Permission and Modality

Paul Portner Georgetown University March 20, 2009 at the University of Connecticut

1

Introduction

Here are the two theoretical ideas in play for the semantic analysis of imperatives: 1. The dynamic theory Imperatives update a component of the discourse context, the To-do List (Portner 2004; 2007), parallel to the way declaratives update the Common Ground. 2. The modal theory Imperatives are modal sentences, containing a phonologically null version of a deontic modal (Schwager 2005b, Grosz 2009, Han 1999, Han to appear). Through some mechanism or another, they are obligatorily performative. In this talk, I’ll investigate the relevance of permission sentences to this debate. One can give a permission by using a modal sentence, as in (1a), or by using an imperative, as in (1b): (1) (a) (b) You may take an apple. Take an apple! (Please!/Go ahead!/If you like.)

When it comes to permission imperatives, the dynamic theory is committed to their receiving the same basic analysis as imperatives which express commands and the like. The difference should lie in some aspect of the context. The modal theory has two branches: Schwager (2005b) treats imperatives as uniformly containing a strong (necessity) modal, and so handles permission pragmatically, like the dynamic theory. Grosz (2009), in contrast, argues that permission imperatives contain a weak (possibility) modal, in contrast to command-type imperatives. I will investigate the question of whether permission imperatives should be represented with a covert weak modal operator. My conclusion is that they should not be. Since the weak-modal analysis of permission imperatives provides one of the two main reasons to suggest the presence of a modal in imperatives generally (the other being conditional imperatives), the feeds into a broader argument against the modal theory. I’ll organize the discussion are the following two problems: 1. Permission is difficult to model formally. Lewis (1979) pointed out the original “problem about permission”: while it is easy to give a 1

formal pragmatic theory, in terms of possible worlds, of commanding, it is not equally easy to give a parallel account of permitting. I will present an analysis of permission in terms of the dynamic theory, building on van Rooij (2000), and argue that the modal theory cannot provide a satisfactory analysis, unless it incorporates the dynamic theory into itself, a move which would undermine the rationale for the modal theory. 2. The permission reading of an imperative need not be formally marked. A modal sentence expressing a command uses a necessity operator, while one expressing permission typically uses a possibility operator: (2) (a) You must bring in a chair from the other room. (b) You may bring in a chair from the other room. In contrast, an imperative can express either reading: (3) Bring in a chair from the other room!

This formal similarity leads is what leads to the claim that permission imperatives are not semantically different from those expressing commands (Sperber and Wilson 1988, Portner 2004; 2007, Schwager 2005b); however, Grosz (2008) points out that certain particles in German can disambiguate imperative sentences, and he takes this to be evidence that imperatives contain either a necessity or a possibility modal operator. I will argue that the dynamic theory is able to explain the data Grosz cites even better than the modal theory.

2

Background on imperative semantics and pragmatics

Assertion is commonly analyzed in terms of Stalnaker’s concept of common ground (Stalnaker 1974, 1978). Parallel to this, asking a question has been analyzed in terms of a second discourse component, what Ginzburg calls the ‘Question Under Discussion Stack’ (Ginzburg 1995a; 1995b, Roberts 1996). Portner (2004) proposes that imperatives are interpreted as contributing to the addressee’s To-Do List. (4) Pragmatic Function of Imperatives a. The To-Do List function T assigns to each participant α in the conversation a set of properties T (α). b. The canonical discourse function of an imperative clause φimp is to add [[ φimp ]] to T (addressee). Where C is a context of the form CG, Q, T : C + φimp = CG, Q, T [addressee/(T (addressee) ∪ {[[ φimp ]] })] The To-do List is similar to ideas in Lewis (1979), Han (1998), Potts (2003) and Roberts (2004). What’s different is the ordering pragmatics for imperatives. In particular, the To-Do List functions to impose an ordering on the worlds compatible with the Common Ground, and this ordering determines what actions an agent is committed to taking (Portner 2004): 2

(5)

Partial ordering of worlds: For any w , w ∈ ∩CG and any participant i, w < i w iff for some P ∈ T (i), P (w )(i) =  and P (w )(i) = , and for all Q ∈ T (i), if Q(w )(i) = , then Q(w )(i) = . Agent’s commitment: For any participant i, the participants in the conversation mutually agree to deem i’s actions rational and cooperative to the extent that those actions in any world w ∈ ∩CG tend to make it more likely that there is no w ∈ ∩CG such that w < i w .

(6)

Portner (2007) points out that the range of pragmatic functions displayed by imperatives is parallel to the readings of priority modals (that is, non epistemic, non-dynamic modals): (7) (8) (9) a. Sit down right now! (order) b. Noah should sit down right now, given that he’s been ordered to do so. (deontic) a. Have a piece of fruit! (invitation) b. Noah should have a piece of fruit, given that it would make him happy. (bouletic) a. Talk to your advisor more often! (suggestion) b. Noah should talk to his advisor more often, given that he wants to finish his degree. (teleological) The differences among the (b) sentences involve the choice of ordering source (Kratzer 1981). Portner (2007) analyzes the differences among the (a) sentences as involving different subsets to the addressee’s To-do List. For example, an imperative expresses an order when it motivates an addition the To-do List on the basis of the speaker’s authority. There are pragmatic links between ordering sources and subsets of the To-do List. The question for today is how permission interpretations of imperatives fit into this picture. I think that the literature would classify the invitation in (8a) as a permission imperative; Portner (2007) analyzes invitation as adding a property to the addressee’s To-do List on the grounds that it will lead to the better satisfaction of her desires. In Portner (2007), I tentatively suggest that permission in a stricter sense can be described as like invitation, with the additional presupposition that the speaker has the authority to prohibit the act in question. An example of this would be the following: (10) Carry rocks every day! [. . . Weeks pass . . . ] Tomorrow, take the day off!

3
3.1

The pragmatics of permission
The dynamic theory

The above account treats permission imperatives as adding a requirement to the To-do List. They differ from strong imperatives like commands only in the grounds which support the requirement. There is a common intuition, however, that permissions are weaker than commands, in the same way that possibility modals are weaker than necessity modals. Lewis (1979) proposes a model of commands for a language game involving three players, the Master, the Slave, and the Kibitzer. The key formal idea is the Sphere of Permissibility: the 3

Slave is responsible for making sure that the actual world remains a member of the Sphere. When the Master utters a sentence of the form !φ, the Sphere is changed as in follows (I’m simplifying Lewis’s theory here): (11) When the Master utters !φ, the Sphere of Permissibility S becomes S ∩ φ.

The problem about permission is that a similar, initially plausible principle, (12), does not work for permission: (12) When the Master utters ¡φ, the Sphere of Permissibility S becomes S ∪ φ.

(12) implies that any world in which φ is true is in the Sphere. Thus, in the case of (10), the Slave would be entitled to drink the Master’s wine, or kill he Master, or take the rest of the week off, provided he takes tomorrow off. Lewis himself suggests, and rejects, several solutions, among them one taken up and developed by van Rooij (2000). The idea is that at any point in time, we have not just a Sphere S, but an ordering which reflects the relative reprehensibility of worlds not in S. Lewis rejects this solution because he has no way of defining the ordering, in particular no way of explaining how it evolves through time. Van Rooij points out that if we replace S with something that determines both an ordering and a Sphere, we should be able to define how a series of utterances affect both things; however, he does not lay out how this works in a clear way. What I will show is that, because the To-do List determines an ordering, it allows us to formalize this solution in a simple way. Let us begin with the following example: (13) Monday carry rocks! Tuesday carry rocks! And Wednesday carry rocks! [. . . Tuesday comes along] Take tomorrow off!

After the first three imperatives are uttered, the Slave’s To-do List becomes {¬K(x, m), C(x, mo), C(x, tu), C(x, we)}.1 The pragmatics of the To-do List implies that the Slave must not kill the Master (not explicitly stated, but surely assumed) and must carry rocks each of the three days. Then on Tuesday, the Master gives permission to take Wednesday off. Thus, the To-do List becomes {¬K(x, m), C(x, mo), C(x, tu), C(x, we), ¬C(x, we)}. It’s impossible to make all five of these propositions true. The pragmatics of the To-do List now implies that the Slave does as well as he can: he either carries rocks Wednesday, or he doesn’t. Either way, he can make four of the five propositions in the To-do List true. (More accurately, after having worked Monday and Tuesday, and not killing the Master, either working on Wednesday or taking the day off makes it the case that there is no better-ranked world.) Crucially, killing the Master is still not an option. Example (10) poses a problem. It seems that after the first imperative is uttered, the To-do List should be {¬K(x, m), ∀d[C(x, d)]}. Adding ¬C(x, we) to this set leads to a situation where never working again is permitted. The problem is that once the Slave takes Wednesday off, ∀d[C(x, d)] is false, and no amount of work can make it true again. Skipping one day is no better than skipping many.
On the analysis of Portner (2004, 2007), the variable x should be abstracted over, to produce a property. I don’t insert the λx, both for simplicity, and to indicate the fact that the points under discussion here don’t depend on the decision to treat imperatives as properties, as opposed to propositions.
1

4

¬K(x,m),C(x,mo),C(x,tu),C(x,we)

¬K(x,m),C(x,mo),C(x,tu),¬C(x,we)

K(x,m),C(x,mo),C(x,tu),C(x,we)

K(x,m),C(x,mo),C(x,tu),¬C(x,we)

Figure 1: The choices on Wednesday

The solution to this problem is to expand the To-do List from {¬K(x, m), ∀d[C(x, d)]} to one more like like {¬K(x, m), ∀d[C(x, d)], C(x, mo), C(x, tu), C(x, we), C(x, th), C(x, f r)}. There might be interesting ways to implement this, for example by giving the quantifier scope over a force operator. But I prefer an uninteresting way: the Slave knows that he is to work on Monday, on Tuesday, on Wednesday, etc. So the To-do List is expanded by inference. In the terms of van Rooij (2000), these are “relevantly entailed” by Carry rocks every day! Once ¬C(x, we) is added to this set, we represent permission to take Wednesday off (or not), but still require work on the other days. The literature has discussed the puzzle of why (14a) entails (14b): (14) (a) (b) You may take an apple and a pear. You may take an apple (while not taking a pear).

This is a real issue for permission sentences involving an explicit modal. However, it seems to me that the corresponding imperative does not have this “disjunctive reading”. (15) (a) (b) Take an apple and a pear! Take an apple (while not taking a pear)!

There are contexts where (15a) allows that the addressee takes only an apple, but these involve a situation where the imperative is designed to overcome the (politness-based) reluctance of addressee to impose on the speaker (as in Sperber and Wilson 1988). That is, we assume a To-do List {¬T (x, a), ¬T (x, p)}. In this context, (15a) adds T (x, a) ∧ T (x, p), and the relevant entailments T (x, a), T (x, p). In this case, taking just an apple is as good as taking both an apple and a pear. In the first case, the addressee partially satisfies politeness (by not imposing) and partially satisfies the explicit imperative. In the second, she totally satisfies the imperative. (In fact, any choice by the addressee is permitted.)2
Note that this is a case where the fact that < i is a partial order becomes relevant; taking both satisfies three properties in the To-do List, as opposed to two when the speaker takes an apple but not a pear; however, these alternatives are not comparable by the ordering. The defined partial ordering makes for an improvement over van Rooij’s system.
2

5

¬T(x,a),¬T(x,p)

T(x,a),¬T(x,p)

¬T(x,a),T(x,p)

T(x,a),T(x,p), T(x,a)∧T(x,p)

Figure 2: All choices equally ok

3.2

The modal theory

How would the modal theory account for the pragmatics of permission? This depends on how it analyzes the performativity of imperatives generally. If it follows Lewis’ lead and tries to provide a pragmatic model, it will end up with something like the account above. This means that a modal sentence, assumed to be of the form M +φ, serves to add φ to the To-do List (or something similar). In other words, it accepts the dynamic account of permission, and must look elsewhere for evidence in favor of a modal operator. If it does not follow Lewis’ lead, it provide a different account of the discourse status of giving permission. The only one I can think of is a classical performative analysis. (16) A sentence φ is performative in context c if, by virtue of its being uttered in c, its truth in the context which emerges just after c is guaranteed.

For example, on a performative use, an utterance of the possibility modal sentence (1a) would automatically make it the case that it is compatible with the deontic background that the addressee takes an apple. A performative necessity modal, like (2a), would make it the case that the deontic background entails bringing a chair. An imperative with a strong reading would be like the strong modal. A permission imperative would be like either the performative possibility or the performative necessity sentence, depending on one’s view (Grosz would take the first option, Schwager the second). This way of thinking of things has several weaknesses. First, it fails to explain why imperatives are universal. If they are nothing but modal sentences, why shouldn’t some language perform this pragmatic function solely with ordinary modals? And second, it doesn’t really explain the discourse function of imperatives. The truth of a modal sentence (even its mutually presupposed truth) does not represent any commitment on the part of an individual. For example, I can explicitly accept that the deontic background entails that I will pay my taxes, and still deny any commitment to paying my taxes: (17) I know I should pay my taxes, but I won’t.

(This point is obscured somewhat if we choose must as the modal, but this merely shows that must 6

itself contributes an imperative-like component of meaning, as discussed by Ninan 2005.) The fact that an individual is committed to some action is precisely what the sphere of permissibility or To-do List represents.3

4
4.1

Evidence for the dynamic theory
Some differences between imperatives and modal sentences

If imperatives contain modals, we expect it to be possible to specify their conversational background explicitly. (18) (19) (20) (21) a. In view of the customs of the tribe, you may wear feathers on your hat. b. #In view of the customs of the tribe, wear features on your hat! a. In view of what the law provides, you must pay your taxes. b. #In view of what the law provides, pay your taxes! a. Legally, you can park here. b. #Legally, park here! a. Morally, you must give to charity. b. #Morally, give to charity!

4.2

Formal marking of permission readings

Grosz (2008) points out that the particles JA and bloß disambiguate imperatives towards a commandtype reading, and ruhig disambiguates towards a permission-type reading (his (2)-(3)):4 (22) (a) Iss *bloß/ *JA/ ruhig den Spinat! Das st¨rt u mich nicht. eat BLOß JA RUHIG the spinach that disturbs me not ‘Eat bloß/JA/ruhig the spinach! That doesnt disturb me.’ (b) Iss bloß/ JA/ *ruhig den Spinat! Sonst wirst du bestraft. eat BLOß JA RUHIG the spinach or.else will.be you punished ‘Eat bloß/JA/ruhig the spinach! Or else youll be punished.’ These particles also occur with performative priority modals. A performative modal is one that determines an additional speech act beyond that of assertion normally associated with declarative
One might suggest that the imperative-modal is necessarily interpreted with respect to a background like “In view of what x is committed to doing”. (Presumably, this background would function as an Ordering Source.) This strategy would amount to accepting the core of the dynamic account, since “what x is committed to doing” is nothing but the To-do List. The idea here would be that we never access the To-do list directly, but only indirectly, through a modal statement made in terms of it. An analogy would be the idea that assertions do not directly access the Common Ground, but rather are implicitly modalized, epistemic or doxastic statements. Some people like this idea about declaratives (e.g., Truckenbrodt 2006). I’d be happy with the conclusion that the dynamic theory of imperatives is as secure as Stalnaker’s classic account of assertion. 4 These forms are not always discourse particles, and we are to ignore the other uses. Furthermore, unstressed ja is assumed to be a different discourse particle.
3

7

clauses (Ninan 2005, Portner 2009). In the case of priority modals, the performativity is imperativelike. (23) You must finish your dinner.

Examples of these particles in sentences containing modals are following (from Grosz’s (8)-(9)): (24) (a) Der soll bloß/ JA den Spinat essen! he shall BLOß JA the spinach eat ‘He shall bloß/JA eat the spinach!’ (b) Der kann/ darf *bloß/ *JA den Spinat essen. he can may BLOß JA the spinach eat ‘He can/may *bloß/ *JA eat the spinach.’ (c) Der kann/ darf sich ruhig was vom Fleisch nehmen. he can may self RUHIG what from.the meat take ‘He can/may ruhig take some meat.’ Note that the JA and bloß occur with necessity modals, and ruhig with possibility modals. Grosz concludes from these facts that the German particles must cooccur with an appropriate modal, and therefore that each imperative sentence contains a covert modal. According to Grosz, ruhig also occurs with sollen, leading to a permission reading (his (9a)): (25) Der soll sich ruhig was vom Fleisch nehmen. he shall self RUHIG what from.the meat take ‘He shall ruhig take some meat.’

This is problematical, because sollen is usually considered a necessity modal; thus, he must claim that it is ambiguous between a necessity and possibility reading. (According to Elena Herburger (p.c.), it’s not entirely clear that (25) really expresses weak modality, but I’ll take Grosz’s claims for granted here. On my account, this can be a permission-giving necessity sentence.)

4.3

The role of performativity

Grosz sees the generalization with these particles as “they must occur with a modal of the right sort”. This requires positing covert modals where there is no other evidence for them. It also doesn’t explain why all of the modals in question are performative. In contrast, I see the generalization as “they must occur in a performative sentence of the right sort”. It happens that some of these sentences contain a modal, while others (imperatives) do not. This analysis is not available to Grosz, because he does not give an account of performativity.5
5 It appears he would prefer an account along the lines of Schwager (2005a,b), but expanded so as to allow for both necessity and possibility operators. See Portner (2007) for criticisms of Schwager’s approach.

8

In Section 3 we saw that an imperative always results in permission. Depending on the nature of the To-do List, sometimes an imperative will additionally result in requirement. Let’s look again at Figure 1. We can say that ¬C(x, we) is permitted, because some best-ranked worlds are ones in which ¬C(x, we) is satisfied.6 (26) S is a permission sentence in context c if an utterance of it in c results, as a matter of its conventional meaning, in a context c in which some best-ranked worlds are in [[ S ]] . More precisely: • ∃w[w ∈ CG ∧ w ∈ [[ S ]] ∧ ¬∃w [w ∈ CG ∧ w < i w ]]

Take tomorrow off ! in (13) is a permission sentence. We can say ¬K(x, m) is required, because all best-ranked worlds are ones in which ¬K(x, m) is satisfied. (27) S is a requirement sentence in context c if an utterance of it in c results, as a matter of its conventional meaning, in a context c in which all best-ranked worlds are in [[ S ]] . More precisely: • ∀w[(w ∈ CG ∧ ¬∃w [w ∈ CG ∧ w < i w ) → w ∈ [[ S ]] ]]

Suppose the master says Thursday carry rocks! Then C(x, th) will become a requirement, because there is no countervailing requirement in the To-do List. Sentences with performative modals also count as permission or requirement sentences, and in fact the modal typically determines which. For example, (23) adds the requirement that the addressee finish his dinner to his To-do List. In terms of the present analysis, this means that it (i) adds the property F (x, d) to the To-do List (Ninan 2005), and (ii) presupposes that, as a result of adding this property, every best-ranked world in the common ground is one in which the addressee finishes his dinner. Similarly, (22a) (i) adds the property E(x, s) to the To-do List, and (ii) presupposes that, as a result, at least some best-ranked worlds are ones in which the addressee eats his spinach. (Normally (ii) will implicate that not all best-ranked worlds are such.) With these ideas in place, we can say that bloß and JA are only allowed in requirement sentences, and that ruhig is only allowed in permission sentences.

4.4

Grosz’s counterargument

Grosz (2009) argues against this analysis on the grounds that the same context could never support equally well both a permission imperative and a requirement imperative. He then gives data showing that the same preceding linguistic context can support both JA and ruhig (his (18)): (28) Context: John is aware of the three different legal systems [. . . ] and asks Sepp what the situation is in Austria. Sepp is a citizen and thus an authority of sorts.

6

I’m making the limit assumption, for simplicity.

9

John (driving): Darf ich das Licht aufdrehen? Gibt es eine Regelung? may I the light turn.on give it a regulation ‘May I turn on the light? Is there a regulation?’ Sepp: a. Ja, dreh das (ruhig/ #JA) auf! Das ist hier kein Problem. yes turn that RUHIG #JA on that is here no problem ‘Yes, turn it [RUHIG/ #JA] on! That’s not a problem here.’ ≈ In view of what the law prescribes, it’s possible that you turn it on. b. Ja, dreh das (JA/ #ruhig) auf! Das ist hier sogar Vorschrift! yes turn that JA/ #RUHIG on that is here even requirement ‘Yes, turn it [JA/ #RUHIG] on! That’s even required here!’ ≈ In view of what the law prescribes, it’s necessary that you turn it on. While technically true, this argument doesn’t take into account the fact that the form of a sentence frequently helps the addressee determine what the intended context is. (29) The water here is so clear. I think that trout can/may live in this river.

According to Kratzer’s theory of modality, can requires that the conversational background be circumstantial, whereas may requires that it be epistemic. If Grosz’s argument were correct, it should be impossible for both of them to be felicitous here. But of course, both are felicitous, because the choice of modal gives the hearer information about what conversational background is assumed. Technically, this is information about what the context is, a very mellow form of accommodation. Similarly, the choice of JA or ruhig gives the hearer information about what To-do List is assumed. In the case of (28a), ruhig shows that the To-do List contains an entry in conflict with ‘x turns the light on’. Perhaps you like to save energy. The resulting To-do List allows John to turn the lights on or leave them off, as he wishes. In contrast, in (28b), the To-do List must not contain an entry in conflict with ‘x turns the lights on.’

References
Ginzburg, Jonathan. 1995a. Resolving questions, part I. Linguistics and Philosophy 18:459–527. Ginzburg, Jonathan. 1995b. Resolving questions, part II. Linguistics and Philosophy 18:567–609. Grosz, Patrick. 2008. Particle-izing imperatives. Ms., MIT. Grosz, Patrick. 2009. German particles, modality, and the semantics of imperatives. To appear in The Proceedings of NELS 39. Han, Chung-Hye. 1998. The structure and interpretation of imperatives: mood and force in Universal Grammar. Doctoral Dissertation, University of Pennsylvania. 10

Han, Chung-Hye. 1999. Deontic modality, lexical aspect and the semantics of imperatives. In Linguistics in morning calm 4 . Seoul: Hanshin Publications. Han, Chung-Hye. to appear. Imperatives. In Semantics: An international handbook of natural language meaning, ed. Claudia Maienborn, Klaus von Heusinger, and Paul Portner. Mouton de Gruyter. Kratzer, Angelika. 1981. The notional category of modality. In Words, worlds, and contexts, ed. Hans-Jurgen Eikmeyer and Hannes Rieser, 38–74. Berlin: de Gruyter. Lewis, David K. 1979. A problem about permission. In Essays in honour of jaakko hintikka, ed. E. Saarinen, R. Hilpinen, I. Niiniluoto, and M. Provence Hintikka, 163–175. Dordrecht: Reidel. Ninan, Dilip. 2005. Two puzzles about deontic necessity. In New work on modality, ed. Jon Gajewski, Valentine Hacquard, Bernard Nickel, and Seth Yalcin, volume 51 of MIT Working Papers in Linguistics, 149–178. Cambridge, MA: MITWPL. Portner, Paul. 2004. The semantics of imperatives within a theory of clause types. In Proceedings of semantics and linguistic theory 14 , ed. Kazuha Watanabe and Robert B. Young, 235–252. Cornell University Linguistics Department: CLC Publications. URL http://semanticsarchive.net/Archive/mJlZGQ4N/. Portner, Paul. 2007. Imperatives and modals. Natural Language Semantics 15:351–383. Portner, Paul. 2009. Modality. Oxford: Oxford University Press. Potts, Christopher. 2003. Keeping world and will apart: A discourse-based semantics for imperatives. URL people.umass.edu/potts/potts-nyu-handout.pdf, talk delivered at the NYU Syntax/Semantics Lecture Series, October 17, 2003. Roberts, Craige. 1996. Information structure in discourse: Towards an integrated formal theory of pragmatics. In Papers in semantics, ed. Jae-Hak Toon and Andreas Kathol, OSU Working Papers in Linguistics, Vol 49, 91–136. Columbus: Department of Linguistics, the Ohio State University. Roberts, Craige. 2004. Context in dynamic interpretation. In Handbook of pragmatics, ed. Laurence Horn and Gregory Ward, 197–220. Oxford and Malden, MA: Blackwell. van Rooij, Robert. 2000. Permission to change. Journal of Semantics 17:119–145. Schwager, Magdalena. 2005a. Interpreting imperatives. Doctoral Dissertation, University of Frankfurt. Schwager, Magdalena. 2005b. Permitting permissions. In Proceedings of the tenth esslli student session, ed. Judit Gervain. Stalnaker, Robert. 1974. Pragmatic presupposition. In Semantics and philosophy, ed. M. Munitz and P. Unger, 197–213. New York: New York University Press. Stalnaker, Robert. 1978. Assertion. In Syntax and semantics 9: Pragmatics, ed. P. Cole, 315–332. New York: Academic Press. Truckenbrodt, Hubert. 2006. On the semantic motivation of syntactic verb movement to c in german. Theoretical Linguistics 32:257–306. 11

Wilson, Deirdre, and Dan Sperber. 1988. Mood and the analysis of non-declarative sentences. In Human agency: Language, duty and value, ed. J. Dancy, J.Moravcsik, and C. Taylor, 77–101. Stanford CA: Stanford University Press.

12