Kiparsky (2002): Event Structure and the Perfect #
This file formalizes [Kip02]'s account of the polysemy of the English perfect. The
readings of [1], with the recent past a special case of the resultative, are the possible
assignments of a verbal predicate's event structure to the temporal parameters E, R and P of the
[Rei47] schema of [2], in which the perfect's E precedes R, [4]: the existential
reading places the event in E, the universal reading makes it coextensive with E, the resultative
reading places the activity of a telic predicate in E and its change of state between E and R
with the result state holding through R, and the present state reading places R in the result
state and leaves the change of state implicit (PerfectReading.mapping). An event structure is
simple, a single run time, or complex, a run time with an activity and a result state
(TemporalDecomposition), so the two result-state readings need the complex decomposition of a
telic predicate (isComplex_of_presentState).
The three arguments for the polysemy rest on one structural contrast: under the existential and
universal readings the event precedes R (precedes_of_existential), under the resultative and
present state readings the result state, and with it the event, extends through R
(le_runtime_of_presentState), so no configuration is of both kinds. A subordinate perspective
time anchored to the main-clause event, [16], therefore precedes the main perspective time in the
first case and includes it in the second, which is why [Dec91b]'s sequence of tense
contrast, [25], separates the existential and universal perfects from the resultative
(sequenceOfTense_of_existential, no_sequenceOfTense_of_resultative). Adverbs specify R, and
the present perfect's R includes P, so [Kle92]'s puzzle dissolves: no adverb anterior to P
is possible, while the past perfect's two readings with such an adverb, [33], are the two sides
of the contrast (present_perfect_puzzle). And the resultative reading is the only one that
assigns the activity and the result state to different parameters, which is what lets the
activity be presupposed and the change of state asserted, the source of the Wh-puzzle of [39]
(resultative_distinguishes).
Implementation notes #
- E, R and P are intervals, [2], and precedence is the strict
NonemptyInterval.Precedes, so the event of the universal reading ends before R; the inclusive boundaries remarked on for [10] are not represented. - The anchoring rule [16a] is read at its tightest, the subordinate perspective time being the main-clause event's trace.
References #
The two phases of a telic event, an activity and the result state it leads to, the activity ending no later than the result state begins.
- activityTrace : NonemptyInterval T
- resultTrace : NonemptyInterval T
- activity_precedes_result : self.activityTrace.toProd.2 ≤ self.resultTrace.toProd.1
Instances For
The event structure of a predicate is simple, a single run time, as for states and activities, or complex, a run time containing an activity and a result state, as for accomplishments and achievements.
- simple {T : Type u_2} [LinearOrder T] (runtime : NonemptyInterval T) : TemporalDecomposition T
- complex {T : Type u_2} [LinearOrder T] (runtime : NonemptyInterval T) (phases : SubeventPhases T) (activity_in_runtime : phases.activityTrace ≤ runtime) (result_in_runtime : phases.resultTrace ≤ runtime) : TemporalDecomposition T
Instances For
The run time of an event structure.
Equations
- (Kiparsky2002.TemporalDecomposition.simple r).runtime = r
- (Kiparsky2002.TemporalDecomposition.complex r phases activity_in_runtime result_in_runtime).runtime = r
Instances For
An event structure is complex when it has an activity and a result state.
Equations
- (Kiparsky2002.TemporalDecomposition.simple r).isComplex = False
- (Kiparsky2002.TemporalDecomposition.complex r phases activity_in_runtime result_in_runtime).isComplex = True
Instances For
The readings of the perfect, [1], the recent past being a special case of the resultative.
- existential : PerfectReading
- universal : PerfectReading
- resultative : PerfectReading
- presentState : PerfectReading
Instances For
Equations
- Kiparsky2002.instDecidableEqPerfectReading x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Kiparsky2002.instReprPerfectReading = { reprPrec := Kiparsky2002.instReprPerfectReading.repr }
In the perfect's temporal schema, [4], the event interval E precedes the reference interval R, and tense relates R to the perspective interval P.
- E : NonemptyInterval T
- R : NonemptyInterval T
- P : NonemptyInterval T
Instances For
The assignment of event structure to the perfect's parameters under each reading, [6], [9], [11] and [13]: the event in E; the event coextensive with E; the activity in E, the change of state between E and R and R in the result state; R in the result state alone.
Equations
- One or more equations did not get rendered due to their size.
- Kiparsky2002.PerfectReading.mapping s Kiparsky2002.PerfectReading.existential x✝ = (x✝.runtime ≤ s.E)
- Kiparsky2002.PerfectReading.mapping s Kiparsky2002.PerfectReading.universal x✝ = (x✝.runtime = s.E)
- Kiparsky2002.PerfectReading.mapping s Kiparsky2002.PerfectReading.resultative (Kiparsky2002.TemporalDecomposition.simple runtime) = False
- Kiparsky2002.PerfectReading.mapping s Kiparsky2002.PerfectReading.presentState (Kiparsky2002.TemporalDecomposition.simple runtime) = False
Instances For
The readings #
The resultative configuration is the present state one with the change of state assigned.
The result-state readings are confined to telic predicates, [11] and [13]: a simple decomposition has no result phase to place R in.
Under the existential reading the event precedes R, [6].
Under the present state reading the result state, and with it the event, extends through R, [13].
The readings are semantically distinct, §1, since no configuration is both existential and a result-state one.
The sequence of tense puzzle, §2 #
Under the existential and universal readings the main-clause event lies in E, which precedes R, so a subordinate perspective time anchored to it, [16a], precedes the main one in the present and past perfects: sequence of tense, [16c], applies, [27].
Under the resultative reading the result state holds through R, which includes P in the present perfect, so the subordinate perspective time includes the main one and [16c] is inapplicable, [26].
The present perfect puzzle, §3 #
Adverbs specify R, so an adverb denoting a time anterior to P makes R precede P, which the present perfect's inclusion of P in R excludes, [12b] and [32a]; the past perfect admits it, and its two readings with such an adverb, [33], are the existential configuration, the event before R, and the resultative one, the result state through R.
The Wh-puzzle, §4 #
Under the existential and universal readings both subevents of a telic predicate lie in E: only the whole event can be asserted or presupposed.
The resultative reading alone assigns the activity and the result state to different parameters, the activity to E and the result state, holding through R, outside E: the activity can be presupposed and the change of state asserted, and questioning the activity is questioning a presupposition, [39].