Context-dependence and the Semantics-Pragmatics Interface

[ Music ] >> This podcast is brought to you by LMU Munich. >> Okay, so let's start. I think it's five. My name is Zidoro Stoyanovic. It's a great pleasure to be here. So just very briefly, my background is in logic, cognitive science, and philosophy, and they specialize in philosophy of language, and to a certain extent, formal semantics, and the semantics, pramantics interface. So here is the plan for this lecture stream. In the first part, which is this lecture and lecture number two, which is taking place on Thursday morning, what I would like to do is introduce two of the most important phenomena that have to do with context dependence, and that relate to the semantics, pramantics interface, namely the phenomenon of presupposition in this lecture, and the phenomenon of indexicality, which will be the lecture number two. And then the second part, so the first part is the two lectures and the two tutorials that everyone attends. And the idea is that for this first part, it's fairly introductory, and my aim is to provide you with certain tools that will hopefully be useful for you, even if you decide never to continue in philosophy of language, or semantics, or areas like that, because in lots of other areas in philosophy, like epistemology, ethics, and so on, you might every now and then stumble upon the notions that come from language-related issues, namely notions of presuppositions, notions like indexical context dependence, and so on. So then the idea is that for all of you, you will get something very useful from this lecture stream. And then the second part, which is lecture number three and four, it's for those of you who want to learn more about issues in the semantics pramantics interface. So the plan for lecture three is to look at something that's sometimes called the dynamic turn, which is the idea of approaching language dynamically. So we will look at things like context update, update semantics, and dynamic semantics. And then the last lecture is going to be about some very recent issues in philosophy of language, which have to do with the question of what kinds of context dependence refined in language. And for some of you, maybe you've recently heard about this debate between contextualist and relativist approaches to a number of matters. So this kind of debate has taken place on matters like disagreement and disputes about taste and so on. So a lot of that comes from philosophy of language, and that's the plan for the last lecture number four. So that's the plan for the lectures. And each lecture is going to be followed by a tutorial, the aim of the tutorial is for you to get back to me with questions about things that you haven't understood in the lecture. So if today, and for any lecture, there is something really unclear and you want to make a quick clarificatory question, don't hesitate to interrupt, but for other sorts of questions that can be left for a tutorial. And at the tutorials, we will do exercises. So the first tutorial is tomorrow morning. And there are handouts with exercises. I've left them there in case some among you need to leave before the end of the lectures. So take them, otherwise at the end of the lectures, I'm going to distribute those exercises. So that's about the plan for this lecture stream. And then so the plan for today is, as I mentioned, to introduce this phenomenon of presupposition. So in the beginning, and it's really not going to be technical, I am aware that for all of you today has been a very long day with like two lectures, two tutorials, very technical stuff. So in the beginning, it's going to be pretty straightforward. And the idea is that you get a feeling about presupposition and how presupposition differs from on the one hand, entailment, semantic entailment or logical entailment. And on the other hand, implicature, which is a kind of entailment that's extremely pragmatic and contextually driven. Then so I'll try to go quickly through this easy part, so to speak. And then we'll start the phenomenon called presupposition projection. And the last bit of today's lecture is to distinguish two ways of accounting for presupposition, namely semantic approaches versus pragmatic approaches. And towards the end, we'll pave the way to dynamic approaches. But that's something that we won't see in this lecture. It will come later in lecture number three. And tomorrow, so a bunch of exercises. So I'm not presupposing pretty much anything, so just a bit about basic logic. But no, I'm not taking that you will be familiar with things we'll talk about. And maybe some of you who have had classes in linguistics will be bored by today's lectures. But I hope not. So presupposition. In the way, so it's a phenomenon that's really ubiquitous that has a lot of literature, both in philosophy and in linguistic, has been written on. But there is really no agreement about what presupposition really is like. And sometimes people use this saying, which is in other occasions, used about other things. But it can be applied to presupposition, namely that we all recognize it when we see it. But we can't exactly say what it is. So here's a slide with some examples. So in language, so in the class, I think that you had about logic. You must have seen the notion of entailment. So the notion of entailment, when you study it from the point of view of logic, is clear and neat. But when it comes to natural language, it becomes trickier. So if you have someone use language, there are lots of various inferences that you are typically able to make. And now the question is among those differences, which ones are entailments and which ones aren't. So here are some examples of inferences that can be made from the use of language. So here we'll be using English all the time. If I tell you that Samira and Mishka are students, you can normally infer that Samira is a student. If I tell you that Mishka lost the keys to his office, you can normally infer that he doesn't have the keys to his office anymore, right? He lost them, so he doesn't have them. If I tell you that Mishka didn't lose the keys to his office, you can normally infer that if he didn't lose them, he still has them. If I tell you that Samira likes to cite her favorite philosopher, you can deduce that she has a favorite philosopher. If I tell you that she doesn't like to cite her favorite novelist, you can normally infer that she has a favorite novelist. So all are seem to be types of inferences. And just to look at those and think for yourself, which whether all of those are on a par, whether they're all in talismans, and we'll come back to them. So focusing on just two among those, which are the second and third. So as I mentioned, if I say Mishka lost the keys to his office, you can infer that he doesn't have them. And if I say that he didn't lose them, you might infer that he still has them. If I tell you that Kabutri lost her passport, you can infer that she doesn't have her passport anymore. If I tell you that she didn't lose it, you may infer that she still has it. So on this type of patterns with lose, it seems that you can have this generalization, namely, that if you lost something, then you don't have it anymore. And if you didn't lose something, then you still have it, right? It seems intuitive, no? So why not intuitive? If you didn't lose something, then you have it. You didn't lose horns. Therefore, you have horns. Is this a good argument? Now, if you have a background in philosophy, you will have seen at some point, probably, say it's this paradox-- I mean, it's not really a paradox. It's more of a puzzle called the paradox of the horned man that comes from the stoics. So something that-- so the stoics, a long, long time ago, So something that's been in discussions or philosophy for a very, very long time and that actually has to do with questions like presupposition. Now, if you didn't lose something, you have it. As we looked previously at this, it seems some kind of generalization that's OK. So that's why the argument may seem appealing that you're willing to say, yeah, that's right. If I didn't lose something, I still have it. I didn't lose horn. But no, I don't want to deduce that I still have horns, right? So then the question is, what's happening in this argument? Why is the first premise still appealing? But the argument has two premises, which are both appealing. And the conclusion is false. So I'm going to leave this as a question that you, after this lecture, you might want to think back at this and decide at which stage you want to block this argument. So people, as I mentioned in the past, they even consider this as a paradox, which would mean that, yeah, we really have two true premises and the false conclusion. So something-- and logically, some arguments or something that's troublesome. But the question is, which one is the one where you want to say that's precisely where the argument breaks? So back to the initial examples, then which are entailments and which aren't. So I hope that for most of you, what have said is the first one, Mishka and Samira are students, hence Samira is a student, is a genuine entailment. Though even that one can be tricky for natural language, because if you consider the same form, the same pattern, Samira and Mishka are enemies. Therefore, Samira is an enemy. You don't want to say that this is a fine entailment, right? But let's take it for granted that this one is these two probably fine entailments. But these two are the ones where we want to say there is something like an inference which normally goes through, but it's not really an entailment. So what kind of inference is it? Well, that's precisely the one that we want to study. It's presupposition. So more examples. So in this part, we're just going through examples to get the feeling of what's happening. So the presupposition we've seen with Luz and the Lost is that you can felicitously use the verb Luz if you presuppose that the person who lost something has had that. But the fact that they've had it is not an entailment. It's something that comes with the triggered by the verb Luz. And that you presuppose. Now, the first this was to test to tear apart presupposition from entailment. And there is another notion important in philosophy of language, which is that of implicature. So let's try to now tear that one apart from presupposition. And look at this example. I'm sorry that I'm late. My alarm clock didn't work this morning. It's a typical excuse. So if a person comes in and says, I'm sorry, that I'm late. There are various things that you might infer. So here is a bunch of possible things that you may infer from that. So you may infer that Al-Sha'i is late. That she's sorry to be late. That the reason why she's late is that her alarm clock didn't work. And so on. Maybe for the sake of time, we won't go in detail through all of that. Just let's look at-- let's contrast that she's sorry to be late and that she's late. And things like that, the reason why she's late is that her alarm clock didn't work. So about three, what kind of inference do you think this one is going to be? If someone says, I'm sorry, I'm late. Is this sequence of sentences in this order? Does it entail that her alarm clock not working? Is the reason that she's late? It doesn't entail. But it's still a sort of inference that you would typically make. So this type of inference that's a contextually triggered inference is what's referred to and known as a conversational implication. Similarly, things like five and six that her alarm clock normally works, that she set her alarm clock the previous evening. So if she said my alarm clock didn't work this morning, you might typically infer well that it was set on the previous evening. So there are all inferences that can come naturally in a sequence of a discourse, but are not entailments. Is that a question there? No. Just like-- I'm sorry. I just want to make sure that there is. What about 10 that Aisha has an alarm clock? Is that an entailment of the above? Think about it. You can-- yeah? Entailments? Yeah? [INAUDIBLE] The simple relationship that she has some follow-up on clock. Mm-hmm. [INAUDIBLE] So you would assume the ownership. Now is that entailment or presupposition? [INAUDIBLE] OK. So so far we haven't really given any tests as to decide when we have a genuine entailment, when we have a presupposition. And I should actually mention that for some of the things, so very often you can have a sentence that both entails generally and presupposes another sentence. And also that there are cases where there is controversy whether they should be treated as presupposition or not. But you are right. Definitely that this one Aisha has an alarm clock. Has a different status than the three, which is that the reason why she's late is that her alarm clock didn't work. Now, so to get sort of to the point with this sort of example, so what was the example about is that you have a statement or a sequence of statements. You can draw a number of inferences, but the inferences have different properties. And there are three main types of inferences you can have, entailments, presupposition, and implicature. So there is something like an easy test to say when an inference is an implicature, which is that it can be canceled without being inconsistent. And this is the way how we can see that the inference that the reason Aisha was late is that her alarm clock didn't work is merely an implicature. Because you can say something like, I'm sorry that I'm late. But actually, that's not the reason why I'm late. I rushed and I would have arrived on time. But there was a traffic jam. So what we do here is we deny the inference. And we don't get anything inconsistent. It's a fine piece of discourse. On the other hand, if you say, I'm sorry that I'm late, but I'm not late, that sounds really bad, right? My alarm clock didn't work, but I don't have any alarm clock. It doesn't sound fine. Now, from-- I'm sorry that I'm late. I'm late is actually a presupposition as we're going to see in a minute from the use of my alarm clock that I have an alarm clock is a presupposition. But even though they're not genuine entailments, you can't consistently deny them right afterwards. One more sort of-- well, now, going back to presupposition versus entailment. So presupposition versus implacature, we saw that there is something like a test, which aims to be working fine, which is that if something is merely an implacature, you can consistently deny it, right? How? about presupposition and entailment. Now things are tricky because in lots of cases, presupposition really is a type of entailment. So how can we tell whether a relationship between P and Q is a case of presupposition or entailment? So one thing that they do have in common is that whenever P is true, Q is true as well. Whenever you say my alarm clock usually works, it's true that I have an alarm clock and so on. But unlike entailments, what happens with presuppositions is that it's also the case that whenever the negation of P is true, Q must be true as well. And this is where we see that from I'm sorry that there arrived late has arrived late as a presupposition, rather just mere entailment. So if you say I'm sorry that I arrived late, yeah, you can infer that I arrived late. But also, if you say I'm not sorry that I arrived late, typically in a normal context, you also infer that I arrived late. If you say, I'm not sorry that I did this stupid thing last summer. You infer that I did something stupid last summer. If you say Mary isn't sorry that she cheated on the exam, you also infer that she cheated on the exam. So from both of those, which is a test, a signal, that what we have is not just entailment, but presupposition. Because from both P and the negation of P, you get it that it inferes, which happens with entailment, only when the second is a technology. Now, sort of to see if we are all together on the same line and getting the intuitions correctly. Look at this example, I shall arrive late again. And this sort of inference is that I shall arrive late, that somebody arrived late, that I shall had arrived late previously in the past, that she has a bad sense of time, and that we could not start on time. So in a suitable context, I might say I shall arrive late again, and you might infer from that that we couldn't start of time. But so very quickly, how would you classify those into entailment, presupposition, and implacature if we have the three types? Which ones are entailments, if any? Number one, everyone agrees? And something else? And number two, yes? And number three? Number three? What's number three? Presupposition? And how do we know that it's a presupposition rather than a regular entailment? Sorry? Oh. In the second instance, again, this has supposed to be a presupposition. OK. But how do we know that it, again, is presupposing? How do we test that? So the test is whether this-- can this be false without this being true, that I shall arrive late again? So even if whether this is true or-- for this to be true, this needs to be true, but also for this to be false, this needs to be true, right? And then what about four and five? Presupposition? Anyone? Implicature, right? Inissutable context. OK. So this is just like summary. As I mentioned, so to tear apart entailment and presupposition, what you typically might do is take the sentence and put it under negation. So I should didn't arrive late again. You can't infer that somebody arrived late. No one needs to arrive late. But you would infer that I should have arrived late previously in the past, right? And of course, you can say I should arrive late, but we didn't wait and we started on time. It shows that this is an implicature, so it can be cancelled. OK. So for the sake of time, we'll skip this and go on further. So now just-- what I hope to have done with this in product is just sort of get you to have a feeling of what distinguishes presupposition from entailment and presupposition from implicature. And implicature is an important notion, but we want to look at more detail really in this course. Maybe we will come back to that in the very last part of the course. So let's now just go quickly through some data about presupposition, namely classes of expressions, the trigger presuppositions, and some terminology about what happens with presuppositions. So what we've seen so far is that there are inferences, but unlike entailments, when you use an expression, whether you use it in a positive form or you use it in an embedded form, for instance, under negation, but also in conditionals. And we are going to see later the expression triggers a presupposition. And this is a phenomenon that we call projection. And unlike conversational implicatures, you can't consistently deny a presupposition right straight out. But the tricky thing is that presuppositions, if suitably cached out in a piece of this course, can be so to speak canceled. So remember that from-- so we had-- I'm sorry that I'm late. Presupposes that I'm late. I'm sorry that I'm not sorry that I'm late. Also presupposes that I'm late. But you can consistently say, well, I'm not sorry that I'm late. I'm not even late. You might say, I should not sorry that she did a stupid thing like summer, because she disagrees that she did anything stupid. In fact, she didn't do anything stupid at all. So with our horn man paradox, you can say, well, of course, I didn't lose any horn. I never had any. And you can say, you can also cancel presuppositions, not only with negation, but with things like, well, if I should sorry that she's late, then she must be crazy, because she's not even late. So you have a presupposition in the antecedent of a conditional, which doesn't go through all the discourse. So this kind of question of how presuppositions can be canceled and prevented from projecting into the whole discourse is what we are going to do in this class. But so to get started, we will first just go through classes of presuppositions and then see how various presuppositions interact with other constructions in the language. And then hopefully there have been a time to go through two main approaches, namely semantic and pragmatic theories of presupposition. So classes of presupposition triggers factors. So sorry regrets, knows, and so on. All those kinds of verbs trigger presupposition. I shall regret that she cheated on the exam. She cheated on the exam. A spectral verbs stop, continue, lose, and so on. If you say that someone stopped smoking, you presuppose that the person smoked. No, for all of those later today or tomorrow, you can check and see why there are presuppositions and not just in the elements. Temporal closes. If you say Samira was prettier before she had another surgery, this presupposes that she had another surgery. Why? Because you used this before. Manor adverbs, if you say someone drove slowly, you can infer that they drove. So this is also an entailment, of course. But it's also a presupposition. position because if you say no, he didn't drive slowly, you presuppose that the person drove. OK? Clifte constructions, if you use this construction in English, like it is Aisha who cheated on the exam, what you presuppose is that somebody cheated on the exam. Here, it's also in the entailments. So to see that it's not just a mere entailment, but a presupposition, again, you look at what happens. If it's not Aisha who cheated on the exam, that's going to presuppose that someone cheated on the exam. International focus, quantifier phrases, and definite descriptions, which is one of the examples where you might depending on your theory of definite descriptions want to say that it's a presupposition, or rather that it's just an entailment, but typically if you use a definite descriptions like the man who proved format's theorem is a genius, you presuppose that there is one such man. Again, why? Because if you say that the man who proved format's theorem is not a genius, well, again, there is the inference that there is someone who proved it. OK? Now, so moving now to presupposition projections, this terminology, which comes from Laurecourt-Tunin, distinguishes various types of expressions as to how they react with presuppositions. We have things that are called plugs, and they are things which block the presupposition from projecting. So there aren't that many expressions of that form, but typically things like said or dreamt are going to be such that if there is an expression embedded further down which triggers a presupposition, the presupposition doesn't project all the way up. So if you say that, Cheng Zhu said that few foreign students took up the exam-- sorry for the typo-- there is no presupposition that there are few foreign students or that there are foreign students at all because said blocks a presupposition. Right? It can be true that she said that few foreign students took up the exam that doesn't entail that there are foreign students at all. Compare this with, unfortunately, few foreign students took up the exam. This presupposes that there are foreign students. So if you want to get the contrast between how said that and unfortunately behave is-- so you can compare those two. From one, you infer that there are foreign students from the second you don't-- so from this, there is the inference that there are foreign students. From this, there is no such inference. Another type of case with the dream. So you can say he dream that he met the man who proved for a March theorem. That doesn't presuppose that there is one and only man who proved for a March theorem because if someone dreamt something with a certain content with a specific man, it doesn't go all the way through. But if you say it perhaps in the future, he'll meet the man who proved for a March theorem. This presuppose that there is a man who proved for a March theorem. So two types of expressions-- plugs, they always block presupposition, holes, they always let presuppositions project. And now the interesting ones are those called filters, which are tricky because sometimes they block presuppositions and sometimes they all add them project. So let's focus on this pair. You say if I can trust my informant, then it's I sure who cheated on the exam. This is typically going to presuppose that somebody cheated on the exam. So what happens with the-- so the cleft, remember, that's the sort of expression that presupposes that someone did the thing that under the cleft. So it is I sure who cheated presupposes that somebody cheated. And the whole presupposition goes up and projects whether or not the antecedent is true. But what about the second one? If anyone cheated on the exam, then it is I sure who cheated on the exam. Does the presuppose that someone cheated on the exam? No, he doesn't, right? So what we have here, what's the interesting case, is a conditional, which sometimes projects the presuppositions from the consequent and sometimes doesn't. And that makes conditional filters. Second type of case, which is somewhat similar in behavior. So remember, 2x is sorry that p presupposes that p, right? Now, if you have a conjunction like I sure is an honest girl, and she's sorry that she cheated on the exam, this presupposes that she cheated on the exam. And the whole conjunction presupposes the cheat sheet on the exam. What about this one? I sure some error cheated on the exam. And I say, sorry that she did. Does this presuppose the cheat sheet on the exam? It does. So this one presupposes, right? I say sorry that she did. Now, who thinks that it does, and who thinks that it doesn't? Let's do a quick test. For the does? One, two, three. Who thinks that it doesn't? I think people are lost. The question is whether the whole conjunction I sure some error cheated on the exam. And I say sorry that she did presuppose that I sure cheated on the exam. [INAUDIBLE] No. So it doesn't literally say it because she cheated on the exam is embedded under the cheat sheet. So if you want to have a better place where it's definitely got-- I mean, where it's not part of the implacature, take Aisha is an honest girl, but she isn't sorry that she cheated on the exam. And Aisha and some error cheated on the exam. But Aisha isn't sorry that she cheated on the exam. Yeah. [INAUDIBLE] No, no, sorry. And that she, Aisha, is-- Yeah, I think he reasons why she says she was sorry. But that doesn't mean that she's needed. Like just because someone may be guilty of doesn't mean that she just saw she cheated on the exam. So actually, it's much more simple than what I think people are trying some sort of to find some complicated reason as to why this wouldn't be presupposed. Yeah? Because the whole conjunction isn't true. It's like going in the right direction, but can the whole conjunction be true without Aisha cheat? [INAUDIBLE] If it is wrong, then it could be wrong. Could the true or not? So the whole conjunction can be false and this not be true. So that's the reason. Maybe I should repeat it. So remember that we have that a big assentance presupposes another if-- so P presupposes Q. If Q is being true is a condition both for P being true and for P being false. Now this one here can be false if Aisha didn't cheat. Because if Aisha didn't cheat, Aisha and Samira cheated is false. So this one can be false with this one being false. Hence, Aisha cheating on the exam is not a condition for the whole sentence being true. And that makes the difference between this one and this one. And again, I think it's easier to see. if you say she is not sorry that she cheated on the exam. Was there a question there? There'll be a bunch of exercises because some of those things are kind of tricky because when you have-- yeah, sorry, it went. So I think like this here, the choice of the example is partly tricky because you get really disimpression that it's entailed and being part of the conjunction. It's entailed all the way through. So you get this blurry distinction between presupposition and entailment. But if you say she isn't sorry, then you get it go all the way through. Quick question. Tomorrow in the tutorial we'll come back to these things, because they're especially with the conjunction, it's weird so why-- I mean, something that's typically a presupposition, you might want to say, well, it goes. But I mean, it doesn't because of this fact that you can have something be false. And we will see, hopefully towards the end of today's class, how various accounts explain that you don't get this presupposition. So let me move on. So as to keep in time to an important generalization, which is the question about presuppositions getting blocked. So we saw that there are a class of expressions, which we call filters, which are tricky because sometimes the presupposition goes all the way up. And sometimes it is blocked. So if you say if anyone cheated on the exam, then it's Aisha who cheated. That doesn't presuppose that somebody cheated on the exam. If you say if Mishka ever smoked, then he must have stopped smoking. That doesn't presuppose that Mishka ever smoked. Because-- well, no, because I sense a mere cheated on the exam. And Aisha is sorry that she did. Doesn't presuppose that I should cheat on the exam. So filters, which block presupposition, land themselves to an important generalization, namely something about the relationship between the first sentence and the presupposition, which is normally carried by the second sentence. What's the relationship? Quickly. Intellment. Well, anyone is tricky because you don't use anyone in a positive form. But if anyone cheated on the exam, then it's Aisha who did. So anyone, it's someone cheated on the exam. Intel, someone cheated on the exam. So if you have an antecedent that entails the presupposition normally carried by the consequence, it's going to be blocked. And similarly, in conjunction, it's the same type of relationship. If you say Mishka used to smoke and he stopped, the whole sentence doesn't presuppose that Mishka used to smoke. It's an entailment, but it doesn't presuppose is because it's part of what's asserted. And the sentence can be true precisely because Mishka never used to smoke. So if you have a conjunction and the first conjunct presuppose entails a presupposition normally carried by the second conjunct, then it's going to be blocked. And that's a generalization that we seem to observe with presupposition projection. Are we together? So when the antecedent in the conditional entails the presupposition, in the consequence, the presupposition is not going to project. When we say it's not going to project, that is the whole conditional doesn't presuppose something that's in the consequence. Some interesting questions about presuppositions with conditional, sorry. We're going to leave this for a second, just for you to-- To think about whether we get the presuppositions. If anyone proved for a month's theorem, I wish I had the person who proved it. I mean, this one is easy. So here we have-- well, no, it's not actually. It's not that easy. So what about this one presupposing that there is one and only one person who proved for a month's theorem? If Mishka posts photos of the Sagrada family on Facebook, his friends will know that he is in Barcelona. Does that presuppose that he is in Barcelona? If Aisha comes to Barcelona for her holidays, then I'm glad that she will visit the Sagrada familia. Does that presuppose that she will get the Sagrada familia? Thanks for you to think about tonight after when you're having beer with friends. And you get-- OK, so-- and we'll get to some of those tricky things. But for the half an hour or a bit less that we have remaining, let's now move to how presuppositions are handled by the different theories. So when introducing the phenomenon, there was something slightly biased when I was talking about the presupposition. I would say things like a sentence presupposes another or P presupposes Q and so on. It's fine in an introductory class. But now what I would like you to do is sort of realize that it's a way of speaking, that carries some presuppositions if you wish. Why? Because so we observe the phenomenon in language. The phenomenon is really that when a person uses a piece of discourse, you can typically normally derive a bunch of inferences from what they say. But there is an old divide, well, old and very present, not just in theories of presupposition, but in lots of questions in philosophy of language, which is a divide between semantic approaches and pragmatic approaches. Now when it comes to presupposition, the semantic approach or the family of semantic approaches, what it holds is that it's a phenomenon, which is a matter of conventional meaning, and that presupposition is best viewed or seen as a relation between sentences as pieces of language. The alternative approach, what it says is that it's really a purely pragmatic phenomenon, and what regulates presupposition. So it's not something about conventional meaning of words. It's something about conversation and communication more generally as an enterprise where you have speakers aiming for a certain goal. So what really regulates things like presupposition are much more general principles that govern communication, that make it a possible rational enterprise. And now what we're going to look into is how those two approaches differ. What they'd like just to tell is that there is a similar divide with respect to many other phenomena, and one of the phenomena where you also get this contrast between semantic versus pragmatic approaches is, for instance, reference. So on the semantic approaches, if you have a word like she, I say, she's filming me today, the fact that the reference of she is this young lady is something that's a matter of conventional meaning of she. Whereas on a pragmatic approach, it has to do with my intentions as a speaker of referring to a certain person. So important divide between semantic versus pragmatic approaches to phenomena. And as it comes to presupposition, it really drives two types of contrasting views about presupposition. On the semantic theories, so they have, Well, it comes from Fraga and then was pursued by Strosson, and we have a Fraga Strosson definition, which really frames presupposition as a sentence. And it's one which we've seen earlier, is that whenever a sentence is either true or false, the presupposition is true, right? Another way of framing the definition would be to say whenever a sentence or its negation is true, the presupposition is true. And you might think for yourself under which conditions the two definitions are equivalent. So that's the semantic approach. It's about the distribution of truth values to sentences and the matter of conventional meaning on pragmatic theories. So in the reading list, that's one of the readings that you've been given, pragmatic presuppositions from Stolnecure. So there are definitions that you get there. One is that Q is presupposed by an assertion that P, just in case under normal conditions, one can reasonably infer that a speaker believes that P. Either from his assertion or his denial that Q. Sorry. I think it should be the other way around. But-- all right, this is. Sorry, yet, well, there is normally-- I mean, it should be the other way around, right? But an alternative definition, which gets more to the heart of a matter, is that presupposition is something that's not asserted. It's something that's really like a set of background beliefs by the speaker, which is that those are the propositions that the speaker takes for granted, or at least pretends to take from granted for granted when he's making his or her statement. So this pragmatic notion of presupposition is much more general than the semantic approach. Because when we speak, we normally take lots of things for granted. I take it for granted that you're in a good hearing conditions, that you understand English, that you are listening, and so on. And lots of those are not presuppositions in the sense in which we saw that loose presupposes that someone had something and so on. So we'll see on some cases that the presupposition in this pragmatic sense-- so that something is going to come in the later lecture, in lecture three, where we have really this idea of a background of a common ground, which is something like a background for a conversation, where a number of things are presupposed when we are delivering a piece of discourse. So in the readings that you were given, you might see some-- I mean, you can look at more detail as to the semantic theories of presupposition. But one of the-- so the driving idea in the semantic theories is, remember, the definition of a presupposition is that we get it if the presupposition is a condition for a sentence to have a truth value, to be either true or false. So what's going to drive the semantic theories of presupposition is going to be a truth conditional approach, where we are going to try to cash out phenomena about presupposition projection by looking at how the sentences behave truth conditionally. So quick reminder, truth value tables for negation, if then, and so on. Everyone is super familiar with that. There is no need to stop at that. But now-- so remember that if a sentence presupposes another sentence, and the second sentence is not true, the first sentence can't be either true or false. So it's not going to have a truth value. And now the tricky question is what you're going to do if you have a sentence as part of a more complex sentence. And one of them doesn't have a truth value. So you have, for instance, P and Q. But P doesn't have a truth value. How do you get to decide whether the whole sentence has a truth value? For instance, if you say, if Samira stops complaining, Mitch Kavill-Tankher, would that be true or false in the state of affairs, where Samira never complained? So again, so if you say Samira's kids are annoying, and Samira doesn't have any kids, is that sentence true or false? Now you want to say, well, it's neither true nor false. I mean, if she doesn't have kids, it can be true that her kids are annoying, and it can be false that her kids are annoying. And so the question is, what happens when we embed such a sentence into a more complex sentence, how does that affect the truth value of a complex sentence? Now you might think, well, it shouldn't be such a difficult question, because if some sentence doesn't have a truth value, it's neither true nor false, the complex sentence shouldn't have a truth value at all. That seems like a possible move. And it's a move known as something that's called a weak three-valued cleaning system. So it's one which looks like this, except that you are considering also the cases where need a true-- well, where P doesn't have a truth value, and Q doesn't. And you simply say that if one of those doesn't have a truth value, the whole sentence doesn't have a truth value. It's a possibility, but it's not a very promising one. And the reason why it doesn't seem really accurate is that you have lots of sentences which seem to have a truth value, even though one of the embedding sentence doesn't have a truth value. So if you say things like, "Mishka believes that some mirror's kids are annoying," some mirror's kids are annoying. What's happening now? Just-- sorry. Let me keep on. So if you say that, "Mishka believes that her kids are annoying, and she doesn't have any kids, does that make the whole sentence truth value less?" If Mishka believes that she has kids, right, he might believe that her kids are annoying, because he believes that she has kids. So that can be true, even though she doesn't have kids. So you don't want to say that if one of the sentences that doesn't have a truth value is going to make the whole sentence truth value less. The power cable was not working. So it shouldn't take long. So-- and the other kind of case where the weak system is, gives wrong predictions, are precisely all of those that we've seen where we cancel a presupposition by the antecedent of the conditional. So if I say, "If Mishka ever smoked, he must have stopped smoking. He must have stopped smoking presupposes that he smoked. He must have stopped smoking. Is neither true nor false if Mishka never smoked." But if Mishka ever smoked, he must have stopped smoking. That can be true, even though Mishka never smoked, right? So what that suggests is that-- and then like a third case, which shows that the weak system is not a good one, you can say, "Somberal kids are annoying unless she has any." Right? So this-- so in the situation where a Samira doesn't have any kids, this is neither true nor false. But the whole sentence becomes true because you say, they're annoying unless she doesn't have any kids. By saying unless she doesn't have any kids, you've made the case that it's possible that she doesn't have any. And then this is trivially true and not true values. If that's unclear, we can come back to that tomorrow. But what we should-- just like handling. So I'm going to skip through those, which is to say that this question of how you calculate the truth value of something more complex. If some of the constituent sentences doesn't have a truth value, it's not just-- it doesn't concern only presupposition. It concerns other things like vagueness and borderline cases of vague predicates. So if he is a borderline bold, you might say that he is bold is neither true nor false. But he is bold and he is not bold. You don't want that to be neither true nor false. You want that to be false. Another type of case where there is a similar question of how you compute the truth value of a complex sentence. So if you have for future contingent-- and we have say there is an exam, but it's not determined yet what kind of questions will be on the exam. There'll be a question about three value truth tables on the exam is something that you might say doesn't have a truth value. But either there'll be a question or there will no be a question at the final exam. You might want to say that that is true and not truth valueless. So you have a similar dilemma with other types of cases. And so this is all motivation to a sort of-- a family of semantic approaches which use a different type of three valued true tables called Strong Cleanie. And so the idea behind those-- so it's a true table like the ones you've seen in any kind of logic. I don't know how many among you have seen true three valued logics and three valued true tables. But basically what the idea is and how you get to have this sort of Strong Cleanie truth tables is that if you have a sentence-- and by knowing the true value of one-- if you know the true value of a certain constituent, and that's enough of a grounds to tell you that the sentence has a certain true value, then you can give it the true value. But if you don't have the relevant knowledge, it's going to be true valueless. So just as an example, consider B and Q. So it's a conjunction. If one of the conjuncts is false, you know that the conjunction as a whole can't be true valueless because in the conjunction, if one of the conjuncts is false, the conjunction itself is going to be false. So if one of the conjuncts doesn't have a true value, but the other one is false, it's going to give you the value of false. However, if you have one of the conjuncts that doesn't have a true value and the other one is true, that's not enough grounds to tell you whether the conjunction is true or false. And that's why it's going to be a true value. So that's the intuitive idea of how you establish a strong three-valued true table. It doesn't have a true value. If one of the constituents doesn't have a true value, and the true value of the other one, is not enough for you to establish the true value of the whole sentence. You can see it with the conditional. So if the consequence is true, as here, here, and here, that's enough for you to know that the conditional is going to be true. Never mind the antecedent, whether the antecedent is false, true or true, or true, it's a lie. But if the consequence is false-- sorry. Yeah. Like here, here, and here. So in some cases, well, you know that it's going to be false if the antecedent is true. If it's the antecedent is false, it's going to be true. But if the antecedent doesn't have a true value, well, you don't know what's going to be the true value of a conditional, because where the antecedent true, it would be false, and where the antecedent false, it would be true. So that's why, because you don't know what the antecedent is, you can't determine what's going to be its true value. Now, using this kind of system to solve the projection problem, so what's going to be the aim of this sort of approach, well, the problem, remember, was one where you are trying to explain why conditionals sometimes let the presupposition project from the consequence, and sometimes they block it. So we had-- sorry. Using this system to solve the projection problem, or to explain the projection problem. So you want to explain why if somebody cheated on the exam, then it is Aisha who cheated, does not presuppose as a whole conditional that somebody cheated on the exam. If you have a semantic approach, so recall that the approach says that S1 presupposes S2. So alias S1 presupposes S2. If whenever S1 is the true or false, then S2 is true. So to show that it doesn't presuppose it, you want to have that S1 is the true or false, and S2 is not true. So what you are seeking is a model where this happens, where we have that S2 is not true, but S1 is the true or false. So what you can simply do is take a model in which S2 is not true. So it's not true that somebody cheated on the exam. This being false gets you a conditional, which then becomes true, and the presupposition is false. So it gives you the right predictions. Now what happens in the other cases where the presupposition does project. So what we want us our system to do is predict accurately that in certain cases the conditional can be true, and the presupposition doesn't go through, whether in other cases from the conditional, the presupposition projects. So we had another case, which is if I can trust my informant, then it is I share who cheated on the exam. So that typically presupposes that somebody cheated on the exam. But is that the predictions of the strong cleaning system are quite tricky, because so what we are now trying to do is find the model in which this one is either true or false, but this isn't true. So for this one to be false, what we need is that it is I share who cheated on the exam needs to be false. So this is like you can just look at this here, because if you have the conditional, well, it's false only in this situation where the antecedent is true, and the consequent is false. If the consequent doesn't have a true value, the conditional doesn't have a true value. So to get this to be false, this one needs to be false, and this is false if it presupposes that I share cheated. So if the conditional is false, it presupposes that I share cheated. Now if the conditional is true, true. So we should either have that this is true or that-- sorry, we should either have that this is true, that it is is aichah Hochitid on the exam, or that this is false, that I can trust my informant. Now, if it's true that it is is aichah Hochitid on the exam, that presupposes again that somebody cheated on the exam. So it's only that if this is false, that we don't get the presupposition. I'm aware that it's difficult to go and digest the reasoning, but still going through that. What this gives you then applying this strong cleaning system is that the conditional, if I can trust my informant, then it is isah Hochitid on the exam. The whole conditional doesn't presuppose that somebody cheated on the exam, but it presupposes something related, which is known as a conditionalized presupposition, namely that if I can trust my informant, then somebody treated on the exam. And that's the prediction that the semantic approach gives for these other sort of cases. So the semantic approach is never going to give you that the conditional as a whole lets all the presupposition from the consequent project. It gives you that the presuppositions are conditionalized, namely if the antecedent is true, then the presuppositions from the consequent. The presupposition is that if the antecedent then what's presupposed by the consequent? In the Stanford Encyclopedia of Philosophy article on presupposition that was in the recommended readings, you can go and look at that and follow the reasoning so that you see how it works.