rel. pron. agreeing in case with head in main clause

[Guy Deutscher]

Dear all,

I wonder if someone could help me with information on any language which
has embedded relative clauses and a relative marker that declines for
*case*, and in which the relative marker agrees in case with the role of
its head in the *main* clause, not in the relative clause.

In such a language, a sentence like 'I know the man who died', should have
the structure 'I know the man REL.ACC (he) died'. Similarly, 'the son of
the man who died', should be something like 'the son of the man REL.GEN
(he) died'. And 'the man whom I saw died', should be 'the man REL.NOM I saw
(him) died'.

Such a system exists in the early stage of ancient Akkadian, and has some
sporadic vestiges in other Semitic languages. There are also some similar
cases in older Indo-European languages (in Greek and Latin they are called
'case-attraction'). But in the Indo-European cases, the examples of this
structure seem to be 'exceptions', i.e.
deviations from the more normal formation (where the relative pronoun marks
the case of the head in the relative clause). I am looking for languages in
which this system is the 'normal' one, and in which it is fully functional.

Many thanks, Guy Deutscher.

=======================================
Dr Guy Deutscher
St John's College
Cambridge CB2 1TP
England

E-mail:  gd116 at cam.ac.uk
Tel:  01223 - 338741
Fax: 01223 - 740540