Morphosyntactic alignment

The best topics from Languages & Linguistics, kept on a permanent basis.
cromulent
Sanci
Sanci
Posts: 33
Joined: Fri Jul 14, 2006 5:51 pm
Location: Deep, DEEP in the heart of my conlanging smithy...and it's going to be a long winter.

Post by cromulent »

Iogkwah is fluid-S/dechticaetiative. I think.

However: in the examples given above: "I taught math" and "I taught math to John"--the argument "math" would take the thematic role in both sentences. For "math" to take the play the patient role would mean to give a lesson to math, not about math. So it might not be a prototypical dechticaetitaive language.

Iogkwah expands the role of the theme quite a bit--it's used in some monotransitive clauses. The most common example is that objects of perception are assigned the theme role. Also, monotransitive verbs that could become ditransitive, behave ditransitively. This category includes "creative verbs"--verbs which result in the creation of the object (or the increased salience or prominence of an object)--"definitive verbs"--these are verbs that involve defining the object in some way, or according the object a certain status--and many (if not most) verbs of directed motion. (Creative and definitive verbs are fuzzy categories which blend together, with no strict rules separarting one from the other. Which doesn't matter, because they both behave in the same ditransitive way. The distinction between the two isn't really necessary.). Iogkwah even assigns the theme role to the sole argument in some intransitive clauses.

In other words, Iogkwah has two object cases. Some verbs act on experiencers, some act on themes--and only in explicitly ditransitive sentences (i.e. in which there is a stated, not theoretical recipient) is the one "primary" and the other "secondary."

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Re: Morphosyntactic alignment

Post by TomHChappell »

Haspelmath and Dryer and so on extended the morphosyntactic alignment question, (which is mostly about how intransitive sentences line up with monotransitive sentences), to typologize how languages line up monotransitive sentences with ditransitive sentences. The main answers are "dative" and "dechticaetiative"; or "directive" and "primative"; or "directive/indirective" and "primative/secundative".

However this typology doesn't apply to absolutely every language, since some languages have no ditransitive sentences. (Most languages do have ditransitive sentences; and in most such languages, most verbs are not ditransitive.) Even though most languages do have ditransitive sentences, it doesn't necesssarily follow that most languages have three grammatical relations ("indirect objects" as well as "direct objects", or, "secondary objects" as well as "primary objects").

A different extension of alignment typology was made by Tadeusz Milewski. He asked how "genitive phrases" might be aligned with intransitive and/or monotransitive sentences.

Has anyone on the board looked into this?

Let me abbreviate the semantic roles this way:
In a monovalent clause, the only participant is "S".
In a monotransitive clause, the Agent is "A" and the Patient is "P".
In a "genitive phrase", the Possessor is "G" (for "genitive") and the Possessum or "thing possessed" is "C" (for "construct").

Milewski pointed out that it is quite rare for these five roles to be represented by five cases. For most languages, either S=A or S=P; and for most languages, either S=G or S=C.

Avoiding the issues of "split-X" languages for the time being, and assuming that A=P is ruled out and G=C is also ruled out, there are the following twenty-six possibilities [see * below];
  1. two cases
    1. S=A=G, different from P=C
    2. S=A=C, different from P=G
    3. S=P=G, different from A=C
    4. S=P=C, different from A=G
  2. three cases
    1. S=A=G, different from P, different from C
    2. S=A=C, different from P, different from G
    3. S=P=G, different from A, different from C
    4. S=P=C, different from A, different from G
    5. S=A, different from P=G, different from C
    6. S=A, different from P=C, different from G
    7. S=P, different from A=G, different from C
    8. S=P, different from A=C, different from G
    9. S=G, different from C=A, different from P
    10. S=G, different from C=P, different from A
    11. S=C, different from G=A, different from P
    12. S=C, different from G=P, different from A
    13. A=G, different from P=C, different from S
    14. A=C, different from P=G, different from S
  3. four cases
    1. S=A, different from P, different from G, different from C
    2. S=P, different from A, different from G, different from C
    3. S=G, different from A, different from P, different from C
    4. S=C, different from A, different from P, different from G
    5. A=G, different from S, different from P, different from C
    6. A=C, different from S, different from P, different from G
    7. P=G, different from S, different from A, different from C
    8. P=C, different from S, different from A, different from G
Milewski knew of only six of these that were actually attested**. If I understand correctly, four of the "four cases" systems (3.e through 3.h) and two of the "three cases" systems (2.m and 2.n) would be unlikely. That still leaves twenty systems, however. Also, I'm not sure which of these are actually attested, and of those, which Milewski knew about.**

Who on the ZBBoard knows more about this?** Are there natlangs that actually attest the four "two cases" systems (1.a thru 1.d)? How about the others? Are there any that actually do have five different "cases" for these five roles?

(English, for nouns, is actually a two-case system of a kind not shown here; S=A=P=C, different from G. Of course some people don't believe English has a genitive; for them English's system is S=A=P=G=C. But for pronouns, English is type 2.b. (S=A=C, different from P, different from G).)

[*]If I've made a mistake, please correct me.** Also, even if I didn't make a mistake, if you know of any other systems, please contribute them. Thank you.

[EDIT]:
[**]chris_notts pointed out the Wikipedia article http://en.wikipedia.org/wiki/Milewski's_typology which lists the six patterns Milewski knew about and some of the languages in some of the patterns.
  • Milewski's Class 1: S=A, different from P, different from G. This could be what I called 2.b., or 2.f., or 3.a.
  • Milewski's Class 2: S=P, different from A, different from G. This could be what I called 2.d., or 2.h., or 3.b.
  • Milewski's Class 3: S=A, different from P=G. This could be what I called 1.b., or 2.e. Apparently Indonesiann and Hopi are examples.
  • Milewski's Class 4: S=P, different from A=G. This could be what I called 1.d., or 2.g. Apparently Inuktitut, some Salishan languges, and some Mayan languages, are examples.
  • Milewski's Class 5: S=A=G, different from P. This could be what I called 1.a., or 2.a. Apparently Nass or Niska or Nisga'a is an example.
  • Milewski's Class 6: S=P=G, different from A. This could be what I called 1.c., or 2.c. Apparently Tsimshian, Tunica, and Guarani are examples.
The difference between, on the one hand, this table from the Wikipedia, and, on the other hand, what I described above, lies in the treatment assigned to the "C", or "thing possessed". Possibly I misunderstood Milewski and this is not important; possibly it is important and he, at a different time, used a somewhat finer-grained typology that took this into consideration, though it would have had to be in "genitive clauses" (if that's what they should be called) rather than in genitive phrases, per se.
[/EDIT]
Last edited by TomHChappell on Mon Feb 12, 2007 6:25 pm, edited 4 times in total.

imploder
Niš
Niš
Posts: 9
Joined: Wed Jun 28, 2006 2:00 pm
Location: Brno, Czech Republic
Contact:

Post by imploder »

I don't understand your request to "contrast" S and T; aren't a(n intransitive)subject and a direct object contrasting anyway? :?
Can they both exist in the same sentence as two distinct arguments? Or do they only distinguish some feature (perhaps existence (in the case of theme) or non-existence (in the case experiencer - that's the meaning of "subject" Sander used) of an agent?) of the same thing?

English merges agents with experiencers (Sander called them "subjects" in his post) into something which is unfortunately also called "subject" - the grammatical subject of the sentence. I think it's one of the sources of confusion here. It would be better to call it "experiencer" to avoid the confusion.
[url=http://awkwords.wsr3.net]Awkwords - word generator[/url] | [url=http://awkwords.wsr3.net/help.html]Documentation & download[/url]

cromulent
Sanci
Sanci
Posts: 33
Joined: Fri Jul 14, 2006 5:51 pm
Location: Deep, DEEP in the heart of my conlanging smithy...and it's going to be a long winter.

Post by cromulent »

imploder wrote:
I don't understand your request to "contrast" S and T; aren't a(n intransitive)subject and a direct object contrasting anyway? :?
Can they both exist in the same sentence as two distinct arguments? Or do they only distinguish some feature (perhaps existence (in the case of theme) or non-existence (in the case experiencer - that's the meaning of "subject" Sander used) of an agent?) of the same thing?
We need to make a distinction here between terms as used in this thread, and in common usage. In the terminology set forth by Sander, "Subject" is simply a catch-all term for sole arguments of intransitive clauses--regardless of how any particular language views them.

Given this definition, Subject and Theme can't occur as separate arguments, because in Sander's terminology (which I find immensely lucid), the Subject is the sole argument of an intransitive verb. However, the Subject and the Theme can be the same thing--the Subject can be a Theme. An example of this would be "a book was given." In a dechticaetiative language, that book is the Theme, and the Subject. Likewise, the Subject can be an Agent, or a Patient, or a Recipient, as in "I ran," "I was killed" "I received," respectively. In fact, the Subject has to be an Agent, Patient, Recipient, or Theme. It can't just be a Subject. Because even in intransitive clauses, that Subject is either doing, being done to, or...being done about, I guess. Again, respectively.

Again, it is critical to note that the term "Subject" as used in this discussion has a very specific definition here that is completely different from what your teachers taught you a "subject" was, i.e. an agent.
English merges agents with experiencers (Sander called them "subjects" in his post) into something which is unfortunately also called "subject" - the grammatical subject of the sentence. I think it's one of the sources of confusion here. It would be better to call it "experiencer" to avoid the confusion.
Actually, Sander was setting forth definitions that took the theta roles out of the accusative-dative perspective (allowing a fresh perspective). He was only merging agents and experiencers only insofar as sole arguments of intransitive clauses have to play one theta role or the other. Notice that there is no Subject in transitive clauses--that's because the Subject in English becomes the Agent, whereas the Subject in Basque becomes the Patient. If we were to call the Subject "the Experiencer," that would be to frame the discussion around the ergative approach, treating it as the norm, or defualt. Which obscures our understanding just as much as framing it around the accusative paradigm does. It is much easier to understand this stuff, and talk about it, if as many distinctions as possible are made--even tho langauges tend to do just the opposite.

cromulent
Sanci
Sanci
Posts: 33
Joined: Fri Jul 14, 2006 5:51 pm
Location: Deep, DEEP in the heart of my conlanging smithy...and it's going to be a long winter.

Post by cromulent »

deleted dupie
Last edited by cromulent on Tue Feb 06, 2007 3:06 pm, edited 1 time in total.

cromulent
Sanci
Sanci
Posts: 33
Joined: Fri Jul 14, 2006 5:51 pm
Location: Deep, DEEP in the heart of my conlanging smithy...and it's going to be a long winter.

Post by cromulent »

I just wrote:However, the Subject and the Theme can be the same thing--the Subject can be a Theme. An example of this would be "a book was given." In a dechticaetiative language, that book is the Theme, and the Subject.
I feel the need to clarify this. Using Sander's strict terminology, there are by definition no Themes in intransitive clauses. Only Subjects. Themes only occur in ditransitive clauses.

But! recognize that I'm only talking about what we call these theta roles. The theta roles themselves can all occur monotransitively, but the terminology (Sander's terminology) is what changes depending on the valnecy of the verb.

Observe: in these sentences,

1. "A book was given,"
2. "A book was given by Fred,"
3. "I was given a book,"
4. "Fred gave me a book."

The book plays the theta role of the theme in all 4 sentences. But as for the terminology (how we refer to that theta role for the purpose of this discussion in such a way as to not frame any particular MSA as the default or norm), we'd say the book was, respectively, the Subject, the Patient, the Theme (of a ditransitive sentence with deleted Agent), and the Theme (of a ditransitive sentence with included agent.)

It helps to think of Subject as containing category that breaks down into Agent and Patient in monotransitives, and Patient in turn breaks down into Theme and Recipient in ditransitives.

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

cromulant wrote:It helps to think of Subject as containing category that breaks down into Agent and Patient in monotransitives, and Patient in turn breaks down into Theme and Recipient in ditransitives.
But why couldn't Agent break down into Donor and Recipient?

cromulent
Sanci
Sanci
Posts: 33
Joined: Fri Jul 14, 2006 5:51 pm
Location: Deep, DEEP in the heart of my conlanging smithy...and it's going to be a long winter.

Post by cromulent »

TomHChappell wrote:
cromulant wrote:It helps to think of Subject as containing category that breaks down into Agent and Patient in monotransitives, and Patient in turn breaks down into Theme and Recipient in ditransitives.
But why couldn't Agent break down into Donor and Recipient?
Good point. There really should be six theta roles discussed here, because the 5 we have can't handle conlangs (and Azoyú Tlapanec, the one known natlang) with a Pegative case.

Here's a paper on Tlapanec's case system. I haven't read it yet but I plan to; I'm quite intrigued.

cromulent
Sanci
Sanci
Posts: 33
Joined: Fri Jul 14, 2006 5:51 pm
Location: Deep, DEEP in the heart of my conlanging smithy...and it's going to be a long winter.

Post by cromulent »

Code: Select all

Accusative-Dative:

    S
   /
  A   P
 /     \ 
D   R   T



Accusative-Dechticaetiative:

    S
   /
  A   P
 /   /   
D   R   T



Accusative-Pegative:

    S
   /
  A   P
    \  \
D   R   T



Ergative-Dative:

    S
     \
  A   P
 /     \
D   R   T



Ergative-Dechticaetiative:

    S
     \
  A   P
 /   /   
D   R   T



Ergative-Pegative:

    S
     \
  A   P
   \    \
D   R   T
I just felt the need to do that for some reason.

Of course as I still know next to nothing about the one "Pegative" language that exists; this chart is theoretical; I don't actually know how such a language works, but pegativity is logically the third corner of the ditransitive allignment triangle, accusativity and dechticaetivity being the first two.

There are other possibilities of course; you could have a tripartite system wherein S aligns with either T or R, P aligns with R or T (whichever one S didn't align with), and A aligns with D. And even more screwball possibilities, such as A crisscrossing to align with T while P crisscrosses to align with D. None of these appeal to me.

I haven't finished reading that paper, but it seems that Tlapanec has a Pegative case and a Dative case, in additon to Absolutive and Ergative.

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

cromulant wrote:

Code: Select all

Accusative-Dative:

    S
   /
  A   P
 /     \ 
D   R   T



Accusative-Dechticaetiative:

    S
   /
  A   P
 /   /   
D   R   T



Accusative-Pegative:

    S
   /
  A   P
    \  \
D   R   T



Ergative-Dative:

    S
     \
  A   P
 /     \
D   R   T



Ergative-Dechticaetiative:

    S
     \
  A   P
 /   /   
D   R   T



Ergative-Pegative:

    S
     \
  A   P
   \    \
D   R   T
I just felt the need to do that for some reason.
Thanks, I think we needed that.
cromulant wrote:Of course as I still know next to nothing about the one "Pegative" language that exists;
The only one that is known to exist, or, the only one whose existence has been published. There are 2,000 to 4,000 languages whose grammars haven't been well-studied and well-published (in addition to the 2,000 whose grammars have been); it's not terribly unlikely that a handful (say, a tenth of a percent) of these have such an alignment.
cromulant wrote:this chart is theoretical; I don't actually know how such a language works, but pegativity is logically the third corner of the ditransitive allignment triangle, accusativity and dechticaetivity being the first two.
There are other possibilities of course; you could have a tripartite system wherein S aligns with either T or R, P aligns with R or T (whichever one S didn't align with), and A aligns with D. And even more screwball possibilities, such as A crisscrossing to align with T while P crisscrosses to align with D. None of these appeal to me.
A=T doesn't appeal to me, and P=D doesn't appeal to me, but those others might!
  • A=D,P=T,S=R ("tripartite dative"?)
  • A=D,P=R,S=T ("tripartite dechticaetiative"?)
  • A=R,P=T,S=D ("tripartite pegative"?)
Also the "split" or "fluid" possibilities; having "A=D-or-R" or "P=T-or-R" or even "R=A-or-P".
cromulant wrote:I haven't finished reading that paper
I haven't even really started it! :oops: Obviously I need to.
cromulant wrote:but it seems that Tlapanec has a Pegative case and a Dative case, in additon to Absolutive and Ergative.
Thanks.

cromulent
Sanci
Sanci
Posts: 33
Joined: Fri Jul 14, 2006 5:51 pm
Location: Deep, DEEP in the heart of my conlanging smithy...and it's going to be a long winter.

Post by cromulent »

TomHChappell wrote:
cromulant wrote:Of course as I still know next to nothing about the one "Pegative" language that exists;
The only one that is known to exist, or, the only one whose existence has been published.
Of course.

EDIT:
Also the "split" or "fluid" possibilities; having "A=D-or-R" or "P=T-or-R" or even "R=A-or-P".
Hmm...I'm pretty sure the MSA I've been describing in my conlang allows all of these possibilities; there's a lot of slip-n-slide involved in assigning surface cases to underlying theta roles. It's not quite "maximally fluid" tho in that there's no equivalent of the Pegative case; no way of particularly emphasizing the extra-agentivity of the Donor.

I'm not sure if Iogkwah actually allows R=A either, but I think it does in that a Recipient can be a "Taker" rather than a "Given-unto." What is the theta role of a Taker, and of the Taken-from?

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

cromulant wrote:What is the theta role of a Taker, and of the Taken-from?
Good questions!
In some languages, of course, the "Taken-From" is in the Ablative case. I don't know whether in some such languages "Taker" can be Dative?
There are languages, however, which make "Taker"=D and "Taken-From"=R and assign "Taken-From" the Dative case.
I wonder if there are languages in which "Taken-From" is Pegative? Obviously, if there are, nobody has published a grammar of any of them yet.

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

I have no evidence that any of these are actually attested in natlangs.

Hierarchical Alignment in Ditransitives
I'd like to mention four possibilities.
In each, suppose for each two of the three participants there is a hierarchy on which one is higher than the other -- whether
  • an empathy hierarchy (1st person > 2nd person > 3rd person),
  • a definiteness hierarchy (pronouns > proper nouns > common nouns),
  • a saliency hierarchy (proximate > obviative),
  • an "animacy" or "likelihood-to-control-or-instigate-or-perform-or-effect" hierarchy (human > animate non-human > inanimate),
  • or some other hierarchy I've forgotten
-- such that the three participants are linearly ordered by hierarchy.
  1. First, suppose only one case appears in ditransitives. Suppose the ditransitive verb can appear in any one of six Voices -- Roman numerals I through VI -- such that:
    • In Voice I, the highest participant is the D participant, the middle participant is the R participant, and the lowest participant is the T participant.
    • In Voice II, the highest participant is the D participant, the middle participant is the T participant, and the lowest participant is the R participant.
    • In Voice III, the highest participant is the R participant, the middle participant is the D participant, and the lowest participant is the T participant.
    • In Voice IV, the highest participant is the R participant, the middle participant is the T participant, and the lowest participant is the D participant.
    • In Voice V, the highest participant is the T participant, the middle participant is the D participant, and the lowest participant is the R participant.
    • In Voice VI, the highest participant is the T participant, the middle participant is the R participant, and the lowest participant is the D participant.
  2. Second, suppose only two cases, Pegative and Accusative, appear in ditransitives. Suppose the D participant is always Pegative, and the R and T participants are always Accusative. Suppose the ditransitive verb can appear in either of two Voices -- "Ditransitive-Direct" or "Ditransitive-Inverse" -- such that:
    • In the Ditransitive-Direct Voice, the R participant is higher than the T participant.
    • In the Ditransitive-Inverse Voice, R participant is lower than the T participant.
  3. Third, suppose only two cases, Ergative and Secundative, appear in ditransitives. Suppose the T participant is always Secundative, and the D and R participants are always Ergative. Suppose the ditransitive verb can appear in either of two Voices -- "Ditransitive-Direct" or "Ditransitive-Inverse" -- such that:
    • In the Ditransitive-Direct Voice, the D participant is higher than the R participant.
    • In the Ditransitive-Inverse Voice, D participant is lower than the R participant.
  4. Fourth, suppose only two cases, Dative and Weirdative, appear in ditransitives. Suppose the R participant is always Dative, and the D and T participants are always Weirdative. Suppose the ditransitive verb can appear in either of two Voices -- "Ditransitive-Direct" or "Ditransitive-Inverse" -- such that:
    • In the Ditransitive-Direct Voice, the D participant is higher than the T participant.
    • In the Ditransitive-Inverse Voice, D participant is lower than the T participant.
"Austronesian/Philippine"-Like Alignment Between Ditransitives and Monotransitives
I'd like to mention three possibilities; they probably aren't the only ones.
In each, I'm ignoring the Monotransitive-to-Intransitive MSA. In each, I'm supposing there are four cases that can appear in ditransitive clauses. In each, suppose the case that the A participant of a monotransitive clause can be marked with is called Ergative (even though it might be Nominative), and the case that the P participant of a monotransitve clause can be marked with is called Accusative (even though it might be Absolutive). In each, suppose that in any ditransitive clause, one participant is marked Ergative and one participant is marked Accusative.
  1. First, the strangest. Suppose the four cases that can occur in ditransitive clauses are Pegative, Ergative, Accusative, and Secundative. If a ditransitive clause has a Pegative participant, then the Pegative participant is the D participant, the Ergative participant is the R participant, and the Accusative participant is the T participant. But if a ditransitive clause has a Secundative participant, then the Ergative participant is the D participant, the Accuasative participant is the R participant, and the Secundative participant is the T participant.
  2. Second, suppose the four cases that can occur in ditransitive clauses are Pegative, Ergative, Accusative, and Dative. If a ditransitive clause has a Pegative participant, then the Pegative participant is the D participant, the Ergative participant is the R participant, and the Accusative participant is the T participant. But if a ditransitive clause has a Dative participant, then the Ergative participant is the D participant, the Dative participant is the R participant, and the Accusative participant is the T participant. In either case, the Accusative participant is the T participant.
  3. Third, suppose the four cases that can occur in ditransitive clauses are Dative, Ergative, Accusative, and Secundative. If a ditransitive clause has a Dative participant, then the Ergative participant is the D participant, the Dative participant is the R participant, and the Accusative participant is the T participant. But if a ditransitive clause has a Secundative participant, then the Ergative participant is the D participant, the Accusative participant is the R participant, and the Secundative participant is the T participant. In either case, the Ergative participant is the D participant.
Last edited by TomHChappell on Sat Feb 10, 2007 4:45 pm, edited 1 time in total.

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

Generalities
  • If a language has ditransitive clauses, there has to be some way to distinguish, in ditransitive clauses, between, on the one hand, the D participant, and on the other hand, the R participant. If this means is not morphosyntactic, it is probably hierarchical. But if this means is morphosyntactic, then either there is a case that can mark the D but can't mark the R, or there is a case that can mark the R but can't mark the D.
  • If a language has ditransitive clauses, there has to be some way to distinguish, in ditransitive clauses, between, on the one hand, the D participant, and on the other hand, the T participant. If this means is not morphosyntactic, it is probably hierarchical. But if this means is morphosyntactic, then either there is a case that can mark the D but can't mark the T, or there is a case that can mark the T but can't mark the D.
  • If a language has ditransitive clauses, there has to be some way to distinguish, in ditransitive clauses, between, on the one hand, the R participant, and on the other hand, the T participant. If this means is not morphosyntactic, it is probably hierarchical. But if this means is morphosyntactic, then either there is a case that can mark the R but can't mark the T, or there is a case that can mark the T but can't mark the R.
  • Every language has monotransitive clauses. There has to be some way to distinguish, in monotransitive clauses, between, on the one hand, the A participant, and on the other hand, the P participant. If this means is not morphosyntactic, it is probably hierarchical. But if this means is morphosyntactic, then either there is a case that can mark the A but can't mark the P, or there is a case that can mark the P but can't mark the A.
  • Every language has intransitive clauses. There has to be a case that can mark the S.
What I've Already Considered
  • What if there's no case that can mark the D but not the R, no case that can mark the D but not the T, no case that can mark the R but not the D, no case that can mark the R but not the T, no case that can mark the T but not the D, and no case that can mark the T but not the R?
  • What if there's no case that can mark the R but not the T, and no case that can mark the T but not the R?
  • What if there's no case that can mark the D but not the R, and no case that can mark the R but not the D?
  • What if there's no case that can mark the D but not the T, and no case that can mark the T but not the D?
  • What if there's no case that can mark the R but cannot also mark either (both the D and the A) or (both the T and the P)?
  • What if there's no case that can mark the D but cannot also mark both the R and the A?
  • What if there's no case that can mark the T but cannot also mark both the R and the P?
Simplifying Assumptions
  1. Let's not consider the possibility that some case can mark both the D and the R.
  2. Let's not consider the possibility that some case can mark both the D and the T.
  3. Let's not consider the possibility that some case can mark both the R and the T.
  4. Let's not consider the possibility that some case can mark both the A and the P.
  5. Let's assume that there must be some case that can mark the D but cannot mark the R and cannot mark the T.
  6. Let's assume that there must be some case that can mark the R but cannot mark the D and cannot mark the T.
  7. Let's assume that there must be some case that can mark the T but cannot mark the D and cannot mark the R.
  8. Let's assume that there must be some case that can mark the A but cannot mark the P.
  9. Let's assume that there must be some case that can mark the P but cannot mark the A.
  10. Let's assume that there must be some case that can mark the S.
If we assume all that, there have to be at least three cases, because some case can mark the D but not the R nor the T, some case can mark the R but not the D nor the T, and some case can mark the T but not the D nor the R.
If we further assume that
  1. There is no case that can mark both the D and the P.
  2. There is no case that can mark both the T and the A.
  3. There are only three cases that can mark the core terms of intransitive, monotransitive, and ditransitive clauses.
then the following are true:
  • Either the case that marks the D also marks the A; or the case that marks the R also marks the A; or both.
  • Either the case that marks the R also marks the P; or the case that marks the T also marks the P; or both.
  • Either the case that marks the D also marks the S; or the case that marks the R also marks the S; or the case that marks the T also marks the S; or some combination of two of those three possibilities; or all three of them.
Definitions of "Split-X" or "Fluid-X"
Definition: Suppose X is one of D, R, T, A, or P, and suppose Y is a different one of those five. If there is a case that can mark both an X and an S but not a Y, and there is also a case that can mark both a Y and an S but not an X, then we say the system is a "split-S" system.

Definition: Suppose X is one of D, R, T, or S, and suppose Y is a different one of those four. If there is a case that can mark both an X and an A but not a Y, and there is also a case that can mark both a Y and an A but not an X, then we say the system is a "split-A" system.

Definition: Suppose X is one of D, R, T, or S, and suppose Y is a different one of those four. If there is a case that can mark both an X and an P but not a Y, and there is also a case that can mark both a Y and an P but not an X, then we say the system is a "split-P" system.

Definition: Suppose X is one of A, P, or S, and suppose Y is a different one of those three. If there is a case that can mark both an X and an D but not a Y, and there is also a case that can mark both a Y and an D but not an X, then we say the system is a "split-D" system.

Definition: Suppose X is one of A, P, or S, and suppose Y is a different one of those three. If there is a case that can mark both an X and an R but not a Y, and there is also a case that can mark both a Y and an R but not an X, then we say the system is a "split-R" system.

Definition: Suppose X is one of A, P, or S, and suppose Y is a different one of those three. If there is a case that can mark both an X and a T but not a Y, and there is also a case that can mark both a Y and a T but not an X, then we say the system is a "split-T" system.

Morphosyntactic Alignments Possible Under Our Simplifying Assumptions
Each of the three cases can mark one and only one of the three participants of a ditransitive clause; each participant of a ditransitive clause can be marked with one and only one of the three cases.

For consistency and simplicity, therefore, I am going to use the following terms:
  1. "Case 1" is whatever case can mark the D but cannot mark the R nor the T.
  2. "Case 2" is whatever case can mark the R but cannot mark the D nor the T.
  3. "Case 3" is whatever case can mark the T but cannot mark the D nor the R.
There are five (5) ways the ditransitive clauses can line up with the monotransitive clauses (note some of these are attested by natlangs but not all of them are, as far as I know):
  1. "Case 1" can mark the D and the A; "Case 2" can mark the R, but not the A nor the P; "Case 3" can mark the T and the P.
  2. "Case 1" can mark the D and the A; "Case 2" can mark the R and the P; "Case 3" can mark the T but not the P.
  3. "Case 1" can mark the D and the A; "Case 2" can mark the R and the P; "Case 3" can mark the T and the P (so this is a "split-P" alignment).
  4. "Case 1" can mark the D, but not the A; "Case 2" can mark the R and the A; "Case 3" can mark the T and the P.
  5. "Case 1" can mark the D and the A; "Case 2" can mark the R and the A (so this is a "split-A" alignment); "Case 3" can mark the T and the P.
[EDIT 3:]Any language that allows Dative-marked As in monotransitive clauses might be considered a "split-A" language; there are several such natlangs. Any language that allows Dative-marked Ps in monotransitive clauses might be considered a "split-P" language; there are several such natlangs.[/EDIT 3:]

There are seven (7) ways the ditransitive clauses can line up with the intransitive clauses (note some of these are attested by natlangs but not all of them are, as far as I know):
  1. "Case 1" can mark the D and the S; "Case 2" can mark the R but not the S; "Case 3" can mark the T but not the S.
  2. "Case 1" can mark the D but not the S; "Case 2" can mark the R and the S; "Case 3" can mark the T but not the S.
  3. "Case 1" can mark the D but not the S; "Case 2" can mark the R but not the S; "Case 3" can mark the T and the S.
  4. "Case 1" can mark the D and the S; "Case 2" can mark the R and the S (so this is a "split-S" alignment); "Case 3" can mark the T but not the S.
  5. "Case 1" can mark the D and the S; "Case 2" can mark the R but not the S; "Case 3" can mark the T and the S (so this is a "split-S" alignment).
  6. "Case 1" can mark the D but not the S; "Case 2" can mark the R and the S; "Case 3" can mark the T and the S (so this is a "split-S" alignment).
  7. "Case 1" can mark the D and the S; "Case 2" can mark the R and the S; "Case 3" can mark the T and the S (so this is a "3-way split-S" alignment).
All this being so, there are thirty-five (35) ways the ditransitive, monotransitive, and intransitive roles can line up; some of these are attested in natlangs, but, as far as I know, not all of them:
[EDIT 3]:
  • Added case names to following table.
  • Added names of monotransitive-to-intransitive alignment systems to following table.
  • Added names of ditransitive-to-monotransitive alignment systems to following table.
Note that the "name" of the system does not completely specify the system.
[/EDIT 3:]
  1. Accusative Directive
    1. Nominative D, A, S
    2. Dative R
    3. Accusative T, P
  2. Ergative Directive
    1. Ergative D, A
    2. Dative R
    3. Absolutive T, P, S
  3. Tripartite Directive
    1. Ergative D, A
    2. Dative R, S
    3. Accusative T, P
  4. Accusative Directive
    1. Nominative D, A, S
    2. Dative R, S
    3. Accusative T, P
  5. Active Directive
    1. Nominative D, A, S
    2. Dative R
    3. Absolutive T, P, S
  6. Ergative Directive
    1. Ergative D, A
    2. Dative R, S
    3. Absolutive T, P, S
  7. Active Directive
    1. Nominative D, A, S
    2. Dative R, S
    3. Absolutive T, P, S
  8. Accusative Dechticaetiative
    1. Nominative D, A, S
    2. Accusative R, P
    3. Secundative T
  9. Tripartite Dechticaetiative
    1. Ergative D, A
    2. Accusative R, P
    3. Secundative T, S
  10. Ergative Dechticaetiative
    1. Ergative D, A
    2. Absolutive R, P, S
    3. Secundative T
  11. Active Dechticaetiative
    1. Nominative D, A, S
    2. Absolutive R, P, S
    3. Secundative T
  12. Accusative Dechticaetiative
    1. Nominative D, A, S
    2. Accusative R, P
    3. Secundative T, S
  13. Ergative Dechticaetiative
    1. Ergative D, A
    2. Absolutive R, P, S
    3. Secundative T, S
  14. Active Dechticaetiative
    1. Nominative D, A, S
    2. Absolutive R, P, S
    3. Secundative T, S
  15. Accusative Split-P
    1. Nominative D, A, S
    2. Primative R, P
    3. Accusative T, P
  16. Ergative Split-P
    1. Ergative D, A
    2. Accusative R, P
    3. Absolutive T, P, S
  17. Ergative Split-P
    1. Ergative D, A
    2. Absolutive R, P, S
    3. Accusative T, P
  18. Active Split-P
    1. Nominative D, A, S
    2. Absolutive R, P, S
    3. Accusative T, P
  19. Active Split-P
    1. Nominative D, A, S
    2. Accusative R, P
    3. Absolutive T, P, S
  20. Ergative Split-P
    1. Ergative D, A
    2. Primative R, P, S
    3. Absolutive T, P, S
  21. Active Split-P
    1. Nominative D, A, S
    2. Primative R, P, S
    3. Absolutive T, P, S
  22. Tripartite Directive Pegative
    1. Pegative D, S
    2. Ergative R, A
    3. Accusative T, P
  23. Ergative Directive Pegative
    1. Pegative D
    2. Ergative R, A
    3. Absolutive T, P, S
  24. Accusative Directive Pegative
    1. Pegative D
    2. Nominative R, A, S
    3. Accusative T, P
  25. Accusative Directive Pegative
    1. Pegative D, S
    2. Nominative R, A, S
    3. Accusative T, P
  26. Ergative Directive Pegative
    1. Pegative D, S
    2. Ergative R, A
    3. Absolutive T, P, S
  27. Active Directive Pegative
    1. Pegative D
    2. Nominative R, A, S
    3. Absolutive T, P, S
  28. Active Directive Pegative
    1. Pegative D, S
    2. Nominative R, A, S
    3. Absolutive T, P, S
  29. Accusative Directive Split-A
    1. Nominative D, A, S
    2. Ergative R, A
    3. Accusative T, P
  30. Ergative Directive Split-A
    1. Ergative D, A
    2. Dative R, A
    3. Absolutive T, P, S
  31. Accusative Directive Split-A
    1. Ergative D, A
    2. Nominative R, A, S
    3. Accusative T, P
  32. Accusative Directive Split-A
    1. Nominative D, A, S
    2. Dative R, A, S
    3. Accusative T, P
  33. Active Directive Split-A
    1. Nominative D, A, S
    2. Ergative R, A
    3. Absolutive T, P, S
  34. Active Directive Split-A
    1. Ergative D, A
    2. Nominative R, A, S
    3. Absolutive T, P, S
  35. Active Directive Split-A
    1. Nominative D, A, S
    2. Dative R, A, S
    3. Absolutive T, P, S
[ADDED IN SECOND EDIT]:
Names of Cases (Using Our Simplifying Assumptions)
  • Definition: Nominative. A case that can mark both the A and the S (but not the P) may be called "Nominative".
  • Definition: Ergative. A case that can mark the A but not the S (nor the P) may be called "Ergative".
  • Definition: Absolutive. A case that can mark both the P and the S (but not the A) may be called "Absolutive".
  • Definition: Accusative. A case that can mark the P but not the S (nor the A) may be called "Accusative".
  • Definition: Pegative. A case that can mark the D but not the A (nor the R nor the T nor the P) will be called "Pegative".
  • Definition: Primative. A case that can mark the R and the P (but not the D nor the T) may be called "Primative".
  • Definition: Dative. A case that can mark the R but not the P (nor the D nor the T) may be called "Dative".
  • Definition: Secundative. A case that can mark the T but not the P (nor the D nor the R nor the A) will be called "Secundative".
    • If Case 1 can mark the D and the A and the S, let's call Case 1 "Nominative".
    • If Case 1 can mark the D and the A but not the S, let's call Case 1 "Ergative".
    • If Case 1 can mark the D but not the A, let's call Case 1 "Pegative", regardless of whether it can or cannot mark the S.
    • If Case 2 can mark the R and the A and the S, let's call Case 2 "Nominative", unless Case 1 is already "Nominative"; in which event let's call Case 2 "Dative".
    • If Case 2 can mark the R and the A but not the S, let's call Case 2 "Ergative", unless Case 1 is already "Ergative"; in which event let's call Case 2 "Dative".
    • If Case 2 can mark the R and the P and the S, let's call Case 2 "Absolutive", unless Case 3 would also be "Absolutive", in which event, let's call Case 2 "Primative".
    • If Case 2 can mark the R and the P but not the S, let's call Case 2 "Accusative", unless Case 3 would also be "Accusative", in which event, let's call Case 2 "Primative".
    • If Case 2 can mark the R but not the A nor the P, let's call Case 2 "Dative", regardless of whether it can or cannot mark the S.
    • If Case 3 can mark the T and the P and the S, let's call Case 3 "Absolutive".
    • If Case 3 can mark the T and the P but not the S, let's call Case 3 "Accusative".
    • If Case 3 can mark the T but not the P, let's call Case 3 "Secundative", regardless of whether it can or cannot mark the S.
[/ADDED IN SECOND EDIT]
[EDIT 3]:I used the following definitions:
  • Intransitive-to-Monotransitive Alignment:
    • Accusative: Some case marks both the S and the A but no case marks both the S and the P.
    • Ergative: Some case marks both the S and the P but no case marks both the S and the A.
    • Active: Some case marks both the S and the A and some case marks both the S and the P.
    • Tripartite: No case marks both the S and the A and no case marks both the S and the P.
  • Monotransitive-to-Ditransitive Alignment:
    • Directive: Some case marks both the P and the T but no case marks both the P and the R.
    • Dechticaetiative: Some case marks both the P and the R but no case marks both the P and the T.
    • Split-P: Some case marks both the P and the T and some case marks both the P and the R.
    • Pegative: Some case marks both the A and the R but no case marks both the A and the D.
    • Split-A: Some case marks both the A and the D and some case marks both the A and the R. (Note: under our simplifying assumptions every split-A system is also Directive.)
[/EDIT 3]
Last edited by TomHChappell on Mon Aug 13, 2007 7:41 pm, edited 5 times in total.

cromulent
Sanci
Sanci
Posts: 33
Joined: Fri Jul 14, 2006 5:51 pm
Location: Deep, DEEP in the heart of my conlanging smithy...and it's going to be a long winter.

Post by cromulent »

Spectacular! Now create a conlang that uses all 35 MSA's, each in highly specific situations, i.e. there is one right way and 34 (very) wrong ways to use each one: Begin.

chris_notts
Avisaru
Avisaru
Posts: 275
Joined: Wed Dec 15, 2004 9:05 am
Location: Nottingham, England
Contact:

Re: Morphosyntactic alignment

Post by chris_notts »

TomHChappell wrote: A different extension of alignment typology was made by Tadeusz Milewski. He asked how "genitive phrases" might be aligned with intransitive and/or monotransitive sentences.

Has anyone on the board looked into this?

Let me abbreviate the semantic roles this way:
In a monovalent clause, the only participant is "S".
In a monotransitive clause, the Agent is "A" and the Patient is "P".
In a "genitive phrase", the Possessor is "G" (for "genitive") and the Possessum or "thing possessed" is "C" (for "construct").

Milewski pointed out that it is quite rare for these five roles to be represented by five cases. For most languages, either S=A or S=P; and for most languages, either S=G or S=C.
I'm not quite sure I understand what this typology is trying to say. For example, most case marking languages I can think of don't exhibit EITHER pattern as a general rule. For example, take Basque:

gizonaren liburua
man-GEN book-DEF.ABS

This seems to exhibit the pattern. But then we have:

gizonaren liburuak
man-GEN book-ERG

gizonaren liburuaz
man-GEN book-INS

gizonaren liburuarekin
man-GEN book-COM

And so on. The point is that the possessed noun "book" can take any appropriate case ending... it is not restricted to taking the absolutive simply because it is possessed. Similarly, so far as I know, for most other case marking languages. In Basque the genitive ending is also distinct from the absolutive ending (similarly in Latin and many other case marking languages).
There are also clear exceptions where the G or C case is the same as one of the cases used to mark A and P, but it is NOT the case which covers S. An example of this, IIRC, is Inuit, where the ergative is also a genitive. This pattern is extremely common in ergative languages incidentally... many ergative languages use their ergative case to also encode either instrumentals, possessors, or both.
Am I missing something here?

EDIT: You've covered the A=G case in your list, so forget I mentioned that. I'm still at a loss to understand the statement that C is patterned after S in many languages though... this hasn't been my experience at all, nor do I know that many languages where G patterns after S.
Try the online version of the HaSC sound change applier: http://chrisdb.dyndns-at-home.com/HaSC

chris_notts
Avisaru
Avisaru
Posts: 275
Joined: Wed Dec 15, 2004 9:05 am
Location: Nottingham, England
Contact:

Post by chris_notts »

I've found the article about the typology on wikipedia but it does not include the construct state in its description of his typology:

http://en.wikipedia.org/wiki/Milewski's_typology

Is this an omission from the wiki, or was the typology later revised?
Try the online version of the HaSC sound change applier: http://chrisdb.dyndns-at-home.com/HaSC

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

cromulant wrote:Spectacular! Now create a conlang that uses all 35 MSA's, each in highly specific situations, i.e. there is one right way and 34 (very) wrong ways to use each one: Begin.
:wink:
You first! :mrgreen:

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Re: Morphosyntactic alignment

Post by TomHChappell »

chris_notts wrote:I'm not quite sure I understand what this typology is trying to say.
The best cure for that is to read Milewski's own work, or a translation of it into a language you know.
chris_notts wrote:For example, most case marking languages I can think of don't exhibit EITHER pattern as a general rule.
Then either I or the Wikipedia have not explained it correctly or you haven't understood one of us or one of us didn't understand Milewski (probably some combination);[Edit] or Milewski didn't explain himself well, since his L1 was Polish but he was writing in French (since French isn't my L1 either, it could be my misunderstanding as much as or more than his misexplaining) [/Edit]; or Milewski was just wrong, which I highly doubt. In any case most languages seem to use at most three "cases" to cover these five roles: agent, patient, possessor (or "nominal attribute"), possessum, and sole participant of an intransitive clause.
chris_notts wrote:For example, take Basque:

gizonaren liburua
man-GEN book-DEF.ABS

This seems to exhibit the pattern. But then we have:

gizonaren liburuak
man-GEN book-ERG

gizonaren liburuaz
man-GEN book-INS

gizonaren liburuarekin
man-GEN book-COM

And so on. The point is that the possessed noun "book" can take any appropriate case ending... it is not restricted to taking the absolutive simply because it is possessed.
What about in sentences such as "The book is the man's" or "That is the man's book"? Naturally if the entire genitive phrase "man's book" is an NP in some clause of a cased language, somehow the case appropriate to that NPs role in that clause must be marked on some part of the NP somehow. I'm sure Milewski was only talking about clauses in which the only nouns were the nominal attribute and the nominal to which it was an attribute.
chris_notts wrote:Similarly, so far as I know, for most other case marking languages. In Basque the genitive ending is also distinct from the absolutive ending (similarly in Latin and many other case marking languages).
Does my previous answer help, or did I shed no light?
chris_notts wrote:There are also clear exceptions where the G or C case is the same as one of the cases used to mark A and P, but it is NOT the case which covers S.
I thought I mentioned some. Is this what you meant by your "Edit" remark?
chris_notts wrote:An example of this, IIRC, is Inuit, where the ergative is also a genitive. This pattern is extremely common in ergative languages incidentally... many ergative languages use their ergative case to also encode either instrumentals, possessors, or both.
Am I missing something here?
EDIT: You've covered the A=G case in your list, so forget I mentioned that.

chris_notts wrote:I'm still at a loss to understand the statement that C is patterned after S in many languages though...
For instance, English. In cased languages in which the case-ending for the S is "zero" and the case ending for "the thing possessed" (C) is also "zero", they pattern together.
chris_notts wrote:this hasn't been my experience at all,
I think it has! Am I wrong?
chris_notts wrote:nor do I know that many languages where G patterns after S.
I can't think of any either, but Milewski's patterns 5 and 6 do that, and he didn't include any patterns he didn't know were attested; so somewhere he published some exemplar-languages of each pattern.
[EDIT]:
Wikipedia wrote:Languages of the 5th class use the genitive not only for the nominal attribute but also for the agent and the experiencer (the "a" marker). The other case, called the accusative, marks only the patient (the "b" marker).
The only language of this class mentioned by Milewski is Nass (Niska, Nisga'a) of the Tsimshianic family.
Languages of the 6th class use the genitive not only for the nominal attribute but also for the experiencer and the patient (the "a" marker"). The other case, the ergative, is used for the agent (the "b" marker).
This group is not too numerous: Tsimshian, Tunica and Guarani belong here.
Retrieved from "http://en.wikipedia.org/wiki/Milewski%27s_typology"
So Nass or Niska or Nisga'a has S=A=G, different from P; and Tsimshian, Tunica, and Guarani have S=P=G, different from A.
[/EDIT]
chris_notts wrote:I've found the article about the typology on wikipedia but it does not include the construct state in its description of his typology:
http://en.wikipedia.org/wiki/Milewski's_typology
Milewski's book is probably a better source.
chris_notts wrote:Is this an omission from the wiki,
Maybe so. His book definitely mentions "la chose possedee" and the possibility that it patterns with the S.
chris_notts wrote:or was the typology later revised?
Not to the best of my knowledge. But the table in the Wikipedia looks familiar; so maybe Milewski wrote more than one article, one of which has that table, and one of which mentions (in French) some concept equivalent to "construct state".

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

cromulant wrote: ... snip ... allows all of these possibilities; there's a lot of slip-n-slide involved in assigning surface cases to underlying theta roles. It's not quite "maximally fluid" ... snip ...
I have thought of another four-case system that's even more fluid.

To show off my truly astounding creativity :mrgreen: , I'm going to name the cases thus:
  1. Alphative
  2. Betative
  3. Gammative
  4. Deltative
All four cases can be used in all three types of clause, but no clause can contain two or more participants marked with the same case as each other.

Here's how the cases are used:
  • If there's an Alphative-marked participant,
    • in a ditransitive clause, it's the D;
    • in a monotransitive clause, it's the A;
    • in an intransitive clause, it's the S.
  • If there's a Betative-marked participant, its use depends on whether or not there is also an Alphative-marked participant.
    • If there is an Alphative-marked participant, then the meaning of the Betative-marked participant is as follows:
      • in a ditransitive clause, it's the R;
      • in a monotransitive clause, it's the P.
    • If there is no Alphative-marked participant, then the meaning of the Betative-marked participant is as follows:
      • in a ditransitive clause, it's the D;
      • in a monotransitive clause, it's the A;
      • in an intransitive clause, it's the S.
  • If there's a Gammative-marked participant, its use depends on whether or not there is also a Deltative-marked participant.
    • If there is a Deltative-marked participant, then the meaning of the Gammative-marked participant is as follows:
      • in a ditransitive clause, it's the R;
      • in a monotransitive clause, it's the A.
    • If there is no Deltative-marked participant, then the meaning of the Gammative-marked participant is as follows:
      • in a ditransitive clause, it's the T;
      • in a monotransitive clause, it's the P;
      • in an intransitive clause, it's the S.
  • If there's a Deltative-marked participant,
    • in a ditransitive clause, it's the T;
    • in a monotransitive clause, it's the P;
    • in an intransitive clause, it's the S.
So, to recapitulate,
  • Four case frames for ditransitive clauses:
    • Alphative=D, Betative=R, Gammative=T
    • Alphative=D, Betative=R, Deltative=T
    • Alphative=D, Gammative=R, Deltative=T
    • Betative=D, Gammative=R, Deltative=T
  • Six case frames for monotransitive clauses:
    • Alphative=A, Betative=P
    • Alphative=A, Gammative=P
    • Alphative=A, Deltative=P
    • Betative=A, Gammative=P
    • Betative=A, Deltative=P
    • Gammative=A, Deltative=P
  • Four case frames for intransitive clauses:
    • Alphative=S
    • Betative=S
    • Gammative=S
    • Deltative=S
Assume that semantic or pragmatic or semantico-pragmatic considerations dictate the choice of case frame.
  • D can be marked with either of two cases, Alphative or Betative;
  • R can be marked with either of two cases, Betative or Gammative;
  • T can be marked with either of two cases, Gammative or Deltative;
  • A can be marked with any of three cases, Alphative or Betative or Gammative;
  • P can be marked with any of three cases, Betative or Gammative or Deltative;
  • S can be marked with any of four cases, Alphative or Betative or Gammative or Deltative.
The Betative case can be used for any role but the T; the Gammative case can be used for any role but the D.

I am aware of natlangs with a three-way split S, but not with a four-way split S as here.
I am aware of natlangs with a two-way split A, but not with a three-way split A as here.
I am aware of natlangs with a two-way split P, but not with a three-way split P as here.
I believe there are natlangs with both a split A and a split P, as here.
I am not aware of any natlangs with a split D nor a split R nor a split T, as here.

How's that for fluidity? :wink:

Gond
Sanci
Sanci
Posts: 21
Joined: Sun Jan 16, 2005 3:06 pm
Location: Lisbon, Portugal

Post by Gond »

I'm not very much into this..., but I feel interested..., I just can't read all this because it's too much ahead of me..., so could anyone just tell how would reflexive sentences be done? The language I'm trying to build is absolutive-ergative dechticaetative(?)...

Thanks in advance :)
Non omne id quod fulget aurum est.

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

Gond wrote:I'm not very much into this..., but I feel interested..., I just can't read all this because it's too much ahead of me..., so could anyone just tell how would reflexive sentences be done? The language I'm trying to build is absolutive-ergative dechticaetative(?)...

Thanks in advance :)
I think this is a really good question.
There may be other threads, possibly in this forum or possibly in the L&L or C&C forum(s), which have some information you'll want.
There are basically two kinds of ways to make things reflexive; you can make "reflexive" be a person, and apply it to pronouns, or you can make "reflexive" be a voice, and apply it to verbs.

If you don't alter the voice or valency of the verb, you can fill in one of the slots with a "reflexive pronoun" or "reflexive anaphor" referring to the occupant of another slot. This will require that the language have "the right kind" of reflexive pronoun; namely, a reflexive that can have any clausemate as an antecedent provided the antecedent is higher on some hierarchy than the slot filled by the reflexive pronoun.
The usual hierarchy is one of grammatical relations;
subject > direct object > indirect object > oblique argument > possessor > object-of-comparison.
(Note this can be refined; possessors are ranked in order of the thing possessed, like
possessor-of-subject > possessor-of-direct-object > possessor-of-indirect-object > possessor-of-oblique-argument > possesor-of-possessor > possessor-of-object-of-comparison.)

But Tagalog and some other languages use a hierarchy of semantic roles; I'm not sure what it is, but it may be something like agent > patient > recipient > beneficiary or maleficiary > causee ("agent of cause") > instrument > location. (Do not trust that list; I don't know how many mistakes I just made. But you can look it up.)

On the other hand, if you don't have two different NPs referring to the same participant, then "reflexivization" counts as a valency-reducing transformation; it probably also counts as a transitivity-reducing transformation. In languages that employ such a strategy, the verb is morphologically marked as being in a reflexive voice.

Examples (made up; not necessarily part of any complete conlang or natlang):
"Aggie bathed Pat" is transitive. Aggie is the Agent, Pat is the Patient.

"Sue bathed-REFL" is intransitive (and reflexive). Sue, the Subject, is both the agent and the patient. The "-REFL" morpheme marks the verb as being in the reflexive voice; that is, the form of the verb lets the addressee know that the subject is both the agent and the patient.

"Aggie bathed herself" is transitive; but it's reflexive because we all know that the pronoun "herself" refers to Aggie.

-------------------------------------------------------------------------------------

In some languages, the rules on use of reflexive pronouns require not only that its antecedent be higher in some hierarchy, but also that its antecedent appear in the clause before the reflexive pronoun.

-------------------------------------------------------------------------------------

If the basic clause one wishes to reflexivize is ditransitive, there are three different things that one can want done. In case the language has the clausemates-only-but-any-higher-role type of reflexives, these can easily be done;
D=R reflexives; "Donny showed Theresa to himself". ("himself" in the Recipient slot co-refers with "Donny" in the Donor slot).
D=T reflexives; "Donny showed himself to Regina". ("himself" in the Theme slot co-refers with "Donny" in the Donor slot).
R=T reflexives; "Donny showed Regina herself". ("herself" in the Theme slot co-refers with "Regina" in the Recipient slot.)

But if it doesn't have that kind of reflexive, it probably has the "subjects-only-but-any-containing-clause" kind of "optionally long-distance" reflexive. This could still work for two of the above clauses, because the antecedent is Donny, the subject.

However, to make R=T reflexives, you would first have to make some voice alteration that would promote either the R or the T to subject; for instance,
"Regina was shown herself by Donny". ("herself" has to co-refer with the subject, which is Regina.)

-------------------------------------------------------------------------------------

Another interesting transformation is the reciprocal.
If "V" is a typical transitive verb and "A V P" is a typical active-voice transitive clause, with two valents, "A" and "P", then:
"A V-REFL" would mean "A V A"; this is the reflexive transformation:
"A&P V-RECIP" would mean "A V P, and P V A" -- this is the reciprocal transformation.

When the clause has more than two slots, interesting complications arise when reciprocation ensues. These are the same complications that arise for the reflexive transformation.
"D married R&T to each other" means "D married R to T and D married T to R".
"D&R gave Ts to each other" means "D gave R a T, and R gave D a T."
"D&T ratted each other out to R" means "D ratted T out to R, and T ratted D out to R".

But when a ditransitive verb undergoes both reflexivization and reciprocation, things can get really complicated. For one thing, it is important to take care whether one must apply the reflexivization first, and then the reciprocation, or whether the other-way-round is meant.
Last edited by TomHChappell on Mon Mar 26, 2007 5:10 pm, edited 1 time in total.

Gond
Sanci
Sanci
Posts: 21
Joined: Sun Jan 16, 2005 3:06 pm
Location: Lisbon, Portugal

Post by Gond »

Thanks, now I understand it a lot better..., though there are a lot of things there that have put me on my nerves..., some serious stuff there... :P
Non omne id quod fulget aurum est.

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

Gond wrote:..., though there are a lot of things there that have put me on my nerves...
1) Like, which things?
2) What do you mean exactly by "put me on my nerves"? Do you mean they make you nervous (anxious or uneasy)? or do you mean they "get on your nerves" (annoy you or piss you off)? or something else?

Gond
Sanci
Sanci
Posts: 21
Joined: Sun Jan 16, 2005 3:06 pm
Location: Lisbon, Portugal

Post by Gond »

TomHChappell wrote:2) What do you mean exactly by "put me on my nerves"? Do you mean they make you nervous (anxious or uneasy)? or do you mean they "get on your nerves" (annoy you or piss you off)? or something else?
Lol, no, I mean, it's just too complicated..., and I get too confuse..., and then I start trying to understand it..., and I usually end up understand its mechanism and the way this works..., but I'll quickly lose concentration, because I'm making a big effort..., and then I have to do all that effort again in order to regain that understanding..., and right now(maybe I'm not yet prepared to understand or know certain things..) I get to tired..., and then I get frustrated..., and that's what I mean by 'getting on my nerves'. Why am I using this expression wrongly?

Sorry for the will... :mrgreen:
Non omne id quod fulget aurum est.

Post Reply