LSA Annual Meeting [Semantics: Tense & modality]

January 7, 2007, Anaheim, California

Predicates of Personal Taste and Epistemic Modals* Tamina Stephenson Massachusetts Institute of Technology web: http://web.mit.edu/tamina/www/; e-mail: [email protected] Goal: Give a unified, relativist account of epistemic modals and predicates of personal taste [PPTs] (fun, tasty), placed within a Stalnakerian pragmatics.

1. The Facts 1.1. Parallel 1: Autocentric Disagreement1 A speaker may make and assess assertions involving PPTs / epistemic modals according to their own knowledge or experience (Lasersohn 2005, Egan, Hawthorne, & Weatherson 2005, MacFarlane 2006). (1)

Mary: Sam: Sue:

Where’s Bill? I’m not sure. (a) He might be in his office. (b) Nuh-uh / No, he can’t be. (c) He never works on Fridays.

(1a) OK if it’s compatible with Sam’s knowledge that Bill is in his office. (1b) OK if it is not compatible with Sue’s knowledge that Bill is in his office. (2)

Mary: Sam: Sue:

How’s the cake? (a) It’s tasty. (b) Nuh-uh / No it isn’t, it isn’t tasty at all!

(2a) OK if the cake tastes good to Sam, (2b) OK if the cake does not taste good to Sue. Why this is puzzling: Each speaker seems to be taking themselves to be the person whose knowledge / experience is relevant, but these items are clearly not indexicals (cf. (3)): (3)

Mary: I’m a doctor. Sam: # No, I’m not!

1.2. Parallel 2: Linked Readings in Attitude Reports When epistemic modals and PPTs are embedded under think, the most natural interpretation links the relevant knowledge / experience to the attitude holder.2 *

Earlier versions of this work were presented at Sinn und Bedeutung 11 and appeared in MITWPL 51. For comments, discussion, and judgments, I would like to thank Pranav Anand, Kai von Fintel, Danny Fox, Valentine Hacquard, Irene Heim, Sarah Hulsey, Sabine Iatridou, Ezra Keshet, Angelika Kratzer, John MacFarlane, Eric McCready, Jillian Mills, Craige Roberts, Robert Stalnaker, Seth Yalcin, the editors and reviewers of MITWPL 51, and audiences at MIT, SuB 11, and SNEWS. 1 Moltmann (2005) independently makes similar observations about the parallels between predicates of personal taste and epistemic modals, and gives brief suggestions towards a parallel analysis. 2 This extends to quantified subjects as in (i)–(ii). (i) [Every boy]i thinks hei must be stupid. ≈ for every boy x, it follows from x’s beliefs/knowledge that x is stupid

Predicates of Personal Taste and Epistemic Modals

(4)

Sam thinks it might be raining. ≈ it’s compatible with Sam’s knowledge/beliefs that it’s raining

(5)

Sam thinks that the roller coaster is fun. ≈ the roller coaster is fun for Sam

2

1.3. Contrast 1: Un-Linked Readings in Attitude Reports The linked interpretation is obligatory with epistemic modals but optional with PPTs: (6)

Mary: Sam:

How’s that new brand of cat food you bought? I think it’s tasty, because the cat has eaten a lot of it.

[tasty = tastes good to the cat] (7)

[Kai von Fintel, p.c.]

Mary: Wow, the dog really likes the dog food you’re feeding him. Sam: (#)I think it might be table scraps. [might ≠ according to the dog’s knowledge]

1.4. Contrast 2: Linking to Salient Referent More generally, PPTs (but not epistemic modals) can be linked to a salient referent: (8)

[Context: Sam is watching his cat eat a new brand of cat food. After a few hesitant bites the cat devours it eagerly. Sam says:] Oh good, the new cat food is tasty. [tasty = tastes good to the cat]

(9)

[cf. Lasersohn (2005): (39)]

[Context: Sam is watching his dog eat dog food. The dog is eating the food with enthusiasm that he usually saves for table scraps. Sam says:] # Oh good, that might be table scraps. [might ≠ according to the dog’s knowledge]

2. Semantic Proposal [Note: My account essentially extends Lasersohn (2005) to epistemic modals, with crucial modifications. It also owes an obvious debt to the context-relativist views of MacFarlane and Egan/Hawthorne/Weatherson. For reasons of time, I focus on my version, with footnotes indicating which parts differ from Lasersohn.]

(ii)

[No boy]i thinks hisi dinner is tasty. ≈ for no boy x does x think that x’s dinner tastes good to x. Speas (2004) and von Fintel & Iatridou (2003) use examples like (i) to make somewhat different points.

3

Tamina Stephenson

2.1. Relativizing Truth Following Lasersohn (2005), add an individual “judge” parameter to the index. In other words: A proposition is a set of world-time-individual triples. (10)

Kaplan (1989): Character

Content Context values of indexicals

Truth Value Index world/time judge of evaluation

2.2. Conventions etc. [[α]]u; w,t,j = the denotation of α w.r.t. context u and index (world w, time t, and judge j)

(11)

For example: [[be-a-doctor]]u; w,t,j = [λxe . x is a doctor in w at t] [[I]]u; w,t,j = the speaker of u

(12)

Rules of interpretation: Given a complex expression α formed from β and γ:

(13) (a)

If [[γ]]u; w,t,j ∈ Dom ( [[β]]u; w,t,j ), then [[α]]u; w,t,j = [[β]]u; w,t,j ( [[γ]]u; w,t,j ).

(b)

If [λw’s . [λt’i . [λj’e . [[γ]]u; w’,t’,j’ ] ] ] ∈ Dom ( [[β]]u; w,t,j ), then [[α]]u; w,t,j = [[β]]u; w,t,j ( [λw’s . [λt’i . [λj’e . [[γ]]u; w’,t’,j’ ] ] ] ).

2.3. Epistemic Modals Basic idea: The relevant “knower” is the judge: (14) (a)

[[might]]u; w,t,j = [λp> . ∃∈Doxw,t,j: p(w’)(t’)(x) = 1]

(b)

[[must]]u; w,t,j = [λp> . ∀∈Doxw,t,j: p(w’)(t’)(x) = 1]

Doxastic Alternatives (Lewis 1979, Chierchia 1989)3: (15)

3

Doxw,t,x = { : for all x knows in w at t, it is possible that he/she/it is y in w’ at t’} Read: “the doxastic alternatives of x in w at t”

For simplicity, my system does not make a distinction between knowledge and belief, but it can be modified to do so. I am also making simplifying assumptions about the general semantics for modals.

Predicates of Personal Taste and Epistemic Modals

4

2.4. Predicates of Personal Taste (PPTs)4 On their basic meaning, PPTs are two-place predicates:5 (16)

[[tasty]]u; w,t,j = [λxe . [λye . y tastes good to x in w at t] ] [[fun]]u; w,t,j = [λxe . [λye . y is fun for x in w at t] ]

Options for argument: • Silent nominal item: (17)

[[PROJ]]u; w,t,j = j

[Note: PROJ is NOT a “pronoun” in the sense of being able to be bound, etc.] • Null referential pronouns.6 (18)

Shorthand: proA = silent pronoun referring to individual A

• Overt PP argument (in some cases) Assume that P’s are vacuous: so e.g., [[fun for Sam]] = [[fun]] ([[Sam]]) 2.5. Attitude Reports Attitude predicates such as think shift the judge parameter to the attitude holder7: (19)

[[think]]u; w,t,j = [λp> . [λxe . ∀∈Doxw,t,x : p() = 1] ] ↑ ↑

So “x thinks that p” ≈ “x thinks that p is true with x as judge”

3. Pragmatic Proposal Basis: theory of conversation from Stalnaker (1978, 2002), etc. (20)

Common ground: set of world-time-judge triples (instead of worlds / world-time pairs), where every j = the group of conversational participants

(21)

Assertion: proposal to restrict the common ground to world-time-judge triples in which the proposition asserted is true.

4

Lasersohn treats predicates of personal taste as directly judge-dependent, giving them lexical entries like the ones I give to epistemic modals. So (e.g.) [[tasty PROJ]] for me is equivalent to [[tasty]] for Lasersohn. 5 There are a number of other ways to encode the difference I am proposing between the two classes. For example, PPTs could be systematically ambiguous, or epistemic modals could take an individual argument that ends up having to always be PROJ to satisfy a selection requirement of some kind. 6 I leave to future work the question of whether these null referential pronouns can be bound like ordinary pronouns. 7 This is based on an earlier proposal by Lasersohn in a pre-published version, somewhat reformulated.

5

Tamina Stephenson

Accepting an assertion: If A’s assertion that p is accepted, then in effect what is added to the common ground is that p is true as judged by the group of conversational participants. [This follows from the definition of the common ground and the role of assertion.] (22)

⇒ Norm of assertion: For speaker A to assert a sentence S in w at t (in context u), it must be the case that ∀∈Doxw,t,A: [[S]]u; w’,t’x = 1 i.e., A must believe that S is true as judged by A (not the whole group)8

4. Accounting for the Facts 4.1. Accounting for Autocentric Disagreement (23)

Mary: Sam: Sue:

How’s the cake? (a) It’s tasty. (b) No it isn’t, (c) it tastes terrible!

[tasty = tasty PROJ] [=(2)]

Semantics: (24)

[[(23a)]]u; w,t,j = [[The cake is tasty PROJ]]u; w,t,j = [[tasty]]u; w,t,j ( [[PROJ]]u; w,t,j ) ( [[the cake]]u; w’,t’,j’ ) = [λxe . [λye . y tastes good to x in w at t] ] (j) (the cake) = 1 iff the cake tastes good to j in w at t

• (23a) = proposal by Sam to restrict the common ground to triples such that the cake tastes good to j in w at t. [Requirement: for all of Sam’s doxastic alternatives the cake tastes good to y in w’ at t’.] • (23b) = objection by Sue to this proposal AND counterproposal to restrict the common ground to triples such that the cake does not taste good to j in w at t. [Requirement: for all of Sue’s doxastic alternatives , the cake does not taste good to y in w’ at t’.] Similarly for (25) with epistemic modals: (25)

Mary: Where’s Bill? Sam: I’m not sure. (a) He might be in his office. Sue: (b) Nuh-uh, he can’t be. (c) He never works on Fridays. [OR] No, he can’t be…. [=(1)]

Semantics: (26)

[[(25a)]]u; w,t,j = [[ might [Bill be in his office] ]]u; w,t,j = [[might]]u; w,t,j ( [λw’s . [λt’i . [λj’e . [[Bill be in his office]]u; w’,t’,j’ ] ] ] )

8

This is a way of formulating Lasersohn’s notion of an autocentric perspective.

Predicates of Personal Taste and Epistemic Modals

6

= [λp> . ∃∈Doxw,t,j: p(w’)(t’)(x) = 1] ( [λw”s . [λt”i . [λj”e . [[Bill be in his office]]u; w”,t”,j” ] ] ] ) = 1 iff ∃∈Doxw,t,j: Bill is in his office in w’ at t’ • (25a) = proposal by Sam to restrict the common ground to triples such that for some among j’s doxastic alternatives, Bill is in his office in w’ at t’. [Requirement: for all of Sam’s doxastic alternatives , there is some among j’s doxastic alternatives such that Bill is in his office in w” at t”.] Æ This amounts to: there is some among Sam’s doxastic alternatives such that Bill is in his office in w’ at t’ (assuming that the doxastic alternative relation is transitive and reflexive). • (25b) = objection and counterproposal by Sue to restrict the common ground to triples such that for all among j’s doxastic alternatives, Bill is not in his office in w’ at t’. [Requirement amounts to: for all of Sue’s doxastic alternatives , Bill is not in his office in w’ at t’.] 4.2. Accounting for Linked Readings in Attitude Reports (27)

Sam thinks it might be raining. ≈ it’s compatible with Sam’s knowledge that it’s raining

[=(4)]

Semantics: [[(27)]]u; w,t,j = [[thinks]]u; w,t,j ( [λw”s . [λt”i . [λj”e [[it might be raining]]u; w”,t”,j” ] ] ] ) ([[Sam]]u; w,t,j)

(28)

= [λp> . [λxe . ∀∈Doxw,t,x : p() = 1] ] ( [λw”s . [λt”i . [λj”e . [[it might be raining]]u; w”,t”,j” ] ] ] ) ([[Sam]]u; w,t,j) = 1 iff ∀∈Doxw,t,Sam : [[it might be raining]]u; w’,t’,y = 1 (a) Æ

= 1 iff ∀∈Doxw,t,Sam : ∃∈Doxw’,t’,y: It’s raining in w” at t” [Assuming that the doxastic alternative relation is transitive and reflexive]

(b) (29)

= 1 iff ∃∈Doxw,t,Sam : It’s raining in w’ at t’ Sam thinks that the roller coaster is fun. ≈ the roller coaster is fun for Sam

[=(5)]

Semantics: (30)

[[(29)]]u; w,t,j = [[thinks]]u; w,t,j ( [λw”s . [λt”i . [λj”e [[the roller coaster is fun]]u; w”,t”,j” ] ] ] ) ([[Sam]]u; w,t,j) = [λps,> . [λxe . ∀∈Doxw,t,x : p() = 1] ] ( [λw”s . [λt”i . [λj”e [[the roller coaster is fun]]u; w”,t”,j” ] ] ] ) ([[Sam]]u; w,t,j)

7

Tamina Stephenson

(a) Æ

= 1 iff ∀∈Doxw,t,Sam : the roller coaster is fun for y in w’ at t’ [Assuming that Sam is aware of his own experience]

(b)

= 1 iff the roller coaster is fun for Sam in w at t

4.3. Freely Linked Readings with Predicates of Personal Taste Since predicates of personal taste have the option of taking a silent referential argument, it is possible to link them to a salient referent, even in an attitude report. In (31), tasty takes a silent argument referring to the cat (“prothe cat”). (31)

[Context: Sam is watching his cat eat a new brand of cat food. After a few hesitant bites the cat devours it eagerly. Sam says:] Oh good, the new cat food is tasty. [tasty = tastes good to the cat]

[(=9)]

Semantics: (32)

[[(31)]]u; w,t,j = [[the cat food is tasty prothe cat]]u; w,t,j = [[tasty]]u; w,t,j ( [[prothe cat]]u; w,t,j ) ( [[the cat food]]u; w,t,j ) = [λxe . [λye . y tastes good to x in w at t] ] ( the cat ) ( the cat food ) = 1 iff the cat food tastes good to the cat in w at t

Note: This is a non-judge-dependent proposition. Similarly in an attitude report: (33)

Mary: Sam:

How’s that new brand of cat food you bought? I think it’s tasty, because the cat has eaten a lot of it. [tasty = tastes good to the cat]

[=(6)]

Semantics: (34)

[[I think the cat food is tasty prothe cat]]u; w,t,j = [[think]]u; w,t,j ( [λw’s . [λt’i . [λj’e . [[the cat food is tasty prothe cat]]u; w’,t’,j’ ] ] ] ) ( [[I]]u; w,t,j ) = [λps,> . [λxe . ∀∈Doxw,t,x : p() = 1] ] ( [λw”s . [λt”i . [λj”e . [[the cat food is tasty prothe cat]]u; w”,t”,j” ] ] ] ) ( [[I]]u; w,t,j ) = 1 iff ∀∈Doxw,t,Sam : the cat food tastes good to the cat in w’ at t’

Note: Shifting the judge has no effect since the embedded clause is non-judge-dependent.

Predicates of Personal Taste and Epistemic Modals

8

4.4. No Freely Linked Readings with Epistemic Modals Epistemic modals do not take an individual argument, so Bill might be in his office, e.g., cannot be made into a non-judge-dependent proposition. This rules out (7) and (9) above.

5. Summary 5.1. Semantic proposal • Lasersohn’s “judge” parameter as part of the index • Predicates of personal taste take a syntactic argument (either referential or PROJ) • Epistemic modals are directly dependent on the judge • think shifts the judge parameter to its subject 5.2. Pragmatic proposal • The common ground consists of triples , where w and t vary and j is the plurality of the participants in the conversation • The norm of assertion is weak in a crucial sense: the speaker only needs to believe that the content of the assertion is true as judged by them 5.3. Main results • Accounts for disagreements with PPTs (like Lasersohn) and with epistemic modals. • Accounts for the contrast between epistemic modals and PPTs in being able to be linked to salient referents. • Accounts for shifted interpretation of PPTs / epistemic modals in attitude reports.

6. References Chierchia, G. (1989). Anaphora and Attitudes De Se. In R. Bartsch, J. van Benthem, & P. van Emde Boas (eds.), Language in Context, Dordrecht: Foris. Egan, A., Hawthorne, J. & Weatherson, B. (2005). Epistemic Modals in Context. In G. Preyer & G. Peter (eds.), Contextualism in Philosophy: Knowledge, Meaning, and Truth, Oxford Univ. Press. von Fintel, K. & Iatridou, S. (2003). Epistemic Containment. Linguistic Inquiry 34;2. Kaplan, D. (1989). Demonstratives. In J. Almog, J. Perry & H. Wettstein (eds.), Themes from Kaplan, Oxford Univ. Press. Lasersohn, P. (2005). Context Dependence, Disagreement, and Predicates of Personal Taste. Linguistics and Philosophy 28;6. Lewis, D. K. (1979). Attitudes De Dicto and De Se. Philosophical Review 88;4. MacFarlane, J. (2006). Epistemic Modals Are Assessment-Sensitive. Ms., http://sophos.berkeley.edu/macfarlane/epismod.pdf. Moltmann, F. (2005). Relative Truth and the First Person. Ms., http://semanticsarchive.net/Archive/mY2NGJhY/. Speas, P. (2004). Person (and Mood and Tense) and Indexicality. Harvard Workshop on Indexicals, Speech Acts, and Logophors, November 20, 2004. Stalnaker, R. (1978). Assertion. Reprinted in P. Portner & B. H. Partee (eds.), 2002, Formal Semantics: The Essential Readings, Oxford: Blackwell. Stalnaker, R. (2002). Common Ground. Linguistics and Philosophy 25;5-6. Stephenson, T. (2005). Assessor Sensitivity: Epistemic Modals and Predicates of Personal Taste. In J. Gajewski, V. Hacquard, B. Nickel & S. Yalcin (eds.), New Work on Modality. MITWPL 51, Cambridge, Mass.: MIT Working Papers in Linguistics.