[Lingtyp] Query about constraints on co-referential arguments in matrix clauses

Wed Jul 24 23:12:29 UTC 2019

Hi All

Thanks for all the suggested tips and reading. I look forward to reading
through what people have suggested.

As Jefferey stated, this is a venerable discussion. However, there are few
things people have picked up on that I will just quickly comment on by way
of response for all your helpful comments. The first is some general points
about syntactic ergativity and the other is co-reference between matrix and
subordinate clauses. These points are both relevant to the Ngkolmpu data
but crucially, its the very specific combination of these that I think
might be unique in Ngkolmpu.

It is clear that the Ngkolmpu data points to a level of syntactic
ergativity; what is more interesting perhaps is the specific manifestation
of this: a restriction on co-reference that applies to the role of
co-referential NP in the matrix clause but not in the subordinate clause.

This brings us to co-reference. This may show ergative alignment (as
described for Yidiɲ in Dixon 1977) or accusative (as with most of the
examples in Austin 1981).

If we take a broader typology of co-reference, one that is classified by
the types of restrictions placed on co-reference independent of alignment,
we can make a few generalisations. Co-reference in this sense might include
switch reference systems (i.e. clause-modifying clauses) & relative clauses
(i.e. noun-modifying clauses).

In cases of clause-modifying clauses (i.e. switch reference), there is a
(gapped) argument in the subordinate clause and there is an marker which
indicates which argument in the matrix clause this is co-referential with.
In all the examples that I have seen so far, this involves matched
alignment, between clauses, i.e. Diyari or the other languages in Austin
(1981); crucially for us is that it is a restriction on both the matrix and
subordinate clause.

By way of note, Ngkolmpu displays a set of simultaneous action clauses very
similar to this. These involve a non-finite clause in which the verb takes
a case marker which indicates whether the omitted S/A argument of the
non-finite clause is co-referential with either the S/A or the O of the
matrix clause, just like Diyari.

In terms of relative clauses, the restrictions have traditionally described
as a restriction of which *role the relativised NP may play in the
subordinate clause *(c.f. Keenan and Comrie 1977). This may be S/A as in
Malagasy or S/O as in Dyirbal. The data from Yidiɲ appears to show a
restriction on which argument may be co-referential in both the matrix
clause and the subordinate clause i.e. S/O for both. Finally, what I still
think is unique to Ngkolmpu (as Peter pointed out) is a restriction on
which argument of the matrix clause may be co-referential without any
restriction on which role it plays in the subordinate clause. I still want
to get more data on this of course.

Jeffrey Heath has also brought up functional motivations for these type of
constraints. I don't have the time to unpack every thing he has said there
but to answer his question:
If you wanted to say  “The dog, who is big, will bite that man” and wanted
to maintain the relative clause type structure you would have to say
something like:
'The dog will bite that man, he (subordinate marker) is big' and this would
be ambiguous between which argument in the matrix clause it refers to: it
could be the dog or the man who is big' or you could say: 'The dog will
bite that man, the dog (subordinate marker) is big'.

Thanks again for all your comments
Matt

On Tue, Jul 23, 2019 at 1:23 PM Eitan Grossman <
eitan.grossman at mail.huji.ac.il> wrote:

> A wonderful article that addresses the issues raised by Jeffrey is
> Marianne Mithun's 2002 paper 'Understanding and explaining applicatives.'
>
>
> http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.691.3166&rep=rep1&type=pdf
>
>
>
> On Mon, Jul 22, 2019 at 5:58 PM Heath Jeffrey <schweinehaxen at hotmail.com>
> wrote:
>
>> This is a venerable discussion, and many contributions old and new mix
>> two issues: a) which underlying categories (e.g. subject, object) in
>> matrix and subordinate clause can be coindexed ina matrix-subordinate
>> construction(such as relative clause and switch-reference, and b) which surface
>> categories can function in this way. Many languages have valency-changing
>> processes (passive, antipassive, applicative, …) that have one or more
>> discourse functions in simple main clauses, but are deployed
>> opportunistically in subordinated clauses solely to mark a specific
>> underlying NP as the coindexed one. This is the functionalist way of
>> describing the phenomena. It differs from the idea favored by typologists
>> that surface constraints are what drive the phenomena, since such
>> constraints by themselves have no functional basis and are far from being
>> universal. What would be the point (i.e., the evolutionary motivation) for
>> a language to require absolutive status for coindexation in
>> switch-reference or relativization, unless this is accompanied by a
>> seamless mechanism to allow underlying transitive subjects (which are of
>> greater referential value than objects) to be marked?
>>
>>
>> Therefore the question for Matthew is how the starred translation *“The
>> dog, who is big, will bite that man” is expressed in Ngkolmpu. It might be
>> better to use a nonadjectival relative clause since presumably “The dog,
>> who is big” can simply be rephased as “The big dog”. It might also be
>> better to use a restrictive rather than parenthetical relative. Something
>> like “The dog that bit he child ran away”.
>>
>> ------------------------------
>> *From:* Lingtyp <lingtyp-bounces at listserv.linguistlist.org> on behalf of
>> Valenzuela, Pilar <valenzuela at chapman.edu>
>> *Sent:* Monday, July 22, 2019 12:11 AM
>> *To:* Seino van Breugel <seinobreugel at gmail.com>; Matthew Carroll <
>> mattcarrollj at gmail.com>
>> *Cc:* <LINGTYP at LISTSERV.LINGUISTLIST.ORG> <
>> lingtyp at listserv.linguistlist.org>
>> *Subject:* Re: [Lingtyp] Query about constraints on co-referential
>> arguments in matrix clauses
>>
>>
>> Hola Matt,
>>
>> Panoan languages have a set of same-subject markers where the matrix
>> clause coreferential argument must be S, and another set where it must be
>> A. The dependent clause argument can be either S or A. Hope this is useful.
>>
>> Saludos,
>>
>> Pilar
>> ------------------------------
>> *From:* Lingtyp <lingtyp-bounces at listserv.linguistlist.org> on behalf of
>> Seino van Breugel <seinobreugel at gmail.com>
>> *Sent:* Sunday, July 21, 2019 9:37:56 PM
>> *To:* Matthew Carroll <mattcarrollj at gmail.com>
>> *Cc:* <LINGTYP at LISTSERV.LINGUISTLIST.ORG> <
>> lingtyp at listserv.linguistlist.org>
>> *Subject:* Re: [Lingtyp] Query about constraints on co-referential
>> arguments in matrix clauses
>>
>> *External Message*
>>
>> Dear Matthew,
>>
>> My 2010 article on attributive clauses, published in Studies in Language,
>> may be useful to you. I have attached a copy.
>>
>> Regards,
>>
>> Seino
>> __________________
>> Dr. Seino van Breugel
>> https://independent.academia.edu/SeinovanBreugel
>> <https://nam01.safelinks.protection.outlook.com/?url=https%3A%2F%2Findependent.academia.edu%2FSeinovanBreugel&data=02%7C01%7C%7Cc8fd8c4dcfaa49e9c22a08d70e5aa84d%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636993654878994873&sdata=UbzB2XoawzM3AcWhxbaixHybN0jO%2FXmrkHjzuj%2BwbOo%3D&reserved=0>
>> https://www.youtube.com/channel/UCHfiZwqyWC7HfZUAQ1RH1ew
>> <https://nam01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.youtube.com%2Fchannel%2FUCHfiZwqyWC7HfZUAQ1RH1ew&data=02%7C01%7C%7Cc8fd8c4dcfaa49e9c22a08d70e5aa84d%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636993654879014864&sdata=7K0kpGepxvUaNoxWI7R3rkT%2F2i77CZvSRBkilE5nsQ4%3D&reserved=0>
>>
>>
>> On Mon, Jul 22, 2019 at 3:30 AM Matthew Carroll <mattcarrollj at gmail.com>
>> wrote:
>>
>> Dear all,
>>
>> I am curious about restrictions on arguments in matrix clauses that are
>> co-referential with those in subordinate clauses.
>>
>> Restrictions on the role that a co-referential argument may play in a
>> subordinate clause are well established in the literature (Keenan and
>> Comrie 1977, and others). Rather I am interested in restrictions that may
>> apply to the role that co-referential argument may play in the *matrix*
>> clause.
>>
>> For example, in Ngkolmpu a Yam language spoken in West Papua that I have
>> been working on, there is a relative clause strategy involving a right
>> adjoined relative clause. The co-referential argument may serve *any
>> role in the subordinate clause* but can only be the *absolutive argument
>> of the matrix clause.*
>>
>> 1.     krar-w               irepe     pi         srampu             [ntop
>> mi                     bori      ye]
>>       dog-sg.erg      man      dist      he:will:bite:him  big       rel.abs
>>          comp    is
>>       'The dog will bite that man *who is big*’
>>       ***’The dog, *who is big*, will bite that man.’
>>
>> Example (1) can only be interpreted as 'the man who is big' and never
>> 'the dog who is big'. This has been confirmed through careful and
>> systematic elicitation on this topic and confirmed by examples in my
>> growing corpus (currently at about 1500 naturalistic utterances).
>>
>> Dixon (1977) notes similar restrictions in Yidiɲ. On page 323 of his
>> grammar he posits the coreferentiality constraint: "*There must be an NP
>> common to the main clause and subordinate clause, and it must be in surface
>> S or O function in each clause." *
>>
>> Unlike the Ngkolmpu example, this applies to both the matrix NP and the
>> subordinate NP which only applies to the matrix NP. Yet, importantly for my
>> purpose, does place a restriction on the role of the matrix NP. I am
>> curious to see if people know of other examples of these kind of
>> constraints in matrix NPs? or perhaps there is a paper that I have missed
>> in my (rather brief) survey of the literature on the topic.
>>
>> Regards,
>> Matt
>>
>> Matthew J. Carroll
>>
>> _______________________________________________
>> Lingtyp mailing list
>> Lingtyp at listserv.linguistlist.org
>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>> <https://nam01.safelinks.protection.outlook.com/?url=http%3A%2F%2Flistserv.linguistlist.org%2Fmailman%2Flistinfo%2Flingtyp&data=02%7C01%7C%7Cc8fd8c4dcfaa49e9c22a08d70e5aa84d%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636993654879024821&sdata=MWwIUyJmOkngPlZMfRy7EqCtEEq8%2FHlEeC%2BfsR5T5AQ%3D&reserved=0>
>>
>>
>> *NOTE: This email originated from outside Chapman’s network. Do not click
>> links or open attachments unless you recognize the sender and know content
>> is safe.*
>>
>> _______________________________________________
>> Lingtyp mailing list
>> Lingtyp at listserv.linguistlist.org
>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20190725/04962b8a/attachment.htm>