Wolfgang Schwarz

Assessment-relativity and pretended dialogs

Does the semantic value of expressions in a language sometimes depend on other things than their utterance context? That depends on what is meant by "semantic value", but for the most part, I think not.

It can appear otherwise if one identifies the content of an utterance with the main proposition it conveys to competent hearers.

Alice, Bob and Carol are searching for honey. Alice sees a bee hive on a tree near Bob and wants to inform both Bob and Carol about this. That is, she wants Bob to acquire the self-locating belief that there is a bee hive on the tree near him, and she wants Carol to acquire the belief that there is a bee hive on the tree over there near Bob. She achieves both goals simultaneously by pointing at the relevant tree and saying, "there's a bee hive on the tree over there".

Since Alice conveys two different (centered) propositions to Bob and Carol with her sentence, one might conclude that her sentence expresses two different contents, one relative to Bob's context of assessment and one relative to Carol's. Content, then, is relative to both an utterance context and an assessment context. However, it is quite implausible that Alice's utterance really has these two propositions as its literal semantic value. Instead, what she expressed was just the proposition that there is a bee hive on the tree she is pointing at, and Bob and Carol figured out the centered propositions they were meant to learn from this information.

In general, the fact that utterances are often used to make the addressee(s) acquire self-locating belief(s) -- even different ones for different addressees -- does not entail that the uttered sentence needs to be semantically evaluated in the context(s) of the addressee(s).

For another illustration, consider the assessment-relative "now" Josh Parsons introduces in his draft "Assessment-contextual indexicals" (PDF). As Josh points out, it can be useful to have a word like "now" that picks out not the time of its production but the time of its reception. That seems true, but we don't need to evaluate anything at a context of assessment to achieve this end, just as having the word "tomorrow" doesn't mean that we have to evaluate sentences both at the context of utterance and at a context located at the following day. The referent of an utterance-contextual indexical need not be physically present at the context of utterance, it only needs to be determined by that context (to the degree it is determined at all): once time, place and world of an utterance is fixed, it is also fixed what day is picked out by "tomorrow". In the same way, we could have an utterance-contextual "now" that always picks out the time at which the addressee receives the message; this, too, is determined by the utterance context.

The main point of uttering "the bees are attacking now" with assessment-relative "now" is of course to convey the centered proposition that the bees are attacking now, but like in the story of Alice, Bob and Carol, it doesn't follow that this is the semantic content of the uttered sentence. Nor does it matter if the hearers immediately grasp the self-locating content without first processing the literal content (that the bees are attacking at the time when this sentence is received), unless "literal semantic content" is somehow tied to a psychological theory of language processing. (And this kind of content will not look much like the content we philosophers like, as relevance theorists are keen to point out.)

What if there is more than one addressee, at different times? An assessment-contextual "now" could then be evaluated at each of the assessment contexts, delivering different propositions. But we can have the same effect with an utterance-contextual "now" that means "at the times of reception". This case resembles those discussed by Andy Egan in his forthcoming "Billboards, Bombs and Shotgun Weddings" (PDF). Unfortunately, the present solution doesn't work there.

Andy looks at cases where people use the second-person singular when addressing several people, as when a billboard or a priest in a sermon says "Jesus loves you". As before, there is the fact that different self-locating beliefs are thereby conveyed to the addressees, and again I think this is perfectly compatible with the sentence expressing a single, assessment-stable content, viz., that Jesus loves all addressees. The main problem here is that I don't see how the sentence could literally express this content, given that the "you" is singular. That it is can be seen plainly in just about every other language where second-person singular and second-person plural don't happen to sound exactly the same, as they do in English ("you"). In German, for example, the billboard could read, "Jesus liebt dich", where the singular "dich", like "I" and "she" in English, can't refer to several people at once. (Andy offers other arguments and examples that bring out the problem more clearly in English. From now on, I will pretend that English has a distinctive singular "you", which I will write as "you*".)

Since "you*" can only pick out a single individual, the literal semantic content of "Jesus loves you*" cannot be that Jesus loves all addressees. So is the referent of "you*" here left open by the utterance context, and only fixed by each context of assessment?

Andy mentions an alternative: one could say that the priest or billboard actually make several utterances, one for each addressee. Indeed, suppose I stand at the road with a "Jesus loves you*" poster, holding it up whenever somebody passes by. Here, I reuse the same sentence token to produce many different utterances. (It wouldn't make a big difference if I held up different posters, or shouted the sentence.) Doesn't the billboard basically do the same as me, displaying itself to everyone who passes by?

I don't think so. The difference shows up with unintended addressees. When Alice, leading bee-hunter and member of the evangelical Exclusive Brethren an exclusive Christian cult that shall remain unnamed here, puts up the "Jesus loves you" poster in her church, she doesn't thereby tell everyone who passes by that Jesus loves them -- including the construction workers that came to remove the electricity lines. Similarly, when I send an email to my friends Bob and Carol, saying "you* are invited to my funeral", I don't thereby invite whoever else may read the email -- Bob's wife, Carol's boss, the CIA, etc. But if the poster and email automatically produced a new utterance whenever they are read, this should be the effect.

The point here is not about speaker meaning. It's true that Alice didn't mean that Jesus loves the construction workers when she put up the poster. But the point is that she didn't even tell the construction workers that Jesus loves them. The situation is not like if she had confused one of them with a believer in her cult and told him, "Jesus loves you". In this case, she might say, "sorry, I didn't mean what I said, I thought you were somebody else". But with the poster, the construction worker was never addressed in the first place, and if he thinks he is, he simply misunderstands the message.

So the poster and billboard and email aren't producing a new utterance whenever someone passes by. Nor do they contain a single utterance that expresses a new proposition relative to everyone who sees it. There is something wrong with both the assessment-relativistic and the multiple-utterances account of Andy's cases. It seems that a single context of production (or placement) fully determines what messages are sent. Once this 'utterance' context is fixed, it is fixed who is said to be loved by Jesus, no matter who later happens to pass by or not. When my email is sent, it is already fixed who is invited to the funeral, no matter who else may read it.

As Christian Nimtz pointed out to me, there is a mirror problem about inattentive addressees: when I send around the email, I thereby invite each addressee, even those who never read the email. If I only produced a new invitation at each reception context, those who didn't read the email would never get invited. But they are, they just don't know about it.

So here is my preferred account of such cases: they are instances of pretended dialogs. The speaker presupposes (in Stalnaker's sense) that there is only one addressee, and the addressees play along. Semantically, "Jesus loves you*" is true iff there is a unique addressee who is loved by Jesus. So the sentence is obviously untrue when uttered by the priest to the crowd. But the hearers realize the failed presupposition and can accommodate -- not, of course, by actually believing or assuming that nobody else is addressed, but by pretending so. It's like the speaker invites the addressees to perform a little play.

Why would people do this? Because pretended dialogs can be useful means of communication. They arguably make some things a lot easier to say and understand. To take another example from Andy: the commanding officer says to 100 soldiers in the briefing room, "the mission you*'re about to undertake is extremely dangerous; it's possible that, two hours from now, you* will be the sole survivor of this group; etc.". This is much easier to understand than the more accurate "...for each of you, it is possible that he or she will be the sole survivor...". More importantly, as the example also illustrates, the psychological effect of being addressed personally in a dialog is much stronger than that of being addressed as part of a crowd. That's why pretended dialogs are so common in propaganda, like in sermons and commercials.

What is communicated to an addressee by an utterance of a sentence S in a pretended dialog is just what S would communicate to them if they were the only addressee. For instance, when I tell you, pretending that you're the only addressee, that you* are invited to my funeral, you can take me to have made the speech act I would have made if we were actually in a dialog: to have invited you to my funeral. This is why one can't properly communicate something in a pretended dialog that is incompatible with several people receiving the message, as when the commanding officer says "two hours from now, you* will be the sole survivor of this group". Given the shared presuppositions, this is perfectly acceptable, but the addressees will find it hard to figure out what the officer is trying to get across.

Making an assertion or invitation doesn't require the addressee to pay attention, and neither does making an assertion or invitation by means of a pretended dialog. So we don't have the problems of unintended and inattentive addressees.

Another nice feature of this account is that it explains why addressing several people in second-person singular requires stage-setting, and more of it the harder it is to ignore the other addressees. You can't just burst into a room full of people and address each of them with the singular "you*"; they will inevitably assume that you are talking to a specific person, and may try to figure out who that is. (Again, this can only be tested easily in languages where "you*" doesn't sound like "you".) With emails and billboards, it is easier to ignore the other addressees -- unless they are explicitly mentioned. Thus in German, when you're sending an invitation to several people and address them with singular "you*" (which is common), you have to use the singular throughout. You can't begin with "dear Alice, Bob and Carol" or "dear friends", and then continue with "you* are invited" ("du bist eingeladen"). You can't make it explicit that you're talking to several people and then pretend that you're not.