Advances in Role and Reference Grammar

TURKISH CLAUSE LINKAGE^557

There are at least two possible ways of avoiding the problem presented
by (66), though both seem unsatisfactory. For one, it is not unreasonable to
assume that -Ip may occur in different kinds of constructions, including
both clausal as in (66) and core junctures as in (60)-(65). The requirement
of shared subjects (in both types of constructions) would then just be a
peculiarity of -Ip. (Or, more correctly, the subject would be shared in the
core construction and simply understood as coreferent in the clausal con­
struction.) This, however, seems unlikely since the two clauses apparently
must always share the modality reading.
On the other hand, one could claim that the temporal setting is not
necessarily a peripheral constituent. Thus it appears to be a core con­
stituent in the istemek construction discussed in 6.3 as an example of core
coordination:

(67) Akşam-lar-i televizyon seyret-mek ist-iyor-um. — (54)

evening-PL-Ac television watch-iNF want-pROG-lsG

"I want to watch t.v. in the evenings."

Here akşamlan "in the evenings" pertains only to the dependent clause.

Yet, this would weaken the claims in F&VV significantly, as a key to distin­

guishing core and periphery would no longer serve. To summarize, either

possibility seems rather unlikely and an adequate solution remains to be

worked out. Suggestions along this line will be made below.

9. Conclusion

This has been a very brief sketch covering only a few instances of Turkish

clause linkage. Yet, the claims made by F&VV regarding the interaction of

operators in clause linkage have been consistently supported by the Turkish

data. This is an impressive result considering the fact that though their

claims are based on investigations of a variety of languages, Turkish was

not one of them. This is a positive sign of the predictive power of the

theory's claims regarding the interaction between operators and the layered

structure of the clause. There remains the problem of the layering of con­

stituents (especially clausal ones) within the clause and their relation to the

operators.^6 In a few instances, the constituents of layers (specifically of the

core and periphery) did not seem to pattern along with the operators of the

layers. F&W suggest a temporal constituent belongs to the periphery and