Permission to Change

In this paper I discuss how to account for the performative effects of imperatives, and concentrate mainly on permission sentences. In the first part of the paper I argue that the performative effects of permission sentences should be accounted for in terms of a context change theory by making use of contraction defined in terms of an ordering relation, and show also how this ordering relation evolves from permission to permission. In the second part a problem for this analysis is discussed, i.e. the problem of conjunctive permission sentences. I develop two ways to solve this problem. First, I suggest that this problem is due to the wrong way of accounting for contraction, and propose an alternative way in which contraction can be defined that accounts for the performative effects of conjunctive permissions in a more satisfactory way. Although the analysis is appealing, I will argue that we should account for the problem by means of a type-shift analysis. 1 I N T R O D U C T I O N According to Austin's (1962) speech act analysis, by using sentences we perform certain linguistic acts. The traditional problem for speech act analyses was to find interesting types of speech acts, and to find necessary and sufficient conditions for the successful performance of the act. In later theories of context change the emphasis was on the essential effects of speech acts. In Stalnaker's (1978) classical analysis of assertions, for instance, the essential effect of an assertion is said to be the change it brings about in what is commonly assumed; the context that represents what is commonly assumed is incremented with the content of the assertion. More recently, the essential effects of other speech acts, like questions and denials, have been analyzed in terms of context change, too. The goal of this paper is to follow up some classical papers of Kamp and Lewis to characterize in a context change theory the essential effects of imperatives, normally stated in terms of command and permission sentences of the form You must/may do A. In sections 2 and 3, I will discuss whether we can account for the performative effects of command/obligation and permission sentences by treating them assertorically, or whether we should treat them as explicit performatives. I will argue for the latter option. This gives rise to the question in what way command and permission sentences change the context. In the third section I will show how we can account for the 120 Permission to Change change when we represent the prior information state by a preference order, and how in terms of this change we can predict most intuitions about coordination connectives and quantificational determiners. In section 4, I discuss the problem how imperatives determine the preference order of the posterior information state, relate the account of the third section with Veltman's analysis of defaults in update semantics, and with analyses of iterated revision. In the sections that follow it, I will concentrate on a particular problem for our analysis that was noted first by Merin (1992): the problem of conjunctive permissions. First, I will show how the standard analysis of contraction used for the analysis of permission sentences gives rise to this problem, and in the last two sections I will discuss some ways in which the problem can be solved when we give up the assumption that all that counts are the propositions that are expressed by the embedded clauses of permission sentences. 2 PERFORMATIVE EFFECT IN TERMS OF ASSERTION According to Austin's classical analysis of speech acts, sentences of the form You must/may do A are not used to describe a states of affairs. In terms of the language game between master and slave as described by Lewis (1970/79), they are typically used by one person, the master, to command or permit another person, the slave, to do certain things. How should we account for these so-called performative effects of the sentences used by the master? One proposal might be to say that command and permission sentences are assertorically used, but that the performative effect is accounted for in an indirect way, due to the fact that we learn, or realize, more about the world (Kamp 1979). One of the things one might learn about the world is what is demanded and permitted. A truth conditional analysis of what is demanded and permitted is given in deontic logic. Standard deontic logic (SDL) was based on the same principles as classical modal logic. Where normal modal logic has the operators • and O standing for necessity and possibility, SDL has the two operators O and P, standing for ought or obliged and for permission, respectively. Just like • 1 There are other truth conditional analysis of deontic concepts, of course. The analyses of Hansson (1990) and Meyer (1993) certainly belong to the most interesting ones. Other approaches claim that to talk about deontic discourse, we should also take the notion of tense seriously. Although these approaches all have something interesting to say, I will not discuss them here, but I should note that they go some (but not the whole) way in solving the problems we will discuss in this paper. For a discussion of a number of approaches towards the closely related analysis of bouletic concepts like 'desire' and 'intention', see van Rooy (1999).