bridget copleybcopley.com/wp-content/uploads/copley.causal.conditionals.2013...causal conditionals...
TRANSCRIPT
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Causal conditionals
Bridget Copley
SFL (CNRS/Universite Paris 8)[email protected]
Atelier How Do We Know What Happens?
April 4-5, 2013 — ENS/SFL, Paris
1/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The bareness of eventive verbs in English allows us to detect athree-way correlation between Aktionsart, temporalorientation, and modal flavor in conditionals (Copley, 2009).
Excluding deontic readings, which occur with both eventivesand statives, we get:
eventives ↔ future orientation ↔ metaphysical modalitystatives ↔ present orientation ↔ epistemic modality
2/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
This three-way correlation, along with an assumption thateventives and statives differ in type, suggests that allconditionals express a causal relationship between events.
3/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
Aktionsart and temporal orientation
4/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
The Bare Eventive Constraint
(1) a. Zoe dances now/tomorrow. atelic eventive: *=, > ok
b. Zoe leaves now/tomorrow. telic eventive: *=, > ok
c. Zoe is home now/tomorrow. stative: = ok, > ok
present orientation future orientation
eventives bad okstatives ok ok
We’ll set aside all generic/habitual readings for now.
5/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
(2) Parsons (1990)
a. “. . . a present tense event sentence construed in the reportive sense can be true only at onespecific instant–the time the event culminates. Such a sentence, then, can be used truly only invery specific circumstances.”(1990: 30)
b. Brutus stabs Caesar = PRES(∃t)[t ∈ NOW & Stabbing(e) & Subject(e, Brutus) & Object(e,Caesar) & Cul(e)] (see discussion pp. 213-4)
(3) Smith (1991)
a. “There is a pragmatic principle of interpretation for sentences about Present time, requiring thatthey be interpreted in a certain way: Present sentences may not include the endpoints ofsituations.” (1991: 110)
b. The viewpoint Perfective is located at interval I; with the condition that the situation S obtainsat I, and there are times ti , tn included in I at which the endpoints of S obtain. (1991: 127)
(4) Kamp and Reyle (1993)
a. The eventuality described by a present tense sentence must properly include the utterance time n.(1993: 536)
b. “[(4a)] indicates that the eventuality must have the status of a state and not of an event. . . . asentence which describes something as going on at a time . . . cannot represent that somethingas an event. For the event would have to be entirely included within the location time and thuswould not extend beyond it.” (1993: 536-7).
6/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
Condoravdi (2001): Temporal orientation facts under modals arethe same.
(5) a. Zoe will/should/may dance now. atelic eventive: *=, > ok
b. Zoe will/should/may leave now. telic eventive: *=, > ok
c. Zoe will/should/may be sick now. stative: = ok, > ok
present orientation future orientation
eventives bad okstatives ok ok
I’m going to argue that there are no real future-oriented statives.
7/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
No real future-oriented statives
Eventives get into the future automatically, while statives need tobe “anchored,” even under will:
(6) a. Zoe will dance.b. Zoe will leave.c. Zoe will be at home. (“. . . when??”)
(7) a. At 4 o’clock, Zoe will be at home.b. When you next see Zoe, she will be at home.c. If it’s 4 o’clock, Zoe will be at home.
8/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
⇒ Eventives are automatically future-oriented with respect to theirtemporal argument t*, while statives are automaticallypresent-oriented with respect to their temporal argument t*, whichmust be shifted forward and bound by the anchor if the stative isgoing to be future-oriented. Will permits (but does not require)this forward-shifting.
(8) Zoe will leave at 4 o’clock. t* = now
(9) Zoe will be at home at 4 o’clock. t* = 4 o’clock
9/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
The same facts hold for other modal contexts, including withinantecedents.
(10) a. If Zoe dances in her office . . . atelic eventive: *=, > ok
b. If Zoe leaves . . . telic eventive: *=, > ok
c. If Zoe is home . . . stative: = ok, > ok? (if not right now, “when??”)
d. If Zoe is home at 4 o’clock . . . stative: > ok with anchor
10/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
A test (particular to antecedents) to see the shifting forward of t*:
(11) a. If it’s true tomorrow that Zoe is sick . . .b. #If it’s true now that Zoe is sick tomorrow . . .
The temporal argument t* = the time at which truth of Zoe is sickis evaluated.
11/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
So, taking into account the idea that t* can be future-shifted, weget the following:
present orientation future orientation
eventives bad okstatives ok bad
This means that the BEC is insufficient to explain the temporalorientation facts in assertions and modal contexts.
12/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
Propositions and non-propositions
If the BEC isn’t sufficient, what would work?
(12) a. #It rains tomorrow.b. If it rains tomorrow . . .c. #If it’s true that it rains tomorrow . . .
(13) a. Zoe dances tomorrow.b. If Zoe dances tomorrow . . .c. If it’s true that Zoe dances tomorrow . . .
Suspicion that will prove useful: Bare eventive phrases such as Itrains tomorrow are not even propositions.
I’ll argue below that Zoe dances tomorrow can be a stative, therefore a proposition on at least one of its readings.
13/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
Let’s assume that propositions are (i) assertable and (ii) fine underit’s true that. Thus:
English statives are propositions.
Bare English eventives are not propositions.
Temporal predicates (anything with tense or aspect) are alsopropositions:
(14) a. Zoe left.b. Zoe has left.
(15) a. If it’s true that Zoe left yesterday . . .b. If it’s true that Zoe has left . . .
(16) Zoe danse.‘Zoe is dancing.’
14/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
The Bare Eventive ConstraintNo real future-oriented stativesPropositions and non-propositions
What could be causing this difference?
It’s plausible that the difference between eventives and propositionsis one of type: eventives are predicates of events while propositionsare predicates of situations. I will make this assumption.
But why does would it matter to temporal orientation thateventives are predicates of events while statives and temporalpredicates are predicates of situations?
15/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Modal flavor and temporal orientation
16/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Propositions might make us think of epistemic modality.
Like assertion, epistemic modality involves a proposition thatis added to the common ground at the current time withrespect to t* (crucially, not a future time with respect to t*).
On our assumption that statives are propositions, maybewhenever they occur there is the addition of a proposition tothe common ground at t*, and this is why there are no truefuture-oriented statives.
In that case, we would be free to invoke a version of the BECfor the lack of present-oriented eventives.
17/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
So maybe . . .
eventives ↔ future orientation ↔ metaphysical modality
statives ↔ present orientation ↔ epistemic modality
And perhaps metaphysical modality has something to do with thetemporal orientation of eventives.
18/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
In order to tell such a story, we would expect that:
Antecedents have their own modal flavor independent of themodal flavor of the consequent.
Epistemic modals in English never take eventives ascomplements.
Metaphysical modality does not involve propositions.
We’ll verify these in turn, and reach the conclusion that
Conditionals are causal.
19/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Antecedents have their own modal flavor
Is it so crazy? We have no evidence for modal flavor apart fromthe judgments of linguists (and now, arguably, the it’s true thattest in antecedents).
(17) a. If it’s raining, Zoe gets sick.b. If Zoe gets sick, it’s raining.
(18) a. If it’s raining, Zoe will get sick.b. If Zoe gets sick, it will be raining.
20/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
No real epistemic eventives
Condoravdi (2001): “settledness”
(19) a. It hasn’t been decided yet who he will meet with.b. He may see the dean. He may see the provost.
(20) a. It has been decided who he will meet with but I don’tknow who it is.
b. He may see the dean. He may see the provost.
cf. futurates as statives, “plannability” (Copley, 2008, 2013)
(21) a. John sees the dean tomorrow.b. #John gets sick tomorrow.
21/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Another fact: t* of consequent is event time of antecedent, soeventive consequent ⇒ consequent event time is after event timeof antecedent.
(22) a. If it rains on Monday, I’ll give her a call (then/*rightnow).
b. If it’s raining right now, I’ll give her a call right now.
22/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
(23) a. If Zoe dances on Monday, I’ll give her a call rightnow.
b. #If it rains on Monday, I’ll give Zoe a call right now.
“event time” of antecedent on the futurate reading is now, notMonday ⇒ time of state of having the plan
(24) If it’s true that Zoe dances on Monday, I’ll give her a callright now.
23/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Just like generic/habitual readings:
(25) If Zoe dances, I’ll give her a call right now.
(26) a. Zoe is dancing these days.b. Zoe is dancing on Monday.
(27) a. Zoe lives in Paris.b. Zoe is living in Paris.
24/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
So: there are no real epistemic eventives. So not only is this thecase:
present orientation future orientation
eventives bad okstatives ok bad
But this is also the case:
eventives ↔ future orientation ↔ metaphysical modality
statives ↔ present orientation ↔ epistemic modality
25/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
No future-oriented statives because: assertion/epistemicmodality, addition of proposition to common ground can onlyhappen at t*.
No present-oriented eventives because: some version of BEC.
(28) BEC, boring type-theoretic version:
a. Assertive contexts and epistemic modals requireproperties of situations.
b. Eventive predicates are properties of events.c. Therefore no eventive predicate can be asserted or be
the complement of an epistemic modal.
Apparent counterexample; narrative present (He shoots, hescores!). However, this is plausibly direct description of an eventrather than a situation.
26/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
But if we use the boring type-theoretic version of the BEC insteadof one of the classic versions, we still need to say whyfuture-oriented eventives are possible. Actually, we only need tosay it for metaphysical modal contexts.
(29) a. Zoe leaves tomorrow. really a stative (futurate)
b. My assistant (now) appears. direct description of event
c. If Zoe leaves (tomorrow ). . . modal context
d. Zoe will/should/may leave (tomorrow). modal context
So we need to understand what these metaphysical modal contextsare doing to future-shift the eventive. This will come out of thenext thing we need to explain . . .
27/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Metaphysical modality without propositions
We noted earlier that metaphysical modality can’t involvepropositions, because bare eventives are not propositions and arenot present-oriented.
Hacquard (2006): accessibility relation takes an event argument
Causal accessibility: e2 is causally accessible from e1 iff e2 isin a causal chains that proceeds from e2
Branching comes from different choices of the initialconditions
ceteris paribus: the speaker chooses initial conditions thatthey think are appropriate, but could be wrong
Not the same as because
28/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
But don’t we need to explain causation in terms of possibleworlds???
That’s only one of two options. Either option is legitimate:
causation is constructed in terms of possibilities, as independency theories of causation, such as Lewis (1975);Dowty (1979)
possibilities are constructed in terms of causation, as inproduction theories of causation: Fair (1979); Talmy (1988)etc.
29/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
(30) λw λp ∈ D〈w,t〉 λq ∈ D〈w,t〉 .
for all w’ metaphysically accessible from w such thatp(w’), q(w’)
Can we just replace worlds with events, like this?
(31) λe . λp ∈ D〈v,t〉 . λq ∈ D〈v,t〉 .
for all e’ causally accessible from e such that p(e’), q(e’)
Not exactly!
30/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
(32) λC λp ∈ D〈v,t〉 λq ∈ D〈v,t〉 . all causally accessible
courses of events from eC that include an e’ such thatp(e’) are such that: all causally accessible courses ofevents from e’ include an e” such that q(e”)
31/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Conditionals are causal
Since conditionals can be mixed (metaphysical antecedent,epistemic consequent, e.g.), and metaphysical and epistemicmodality involve different types of arguments, what is the meaningof conditionals?
Either:
Option 1: Conditionals take predicates of situations (orworlds, as in (30)). Eventives get type-shifted from predicatesof events to predicates of situations (or worlds).
Option 2: Conditionals take predicates of events, as in (32).Statives and temporal predicates get type-shifted frompredicates of situations to predicates of events.
32/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Option 1: Conditionals take predicates of situations (or worlds).Eventives get type-shifted from predicates of events to predicatesof situations (or worlds).
Need some way to relate an event to a situation. Remember,events can’t be situations if the reason why bare eventives can’t beasserted is a type difference.
Tricky (though see Copley and Harley (2012)). But even if it canbe done, it doesn’t help us with the temporal facts right: why areeventives forward-shifted?
33/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Option 2: Conditionals take predicates of events. Statives andtemporal predicates get type-shifted from predicates of situationsto predicates of events.
(33) Add(e) =λC λp ∈ D〈s,t〉 . e is an event of adding p to CGC
Note that Add:
is, type-wise, a kind of opposite of aspect
could easily be used for evidentials if manner is predicated ofthe event
34/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
If Option 2 is correct, all conditionals are causal.
(34) If a creature has a heart, it has a liver.
‘all currently causally accessible courses of events that include anevent e’ of [adding the proposition this creature has a heart to thecommon ground], are such that all courses of events causallyaccessible from e’ include an event of [adding the proposition thiscreature has a liver to the common ground].’
35/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Antecedents have their own modal flavorNo real epistemic eventivesMetaphysical modality without propositionsConditionals are causal
Option 2 explains the temporal facts too:
Only forward-shifted eventives: Since events are in a causal chain(of arbitrary length) from the contextually salient event at t*, theyoccur after t*.
Only present-oriented propositions: If we assume that adding aproposition to the common ground is instantaneous, then althoughthe addition is a causally subsequent event to the contextuallysalient event, it still gives the impression of being close to t*; andin any case the propositions themselves are evaluated with respectto C (i.e., where t* comes from).
36/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Conclusions
Bare eventives are future-oriented because they describeevents that are causally accessible from the present.
Statives and temporal predicates are present-oriented becausethey are propositions that are added to the common ground;they are evaluated at the present time.
Will can but need not shift the temporal argument of itscomplement; the futurity of eventives does not come from thisshifting.
Futurates are stative.
Conditionals are causal, with a type-shifting operation(“Add”) to turn propositions into descriptions of the event ofadding the proposition to the common ground.
37/38
IntroductionAktionsart and temporal orientation
Modal flavor and temporal orientationConclusionsReferences
Condoravdi, C. (2001). Temporal interpretation of modals. In D. Beaver, S. Kaufmann, B. Clark, and L. Casillas(Eds.), Stanford Papers on Semantics. Palo Alto: CSLI Publications.
Copley, B. (2008). The plan’s the thing: Deconstructing futurate meaning. Linguistic Inquiry 39(2), 261–274.
Copley, B. (2009). Temporal orientation in conditionals. In J. Gueron and J. Lecarme (Eds.), Time and Modality,pp. 59–77. Springer.
Copley, B. (2013). Causal chains for futurates. To appear in Future Time, Future Tense, Oxford University Press.Philippe de Brabanter, Mikhail Kissine, & Saghie Sharifzadeh, eds.
Copley, B. and H. Harley (2012). A force-theoretic framework for event semantics. ms. submitted to Linguisticsand Philosophy.
Dowty, D. (1979). Word meaning and Montague Grammar. Dordrecht: Reidel.
Fair, D. (1979). Causation and the flow of energy. Erkenntnis 14(3), 219–250.
Hacquard, V. (2006). Aspects of Modality. Ph. D. thesis, MIT.
Kamp, H. and U. Reyle (1993). From Discourse to Logic. Dordrecht: Kluwer.
Lewis, D. (1975). Counterfactuals. Cambridge, Mass.: Harvard University Press.
Parsons, T. (1990). Events in the Semantics of English: A Study in Subatomic Semantics. Cambridge, Mass.: MITPress.
Smith, C. (1991). The parameter of aspect. Dordrecht: Kluwer.
Talmy, L. (1988). Force dynamics in language and cognition. Cognitive Science 12(1), 49–100.
38/38