472 MICHAEL SILVERSTEIN
Figure 2
The respective complements of these categories gives pragmatically deduci-
ble "present" and "past"; and the "future" reading is either based on a
modal category or is classified regularly with one of these, depending on the
aspectual properties of the interval of speaking as lexicalized in the verba
dicendi.
The next most simple system includes a special marking for a true tense
opposition, having marked form for "past" tense, i.e., a specific marking
for the case where the narrated event is to be evaluated for a point in an
interval earlier than — i.e., which concludes by — the moment of speaking.
This is a true pragmatic, tense category as distinguished from the pragmatic
implementation of semantic categories of aspect. See Figure 3. In fact, the
construction of the necessary interval as completed by the moment of
speaking depends on the aspectual characteristics of events, though the
actual tense morpheme or equivalent adds to this the encompassment of
both some predicated event and the speech event in an ordered relationship
in some larger interval. When a tense-aspect system crosses this "past/non-
past'marking with its aspectual category, one can have an "imperfect" past
as opposed to a "simple" or "perfect" past, of course. It is important to see,
with Bull, that this notion of the past being ordered within an interval with
respect to the moment of speaking implies that the simplest "past" is a vecto
rial (minus vector) measure of pastness from the present as the point of ref
erence. Similarly, a slightly more complicated system has a vectorial future
in addition to a vectorial past, defining a temporal relationship of sequen-