Page 4 of 9

Re: Grammaticalization Quickie Thread

Posted: Sat Jul 03, 2021 6:52 am
by Imralu
Richard W wrote: Mon Jun 28, 2021 7:09 pm
Pabappa wrote: Sun Oct 13, 2019 7:58 pm I typed the example in English because I dont actually know a language that uses person marking inside of a relative clause.
English can. In mine, it generally agrees in person with an antecedent that is a nominative pronoun. From prescriptive statements that the relative pronoun agrees in person with its antecedent, Latin or Greek must also do so.
The Our Father in German begins with
Vater
father(M)
unser
1P.GEN
,
,
der
REL.NOM.M
du
2S.NOM
bist
be.PRES.2S
im
in+DEF.DAT.M
Himmel
sky/heaven
,
,
...
...

Our father who art in heaven ...

Not only is the verb conjugated in the second person, a resumptive second person plural pronoun is also present in the relative clause. (The word order is old fashioned, both the non-final placing of bist and the use of the postposed, invariable genitive pronoun unser rather than the equivalent preposed nominative masculine possessive article unser-∅ (POSS.1P-NOM.M)

In Swahili, the subject has to be indicated on verbs, whether relativised or not.

Unajua.
u-na-ju(a)
2S-PRES-know

You know.

Wewe ndiye unayejua.
wewe
2S
ndi-ye
FOC.COP-CL1
u-na-ye-ju(a)
2S-PRES-REL.CL1-know

You're the one who knows.

Re: Grammaticalization Quickie Thread

Posted: Sun Jul 04, 2021 7:57 am
by Znex
I don't really know where else to post this, but I'm working on a minimal phoneme conlang atm and I had the idea that syllable structures could be shiftable, with segments being reinterpreted as consonantal or vocalic alternately depending on where the stress is.

It's simpler to demonstrate what I mean. So I have the phonemes /c~i k~u h~a t~ə/. If I have a set of 6 segments for a hypothetical word root:

/ckhtck/

If we syllabify it into two syllables of three segments, counting from the left, ckh.tck, we have new nuclei that can become vocalic:

[cuhtik]

If we add a suffix, the exact order remains the same and I can have the same syllables. But if I add a prefix and do the same process:

/ht-ckhtck/ = /htc.kht.ck/ > [həckatcu]

If I have no consonant clusters, it may look even less recognisable.

/ckh.tck/ > [cu:tiu]
/ht-ckhtck/ = /htc.kht.ck/ > [hikacu]

Now I'm aware my conlang already is not exactly naturalistic. I guess my real inquiry is could it conceivably work as a language process and even be considered predictable? Or is it too much change?

Re: Grammaticalization Quickie Thread

Posted: Sun Jul 04, 2021 8:11 am
by bradrn
I tried something like that once; see it here. You may judge for yourself how well it worked.

Re: Grammaticalization Quickie Thread

Posted: Sun Jul 04, 2021 8:37 am
by Znex
bradrn wrote: Sun Jul 04, 2021 8:11 am I tried something like that once; see it here. You may judge for yourself how well it worked.
It looks like a mess at first glance, but you see consistencies still. Even if you're not really aware of the consonant/vowel allophony, you can start to pick up that each word has different stems that alternate predictably with prefixes.

I will think more on this, the example is helpful.

Re: Grammaticalization Quickie Thread

Posted: Sun Jul 04, 2021 9:10 am
by Creyeditor
Znex wrote: Sun Jul 04, 2021 7:57 am I don't really know where else to post this, but I'm working on a minimal phoneme conlang atm and I had the idea that syllable structures could be shiftable, with segments being reinterpreted as consonantal or vocalic alternately depending on where the stress is.

It's simpler to demonstrate what I mean. So I have the phonemes /c~i k~u h~a t~ə/. If I have a set of 6 segments for a hypothetical word root:

/ckhtck/

If we syllabify it into two syllables of three segments, counting from the left, ckh.tck, we have new nuclei that can become vocalic:

[cuhtik]

If we add a suffix, the exact order remains the same and I can have the same syllables. But if I add a prefix and do the same process:

/ht-ckhtck/ = /htc.kht.ck/ > [həckatcu]

If I have no consonant clusters, it may look even less recognisable.

/ckh.tck/ > [cu:tiu]
/ht-ckhtck/ = /htc.kht.ck/ > [hikacu]

Now I'm aware my conlang already is not exactly naturalistic. I guess my real inquiry is could it conceivably work as a language process and even be considered predictable? Or is it too much change?
I had this idea several times (sometimes as part of bigger autosegmental concepts) and I would say it works, but it's a lot of work for a conlanger. Too much work for me.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 1:57 am
by bradrn
What pathways, if any, are available to go from SVO to SOV basic word order? And has this development been attested in any natlangs?

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 2:01 am
by Rounin Ryuuji
What other structural changes are going on?

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 2:05 am
by bradrn
Rounin Ryuuji wrote: Mon Oct 04, 2021 2:01 am What other structural changes are going on?
The structural changes are exactly what I’m trying to figure out. Aside from wanting to go from SVO to SOV, all I know so far is that the language turns its SVCs into compound verbs, and loses most 2P clitics.

EDIT: Also, at one point I considered developing some sort of inverse marking or ergativity, which would be areally reasonable. This might end up with SOV word order in the process, but unfortunately I can’t find any reasonable starting point from which I might get some sort of interesting alignment system. (The most promising would be instrumental SVCs — ‘I take ball throw’→‘I ACC-ball throw’ — but that gives me accusative case-marking, and besides it would most probably end up as an instrumental applicative prefix on the verb.) I’m mostly just interested in knowing what possibilities I have to get from SVO to SOV.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 2:40 am
by Rounin Ryuuji
bradrn wrote: Mon Oct 04, 2021 2:05 am The structural changes are exactly what I’m trying to figure out. Aside from wanting to go from SVO to SOV, all I know so far is that the language turns its SVCs into compound verbs, and loses most 2P clitics.
The only language I know of that does this (and still has serial verbs) is Japanese, which already has the verb at about the position you want it. I think one of the biggest reasons Japanese would be resistant to changing to, say, V2, is that placing a verb before a nominal effectively turns the verb-nominal sequence into a relative clause (adjectives also work this way):

雨は冷たい Ame wa tsumetai "The rain is cold"
冷たい雨 Tsumetai ame "Cold rain"
人が走ってる Hito ga hashitteru "The person is running"
走る人 Hashiru hito "A person who runs"

If you end up with this pattern, there's a chance that, if the verb already can optionally come after the nominals but this is a marked form, having this very common adnominal marker sound like some other conjugated form (or have your relative clause markers erode with the second-person clitics, so you end up with ambiguous adnominality) will likely push the verb to be somewhere that there won't be any ambiguity about what it's doing. You could speed this along further by having common structure verbs erode into case particles in some contexts (Japanese also did this, -ni, -nite/-de, and -no were probably all originally verbs themselves), such that you have to switch the order to from this is that to this that is in order not to have ambiguity with such a structure as this of that, this with that. Since clitics are eroding, I expect other words must be, too, so some new case markers sound likely to form.

I wish I could tell you all this precipitated the change in Japonic, but Old Japanese was already subject-object-verb, as far as I'm aware. (It's also an extremely resilient feature once you have it — Ineshîmé, having a longer time depth than Old Japanese to modern in its setting, didn't dis-evolve it, but Japonic syntax seems to tend towards conservatism, in my experience.)

More hypothetically, it could also be that clauses could originally only tolerate either a main verb or serial verbs, and the development of verbs into markers did not result in a new slot for verbs within a noun phrase, so that your this-with-that structure, with the with word being formerly a verb, cannot be split (if I recall correctly, the "rule" against splitting infinitives in English was once actually descriptively correct — it seems to have at least been uncommon before sometime in the Eighteenth Century, when people started worrying about it — so if you have this noun-marker-noun structure require the verb come after, the verb could possibly fossilise into this position, especially if verbs tend to be big snarls of affixes).

As far as serial verbs becoming compound verbs, Japanese also seems to have done that historically through attaching a stem called the ren'yōkei, which usually ends in -i, but in some verbs -e to whatever other form, e.g. 走り回る hashirimawaru "run around", where the hashiri- is the ren'yōkei of 走る hashiru "run, drive, travel", and appended to it is the further conjugable 回る mawaru "turn, revolve"; verbs could formerly, and in poetry sometimes still, use the ren'yōkei in place of the te-form to mean "do something and something else" (most of the fossilised forms mean something slightly different from their components nowadays).

I hope at least some of this is helpful (I'm starting to feel it's getting a bit rambly). There's also the possibility of having that structure simply common in surrounding language. Areal features are well-documented, and don't tend to require internal justification.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 2:56 am
by bradrn
Rounin Ryuuji wrote: Mon Oct 04, 2021 2:40 am
bradrn wrote: Mon Oct 04, 2021 2:05 am The structural changes are exactly what I’m trying to figure out. Aside from wanting to go from SVO to SOV, all I know so far is that the language turns its SVCs into compound verbs, and loses most 2P clitics.
The only language I know of that does this (and still has serial verbs) is Japanese, which already has the verb at about the position you want it.
Wait, Japanese has serial verbs? Converbs, yes, light verbs, yes, but not serial verbs. At least, I’m pretty sure I would have heard about it if it did.
More hypothetically, it could also be that clauses could originally only tolerate either a main verb or serial verbs, and the development of verbs into markers did not result in a new slot for verbs within a noun phrase, so that your this-with-that structure, with the with word being formerly a verb, cannot be split
Sorry, but I have no idea what you’re trying to say here. Serial verbs already cannot be split. Indeed, it seems that this is universally taken as a defining feature. There’s no such thing as a ‘serial verb’ as distinct from a ‘main verb’ — there’s just ‘verbs’, which can occur in serial verb constructions. I suspect you are getting confused between serial verbs and clause chaining here. (Not that you’re the first person to do so. As I have complained about before, reading the literature on SVCs is an exercise in frustration.)
There's also the possibility of having that structure simply common in surrounding language. Areal features are well-documented, and don't tend to require internal justification.
This is indeed the case. I’d just like to do something more interesting with this, if I can.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 3:12 am
by Rounin Ryuuji
bradrn wrote: Mon Oct 04, 2021 2:56 am
Rounin Ryuuji wrote: Mon Oct 04, 2021 2:40 am The only language I know of that does this (and still has serial verbs) is Japanese, which already has the verb at about the position you want it.
Wait, Japanese has serial verbs? Converbs, yes, light verbs, yes, but not serial verbs. At least, I’m pretty sure I would have heard about it if it did.
Then what would you call its verb-te sequences? I did find it in the Wikipedia article called a "serial verb", though I'm ready to accept that I and somebody else could misunderstand what a "serial verb" is.
More hypothetically, it could also be that clauses could originally only tolerate either a main verb or serial verbs, and the development of verbs into [case] markers did not result in a new slot for verbs within a noun phrase, so that your this-with-that structure, with the with word being formerly a verb, cannot be split
Sorry, but I have no idea what you’re trying to say here.
I missed the word case in there. I don't know if you have this happening, however.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 4:06 am
by Moose-tache
I was of the impression that -te constructions were converbs. Subordinating one verb and moving all TAM information to the other doesn't sound like an SVC to me, but it's tricky to say exactly since so many languages with SVCs have virtually no verbal morphology.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 4:14 am
by Rounin Ryuuji
Are these things mutually-exclusive?

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 4:17 am
by bradrn
Rounin Ryuuji wrote: Mon Oct 04, 2021 3:12 am
bradrn wrote: Mon Oct 04, 2021 2:56 am
Rounin Ryuuji wrote: Mon Oct 04, 2021 2:40 am The only language I know of that does this (and still has serial verbs) is Japanese, which already has the verb at about the position you want it.
Wait, Japanese has serial verbs? Converbs, yes, light verbs, yes, but not serial verbs. At least, I’m pretty sure I would have heard about it if it did.
Then what would you call its verb-te sequences? I did find it in the Wikipedia article called a "serial verb", though I'm ready to accept that I and somebody else could misunderstand what a "serial verb" is.
I agree with Moose-tache here. I just finished reading a chapter about Japanese converbs, and they specifically treat the -te form as a converb. Serial verbs are things like these (Aikhenvald 2006):

n=babas
3SG=bite
welik
pig
n=mot
3=SF=die
do
REAL

It bit the pig dead (Taba)

fu
he
fase
letter
naebe
NEG
fi
sit
isoe
write

He did not sit and write a letter (Barai)

abó
gf.
lori
take
tudik
knife
ko’a
cut
paun
bread

Grandfather cut the bread with a knife (Tetun Dili)
Rounin Ryuuji wrote: Mon Oct 04, 2021 4:14 am Are these things mutually-exclusive?
Yes, certainly. No-one can agree on the definition of either, but I’ve yet to see anyone suggest that they might overlap. (Actually, they’re about as far away from each other as two constructions can get: converbs are diclausal, SVCs are monoclausal.)

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 4:23 am
by Rounin Ryuuji
That will teach me to trust Wikipedia, then.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 4:28 am
by bradrn
Rounin Ryuuji wrote: Mon Oct 04, 2021 4:23 am That will teach me to trust Wikipedia, then.
I’ve said it before and I’ll say it again: never, ever, ever rely on Wikipedia for linguistic information!

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 4:32 am
by bradrn
bradrn wrote: Mon Oct 04, 2021 4:17 am Yes, certainly. No-one can agree on the definition of either, but I’ve yet to see anyone suggest that they might overlap. (Actually, they’re about as far away from each other as two constructions can get: converbs are diclausal, SVCs are monoclausal.)
This deserves a slight correction: I’ve just remembered that in some cases converbs can cover the same semantic relations as SVCs, and vice versa some languages have SVCs which are used almost for clause chaining. That being said, the prototypical cases are still very far apart.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 4:40 am
by Rounin Ryuuji
This might be why "converbs" still look to me like auxiliaries with delusions of grandeur.

Either way, maybe post a sketch of the language. I seem to be good at filling languages with phonemes you hate making languages do things, even if I don't know what to call the things I'm doing half the time.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 5:35 am
by bradrn
Rounin Ryuuji wrote: Mon Oct 04, 2021 4:40 am This might be why "converbs" still look to me like auxiliaries with delusions of grandeur.
Sorry, what? Auxiliaries are nothing like converbs, e.g. English has both:

Being hungry, I will eat something

(First underline is a converb; second is an auxiliary.)

If anything, auxiliary verb constructions are more like SVCs: indeed, I’ve never even been particularly sure what the difference is. (Auxiliaries are more closed, perhaps.)
Either way, maybe post a sketch of the language. I seem to be good at filling languages with phonemes you hate making languages do things, even if I don't know what to call the things I'm doing half the time.
I’ve been meaning to write a post properly fleshing out the SVC system, but at the moment the only thing I’ve publicly written about it is one annotated text.

Re: Grammaticalization Quickie Thread

Posted: Mon Oct 04, 2021 6:32 am
by Rounin Ryuuji
bradrn wrote: Mon Oct 04, 2021 5:35 am Being hungry, I will eat something
I feel like I've just had some sort of very massive brainfart.
Either way, maybe post a sketch of the language. I seem to be good at filling languages with phonemes you hate making languages do things, even if I don't know what to call the things I'm doing half the time.
I’ve been meaning to write a post properly fleshing out the SVC system, but at the moment the only thing I’ve publicly written about it is one annotated text.
Well, looking at this, if I've understood it correctly...

Sasat thaŋmu tlaqufli Qalit thaŋ waqli gilut fawetlli, ‘naqeŋ khayiqenebey tshaal pawtiq?’,
/ˈsasat ˈtʰaŋmu ˈt͡ɬaʔufli ˈʔalit tʰaŋ ˈwaʔli ˈɣilut ˈfawət͡ɬli | ˈnaʔəŋ ˈkʰajiʔənəbəj t͡sʰaːl ˈpawtiʔ/
sasat thaŋ=mu tlaquf-li qalit thaŋ waq-li gilut fawetl-li, ‘naqen khayiq-ene=bey tshaal pawtiq?’,
wind DEF.SG=PST follow-CONT sun DEF.SG do.IPFV-CONT argument say.IPFV-CONT, ‘who strong-NMLZ=Q leave all?’,

The first thing that strikes me as probable is that this giant crufty qualifier thaŋmu tlaqufli would like to be smaller, and fall together into a single word; a bit of assimilation and loss of unstressed syllables and glottal stops, and we have "thamtlau", possibly simplifying further to mtlau, mblau, mlau, blau (I hope all that reduction won't make you blue, German or otherwise).

This change should suggest that verbs intimately connected with a nominal ought to follow it, inverting the order of waqli gilut to gilut waqli. This could, in turn, trigger the sequence of verbs to merge together — waq fawetlli > (disinflect the first verb) wappautli, and then treat the whole phrase gilut wappautli as a modifier of the entire quote, and you get this:

Sasat blau Alit thaŋ "Naeŋ khayienebey tshaal pawtî?" gilut wappautli.

I just hope I understood what each word was supposed to be (and what your objective actually was).

Areal influence would presumably also encourage the change, but the reinforcement of the "thing which modifies noun must follow it, and a connected chain of nominals are unsplittable" will certainly help.