538 JAMES K. WATTERS
inner Operators to be re-analyzed as operators over outer layers"
(1984:216). For our purposes here it will be important to recognize that
-ml§ may be a nuclear operator (perfect aspect) in some instances and a
clausal operator (tense or evidential) in others. Again, matching the claims
in F&VV regarding the scope and ordering of operators we find that when
-ml§ is followed by the tense suffix -dl it has only the aspectual reading (see
Lewis 1967:123). That is, when it occurs "within" tense it is stripped of its
tense and evidential interpretation:
(1) Gel-iş
come-rIş
"I gather that he has come."
(3) Gel-iş-
come-rIş-PT
"He had come."
Though traditionally English grammars of Turkish have applied the
term "tense" to several verb suffixes, I will assume that only two suffixes
regularly have a basic tense function: -dl "past", and -ecek, "future".^1
These are the only suffixes that consistently have the deictic function
characteristic of true tense: marking the time of the event in relation to the
speech act. Other grammars have considered -lyor and -Er (or -Ir) to be
instances of present and aorist tense, respectively (though Underhill 1976
calls them "progressive tense" and "present tense"). This position is dif
ficult to maintain when faced with instances of these suffixes occurring with
the past tense suffix, -dl:
(3) Çalιş-ιyor-um
WOrk-PROG-lSG
"I'm working."
(4) Gel-iyor-du-m.
come-PROG-PT-lsG
"I was coming."
(5) Çahş-ir-di.
WOrk-AOR-PT
"I used to work."
What instead appears to be the case is that -lyor functions to mark present
tense, as in (3), when there is no "pure" tense suffix (or other temporal
deixis) present.^2 When it occurs within the past tense suffix as in (4), how-