Arguments vs. adjuncts as heads of relative clauses

[David Gil]

Dear Whitney and ANLANG members,

I must confess that it was not clear to me, originally, whether the
query was about the grammatical role of the NP in the embedded clause
(about which much has been written both before and after the well-known
Keenan and Comrie study), or the grammatical role of the NP in the main
clause (about which there is much less in the general literature).  Now
that is clear to me that the query is about the latter, I would like to
chip in with my two-Rupiah worth, with some data from Riau Indonesian.

In Riau Indonesian, internally-headed relative clauses are possible,
such as the following, from a naturalistic corpus [in which the
internally headed relative clause is enclosed in square brackets]:

(1)	[Kemarin ada kawan sini] mana?
	previous.day exist friend here where
	[Hotel receptionist, at hotel, where, last time, I had come with friend]
	"Where’s your friend who you were with last time?"

(2)	[Damsir beli celana sama si Man] sudah bulu-bulu
	Damsir buy trousers accompany PERS:NON.VOC FAM-Mansudir PFCT DISTR-feather
	[About a pair of trousers]
	"The trousers that Damsir bought with Mansudir are already frayed"

(3)	[Saya ada beli kaca mata dulu], mana dia?
	1:SG exist buy glass eye PST:DIST where 3
	[Looking for glasses]	
	"The eye-glasses I bought before, where are they?"

(4)	[Ada perempuan tadi], ininya nampak
	exist woman PST:PROX DEM:PROX-ASSOC AG-see
	[Commenting on a women with a low front to her blouse, speaker points
to his own chest and says]
	"The woman before, her [this] was showing"

Anyway, the crucual point is that in all the examples in my corpus, the
internally-headed relative clause occurs in what seems to be either
subject or (perhaps more relevantly) topic position in the main clause.
This would thus appear to be a constraint of the general type suggested
by the original query, namely of the grammatical role of the relativized
NP in the matrix clause.

Similar constructions seem to occur in other dialects of colloquial
Indonesian, and also in nearby Singlish, as in the following (elicited) example:

(5)	[Lisa choose the house yesterday], Ah Chew buy
	"Ah Chew bought the house that Lise chose yesterday"

Here, the internally headed relative clause is the object of the matrix
clause, but has been fronted to topic position: the unfronted version,
*Ah Chew buy Lisa choose the house yesterday, would be ungrammatical.

So it seems as though internally-headed relative clauses, in Riau
Indonesian and in Singlish, must (or at least strongly prefer to) occur
in topic- or subject-like position in the matrix clause.  Interestingly,
this constraint applies only to internally-headed relative clauses,
left-headed ones, which are common in both languages, may occur in any
position in the matrix clause.

Greetings,

David

--
David Gil

Department of Linguistics
Max Planck Institute for Evolutionary Anthropology
Inselstrasse 22, D-04103 Leipzig, Germany

Telephone: 49-341-9952321
Fax: 49-341-9952119
Email: gil at eva.mpg.de
Webpage:  http://monolith.eva.mpg.de/~gil/