	
-       kore       koi          kuri
TOP     kore wa    koya         kuree

Particle differences also exist between them. For example, the standard Japanese construction "verb + ni + iku" (going to do X) becomes "verb + ke + iʔ" in Kagoshima and "verb + ga + ichun" in Okinawa. But to indicate location (i.e. going to X), the standard construction "place + ni + iku" becomes "place + i/see + iʔ" in Kagoshima and "place + nkai + ichun" in Okinawa.

[clawgrip]

I can only answer part of your question. All the material I've seen on Classical Japanese seems to treat が ga and の no as mostly interchangeable (Edit: Hakaku has shed some light on this). They often divide the functions of these particles into these five categories:
1. topic marker (主格)
2. adnominal modifier (連体修飾格)
3. appositive marker (同格)
4. nominalizer (準体法) (I'm guessing nominalizer is the best translation here)
5. simile marker (比喩)

To get the easy ones out of the way, In modern Japanese:

#1 is always が ga;
#3 Is で de, e.g. 大きい街で、人口は大阪市に匹敵する Ōkii machi de, jinkō wa Ōsaka-shi ni hitteki suru "it is a large city whose population rivals Osaka" (example lifted from the Internet)
#4 is interchangeable with のこと no koto or のもの no mono with nouns, e.g. 私の watashi no (私のこと watashi no koto "me" or 私のもの watashi no mono "mine"), and with verbs, it is just の no, e.g. 食べるの taberu no (食べること taberu koto "eating"), 赤いの akai no (赤いもの akai mono "the red one").
#5 is indicated with のように no yō ni.

#2 is generally の. It seems straightforward at first, since it includes familiar phrases such as possessives and other similar things (e.g. 友達の本 tomodachi no hon "(my) friend's book"; 最大の鳥 saidai no tori "the biggest bird"). However, it also also appears unexpectedly in subject position of relative clauses where you would otherwise expect が ga (山のない国 yama no nai kuni "a country with no mountains"; 鼻の高い人 hana no takai hito "a person with a large nose"). が only remains in place names such as you mentioned (e.g. 茅ヶ崎 Chigasaki, 自由が丘 Jiyūgaoka).

So this solves the question of how they were split up (though not why). But it also doesn't explain how they came to be so interchangeable in classical Japanese, and what they were before that. Maybe someone else knows.

[Pabappa]

I think thre are some langs where the default verb in a sentence w one animate and one inanimate noun is "have" .

E.g. "I story" = "i have a story.". This could be the source of thepattern in which genitives and ergatives are the same

[GBR]

Super interesting stuff guys.
I originally thought ga was explicitly used for genitive historically, and that maybe some sort of chain shift had occurred, whereby an innocuous locative or some such drifted, and that pushed the other particles into new roles. The thought of having two particles that both mark genitive and nominative is fairly mind blowing. I guess if you have topicalisation too, like Japanese, that would help alleviate some of the potential ambiguity. I just spoke to my boyfriend, by the way, and he said ga is still sometimes used for members of groups as well, though mostly in manga and the like, to try and sound cool e.g. お庭番集が一人ひょっとこ oniwabanshuu ga hitori hyottoko Hyottoko of the Garden Gaurds

Clawgrip, do you have any good sources on old Japanese?

Anyone else got examples of grammatical particles/cases etc shifting meaning?

[clawgrip]

My only real resources are the Internet and some books I have bought in bookstores. The problem is that a large percentage of the books on Classical Japanese grammar that you will find in bookstores in Japan are aimed at high school students, so the books often avoid getting into any more detail than is absolutely necessary.

[Niedokonany]

As for the (perceived) genitive > ergative development, I can see at least two hypothetical scenarios:

1) both genitive and ergative derive from a sort of ablative case:

Y verb-INTR from X > X-erg verb Y
Y from X > X's Y

2) the language's basic simple clause structure derives from an action nominal construction

so instead of “X verbsY” people say "X's verbing Y-Obl is happening" more and more, eventually the latter becoming the only structure.

[Salmoneus]

Don't ask me to remember how but genitives sometimes become ergatives.

This is exactly right. Cross-linguistically, subjects and possessors are often marked the same way, particularly in languages with ergativity.
One observed way in which this can happen is by having ergative-alignment verbs originate from verbal nouns and gerunds: "my eating cake" is reanalysed as ergative-verb-absolutive, so the new 'ergative' is the old 'genitive', which may continue also being the genitive.

More generally, I think there's a strong semantic thing underpinning this, in which an action in some way 'belongs to' or is under the control of its subject in the same way that a possessum belongs to or is under the control of its possessor.

[Vardelm]

I've been reading about how ergative systems can originate in the passive voice of accusative languages. There's a preview on Google Books of Passive and Voice, which is a series of papers assembled & edited by Masayoshi Shibatani. Within, starting on pg. 441 is a paper called "Formal and functional aspects of the development from passive to ergative systems". It discusses, at least to some extent, how ergative cases can arise from genitives. Unfortunately, it's a preview like anything else on Google Books & some pages are missing. I would kill to get my hands on the whole thing, but $300+ for a single book is a bit more than I'm willing to drop on my conlanging habit.

Basically, the connection between ergative & genitive cases is in part due to genitive case being used for the oblique agent argument in a passive.

[Cedh]

Possibly relevant: a very interesting paper about the diachronics of ergativity.

[Vardelm]

Possibly relevant: a very interesting paper about the diachronics of ergativity.

Nice find!

After a quick skim through that, 1 thing that the "Passive and Voice" article mentioned was that a verbal marker for passive voice can be reanalyzed as a transitivity marker. That's exactly how my Tibetan Dwarvish language (undergoing a revamp) will work. I'm curious if maybe languages that reinterpret the passive marker as transitivity might take a more direct route from passive to ergative.

[Zhen Lin]

の can substitute for が sometimes in modern Japanese in relative clauses. Given that the modern predicative form (終止形) was historically the attributive form (連体形), i.e. the form used to construct relative clauses, perhaps the shift of が from "genitive" to "nominative" isn't that far-fetched.

[clawgrip]

の can substitute for が sometimes in modern Japanese in relative clauses.

I mentioned this in my description of the adnominal modifier function of historic の and が, namely that の had taken over this function, which could formerly employ either one.

the modern predicative form (終止形) was historically the attributive form (連体形)

This is only partially true. Modern predicative comes from the attributive for shimo- and kami-ichidan, and ku and su irregular verbs. Shimo- and kami-nidan verbs seem to use the adverbial form (連用形), while -nu irregular still uses the historic predicative. Yodan predicative and attributive forms are identical, so it doesn't matter here.

Anyway, it is clear that の and が were able to perform the same functions in Middle Japanese (with perhaps some differences in register), and that their various functions were eventually redistributed, making の and が completely separate in Modern Japanese except for some close overlap in relative clauses a few archaicisms such as 我が for 私の.

The question I wonder now is what these two particles indicated (in Old Japanese?) before they converged.

[Radius Solis]

The affinity goes beyond just the context of ergativity, too. For example in Austronesian alignment ("trigger" systems), I have read that subjects (triggers) are typically marked the same way as possessors - and that clauses may be analyzable as noun phrases in zero-marked existentials, like "my eating cake (is the case)" or "the cake's being-eaten (is the case)" or at least have comparable semantics.

[clawgrip]

A paper by Bjarke Frellesvig suggests that the Old Japanese particles ga, i, and rwo are the only ones that "lack a transparent internal source" and suggests, with caution, that the possibility is open that they may have come from Old Korean. It's a bit hard for me to imagine how a language could adopt grammatical case markings from another language, but a few points from this paper are (indirectly) suggestive of how this could happen. The paper first mentions that Old Japanese core case markers were optional (and in informal Modern Japanese, they still are), and also interestingly that there is no dedicated marker for subjects (leaving an opening that something could potentially fill), and it also mentions that many particles originated from nouns or verbs. As such, these particles might possibly have been borrowed in a similar capacity and eventually grammaticalized as particles/case endings like the rest of them. I don't know though.

If this is the case, it could explain how there are two words with essentially the same meaning (though this paper does mention a couple ways they are different, such as how personal pronouns always used ga, e.g. wa-ga, ni-ga, and demonstratives always used no , e.g. ko-no, so-no). Additionally, it seems like in Old Japanese (as in Modern Japanese) only no could used for meaning #4 in my list above. If this Korean thing is true, then it seems to make sense that the native no would have a wider variety of uses basic than a loanword.

All purely hypothetical, based on my own knowledge and some hints from this paper.

[Hallow XIII]

It's a bit hard for me to imagine how a language could adopt grammatical case markings from another language, but a few points from this paper are (indirectly) suggestive of how this could happen.

English borrowed are and its entire set of pronouns from Old Norse - and creoles and pidgins do stuff like this all the time, so why the heck not?

[Drydic]

It's a bit hard for me to imagine how a language could adopt grammatical case markings from another language, but a few points from this paper are (indirectly) suggestive of how this could happen.

English borrowed are and its entire set of pronouns from Old Norse - and creoles and pidgins do stuff like this all the time, so why the heck not?

Your facts are extremely wrong. Are has an Old English antecedent as well. And English borrowed one pronoun from Old Norse.

But your point is still accurate. A better example are the extra not-of-IE-vintage local cases in dialectal Lithuanian. They are quite clearly borrowings of Finno-Ugric ones (matching up to Livonian forms quite well.)

[Radius Solis]

Well, one set of pronouns. They/them/their and the forms based on them are all from Old Norse though the rest are not.

Wikipedia does also include "are" as being from ON, but the case there is murkier, as it is listed as a "merger" of the OE and ON forms.

[GBR]

A paper by Bjarke Frellesvig

Great paper. Where did you find it? As a general note, where do you guys find this stuff? My google-fu is terrible.

[clawgrip]

Google search: old japanese particles

First result.

[Hallow XIII]

Your post is a bullshit, but the point is more or less there

I know. I was just too lazy too google this stuff because I (correctly, as it seems) assumed that you'd all know this by heart at this point anyway.

[Vardelm]

I (correctly, as it seems) assumed that you'd all know this by heart at this point anyway.

ZBB > Google.

[Hallow XIII]

I (correctly, as it seems) assumed that you'd all know this by heart at this point anyway.

ZBB > Google.

Obviously. I can't google derisive comments on my conlang.