TURKISH CLAUSE LINKAGE^547(29) işi- gid-er-ken ben ev-de otur-uyor--.
PRO work-DAT go-AOR-ken I home-Loc sit-iMPF-miş-lsG
"(They say) when he was going to work, I was staying home."
(30) işi-ne git-iş- ben ev-de otur-uyor-du-m.
PRO work-DAT go-miş-ken I home-Loc sit-IMPF-PT-lsG
*"(They say) when he was going to work, I was staying home."
"With him having gone to work, I was staying home."Thus (28) and (29) are synonomous; whether or not the dependent clause
has the -mlş suffix, it is included in the scope of the evidential in the main
clause. The sentence in (30), on the other hand, was rejected as ungram-
matical by my informants, given a "hearsay" reading. The -ken clause can
not have an evidential reading that is independent from the main clause,
which would be the case in (30). However, as Karl Zimmer has since
pointed out to me, (30) is grammatical if -mlş is only given its aspectual
reading: "While he has gone to work, I was staying home." (See similar
form in Lewis 1967:190 and Underhill 1976:401,2.) This confirms the
analysis in the sense that the construction of clausal cosubordination pre
vents -mlş from having the independent evidential reading in the -ken
clause.
Thus, a -ken clause has independent core and clausal arguments but is
dependent on the main clause for its tense and evidential operators. On the
basis of such evidence we conclude that -ken constructions are instances of
clausal cosubordination. In cases of clausal juncture (including cosubordi
nation) we would expect the two clauses to be independently specified for
the main core-level operator, modality. This is, in fact the case for the -ken
construction with -Ebil, "abilitative":
(31) Ben -abil-ir-ken siz sus-mali-sιnιz.
I speak-ABIL-AOR-KEN 2PL be.quiet-mElI-2pL
"When I'm able to speak, you(pl) should keep quiet."
Note that this same construction cannot occur with the other modality
operator, -mEU "obligation":
(32) * Ben -mali-yir-ken siz sus-mah-siniz.
I speak-oBL-AOR-ken 2PL be.quiet-mElI-2pL
"When I have to speak, you should keep quiet."
However, as was the case with the -ince construction above, this difference
between -Ebil and -mEU is due to morphological rather than semantic con-