TURKISH CLAUSE LINKAGE^553clause is case-marked with the accusative and functions as the direct object.
However, when the subjects are coreferential the resulting form does not
appear to be an instance of subordination. The lack of case-marking on the
nonfinite clause suggests that it is not filling an argument position of the
main clause:
(52) Çalιş-abil~mek ist-iyor-um.
WOrk-ABIL-INF Want-PROG-lSG
"I want to be able to work."
(53) Bu kiş ka-yabil-mek ist-iyor-um.
this Winter Ski-ABIL-INF Want-PROG-lSG
"I want to be able to ski this winter."
If such a construction does not involve subordination, it must be an instance
of either core coordination or core cosubordination. An argument against
considering this to be an instance of core cosubordination is the fact that the
dependent clause can be independently specified for modality as in (52) and
(53). If it were cosubordination we would expect the core operator of
modality to be shared by both clauses. I conclude, then, that the infinitive-
istiyor construction is an instance of core coordination. Though the two
clauses share the subject argument as well as the clausal operators of evi-
dentiality, tense and status, they are independently specified for the core
operator of modality.
It should be pointed out that this construction brings up two problems
in regard to the claims made by F&VV. First, we find what they claim to be
peripheral elements occurring within the core. Note the following exam
ples, as well as (53), above:
(54) Akşam-lar-i televizyon seyret-mek ist-iyor-um.
evening-PL-DAT television watch-INF want-PROG-lsG
"I want to watch television in the evenings."
As shown above this is an example of core coordination; hence, we would
expect the two clauses to share the peripheral elements, including the tem
poral adjunct. However, akşamlan as a temporal modifier only pertains to
the dependent clause; i.e., it specifies the time of watching television, not
the time of the desire. There are other constructions which also seem to
involve a temporal adverbial within a core rather than as a clausal modifier.
It appears that we will have to discard the strong claim of F&VV that the
operators of each level are tied to their corresponding arguments on the