A way of sorting contextual possibilities

In general, an ordering source (OS) contains propositions which represent some kind of ideal. A world where the propositions in an OS are true is higher-ranked by some standard represented by that OS than worlds where some or all of them are not. We want, or expect, the propositions in an OS to be true at our world, but we know that not all of them are.

The ability to sort contextual possibilities is the function we require of an ordering source. An epistemic modal base tells us what propositions are known to be true by some individual, and we derive from that set the set of all worlds which are compatible with it, what he have called the set of accessible worlds. The ordering source orders those accessible worlds based on how ideal they are according to some standard. This division of labor between the modal base and ordering source allows us to derive human (maximally ideal) and slight (not maximally ideal) possibilities.

With deontic modals it is self-evident that some kind of "legal" or "regulatory" ideals are involved. A deontic conversational background is a set of propositions that “represent the content of a body of laws or regulations”

(Kratzer, 2012, p. 37). A proposition in a deontic conversational background sorts contextual possibilities in terms of legality and illegality. The maximally legal possibilities correspond to those propositions which are compatible with the max set based on a law- or rule-based deontic ordering source, identified as a conversational background.

Deontic conversational backgrounds tend to be finite lists provided by regulatory bodies such as governments, authority figures (teachers, parents), households, associations, etc. Typical laws and rules limit the behavior of individuals or the status of happenings of certain kinds, e.g. (i) people under 20 may not drink alcohol, (ii) only citizens may vote, (iii) only students who have finished their homework may participate in recess, (iv) smoking is prohibited on public property, etc. These laws in themselves only apply to certain individuals or situations. For instance, (i) tells us that it is illegal for people under 20 to drink. It does not tell us that 21 year olds

57

may drink alcohol, though we will usually assume that this is the case. In order to regulate society, laws pick some subclass to which they should apply.

In the case of a stereotypical ordering source, Kratzer describes the propositions as reflecting what is normal in the evaluation world according to "some suitable normalcy standard" in that world. Normalcy is not typically determined by explicit lists provided by regulatory bodies, and accordingly the exact nature of the propositions has been left (purposely) to the imagination,

"What is to count as normal? Definition (e) [i.e. the quoted portion above] is deliberately vague and non-committal about what suitable standards of normalcy are and where they may come from."

(Kratzer, 2012, p. 37)

There are many different standards one can use for determining that something is "normal," "usual," or

"stereotypical." The standards will vary depending on who is evaluating whatever it is that is being evaluated. Even so, we can find some empirical necessities in order for a proposition to represent a stereotypical state of events, which turn out to be very similar to the characteristics we saw in a deontic conversational background.

When we deem something "stereotypical" or "normal" we do so because we perceive it to belong to some class of individuals, events, or states which we perceive to usually behave in or appear in a certain way. For example, the proposition, “that it rains,” cannot indicate stereotypicalness without a subclass of states in which rain is supposed to occur, be it a subclass of time such as one of the months of the year, a subclass of environment such as a configuration of the sky, a subclass of personal experience such as when an individual goes to the beach. Likewise, the proposition,

“that people eat 3 meals a day” can only tell us eating 3 meals a day is stereotypical if we qualify which people it is that are supposed to eat 3 meals a day, whether it be all people, healthy people, people of a certain culture, etc.

Thus, we can see that both deontic and stereotypical ordering sources pick out some subset of states or individuals and associate that subset with a proposition. For purposes of analysis, conditionals are an optimal way of accomplishing such situational qualification because they have well-studied logical characteristics. Throughout this

58

dissertation, deontic and stereotypical ordering source propositions shall be expressed as conditionals, where the antecedent qualifies the situation and the consequent says what is legal/stereotypical of said situation³⁹. Using conditionals in the ordering source will allow clear predictions for truth conditions and inter-propositional relations.

For example, the following propositions are probably contained in the stereotypical conversational background that many individuals would generate the following, where each value for x and t determines a separate ordering source proposition.

g={ p1| If it’s thundering in x at t, it's raining in x at t ; p2| If it’s sunny in x at t, it's not raining in x at t}⁴⁰

Stereotypical states of events are in essence generalizations which individuals draw from their experiences.

Any given state of events is composed of an infinite number of characteristics, and any of those characteristics can be associated by an individual with some other characteristic. For individuals who count the above propositions as stereotypical states of affairs, most or at least many of their experiences with thunder will have coincided with rain, and most or at least many of their experiences with sunny weather will have coincided with no rain⁴¹. Thus, we get common associations like those in g above, but there may also be unfortunate individuals who consider "If I go to the beach at t, it rains at t," or "If my right knee hurts at t, it rains at t" a stereotypical state of affairs. There are countless more stereotypical states of affairs concerned with raining and not raining that could be included in g.

Returning to the problem of truth conditions, this multitude can make it difficult for Hearer or a researcher to pinpoint what states of affairs are relevant for a given modal utterance. In fact, it is often the case that Speaker himself does not have a clear idea of what he considers normal or why, even though he has an intuition that it is.

This does not hamper the ability of an ordering source to sort contextual possibilities.

When Speaker says "It might rain," it is not necessarily important why he thinks that "it rains" is part of a

39 Some ordering sources, such as bouletic ones, are not dependent on external conditions. Though they will scarcely be mentioned in this dissertation, where they are they will not be expressed as conditionals.

40 In this case the variable ranges over places and times, but there could also be variables for any type entity.

41 Nowhere does this experience require that thunder and sun never coincide, so we get two generalizations which can collide in the actual state of events. When they don’t collide, of course we’ll choose the generalization that fits the situation, but when they do collide will typically be able to assess the situation based on different generalizations.

59

stereotypical state of affairs. Speaker may feel entitled to make the assertion because he sees someone with a wet umbrella, but Hearer may agree because she thinks the air is moist, having not even noticed the wet umbrella. On the other hand, she may disagree with him because she can see the cloudless blue sky out the window. In either case, Speaker's utterance has indicated that he believes that "it rains" is a stereotypical possibility⁴². For the stereotypical ordering source to sort contextual possibilities, Hearer need only recognize that Speaker's utterance is relative to a stereotypical ordering source. Hearer will have her own set of stereotypical states of affairs which she can use to evaluate his utterance.

Thus, maintaining flexibility as to who or what determines normalcy is an important consideration to avoid evaluating the truth of epistemic possibility/necessity utterances according to standards by which Speaker did not intend them to be evaluated. As researchers we are better off not generating an explicit list of propositions expressing stereotypical states of affairs. Instead, we can deduce them based on the utterance and its surrounding context, just as Hearer does. This does not hamper their ability to sort slight and human possibilities.