Grammaticalization Quickie Thread
Grammaticalization Quickie Thread
Much along the lines of the Sound Change Quickie Thread, I thought I would create a place where we could constellate the many questions and ideas we have about grammatical change over time, including, but not limited to, those about the process of grammaticalization.
Let me get the ball rolling by both seeking advice and posing a question:
First, how might a language go about switching from using prepositions to case suffixes? Are some of the extant prepositions moved to the right of nouns, or are entirely new forms created for the suffixes? Have we seen this happen in any natlangs?
And my question: If we agree that grammaticalization is largely unidirectional, (and my brief skimming of an etymological dictionary shows that it seems to be, but I'd love to be challenged on this front), does this imply a state, perhaps hundreds of thousands of years ago, shortly after some 'dawn of language', when languages had many words for concrete concepts but were still in the process of creating those first words for abstract and grammatical notions? Does this imply that (nearly?) every abstract or grammatical word or affix evolved ultimately from words denoting entirely concrete concepts?
Let me get the ball rolling by both seeking advice and posing a question:
First, how might a language go about switching from using prepositions to case suffixes? Are some of the extant prepositions moved to the right of nouns, or are entirely new forms created for the suffixes? Have we seen this happen in any natlangs?
And my question: If we agree that grammaticalization is largely unidirectional, (and my brief skimming of an etymological dictionary shows that it seems to be, but I'd love to be challenged on this front), does this imply a state, perhaps hundreds of thousands of years ago, shortly after some 'dawn of language', when languages had many words for concrete concepts but were still in the process of creating those first words for abstract and grammatical notions? Does this imply that (nearly?) every abstract or grammatical word or affix evolved ultimately from words denoting entirely concrete concepts?
Re: Grammaticalization Quickie Thread
My answer is only slightly related to your exact question, but:sucaeyl wrote:First, how might a language go about switching from using prepositions to case suffixes? Are some of the extant prepositions moved to the right of nouns, or are entirely new forms created for the suffixes?
In my most recent revision of Kala grammar I have am calling "prepositions" "locative verbs". Let me explain...
mita tsakahue
dog house-LOC
The/A dog is at the/a house.
mita tsakahue ina
dog house-LOC eat
The dog is eating at the house.
mita tsaka nahe
dog house inside
The/A dog is inside the/a house.
mita nahe tsaka ina
dog inside house eat
The dog is eating inside the house.
mita tsaka ka'ela
dog house toward-go
The dog is going toward the house.
mita ka'e tsaka yala
dog toward house go
The dog is going toward the house.
What you're asking about seems like a similar concept and works well to free-up word order.
Re: Grammaticalization Quickie Thread
This seems cool! What motivates a choice between these?
a. mita tsakahue
b. mita tsaka nahe
c. ? mita nahe tsaka
Was either one of ‹-hue› or ‹nahe› used before the other? If you have them, what are their etymologies? Thanks!
a. mita tsakahue
b. mita tsaka nahe
c. ? mita nahe tsaka
Was either one of ‹-hue› or ‹nahe› used before the other? If you have them, what are their etymologies? Thanks!
Re: Grammaticalization Quickie Thread
A is fine by itself, as is B, C would need a verb to end the clause. B would be glossed [dog house be.inside] because nahe is acting in the "locative verb role" rather than as an adposition.sucaeyl wrote:What motivates a choice between these?
a. mita tsakahue
b. mita tsaka nahe
c. ? mita nahe tsaka
Well, -hue is a general locative (LOC) while nahe specifies inside; within; or into, also, -hue is never in the verb role and never takes the motive suffix -la.sucaeyl wrote:Was either one of ‹-hue› or ‹nahe› used before the other?
mita tsaka nahela
dog house inside-MOT
The dog goes into the house.
I took nahe from here.sucaeyl wrote:If you have them, what are their etymologies?
-hue was a choice based on phonotactics and euphony. Pretty much every adposition in Kala ends with 'e, ye, he, or ue.
You too. This is a great topic and has helped me to focus on finishing my Kala grammar...a never-ending pursuit.sucaeyl wrote:Thanks!
Re: Grammaticalization Quickie Thread
I assume that in general case suffixes predate prepositions rather than being derived from them, either from when prepositions were still postpositional, or before they were grammaticalized in the first place. According to this WALS map, overlap of the two are sporadic except for European Indo-European languages and a cluster of languages north of Lake Victoria in Africa, which from what I've gathered glancing at other word-order-influenced properties is a mess of languages that can't decide if they're SOV or VSO. But looking deeper into a few of those other scattered languages might be a place to start.sucaeyl wrote:First, how might a language go about switching from using prepositions to case suffixes? Are some of the extant prepositions moved to the right of nouns, or are entirely new forms created for the suffixes? Have we seen this happen in any natlangs?
I've seen a couple articles on supposed degrammaticalizion; this has a few examples starting about halfway through.And my question: If we agree that grammaticalization is largely unidirectional, (and my brief skimming of an etymological dictionary shows that it seems to be, but I'd love to be challenged on this front)
I'm not sure it implies abstract notions came from concrete notions. But I believe the only class of morphemes I've seen claimed have no known diachronic source from a different class of morpheme is interrogative pronouns, which may be reinforced by other morphemes but have never been grammaticalized from a non-interrogative (this coming from a single half-remembered source, though, and I don't remember how comprehensively the source sampled world languages).does this imply a state, perhaps hundreds of thousands of years ago, shortly after some 'dawn of language', when languages had many words for concrete concepts but were still in the process of creating those first words for abstract and grammatical notions? Does this imply that (nearly?) every abstract or grammatical word or affix evolved ultimately from words denoting entirely concrete concepts?
Re: Grammaticalization Quickie Thread
Where do evidentiality morphemes come from?
Re: Grammaticalization Quickie Thread
Aren't there cases of languages switching from being largely prepositioning/prefixing to being mostly postpositioning and/or suffixing? And the other direction as well? Well, I'm not sure if one direction is more likely, but I recall reading of a (mostly theoretical, I grant you) macrocycle involving circular change between largely isolating, head initial languages, to largely suffixing head final languages, to languages with high degrees of synthesis and freer word order, and back to the first, in what I believe was in that order. Is this totally BS? Or at least just one of many possible large scale morphosyntactic changes that occur?vokzhen wrote: I assume that in general case suffixes predate prepositions rather than being derived from them, either from when prepositions were still postpositional, or before they were grammaticalized in the first place.
Anyways, where in general do basic case markers, whether before or after a noun, whether or to what degree they are phonologically bound, come from? What in the heck grammaticalizes into an accusative suffix? This baffles me.
I guess this does sort of put a hole in the idea that any significant number of languages are in the process of switching their overall morphosyntactic 'type', if I may use such phrasing. What gives? Surely agglutinating languages weren't always agglutinating, isolating languages not always with such low morpheme/word ratios and the typical prepositions and SVO order, surely polysynthetic languages came from languages with simpler and totally different morphologies. Am I missing something?vokzhen wrote:According to this WALS map, overlap of the two are sporadic except for European Indo-European languages and a cluster of languages north of Lake Victoria in Africa, which from what I've gathered glancing at other word-order-influenced properties is a mess of languages that can't decide if they're SOV or VSO. But looking deeper into a few of those other scattered languages might be a place to start.
Well, if interrogative pronouns (also, I've heard demonstratives) are the only words that don't ever seem to be grammaticalized from somewhere else, doesn't this still keep the possibility open for every other word and affix type to ultimately come from a less grammatical, less abstract source? To back the notion up a bit, and I'll concede this only comes from reading the etymologies of a select group of English words, almost every word for an abstract concept seems to come from a concrete source. Likewise, many affixes, if not present in PIE, seem to have come from some abstract but phonologically unbound word. The only conclusion I can really draw is that most all words and affixes for abstract concepts or grammatical functions came from a word denoting something concrete. Maybe I'm misunderstanding something huge.vokzhen wrote: I'm not sure it implies abstract notions came from concrete notions. But I believe the only class of morphemes I've seen claimed have no known diachronic source from a different class of morpheme is interrogative pronouns, which may be reinforced by other morphemes but have never been grammaticalized from a non-interrogative (this coming from a single half-remembered source, though, and I don't remember how comprehensively the source sampled world languages).
I don't know too much about the origin of this in natlangs, unfortunately. Hopefully somebody else will know more. I would say that English 'must' and, to a lesser extent, 'I heard', serve somewhat grammaticalized evidential functions.TaylorS wrote:Where do evidentiality morphemes come from?
a. He must've eaten the whole sandwich. (inferential)
b. I heard she doesn't ever eat from restaurants. (hearsay)
Aren't these pretty much evidentials? Apply some more phonetic erosion, analogical extension and obligatory inclusion and voilà. Or so, as far as I know.
Re: Grammaticalization Quickie Thread
Honestly not sure, I haven't looked into it much. It wouldn't surprise me though if one source is "movement" of cases from more concrete, semantic relations to more abstract, grammatical ones, such as a postposition for "towards" being used to mark the endpoint of a verb of movement (allative case), then expanded to include a metaphoric endpoint (dative), and then finally used for grammatical objects in general (accusative), possibly being pushed by newer methods for showing the previous relationships.sucaeyl wrote:Anyways, where in general do basic case markers, whether before or after a noun, whether or to what degree they are phonologically bound, come from? What in the heck grammaticalizes into an accusative suffix? This baffles me.
Yes, I think you are. The languages that have overlap with prepositions and case suffixes are rare, because older case suffixes are lost as part of the process of becoming more isolating, head-initial languages. Head-final with case suffixes > intermediate state with both old case suffixes and new prepositions > head-initial with prepositions. You can see it pretty clearly in English and the Romance languages, where an old case-marked SOV language changed to a head-initial language that marked with prepositions. The case suffixes are lost because they're redundant, lost to phonological reduction, fossilized, or so on.I guess this does sort of put a hole in the idea that any significant number of languages are in the process of switching their overall morphosyntactic 'type', if I may use such phrasing. What gives? Surely agglutinating languages weren't always agglutinating, isolating languages not always with such low morpheme/word ratios and the typical prepositions and SVO order, surely polysynthetic languages came from languages with simpler and totally different morphologies. Am I missing something?
Yea, that was my point, perhaps poorly worded: all grammatical morphemes seem to have at least the possibility of having lexical etymologies, except potentially interrogatives and demonstratives, for which we have no examples except for reinforcement with additional morphemes (incidentally, it seems to me that if someone were to take Proto-Human seriously, they would be the place to start for that reason: the possibility that they only arose a single time with a minimum of semantic change).Well, if interrogative pronouns (also, I've heard demonstratives) are the only words that don't ever seem to be grammaticalized from somewhere else, doesn't this still keep the possibility open for every other word and affix type to ultimately come from a less grammatical, less abstract source?
At a guess, because I'm also not sure, it certainly seems possible. But a more interesting English example I think is perfect tense, since it doesn't get mixed up in with modal verbs' possibility marking. Someone walks in disheveled, smiling, and smelling of someone else's cologne/perfume. You saying "You've had a fun night" is a clear inferential evidential. Georgian's perfect is the same, coexisting alongside the general past (perfective) and a past imperfective. So it seems like one source is a language having multiple ways to refer to past events, and one of them gains an inferential meaning to it while another other stays an unmarked past tense. Wouldn't be surprised if Turkic's pervasive evidential marking as part of the tense system is a result of something similar, but I know nothing about Turkic diachronics. But again sourcing WALS, evidentiality as part of the tense system is rare outside Turkic. Might be that evidential-marking tenses are disassociated with the rest of the tense system and end up able to coexist with each other, though I'm guessing other sources are more common.I don't know too much about the origin of this in natlangs, unfortunately. Hopefully somebody else will know more. I would say that English 'must' and, to a lesser extent, 'I heard', serve somewhat grammaticalized evidential functions.TaylorS wrote:Where do evidentiality morphemes come from?
a. He must've eaten the whole sandwich. (inferential)
b. I heard she doesn't ever eat from restaurants. (hearsay)
Aren't these pretty much evidentials? Apply some more phonetic erosion, analogical extension and obligatory inclusion and voilà. Or so, as far as I know.
EDIT: Quote tags are hard
Re: Grammaticalization Quickie Thread
This makes sense! I'd be interested in seeing a cross-linguistic comparison of how these things arise.vokzhen wrote:Honestly not sure, I haven't looked into it much. It wouldn't surprise me though if one source is "movement" of cases from more concrete, semantic relations to more abstract, grammatical ones, such as a postposition for "towards" being used to mark the endpoint of a verb of movement (allative case), then expanded to include a metaphoric endpoint (dative), and then finally used for grammatical objects in general (accusative), possibly being pushed by newer methods for showing the previous relationships.
But where do these head final, case-suffixing languages come from? Can languages of the Standard Average Altaic breed, that is, head final, SOV, case-suffixing, agglutinating with low degrees of fusion, ever come from head initial, isolating, prepositioning languages?vokzhen wrote:Yes, I think you are. The languages that have overlap with prepositions and case suffixes are rare, because older case suffixes are lost as part of the process of becoming more isolating, head-initial languages. Head-final with case suffixes > intermediate state with both old case suffixes and new prepositions > head-initial with prepositions. You can see it pretty clearly in English and the Romance languages, where an old case-marked SOV language changed to a head-initial language that marked with prepositions. The case suffixes are lost because they're redundant, lost to phonological reduction, fossilized, or so on.
That makes a lot of sense! Though it seems like basic, concrete vocabulary terms like 'water' and 'arm' are relatively stable. At least, I don't find them likely to be the end result of a grammaticalization cline. I do wonder, when new terms for these concepts arise, how and why they do so, besides a sudden increase in the influence of another language that is.vokzhen wrote: Yea, that was my point, perhaps poorly worded: all grammatical morphemes seem to have at least the possibility of having lexical etymologies, except potentially interrogatives and demonstratives, for which we have no examples except for reinforcement with additional morphemes (incidentally, it seems to me that if someone were to take Proto-Human seriously, they would be the place to start for that reason: the possibility that they only arose a single time with a minimum of semantic change).
This is interesting! I wonder though if your 'you've had a fun night' is an example of inferential evidentiality expressed through the actual morphemes (I don't know if this is the right terminology) or more pragmatically, through context. Any past tense would accomplish the same inferential feeling, wouldn't it? 'You had a fun night' seems to accomplish the same. 'You've eaten the whole pie' and 'You ate the whole pie' likewise carry this inferential meaning; is it really due to a grammaticalization of the perfect tense towards an inferential evidential?vokzhen, on the topic of evidentiality, wrote:At a guess, because I'm also not sure, it certainly seems possible. But a more interesting English example I think is perfect tense, since it doesn't get mixed up in with modal verbs' possibility marking. Someone walks in disheveled, smiling, and smelling of someone else's cologne/perfume. You saying "You've had a fun night" is a clear inferential evidential. Georgian's perfect is the same, coexisting alongside the general past (perfective) and a past imperfective. So it seems like one source is a language having multiple ways to refer to past events, and one of them gains an inferential meaning to it while another other stays an unmarked past tense. Wouldn't be surprised if Turkic's pervasive evidential marking as part of the tense system is a result of something similar, but I know nothing about Turkic diachronics. But again sourcing WALS, evidentiality as part of the tense system is rare outside Turkic. Might be that evidential-marking tenses are disassociated with the rest of the tense system and end up able to coexist with each other, though I'm guessing other sources are more common.
On the other hand, 'have' as a marking of obligation, as well as many other markings of obligation, including the aforementioned 'must', 'should' and others, have seemed to grammaticalize towards marking inferential evidentiality. Interestingly, these all seem to work the best in the perfect tense, which lends some credence to your idea, but is it really the perfect tense that gives it its inferential meaning, or is it the word that previously marked obligation? Maybe obligation morphemes, when in perfect tense, are prone to this grammaticalization. After all, most of these don't seem to make much sense to me if they're in anything besides the perfect.
a. He has to have arrived by now.
b. He must have arrived by now.
c. He should have arrived by now.
d. He has got to be here by now.
e. He ought to be here by now.
-
- Niš
- Posts: 10
- Joined: Sun Feb 05, 2012 4:36 pm
Re: Grammaticalization Quickie Thread
How do noun class/gender systems evolve? What starts them off and how do they spread throughout a lexicon? Thanks!
Re: Grammaticalization Quickie Thread
What is the most basic form of relative clause structure/formation?
- communistplot
- Avisaru
- Posts: 494
- Joined: Thu Oct 07, 2010 6:49 am
- Location: La Ciudad de Nueva York
- Contact:
Re: Grammaticalization Quickie Thread
Generally it's thought (and Zomp actually writes about this in the ALCK) that they derive from Nominal Classifiers (of the type seen in Sino-Tibetan languages) that become grammaticalised for agreement rather than just counting. (e.g. they become required on descriptives (adjectives, descriptive verbs, &c.)). I'm actually working on a language where such a grammaticalisation has already occurred, creating a language that's like a cross between Tibeto-Burman and Bantu.Jana Masala wrote:How do noun class/gender systems evolve? What starts them off and how do they spread throughout a lexicon? Thanks!
EDIT: I should probably give an example of what a system in transition might look like.
From Apanic:
Cìt dès pwỏ-ñyo sà zúu sà dès wáuu sà wỏa.
[t͡ɬʰîtʰ tʰēsʰ pʰɰoʔ.ɲjō sʰâ sǔ sʰâ tēsʰ ɰǎǔ sʰâ ɡoaʔ]
run IND | be fifteen IV | be beautiful IV IND | dog IV TOP
Fifteen beautiful dogs run.
You can omit the class marker after the descriptive verb still, but for many speakers it is obligatory, enough so that I imagine descendants acquiring full class systems.
The Artist Formerly Known as Caleone
My Conlangs (WIP):
Pasic - Proto-Northeastern Bay - Asséta - Àpzó
My Conlangs (WIP):
Pasic - Proto-Northeastern Bay - Asséta - Àpzó
Re: Grammaticalization Quickie Thread
I have really no idea, so I would refer to Vidurnaktis, but do I remember reading something theorizing about how it may have started in German. I'll try to find it!Jana Masala wrote:How do noun class/gender systems evolve? What starts them off and how do they spread throughout a lexicon? Thanks!
This is so cool! I was JUST thinking about making a conlang where classifiers became noun class markers. What I was thinking was, at the earliest stage, classifiers, of which there were many, were generally only used when showing specific numbers greater than one. If there was only one, no classifier was necessary.Vidurnaktis wrote:Generally it's thought (and Zomp actually writes about this in the ALCK) that they derive from Nominal Classifiers (of the type seen in Sino-Tibetan languages) that become grammaticalised for agreement rather than just counting. (e.g. they become required on descriptives (adjectives, descriptive verbs, &c.)). I'm actually working on a language where such a grammaticalisation has already occurred, creating a language that's like a cross between Tibeto-Burman and Bantu.
EDIT: I should probably give an example of what a system in transition might look like.
From Apanic:
Cìt dès pwỏ-ñyo sà zúu sà dès wáuu sà wỏa.
[t͡ɬʰîtʰ tʰēsʰ pʰɰoʔ.ɲjō sʰâ sǔ sʰâ tēsʰ ɰǎǔ sʰâ ɡoaʔ]
run IND | be fifteen IV | be beautiful IV IND | dog IV TOP
Fifteen beautiful dogs run.
You can omit the class marker after the descriptive verb still, but for many speakers it is obligatory, enough so that I imagine descendants acquiring full class systems.
hin.tsɛːg os
dog one
One dog
hin.tsɛːg-e kam hɔp
dog-GEN CL two
Two dogs
At a later stage, two things happened; these classifiers grammaticalized to mark for agreement, but only in the plural, and talking about 'one' of an object began to require needing a classifier, as well, but this classifier is the same for most all nouns.
çĩ.seːj-i ɭɔ us
dog-GEN CL one
One dog
çĩ.seːj-i kǝm ɔp
dog-GEN G6 two
Two dogs
In the protolanguage, plurals were expressed by a classifier and a determiner:
hin.tsɛːg-e kam ɔjɔ
dog-GEN CL many
Many dogs
But now, just including the noun class, along with a useless gentivie, shows plurality:
çĩ.seːj
dog
Dog
çĩ.seːj-i kǝm
dog-GEN G6
Dogs
So now we have a noun class system that's only marked on plurals! At a current stage, this has evolved into a sort of collective-singulative system, with a very divergent collective form: The noun class marker on the plural has fused with the stem, and additionally, this form is now used for the noun when plurality is unspecified. On other hand, expressing the singulative requires the classifier:
s̩.sẽː.xǝm
dog.G6.COL
Dog(s)
s̩.sẽːj-iɽɔːs
dog-SGV
One dog
Still, the now fully engrained noun class system is only visible on collective nouns. To show the change with one more example:
Early:
hɛ.paɕ.kɔ os
seagull one
One seagull
hɛ.paɕ.kɔ-e sint hɔp
seagull-GEN CL two
Two seagulls
hɛ.paɕ.kɔ-e sint ɔjɔ
seagull-GEN CL many
Many seagulls
Middle:
ɛ.px̩.kɔ-i ɭɔ us
seagull-GEN CL one
One seagull
ɛ.px̩.kɔ-i sĩːθ ɔp
seagull-GEN G3 two
Two seagulls
ɛ.px̩.kɔ-i sĩːθ
seagull-GEN G3
Seagulls
Modern:
ɛ.pəx.sœ̃ːθ
seagull.G3.COL
Seagull(s)
ɛ.pxɔ-iɽɔːs
seagull-SGV
One seagull
- StrangerCoug
- Avisaru
- Posts: 269
- Joined: Sat Aug 02, 2014 8:56 pm
- Location: El Paso, TX
Re: Grammaticalization Quickie Thread
What are some ways within the scope of this thread to get rid of cases you don't want anymore? I have a protolanguage with 18 cases, and in my notes many of them are supposed to be re-adapted, merged, vestigal, or even gone entirely in daughter languages, in some cases depending on the register.
If it's needed, my notes call six of the cases "core cases" (i.e. they inflect for gender): active, passive, instrumental, genitive, comitative, and benefactive. The remaining twelve, which do not inflect for gender, are allative, ablative, illative, elative, perlative, inessive, adessive, antessive, postessive, superessive, subessive, and apudessive. This is a fluid-S, secundative language, but there is some allowance for change.
If it's needed, my notes call six of the cases "core cases" (i.e. they inflect for gender): active, passive, instrumental, genitive, comitative, and benefactive. The remaining twelve, which do not inflect for gender, are allative, ablative, illative, elative, perlative, inessive, adessive, antessive, postessive, superessive, subessive, and apudessive. This is a fluid-S, secundative language, but there is some allowance for change.
Re: Grammaticalization Quickie Thread
With that many cases, I have a feeling they would need to be gotten rid of in stages. Unlike the English or Romance ones, I think they'd resist wholesale phonological reduction to zero, unless it somehow gained extensive "double-marking" with adpositions as well, though merging a few cases that are phonologically similar could work. For grammatical changes, "dative" cases often fulfill benefactive and allative roles, and genetive as well (see page 8 here, and page 11 for various allative functions) for some possible mergers. Instrumental and comitative often use the same marker. If you're going for reduction it seems that a number of the locational/directional cases could be merged as well, but I'm unsure how likely they are to merge in the first place.
Most of the Northeast Caucasian languages have an extensive set of locational-directional suffixes, but a few (Chechen-Ingush and Bats for sure) have a severely reduced case system of "only" 8-10 cases. If you can find any diachronic information on them, that might be a place to get further ideas. It wouldn't surprise me if there's similar in Uralic languages but I know less about them.
However, in order to get a family-wide reduction in case uses like you appear to be talking about, it seems likely that the process would need to start in the proto-language. Perhaps certain cases are very limited in use, such as the a case being primarily restricted to set sayings and lexicalized adverbials such as "he acted out of anger"/"he acted in anger"/"he acted with anger" with the role primarily being picked up by a different marker, and already starting to use body parts in adposition-like constructions. The other option that would make it likely for all branches to lose a lot of cases would be to have multiple branches have long-term cultural and linguistic domination by another, non-case-marked language. Even then, having it apply to the entire family - albeit in different ways - seems unlikely. I'd think a number of branches would probably keep the case system going strong and maybe even innovate further cases.
Most of the Northeast Caucasian languages have an extensive set of locational-directional suffixes, but a few (Chechen-Ingush and Bats for sure) have a severely reduced case system of "only" 8-10 cases. If you can find any diachronic information on them, that might be a place to get further ideas. It wouldn't surprise me if there's similar in Uralic languages but I know less about them.
However, in order to get a family-wide reduction in case uses like you appear to be talking about, it seems likely that the process would need to start in the proto-language. Perhaps certain cases are very limited in use, such as the a case being primarily restricted to set sayings and lexicalized adverbials such as "he acted out of anger"/"he acted in anger"/"he acted with anger" with the role primarily being picked up by a different marker, and already starting to use body parts in adposition-like constructions. The other option that would make it likely for all branches to lose a lot of cases would be to have multiple branches have long-term cultural and linguistic domination by another, non-case-marked language. Even then, having it apply to the entire family - albeit in different ways - seems unlikely. I'd think a number of branches would probably keep the case system going strong and maybe even innovate further cases.
- StrangerCoug
- Avisaru
- Posts: 269
- Joined: Sat Aug 02, 2014 8:56 pm
- Location: El Paso, TX
Re: Grammaticalization Quickie Thread
Would it be more plausible to just limit some of the location/direction "cases" for derivational morphology, then, along the lines of the "super-" and "sub-" prefixes we got from Latin? (I like your idea; I'm just seeing what I can work with.)
Edited to add: Now that I think of it, where is the line between a separate case and just something with a marker stuck to it?
Edited to add: Now that I think of it, where is the line between a separate case and just something with a marker stuck to it?
Re: Grammaticalization Quickie Thread
An easy way to selectively get rid of certain oblique cases is to have the speakers start to prefer the use of adpositions to mark their functions.
Not necessarily, it can also be a later areal process that spreads from branch to branch, provided that the languages stay in contact.vokzhen wrote:However, in order to get a family-wide reduction in case uses like you appear to be talking about, it seems likely that the process would need to start in the proto-language.
Could you clarify what you mean by "something with a marker stuck to it"?StrangerCoug wrote:Now that I think of it, where is the line between a separate case and just something with a marker stuck to it?
Re: Grammaticalization Quickie Thread
While it, of course, depends on the time depth we're talking about, it seems unlikely to me that such an extensive case system would be lost in a "reasonable" time frame using purely internal changes. Do we have any evidence of a case system of such a high number reduced almost entirely? To be honest the only ones I know of are various Indo-European languages, which have far fewer cases to begin with than the language in question, and the Nakh branch of NEC that I already mentioned that reduced a complicated locational-directional system but kept the core case system by and large intact.gach wrote:Not necessarily, it can also be a later areal process that spreads from branch to branch, provided that the languages stay in contact.vokzhen wrote:However, in order to get a family-wide reduction in case uses like you appear to be talking about, it seems likely that the process would need to start in the proto-language.
I believe he means what's the difference between a case marker and any other nominal affix. I believe the core of it is that a case marker puts the marked noun in a relationship with a head, without being used for derivation. Though I have to admit I've seen examples that seem to do just that but are claimed to be outside the case system (off the top of my head, comitatives and various locational markings seem to fall here), and I've seen examples that seem to be used derivationally still called cases (again with locationals). I'm not sure if that means there's something I'm missing, or if it's that different authors have more and less restrictive standards for what a "case" is. Probably a bit of both.gach wrote:Could you clarify what you mean by "something with a marker stuck to it"?StrangerCoug wrote:Now that I think of it, where is the line between a separate case and just something with a marker stuck to it?
Re: Grammaticalization Quickie Thread
Loosing a rich case system should certainly take time. I was just saying that the process doesn't need to have a genetically unified starting point but can spread as a borrowed feature, like increased preference to adpositional phrases, initiated by one innovative branch.vokzhen wrote:While it, of course, depends on the time depth we're talking about, it seems unlikely to me that such an extensive case system would be lost in a "reasonable" time frame using purely internal changes. Do we have any evidence of a case system of such a high number reduced almost entirely? To be honest the only ones I know of are various Indo-European languages, which have far fewer cases to begin with than the language in question, and the Nakh branch of NEC that I already mentioned that reduced a complicated locational-directional system but kept the core case system by and large intact.
Often there's no clear cut distinction between cases and derived adverbs or adjectives. You could say that a case has to be productive and any adverbial affixes failing that should be called adverb derivations. Finnish for example has a very case like prolative suffix -tse (meritse, "via sea") that would group nicely with the local cases and in fact forms miniature paradigms with them on some postpositions. It's not usually considered a case though, since it's only available for a small number of nouns. Then again, some Hungarian grammars will present you a list of "unproductive cases" even though these would be better to consider adverb derivations as well.I believe he means what's the difference between a case marker and any other nominal affix. I believe the core of it is that a case marker puts the marked noun in a relationship with a head, without being used for derivation. Though I have to admit I've seen examples that seem to do just that but are claimed to be outside the case system (off the top of my head, comitatives and various locational markings seem to fall here), and I've seen examples that seem to be used derivationally still called cases (again with locationals). I'm not sure if that means there's something I'm missing, or if it's that different authors have more and less restrictive standards for what a "case" is. Probably a bit of both.
Fully productive manner adverbs like English -ly are an important exception to this convention as they are regularly left outside case paradigms. I'd expect that there's a good syntactic reason for this.
- StrangerCoug
- Avisaru
- Posts: 269
- Joined: Sat Aug 02, 2014 8:56 pm
- Location: El Paso, TX
Re: Grammaticalization Quickie Thread
For example, Japanese uses markers for what seems like a lot of things, yet I come across very few people talking about that language as having cases. vokzhen describes what I'm going for better than I can.gach wrote:Could you clarify what you mean by "something with a marker stuck to it"?StrangerCoug wrote:Now that I think of it, where is the line between a separate case and just something with a marker stuck to it?
I'm not trying to force an English-like noun system on my daughterlangs (if I remember correctly, the inspiration for such an extensive system was Finnish)—my idea was that the most used ones stick around—but I think this rich a starting system is making me think quite a bit about how to go about things.
Re: Grammaticalization Quickie Thread
I believe these are clitics: Syntactically they behave like separate words, but they are phonologically bound like affixes.StrangerCoug wrote: For example, Japanese uses markers for what seems like a lot of things, yet I come across very few people talking about that language as having cases. vokzhen describes what I'm going for better than I can.
This brings up a question; namely, do cases have to be marked with an affix to be called a case? Theoretically, I don't see why they would have to; whether a given case is marked morphologically, through a clitic, a separate word, or is shown purely through word order, it's still the same case isn't it? In the phrase 'in the house', isn't 'house' in the locative case just as much as that noun would be in a language that happens to have case suffixes? It's just that instead of being marked with a case suffix, it is indicated with a preposition. What is the difference? StrangerCoug is right, I don't hear people talking about 'case' much in these contexts. I don't really understand why.
Re: Grammaticalization Quickie Thread
I have wondered this, myself. Case and adpositions seem to just be different strategies for marking the same syntactic information, with case-marking developing from cliticized adpositions via analogical extension onto adjectives and articles.sucaeyl wrote:I believe these are clitics: Syntactically they behave like separate words, but they are phonologically bound like affixes.StrangerCoug wrote: For example, Japanese uses markers for what seems like a lot of things, yet I come across very few people talking about that language as having cases. vokzhen describes what I'm going for better than I can.
This brings up a question; namely, do cases have to be marked with an affix to be called a case? Theoretically, I don't see why they would have to; whether a given case is marked morphologically, through a clitic, a separate word, or is shown purely through word order, it's still the same case isn't it? In the phrase 'in the house', isn't 'house' in the locative case just as much as that noun would be in a language that happens to have case suffixes? It's just that instead of being marked with a case suffix, it is indicated with a preposition. What is the difference? StrangerCoug is right, I don't hear people talking about 'case' much in these contexts. I don't really understand why.
Re: Grammaticalization Quickie Thread
Exactly! I'm guessing something rooted in the traditions of grammatical description... But maybe there is a justification. Anyway, you talk about case affixes ending up on adjectives and articles. Out of curiousity, do you know how widespread this is? That is, of the languages that morphologically mark case on nouns, how many of them show agreement for case on adjectives and articles?TaylorS wrote:I have wondered this, myself. Case and adpositions seem to just be different strategies for marking the same syntactic information, with case-marking developing from cliticized adpositions via analogical extension onto adjectives and articles.
Re: Grammaticalization Quickie Thread
Case and adpositions certainly fall on the same spectrum of strategies for marking grammatical relations and examples of adpositions getting cliticised and developing into bound cases show that the spectrum is a continuous one. Language internally though, adpositions and bound cases are typically distinct enough categories to motivate the use of differentiating terminology between them.
"Case" usually refers to bound morphological marking so it's a bit out of place to begin to call adpositions case marking despite the fact that functionally there would be no distinction between them. Also by the above argument it doesn't make sense to call prepositions case in English since prepositional phrases are structurally clearly different from the use of the genitive 's or cases on pronouns (I ~ me ~ my etc.).sucaeyl wrote:In the phrase 'in the house', isn't 'house' in the locative case just as much as that noun would be in a language that happens to have case suffixes?
Re: Grammaticalization Quickie Thread
I think what we're dealing with is that 'case' has two different meanings depending on context: it can refer to, to quote wikipedia, "a grammatical category whose value reflects the grammatical function performed by a noun or pronoun in a phrase, clause, or sentence". In this sense, 'in the house' is still in the locative case. At the same time, 'case' can refer specifically to affixes and other morphological changes that display the first meaning. Is there a different word that can refer to one of these two meanings? Perhaps just 'syntactic role' for the former? Or, on the other hand, and my preferred option, 'case affix' for the latter, letting 'case' include syntactic roles marked by means other than morphology.