NELS 38 [Semantics Workshop]
October 27, 2007, Ottawa, Ontario *
Relativism and the De Se Interpretation of PRO
Relativism and the De Se Interpretation of PRO
1.4. Attempts (5)
Tamina Stephenson University of British Columbia web: http://alum.mit.edu/www/tamina; e-mail:
[email protected] Goals of talk: Propose a view of PRO in attitude reports using semantic relativism. Evidence from: A restriction on the interpretation of words like tasty under control verbs A parallel between PRO and epistemic modals
[Context: Sue hates spending time alone. When she’s with other people, they always seem to enjoy her company, appreciating her stories and laughing at her jokes, but she just can’t ever seem to entertain herself, and wishes that she could.] (a) # Sue wants to be entertaining. (b)
Sue wants to be entertaining to herself.
This is clearer with reciprocals (Henry Davis, p.c.): (6)
Sam and Mary want to be entertaining. ≠ Sam and Mary want to be entertaining to each other.
We can also be creative: Imagine a magical living cake (“Mr. Cake”), who • walks, talks, thinks, feels, eats • needs to regularly have slices cut off of him and eaten (which he regenerates) • if no one else is around to take slices, he’ll eat them himself (since he doesn’t like to see them lying around) – but oddly enough, they always taste terrible to him
1. Introduction: A Puzzle 1.1. Predicates of personal taste tasty, fun, entertaining, exciting, boring, … Observation: Under attitude predicates, these seem to always have a salient reading that means “tasty/fun etc. for the attitude holder” (Lasersohn, 2005, 2006; Stephenson, 2007): (1)
Sue thinks the cake is tasty. [most natural reading] = (Sue thinks) the cake tastes good to her (Sue)
(2)
Bill thinks the roller coaster is fun. [most natural reading] = (Bill thinks) the roller coaster tastes good to him (Bill)
(7)
[Context: Mr. Cake was starting to feel stale. He called his friends to see if they would come by and have a slice, but none of them were home. Since he had no other choice, he cut off a few small pieces of himself and ate them. As usual, they tasted terrible to him. This was very frustrating because all of his friends who had tasted slices of him had always enjoyed them.]1 (a) # This was very frustrating. Mr. Cake wanted to be tasty!
cf.
(b)
This was very frustrating. Mr. Cake wanted to taste good to himself!
Generalization: Under control verbs, predicates of personal taste cannot be linked to the attitude holder. [Even though with non-control attitudes the “linked” reading is the default!]
[Note: If they acquired their belief through direct experience, then they can’t be wrong about their own experience. (hence the parentheses around Sue thinks/Bill thinks)] 1.2. Control verbs
1.5. A parallel
Under want (and similar control verbs) the implicit subject (PRO) must be the attitude holder:
Assumption: tasty, fun, etc. have some kind of covert argument (e.g., Epstein, 1984).
(3)
This is exactly the restriction we would expect if this covert argument were a non-reflexive nominal coreferential with PRO. [“Principle B”]
Sue wants/hopes [PRO] to go on the roller coaster. PRO = Sue
1.3. An expected reading
(8)
Given these two observations, we would expect to find sentences of the form “X wants [PRO] to be tasty / fun / exciting / …” meaning “X wants to be tasty / fun / exciting / … to themselves”
(9)
* [Sam and Mary]j⊕m are entertaining to themj⊕m.
(10)
* Mr. Cakei tastes terrible to himi.
(4)
Suggestion: The restricted reading is an instance of Principle B but first…
# The cake wants to be tasty. ≠ the cake wants to taste good to itself
2
[pragmatically implausible]
* Suei is entertaining to heri.
[compare to (5a)] [compare to (6)] [compare to (7a)]
But we still don’t see the relevant reading when we come up with more plausible examples: 1 *
Some of this material appeared in my Ph.D. thesis (Stephenson, 2007, Ch. 4). For discussion and helpful suggestions, I would like to thank my thesis committee – Kai von Fintel, Irene Heim, Danny Fox, and Agustin Rayo – as well as Martina Wiltschko, Henry Davis, and other audience members at a UBC research seminar. Any errors are my own.
Kai von Fintel (p.c.) points out the similarity of a story in Douglas Adams’s Restaurant at the End of the Universe; however, I gather that the cow in question doesn’t eat itself (or, sadly, get regenerated).
3
Tamina Stephenson
Relativism and the De Se Interpretation of PRO
2.3. Attitude reports
2. Background on Predicates of Personal Taste
Using doxastic alternatives (Lewis, 1979; Chierchia, 1989) and analogous notions:
2.1. Two uses Tasty, fun, etc. have a use where they just seem to take an implicit, contextually salient argument: (11)
[Context: Sam is watching his cat eat cat food. The cat seems to enjoy it. Sam says:] Oh good, the new cat food is tasty. [tasty = tastes good to the cat]
Mary: (a) Sam: (b) Sue:
How’s the cake? It’s tasty. No it isn’t!
Note: Each speaker seems to be taking themselves to be the person whose knowledge / experience is relevant, but these cannot contain 1st-person indexicals (cf. (13)): (13)
Mary: I’m a doctor. Sam: # No, I’m not!
2.2. Analysis of predicates of personal taste [Judge-dependent approach] (Stephenson, 2006, 2007; based on Lasersohn, 2005) • Propositions treated as sets of world-individual pairs rather than sets of worlds [or world-timeindividual triples rather than sets of world-time pairs] ⇒ truth is relativized to individuals ⇒ propositions = type
[Individual parameter = Lasersohn’s “judge parameter”] • Predicates of personal taste are basically 2-place predicates: (14)
[[tasty]]w,j = [λy . [λx . x tastes good to y in w] ] [[fun]]w,j = [λy . [λx . x is fun for y in w] ] [[entertaining]]w,j = [λy . [λx . x is entertaining to y in w] ]
The first argument is either filled in by a contextual argument / “silent pronoun” (I’ll write proName) or by PROJ, which refers to the “judge”: w,j
(15)
[[PROJ]]
So:
[[tasty proSam]]w,j = [λx . x tastes good to Sam in w] [[tasty PROJ]]w,j = [λx . x tastes good to j in w]
=j
(16)
[[think]]w,j = [λp . [λze . ∀∈Doxw,z: p(w′)(y) = 1] ]
(17)
Doxastic Alternatives: Doxw,x = {: it is compatible with what x believes in w that he/she/it is y in w′}
(based on examples by Lasersohn, 2005)
But crucially they also have a use that seems to be doing something more complicated: (12)
4
In other words: [[z thinks (that) φ]]w,j = 1 iff ∀∈Doxw,z, [[φ]]w′,y = 1 ↑ ↑ ↑
(18)
↑
Similarly for other attitudes, for example: (19)
Want Alternatives:
(20)
[[z wants φ]]w,j = 1 iff ∀∈Doxw,z, [[φ]]w′,y = 1 ↑ ↑ ↑ ↑
Wantw,x = {: it is compatible with what x wants in w for x him/her/itself to be y in w′}
Result:
“z thinks that φ” ≈ z thinks that φ is true with z as judge “z wants φ” ≈ z wants φ to be true with z as judge
In particular: Sam thinks that the cake is tasty [PROJ] ≈ Sam thinks that the cake tastes good to him [Note: I’m focusing on judge-dependent propositions that are embedded in attitude predicates, which in effect “bind off” the judge parameter. I discus matrix cases elsewhere (Stephenson, 2006, 2007).]
3. Proposal • “PRO” (in these kind of constructions) is PROJ. • PROJ is a non-reflexive nominal (i.e., subject to Principle B) * [PROJ] is
tasty fun entertaining …
[PROJ] [Principle B]
Essentially there is nothing wrong semantically with the disallowed readings, but they contain two instances of PROJ that are in a Principle B configuration: For example, (5a), Sue wants [PROJ] to be entertaining [PROJ]: [[Sue wants [PROJ to be entertaining PROJ] ]]w,j = 1 iff ∀∈Wantw,Sue: y is entertaining to y in w′
5
Tamina Stephenson
Relativism and the De Se Interpretation of PRO
4. Predictions
This follows from the compositional semantics:
4.1. De se interpretation of PRO
(25)
[[(24)]]w,j = [[want]]w,j ( [[Sue]]w,j ) ( [λw′ . [λj′ . [[Bill to want PROJ to be famous]]w′,j′ )
Old observation (Morgan, 1970; Chierchia, 1989): PRO must be interpreted de se:
= [[want]]w,j ( Sue ) ( [λw′ . [λj′ . ∀∈WantBill, w′: z is famous in w″] ] )
(21)
= 1 iff ∀∈WantSue,w: ∀∈WantBill, w′: z is famous in w″
Pavarotti wants to be famous. [obligatorily de se] ≠ P. wants the guy he heard on the radio to be famous (based on examples by Morgan, 1970)
Pavarotti thinks, “It would be great if I became famous” Æ (21) is true. Pavarotti listens to recording, thinks “it would be great if that guy became famous” Æ (21) is false.
4.3. Non-Principle B Violations Predicates of personal taste can be linked to “controller” if the two instances of PROJ aren’t in a configuration that violates Principle B [for my purposes: if they aren’t co-arguments]. (26)
Similarly for Italian credere, ‘think’ with infinitive complement: (22)
Sam wants to eat a cake that’s tasty. Sam wants [S PROJ to eat [NP a cake [that’s tasty PROJ] ] ]
Parallel to (27):
[Italian] Pavarotti crede di essere un genio. Pavarotti believes COMP be a genius.
(27)
‘Pavarotti believes that he’s a genius.’ [Lit.: “Pavarotti believes to be a genius”] (Anand, 2006, no. 1a; based on Chierchia, 1989)
Sami wants to eat a cake that tastes good to himi.
5. Further Evidence: Parallel with Epistemic Modals
Pavarotti thinks, “I’m a genius” Æ (22) is true. Pavarotti listens to recording, thinks “that guy is a genius” (22) is false.
5.1. Judge-dependent view of epistemic modals
[A standard view (Chierchia, 1989): treat the embedded proposition as a property and make attitudes like want and credere, ‘think’ ambiguous between a proposition-taking and a property-taking view.]
(28)
[[might φ]]w,j = 1 iff ∃∈Epistw,j: [[φ]]w′,x = 1
(29)
Epistemic Alternatives Epistw,z = { : it is compatible with x’s knowledge in w that x is y in w′}
My proposal: immediately derives the de se interpretation of PRO in these attitude reports while keeping all propositions the same type (now ) and giving each attitude report only one meaning for both thatclauses and infinitive clauses.2 For example: (23)
= 1 iff ∀∈WantPavarotti,w: [[PROJ to be famous]]
=1
= 1 iff ∀∈WantPavarotti,w: y is famous in w′ 4.2. A locality requirement Another property of PRO: must be linked to (controlled by) the nearest attitude holder:
2
Argued in recent work (Stephenson, 2005, 2006, 2007): Epistemic modals are also judge-dependent:
Crucial difference between epistemic modals and predicates of personal taste: Epistemic modals have judge dependency built into them (cannot take a contextually salient argument) 5.2. Epistemic modals in attitude reports
[[(21)]]w,j = [[want]]w,j ( [[Pavarotti]]w,j ) ( [λw″ . [λj″ . [[PROJ to be famous]]w″,j″] ] ) w′,y
(24)
6
Sue wants Bill to want PRO to go to the party. PRO = Bill (not Sue)
Of course, there is an obvious connection between the property-view and the judge-dependent view, since sets of world-individual pairs are exactly what “properties” are standardly taken to be. The difference here is that there are independent reasons to think that the individual “judge” parameter is needed (as argued by Lasersohn, 2005, 2006; Stephenson, 2006, 2007).
(30)
[[z thinks that might φ]]w,j = 1 iff ∀∈Doxz,w: ∃∈Epistw′,x: [[φ]]w″,y = 1 = z thinks that it’s compatible with z’s own knowledge that φ
Note: This is the only available reading since epistemic modals have the judge dependency built right in. In attitude reports, epistemic modals are de se with the same locality requirement.
7
Tamina Stephenson
Relativism and the De Se Interpretation of PRO
5.3. Obligatory de se interpretation
6. Conclusion
Epistemic modals in attitude reports are also obligatorily de se (Pranav Anand, p.c.):
Essential proposal: New twist on the property view of controlled PRO using the “judge” parameter.
(31)
Advantages:
[Context: Sam, a disgruntled NASA spokesperson, makes a false announcement on a radio show that there is new evidence of water on the moon. He is fired, goes home, and gets so drunk that when he hears a clip of his own interview on the radio, he doesn’t recognize it as himself. He says to himself, “That idiot doesn’t know that there’s no water on the moon!”]
[This is predicted from (30)]
• All sentences / clauses have the same semantic type (since propositions have an individual parameter anyway) • The semantics of propositions (type ) naturally fits together with doxastic alternatives and related notions (which are independently needed for de se attitudes) • Captures parallel between PRO and epistemic modals in attitude reports
5.4. Control-like locality requirement
A remaining question:
Epistemic modals in attitude reports must be linked to the closest attitude holder:
• Other types of control: Object control (e.g., persuade): can be analyzed as involving an attitude of the object BUT: extensional adjuncts (e.g., while singing) behave differently. [General Q. To what extent is “control” really a uniform phenomenon?]
# Sam thinks there might be water on the moon.
(32)
Sue thinks that Bill thinks it might be raining. might = compatible with Bill’s (not Sue’s) knowledge
This follows from the compositional semantics: (33)
[[(32)]]w,j = [[think]]w,j ( [λw′ . [λj′ . [[Bill thinks it might be raining]]w′,j′ ] ] ) ( Sue ) w,j
= [[think]]
([λw′ . [λj′ . ∀∈DoxBill,w′: ∃∈Epistw″,y: it’s raining in w′′′] ]) (Sue)
= 1 iff ∀∈Doxw,Sue: ∀∈DoxBill,w′: ∃∈Epistw″,y: it’s raining in w′′′ 5.5. Contrast: Shifting indexicals Shifting indexicals are also obligatorily de se (Schlenker, 2003; Anand, 2006, etc.).3 But: They are not generally subject to a control-like locality requirement: (34)
[Zazaki] Ali Fatima-ra va kε Rojda Ali Fatima-to said that Rojda
Bill-ra va εz to-ra miradiša Bill-to said I you-to angry.be.PRES
(1 reading =) ‘Ali said to Fatima, “Rojda said to Bill that I am angry at you” ’ [Lit: Ali said to Fatima that Rojda said to Bill that I am angry at you] (Anand & Nevins, 2004) Suggestion: There are (at least) two separate kinds of de se items4, some of which are derived via Kaplanian “monsters” and some of which are derived by relativist semantics. A requirement akin to “local control” is a diagnostic for the second kind.
3 4
West African-type logophors pattern the same way (Schlenker, 2003; citing, e.g., Kusumoto, 1998, for Bafut). Anand (2006) also argues for more than kind of de se item, for different reasons.
References Anand, Pranav (2006). De De Se. Ph.D. thesis, MIT. Anand, Pranav & Andrew Nevins (2004). Shifty Operators in Changing Contexts: Indexicals in Zazaki and Slave. SALT 14. Chierchia, Gennaro (1989). Anaphora and Attitudes De Se. In R. Bartsch, J. van Benthem & P. van Emde Boas (eds.), Language in Context. Epstein, Samuel David (1984). Quantifier-pro and the LF Representation of PROarb. LI 15. Kusumoto, Kiyomi (1998). Tenses as Logophoric Pronouns. Handout of talk given at the MIT/UConn/UMass Semantics Workshop, University of Connecticut, Storrs. Lasersohn, Peter (2005). Context Dependence, Disagreement, and Predicates of Personal Taste. L&P 28. Lasersohn, Peter (2006). Relative Truth, Speaker Commitment, and Control of Implicit Arguments. NELS 37. Lewis, David (1979). Attitudes De Dicto and De Se. Philosophical Review 88. Morgan, Jerry (1970). On the Criterion of Identity for Noun Phrase Deletion. Proceedings of CLS 6. Schlenker, Philippe (2003). A Plea for Monsters. L&P 26. Stephenson, Tamina (2005). Assessor Sensitivity: Epistemic Modals and Predicates of Personal Taste. In J. Gajewski, V. Hacquard, B. Nickel & S. Yalcin (eds.), New Work on Modality. Stephenson, Tamina (2006). A Parallel Account of Epistemic Modals and Predicates of Personal Taste. SuB 11. Stephenson, Tamina (2007). Towards a Subjective Theory of Meaning. Ph.D. thesis, MIT.
8
