← New Books Network
New Books Network
Daniel Cohnitz and Jussi Haukioja, "Foundations for Metasemantics" (Oxford UP, 2025)
01:09:28
Loading summary
Transcript52 lines
- [00:01]
A
Welcome to the New Books Network.
- [00:05]
B
Hello and welcome to New Books and Philosophy, a podcast channel with the New Books Network. I'm Carrie Figdor, professor of Philosophy at the University of Iowa. I'm co host of the channel along with Sarah Tyson and Blaine Neufeld. Together we bring you conversations with philosophers about their new books in a wide range of areas of contemporary philosophical inquiry. Today's interview is with Daniel Koenitz, professor of Philosophy at Utrecht University, and Jusi Hakioja, professor of Philosophy at the Norwegian University of Science and Technology. Their new book, Foundations for Meta Semantics, is just out from Oxford University Press. A theory of semantics is a theory of what a person means when they use a word. For example, that when someone uses the word cat, they refer to cats or the name Beyonce to refer to Beyonce. A metasemantic theory explains why a word means what it does. What makes it the case that when an English speaker says cat, they refer to cats? Externalism claims that facts about the world determine the semantic facts, while internalists hold that the facts about the speaker determine them. This internalist externalist debate has also been a leading edge in the use of experimental methods in philosophy, also called X phi, to provide empirical evidence from ordinary speakers that might be relevant to different philosophical theories. Konitz and Haukioia approached the externalist internalist debate from the perspective of what they call meta metasemantics, which is a theory of what makes it the case that externalism or internalism might be true of a speaker's uses. In Foundations for Metasemantics, they argue for a view they call dispositionalist meta internalism, in which the internal dispositions of a speaker determine whether their uses are best explained by internalism or externalism. These dispositions include dispositions to use a term in a certain way, as well as dispositions to revise usage in response to new information. They also critically considered the role of thought experiments in these semantic debates and how XY methods might be revised to get better data. Let's turn to the interview. Hello, Daniel Konitz and Jusi Haukioja. Welcome to New Books in Philosophy.
- [02:43]
C
Hello Carrie. Thank you so much for inviting us.
- [02:47]
B
So this will be a very interesting three part interview here with two authors of Foundations for Metasemantics. Before we get into the details of the book, perhaps you can tell us a bit about yourselves and as philosophers and then how this collaboration came about.
- [03:09]
C
Yeah, so maybe I start with and telling a bit about myself. So my name is Daniel Koenins. I started being interested in philosophy sort of already in high school. I studied philosophy in Germany, in Dusseldorf, primarily sort of focusing on philosophy of science and the social sciences and humanities in particular. And I was always curious about foundation questions in philosophy itself. So methodology of philosophy, for example. And the. And wrote my dissertation then in 2005 on the role of thought experiments in philosophy, which shortly after that became sort of a really interesting topic in the philosophical discussion. This field that's now called metaphilosophy. Yeah. So my interests are largely in that area. And then in 2000 I think six, I got a professorship in Estonia, that's in the Baltics, in Tartu, which at the time was or still is a partner department of Turku University Town in Finland. And that's also where I met uc, who was affiliated with that university at that time.
- [04:41]
A
Yeah.
- [04:41]
C
Then I worked 10 years there as a professor of theoretical philosophy. And then in 2016 I got a professorship in Utrecht in the Netherlands. And now I'm working here in Utrecht in the Netherlands as a professor of theoretical philosophy. Still interested in. In these kind of methodological, largely methodological issues in philosophy.
- [05:03]
A
Yeah.
- [05:03]
B
Okay. Jussi.
- [05:05]
A
Yes, my name's Jussi Hauke. So I wasn't really exposed to philosophy until I was already at university. So I started studying physics and mathematics, but then started taking some philosophy courses on the side just because I was curious and then very quickly just react. Realized that philosophy is what I should be doing and then eventually changed to change that and did all my studies in philosophy and studied in. Daniel pointed out I studied in Finland and after getting my PhD, I worked in temporary research positions in Finland and that's when I also first met Daniel. I'm currently at the Norwegian University of Science and Technology at Trondheim. I arrived here in 2010 and yeah, have been here ever since. And also I've been focusing on philosophy of language more or less throughout all my studies and my research.
- [06:17]
B
Okay, so I guess maybe back to Daniel for how this book came about. Collaboration.
- [06:25]
C
Yeah, yeah. Thank you. Yes, as I said, so we met already almost 20 years ago in this departmental collaboration workshops and I was interested in this early X file stuff because of the sort of methodological challenges that came with that. And Jussi was I think at that time already doing some work in experimental philosophy and thinking about that. And we were both interested also in two dimensional semantics on all these issues around this. And then I think back then we already started to discuss a little bit, let's say, the meta issues about that. So the questions. Okay, so if experimental philosophy is supposed to be a methodological challenge for philosophy of language. What kind of challenge precisely is that? How is evidence about intuitions of speakers related to certain kind of externalist and radical externalist theories? Does that, like, what should be the relevance of that in the first place, if one is externalist of a certain kind? And we did then, I think, intensify that collaboration in the context of a European Science Foundation Euro course project that was the main initiator was, I believe, Oza, Wikforce and some other colleagues from Stockholm that we also collaborated with. And that was an interesting context for discussing these topics because they were sort of semantic internalists and semantic externalists in this team. And there were cognitive scientists in this team and psychologists in this, in this team that were doing experiments on classification intuitions and stuff like that. And we kind of were interested in, okay, so what could be a common framework for all that stuff to really collaborate and maybe make progress on some of these questions in philosophy of language, like, for example, whether the internalist or the externalist is right about, for example, reference or meaning. And that's how we then started collaborating on papers on these questions. And then we tried to, yeah, came to that view that we now defend in the book, this meta internalist view. And I think the idea to write a book was really that we saw, okay, so there are these interesting connections of that view that are maybe important to spell out to show how fruitful that idea is. And that view is about methodology and maybe even consequences for discussions of certain issues in metaphysics and so forth. And that we felt that to be increasingly difficult to explain in a paper because you always had to sort of explain all that story about sort of, what is that meta. Metasemantic level? And what is meta externalism? What is meta internalism? Why do we think that one is correct, blah, blah, blah. And then we could eventually come to a point where we could then explore some of these further consequences. And we thought, okay, so it's just easier to do that in a book than try to do that in a paper.
- [09:44]
B
Excellent, excellent. So, yeah, so the title is Foundations for Meta Semantics. And as you just kind of mentioned, it's kind of a meta. Meta semantics, right? So there's, you know, it goes first order, second order, third order, third order. And as one of the questions that you mentioned in the book, that has been raised as well, why not keep going forever? I'm not that interested in that particular question, although it certainly could be asked, but what is metasemantics? I mean, could you, yeah, explain what metasemantics are actually meta reference and what the motivation is for, you know, having this second or third order level for thinking about reference.
- [10:36]
C
Yeah, right. So the, so semantics, the way that we use these terms. So there's a little bit of inconsistency, I think, in the discipline about how to use these meta and terms. But in any case, the way that we use it is that we think, okay, semantics is a theory that assigns meanings to expressions. So that either for concrete token expressions, a concrete token meaning for that expression, or typically in philosophy, at least at a more abstract level, that for types of expressions, certain types of meanings. Right. So that I don't know that names, for example, refer to persons. And then at a metasemantic level, these theories, they try to explain how that connection is made. Right, so between, for example, a name and particular person that that name refers to. Right. So these are these familiar theories that we know from Kripke and Putnam on the one hand, or Frigg and Russell on the other. So that either, for example, that we think, okay, so name refers to a particular individual because that individual best satisfies the cluster of descriptions that the speaker associates with the name, or that competent speaker of the language associates with the name, or something like that, or in the causal historical story, that the name refers to a particular individual, because current usage of that name is causally and historically linked to, I don't know, maybe a baptism event in the past where that name was introduced for a specific individual. Right? So that's what we call the metasemantic level. So a theory that explains how the semantic links are established, and then the meta, Metasemantic level or the foundational level that we are concerned with is then the question, okay, so, but given that there are these alternative theories, like, for example, the two types that I just sketched, the descriptivist theory on the one hand, the causal historical on the other, what actually makes it the case that one of these is true for expressions? Right, so four names, for example, or natural kind terms or artifact terms, artifacts in terms of. And so forth. So just asking sort of the same kind of why question to some extent, just at the next level. So as you say this, you can to some extent go higher up. We stop at that. We don't really stop at that point because we also have some meta considerations like, okay, so how do we want to answer that particular question? Of course, in the book. Right, so we reflect on that next level too, if you will. But in philosophy, I think you can always just continue asking why questions. But then you get at some point I think. And that's what we say also in the book, I believe, I think outside the domain of philosophy of language at some point and then just end up in a very general philosophical discussion.
- [13:59]
B
Okay, let me just ask a question. So you have these alternative theories, you know, well known, the causal, historical versus descriptivist and then some hybrids and things like that. And the meta, meta view is, you know, what is it that makes these true? The answer would seem to be, well, the semantic facts of some sort. So is that, is it, is it a matter of what are the criteria for deciding which is true or what the facts are that make them true?
- [14:39]
C
I think both. Like it's on one hand, it's first of all the question, okay, so what kind of facts, like what kind of theories are these in the first place? Right. So there's, I think one of the first persons who wondered about that was probably Stephen Stitch. So I think in Deconstructing the Mind, he has that question. Right? So he says, okay, so there are these different theories of reference and how should we think about these? Like, what are they actually trying to do in the first? What is it that they are trying to explain? What is it that they're trying to capture? Precisely. Should we, I think in his terminology back then, should we think about these as sort of folk, psychological, folk theories of reference, or is it something else? If we think about these more in a more naturalist way, so not so much trying to capture the intuitions of speakers and the way that Stitch at the time was thinking about this, but more about a natural phenomenon in the world that these theories try to explain. So that's more question about the metaphysics. Right, so what are the facts that these theories are concerned with? But of course, we think that that's sort of prior to a question that you have to have an answer to prior to the question. Okay, so what are the criteria for favoring one theory over the other? So what is the evidence then for. For these theories? And can we improve, for example, methodology in philosophy of language to get better evidence for one of these two questions and first have to know, in order to know what better evidence would be, what kind of theory is it in the first batch? Right, so what is it? What are the facts that we want to get at in order to know which theory is Right?
- [16:33]
B
Okay, you see. Do you have anything you want to add to that?
- [16:40]
A
Yeah, so one of the central points of the book is to stress that these meta level questions and meta level theories and these meta. Meta level or metametasemantic theories and questions that they are important different. The questions that these theories are concerned with are. Are different. So these metasematic theories and metametasematic theories are trying to answer completely different questions. And our claim is that in much of the literature these two levels have not been sufficiently clearly distinguished and that that has led to some confusion. And I was thinking, one way, I think, to illustrate the importance of the meta. Metasemantic questions is to notice that at least most of us would think that different expressions often can get their meanings and extensions determined in different ways by different kinds of mechanisms. So to take an example, most people today are semantic externalists, or what we call first order externalists, about natural kind terms. So we think that, for example, example, the extension of the term water is at least partly determined by some factors external to the individual speaker using the term water. Whereas to take another example, the term bachelor is. The extension of the term bachelor is not, most people would say, determined in the same way. So we have two terms, the term water and the term bachelor, and their extensions are determined in different ways. What is it the case, or what makes it the case that the extension of one term is determined in one way while the extension of another is determined in some other way? So this extension determination has to be based on some further facts and metasemantic questions which are our focus in this book. Meta. Metasemantic theories are theories that try to answer this further question. What makes it the case that the extension of a particular term as used by a particular speaker is determined in one way rather than another? When we see that the extensions of different terms can apparently be determined in different ways? So there's not one uniform mechanism for extension determination across all linguistic expressions.
- [19:27]
B
Okay, good. Thanks. That was, that was helpful. So, yeah, so pursuing this, you know, you distinguish, you know, again, you said, you know, first order externalism versus, you know, first order internalism, and then you go, you know, again, the meta. Meta externalism and meta internalism. Could you explain those differences? Because you are defending meta internalist view, right? So maybe we should, you know, what are these two different views and what is your view and why do you hold that view?
- [20:09]
A
Yeah, so as you said in the book, we defend meta internalism, but we also combine that with, we call first order externalism, which is really just our term for what typically gets called semantic externalism. But we use the term first order externalism and internalism and meta internalism and meta externalism to keep these two, two distinctions separate. So one of our central points is that just as in the case of metasemantics, you can be a semantic externalist or a semantic internalist. You can also be an internalist or externalist on the level of meta. Meta semantics. So we, we distinguish between meta internalism and meta externalism. Meta internalist, which is our view, meta internalist says that the question of how the. How the extension of that expression used by a speaker. Sorry, let me backtrack. So the metain journalist claims that when a speaker uses a term, the question of how that term's extension is determined is determined by sound facts internal to the speaker at the time of making the utterance. So, for example, I will probably get back to this later when I use the term water, supposing that first order externalism is true about the term water, then our view is that something about me when I use the term water makes it the case that its extension is determined partly by external facts in my environment, such as the molecular structure of the watery stuff that I'm closely interacting with, or something similar. A meta. Ex journalist view, by contrast, would be a view on which the semantic facts are. Sorry, the meta. Meta. Semantic facts are somehow externally determined so that what I do and what I think, how I speak is not enough to determine how the extensions of my terms are determined. But the extension determination is at least partly due to some external facts of which I might not even be aware of and to which I might not even have cognitive access. And an important theme in our book is that these two levels are set logically independent, so that the fact that you are externalist on the metasemantic level, that is a first order externalist, that doesn't commit you to being a meta externalist. And indeed, as I said, we combine meta internalism, that is internalism in meta metasemantics with first order externalism, that is externalism in metasemantics.
- [23:09]
B
Okay, okay, so it sounds like the sorts of facts that would distinguish a meta internalist from a meta externalist would be something about my relation to other speakers. You know, the, the language that I'm speaking and my maybe deference to certain people could. Is, is that, is that a correct way to kind of distinguish between the two? I mean, you gave a good explanation, but I'm wondering if there could be an example that would show what this psychological, you know, the internalist is depending. It's just facts on me. The externalist. It's not just facts on me of me. What are these other facts that the internalists. The meta internalist says? No, it's really just up to me.
- [24:11]
C
Yeah. So there are different meta externalist options. Right. So we discuss, I think, two candidates that we think we can find in the literature of relatively clear meta externalist positions. So one is this idea that goes back to paper by David Lewis where it's not so clear whether David Lewis really had that position, but Ted Seider endorses it in any case, that's this sort of reference magnetist view. Right. So in this case, the idea is that the metasemantics are somehow settled by the world, so really completely independently of speakers to some extent. Right. So it's the fact that there are certain joints in nature that work as reference magnets for certain kinds of expressions that are used by speakers, on the other hand. And that is what determines then that a certain expression, namely that there is this joint, determines that a certain expression refers in a certain way to that joint. So that would be a way where you'd sort of locate the metametasemantic facts really in the external world. And then there's another theory that is a bit more complicated. We think that we can see that in the writings or find that in the writings of Williamson and Hermann Capan, where the idea seems to be that the relevant meta semantic facts do supervene to some extent on usage in a linguistic community. Right. So on the causal interactions of language users in this community, but are not directly reducible to sort of individual dispositions or any kind of what they call common creed amongst the speakers of the language. Right. So where it's just some sort of higher level, let's say, let's call it like a social fact that emerges from these individual interactions, but it's not really, cannot be related back to a common creed among the speakers. So in this case. Yeah. So the facts that would Adam sort of meta externalistically determine metasemantics are still in some sense facts about the linguistic community, but not directly about sort of linguistic users, but it's something that sort of supervenes on the complex social interaction and linguistic interaction in this particular case of the speakers in the linguistic community.
- [27:11]
B
Okay, as you mentioned, the two of you defend a meta internalist view, and it's specifically dispositionalist. So could you say what that is and why you think that's the best way to go?
- [27:33]
A
Yes, I'll answer this one. This also connects to what we were asking about earlier, about difference. So I think it's useful to motivate our dispersionless meta internalist view in kind of three steps. So first we just start with the observation that it just seems overwhelmingly natural to think that our dispositions to use terms have something to do with what they mean and what they refer to. So there has to be some clear or close connection between, for example, what my word table means and what I'm disposed to use the word table to, and how I'm disposed to act when I hear other people use the term turntable. But then of course, we obviously can't just equate meaning with actual use. For one thing, we often make mistakes. We call non tables tables, or we omit to call tables tables. And one way to try to account for that, which is what we also do, is to bring in what we call directive dispositions. So sometimes I may apply the term table to something that's not a table. But then it makes sense to think that my word table doesn't correctly apply to that object. Because if I were to reconsider my application of the word table to this object, I would retract that and say that no, it's not really a table table after all. So far, all of this is quite familiar from discussions of the rule following problem and Kuipers, Wittgenstein and so on. But where our view brings in something novel is the following and the third step. So we explain with our dispositionalist meta internalism, we explain why what kinds of dispositions on the part of speakers are required to make first order externalist theories true, when they in fact are true. And here we bring in what we call burden shifting dispositions to do this. So it appears that with some terms, at least, for example natural kind terms, we seem to be systematically disposed to revise our patterns of using those terms in response to getting new information about the world or new information about how other people use the same term. So, for example, to take the example of the term water as Putnam's twin earth thought experiment makes explicit, or at least tries to convince us of
- [30:37]
B
if,
- [30:38]
A
for example, I were to find out that the stuff around me that I've been calling water has a completely different underlying structure than I've been assuming. Suppose I find out that I'm really on twin earth and the watery stuff around me is XYZ instead of H2O, then I think I would be disposed to revise my patterns of using the term water both in actual and in counterfactual conditions. So what this means is then that new information about the about my physical surroundings makes a difference to how I'm disposed to apply the term. So it makes a difference. It makes me revise my systematic patterns of using that term. And so the Twinner thought experiment, in our view, it's really a vehicle for bringing these dispositions that we have. To our attention. And in the book we then explain how systematic dispositions of this kind, when present in actual speakers, makes it the case that first order externalist views are correct or are true of those those terms. And we call these burden shifting dispositions because in effect, these are dispositions to let the let the external world do some of the work, share some, or bear some of the burden of determining the metasemantic facts, the work of determining the extension of the term water in this case. So you asked why we should what's attractive about dysfunctionalist meta internalism? What I think is attractive about this is on the one hand, it makes sense of our ordinary practice of for example, it makes good sense of our use of thought experiments in philosophy of language, because on our view, famous thought experiments in philosophy of language, like Twin Earth or like Kripke's thought experiments in naming and necessity, they are really, as I said, vehicles for pointing our attention at certain kinds of linguistic fairly complex linguistic dispositions that we already possess. And it's these dispositions, in our view that in fact make it the case that semantic externalism, or first order externalism as we call it, is true. Oh, and if I can come back to what you asked about earlier, you mentioned deference. So on our view it's really an empirical question whether we in fact defer to other speakers and that we think that there's very, very good reason to think that semantic difference is highly prevalent in actual speakers. But what we then say is that an individual speaker defers to other speakers and thereby it's the case that the extensions of that speaker's words are partly determined by how other speakers in the community use the same word. That can only be the case if that speaker is in fact disposed to let information about other speakers usage of the same term affect her own usage, and if that speaker is disposed to revise her own patterns of using the term in response to learning or getting new information about how other people in in the same community are using that term. So a speaker is deferring and therefore some kind of social externalist view is is true of can be a true of of of a speaker's usage of a term. But that can only be the case if the speaker herself at the time of usage is disposed to let other speaker other speakers usage be relevant for her own usage.
- [34:53]
B
Okay, should I want to ask about X phi and the thought experiments, because you both mentioned that was kind of a part of the genesis of the book way back when. But let me just. Is it possible to be a meta internalist who does not have these burden shifting dispositions? I mean, the worry, I suppose is just that language use is not something that we have total control over. And the burden shifting disposition sort of says, yeah, we don't. But I'm just wondering if there could possibly be an internalist who isn't a dispositionalist or who doesn't have these burden shifting dispositions at all. It seems like that's the only coherent meta internalist position to have, which the. Which is a dispositional. The person who has these burden shifting dispositions. It's hard to imagine somebody who doesn't.
- [36:17]
C
I don't know whether that's. So here's an example. So that shows two things, I think. So one, that it's possible to have these. And to another also why meta externalism is wrong. So when we wrote the book, I think when we were writing the last chapter of the book, I think that was when already ChatGPT emerged on the scene. And I was playing around a little bit with that because we were also in the book saying a little bit about, okay, so how would you know what kind of meta. Metasemantic theory is true of a large language model and these kind of things? So can we apply sort of these kind of theories also to these sort of artificial or AI chatbots and so forth? And that was interesting because what I tried to do was first like to give ChatGPT some philosophical thought experiments and see how ChatGPT evaluates these. And that was kind of what you would expect, that ChatGPT recognized these as philosophical thought experiments and then said what the philosophers said about this, right? So that was not so super interesting. But then I tried to do this in a more conversational way, right? So give ChatGPT certain kind of information about water and what scientists have just recently found out about the watery stuff on our planet and stuff. And to see whether ChatGPT would be in any way disposed to change its use of the term water in any way in response to that new information. And what I found at that time, at least in this conversation that I had back then with ChatGPT, I don't know whether that's still the case and haven't tried it recently, is that ChatGPT would always default to a definition of water. Just always say no, by definition, water is H2O. There was just no way to so push ChatGPT to, to accommodate any of that new. So ChatGPT just wasn't disposed in that case to defer, let's say, to anything for the tempo. So it was just sort of sticking to the definition that it had. So that's the way our speaker could be, right? So we would think that chatgpt. And that's now the second part. Like why is externalism non explanatory? So we would think, okay, for whatever reason, so whether we are meta externalists or meta internalists, in any case, we would think that ChatGPT is mistaken here, right? So that we would think that, okay, so it's just any case having the wrong metasemantic theory for water. So it's not just by definition that water is H2O maximum, because we all learned from Kripke and Putnam that's because of the causal historical link to the actual worry stuff in our world that what is H2O? But nevertheless, ChatGPT is using language in this respect, nevertheless in an internally coherent way. It's not coherent with our usage of the word, but it's an internally coherent way to do it. And you could imagine a linguistic community where every speaker is just like ChatGPT, right? So that would not be our linguistic community, but it would be a linguistic community. And in that community, these sort of chatgpt like speakers, they would presumably have no issues in communicating, right? So in transferring information from one speaker to the other. In a successful way. Now, still, like a meta externalist would look at that community and would think for, and could think, they're still wrong, they're still getting it wrong. So magnetist, for example, would say they're still wrong. So water is not by definition H2, okay? And the way that they sort of stick to the definition is a mistake. But that idea of reference that the meta externalist would have, who looks at that, that reference or that notion of reference would play no role whatsoever in explaining why these speakers in this like imaginary chatgpt community or chatgpt like community are communicating successfully. Because that would be just a fact that's also there, but it's not playing any explanatory role. And that's why we think that meta internalism must be right. So because meta internalism says no, no, it's like this notion of reference is just not linked to what is actually affecting the speaker. It's not linked to what can possibly explain success or non success in linguistic communication. And that's why we need at that level, right? So at this meta metasemantic level, an internalist account. Right, so that somehow explains. Okay, so how does information and communication affect the language users?
- [41:44]
B
Well, I mean, the problem there is that we would not think ChatGPT is referring to anything to begin with. I mean. Right. It's not really a speaker.
- [41:58]
C
Yeah, yeah, that's why I said, so the ChatGPT, like speakers. Right. So you would then, I think, presumably think about like just a linguistic community of humans that, however, that have the dispositions that ChatGPT actually had in that conversation that I then had with ChatGPT. Right, so it's not so. Right, so speakers that are in a. In a. In an environment with actual water, that are in closer contact with actual water and, and so forth. All the kind of things that ChatGPT in fact is not. Of course. Right, so the.
- [42:29]
B
Right, okay, so let me, Let me, I mentioned. You've mentioned and I said I wanted to ask about the X phi. All right, so one of the. Yeah, a lot of this, you know, sort of meta thinking about theories, you know, you know, stitch, but also the, you know, famous study of, you know, by Edouard Macherie and others, you know, testing, you know, using these thought experiments as, you know, essentially as, you know, sort of social psychology or sociolinguistics to see, you know, what I forget was it Hong Kong? I think they tested intuitions.
- [43:09]
A
Right.
- [43:10]
B
So the idea is, let's use these as actual experiments to see what people think. And are these intuitions just local, Western, you know, or maybe just, you know, Anglo American communities? Are they universally human, those sorts of things? So you've both been immersed in this, you know, XY language community for a while. So what do you, you know, what do you think about, you know, the role of intuitions in. In the broader question of the foundations of metasemantics. And you also make a recommendation for how these XFI methods might be improved. So could you tell us a bit about the role of XFI in your theorizing and then where you think it could be improved?
- [44:17]
C
Maybe I start with saying a few things about the intuitions thing and then give it over to Jussi for saying a bit more about this sort of potential improvements of methodology. Right, so the X File story is of course, a little bit complex because in the beginning. So in 2004, when they first started thinking about this, it was, I think, primarily intended as an attack on the methodology of philosophy of language. Right. So it was not meant to improve the methodology, but it was like at the time with really sort of really skeptical view on the way that philosophy of language works, and also skeptical about whether that's a field of study that's worth doing in the first place. So there are some papers by that group, like Masher Eduard and some other colleagues. Yeah, that had really a sort of very skeptical view on this whole business of trying to find a theory of reference. These kind of things that have changed. So nowadays there's a very active field, experimental philosophy of language, where people try not just to sort of show that philosophy is sort of methodologically bankrupt or something, to show. No, look, we can, yeah, do interesting experiments that might inform us about how reference actually works. And what we try to do in the book, in one part is like, first, to clarify a little bit more, when we talk about intuitions in philosophy of language and the role of thought experiments in philosophy of language, then what are we really talking about? What's their function there? And how do they presumably work? And why do they maybe still tell us something about reference, despite the skepticism that the experimental philosophers have voiced in the beginning, like in 2004 and so forth? And there we make a distinction that you can make maybe between thought experiments where you look at the imaginary case or the counterfactual case as a sort of outside observer, where you then try to come with a judgment about the case, or a thought experiment where you put yourself sort of in imagination, also in the case, and sort of simulate the case in a way and come up with a judgment that way. So in the case of philosophy of language, for example, it's very often the question of, okay, so here's a hypothetical situation described. And now how would you use a certain expression in that kind of situation? And the way that we think about it is that it's plausible to think that what these thought experiments ask us to do is to sort of use our linguistic competence, right? So the way that we normally just use language in producing utterances and interpreting utterances to also do that in this hypothetical case. So to simulate the case and see, okay, so what is it that we would say, or what is it that we would understand in such a situation? And if understood that way, then, or in this way, then we are talking about, to some extent, the exact same dispositions that Jussi was talking about earlier, right? So, namely, those dispositions, too, produce utterances in application to certain situations or to interpret utterances in certain situations. And if understood that way, then thought experiments are not maybe a perfect source of information because there's like. Because it's complicated, right? It's difficult to get the case right. To put yourself into the right kind of situation. But it's relevant data, right? So it's relevant information that if meta internalism is right, that would inform us about, okay, so what exactly are these dispositions then? And that should inform us a little bit in any case about which theory of reference is right. And so that's, that's how we would interpret some thought experiments in philosophy of language, like for example, the most famous thought experiments by Kripke and Putnam in any case as thought experiments that invite us to use the linguistic competence that we have in order to come up with judgments about that case. And if you interpret thought experiment that way and the use of intuitions that way, then that's also what we argue in the book. Then that also clarifies a little bit around this discussion of this experimental philosophy X File challenge. Right. So there's this famous expertise defense, for example, right. So where people were saying, oh, all this X file stuff is completely irrelevant because they were asking lay speakers or laypeople about their intuitions and why should that matter? Right. So we should ask experts, and experts in this case are philosophers for language. And so what they into it's the laypeople and that's not interesting. But if our reading is right, then that's not the right move. Right? Because what we're interested in, the data that we're interested in, is data about the dispositions of just ordinary competent speakers. And an expert in philosophy of language is not necessarily a better speaker and doesn't have better competence in intuitively interpreting utterances or something like that. What they may have, and that's also what we allow in the book and why we think about why we, what motivates us to think about alternatives to this thought experiment methodology in general is that that an expert philosopher of language, of course, might be better in knowing what's at stake. Right. So they might be better in putting themselves in the right relevant imaginary case that's relevant and right for evaluating that particular theory, for example, they might be better in reporting the intuitive response that is of relevance for evaluating that theory and so forth. So at that level, an expert might have an advantage over a non expert given that particular very demanding, complicated methodological tool, thought experimentation. And that's why we think, okay, so maybe there are methodological tools that are less demanding or ways to use a modified way of considering these kind of counterfactual or hypothetical cases that make it a little bit easier to test more broadly what just ordinary speakers, how they would use language. Yeah. And then maybe jussi can say a bit about that.
- [52:03]
A
Yeah. So in the book we also discuss, maybe present some general recommendations for X5 methodology. So as you both mentioned, a lot of the early work in experimental philosophy of language used thought experiments that were fairly directly adapted from these well known and influential thought experiments in philosophy of language, like from Putnam and Kripke. But then we should keep in mind that these thought experiments were originally developed for a very specific purpose, that is for the purpose of convincing other philosophers who are already familiar with the issues, with the relevant distinctions, with the questions that we are asking, with the theories that are on the table. And there's no reason to think that, at least no obvious reason to think that thought experiments that work well for that purpose, as these famous thought experiments obviously did, that they would also work well, or let alone optimally as tools for studying the linguistic or the language use and language understanding of ordinary speakers. So of course, when people started doing experiential philosophy of language, they had to start somewhere. And so it was natural to start with something that we were already familiar with as philosophers. But one thing that we recommend, and one thing that's already, I think been happening for a number of years now, is that in experimental philosophy of language, people are distancing themselves more and more from these well known Thor experiments and developing experimental setups that were, that are kind of specifically developed for testing lay speaker usage. And that's something that we very much support and applaud and also something that I've been involved in my own experimental work. So we think that, we definitely do think that there's a place for survey methods. But they, of course there are better and worse ways of conducting these surveys. And there's a lot of exciting work, I think, going on where people are developing better survey methods. At the same time, we think that we shouldn't just stick to these survey methods. There's a lot of other experimental methods that we could be using. And here we think that we should be looking to psycholinguistics for ideas and inspiration. And in fact, in psycholinguistics for some decades now, people have been studying related issues like also for example, the reference of pronouns and, and definite descriptions using methodologies such as eye tracking or elicit production or EEG measurement of language processing in the brain. And some of this is also happening now in experimental philosophical language. So people have been using elicit production. There's now at least one study using eye tracking. So, so this is very clearly an ongoing process in experimental work. Now
- [55:42]
B
how is eye tracking is the idea that where people are looking is where they're seeking information and that is where that will somehow inform what their reference or referential dispositions are, right?
- [56:01]
C
Yeah. So we got our inspiration already many years ago for these type, for these kind of experiments from our collaboration with Peter Bosch, cognitive scientist, who was also partnering our Euro course project. They had sort of, they were working in what's called the is it visual world paradigm or something, right? So, so it's the idea that you see a certain visual scene, right? So you get a screen that displays certain objects and then you hear a sentence uttered with referential expressions in it. And then what you can measure is indeed fixation on objects on the screen after a certain kind of expression was appeared in that sentence that you hear. And that gives you evidence on the one hand, indeed like how you intuitively interpret a referential expression, right? So it's like where the fixation probability goes up for on the screen, but you get also sort of evidence about processing, right? So like how quickly that is and whether that follows first fixation on some other object maybe, and then you fixate on the this and that kind of evidence. That's really interesting because it circumvents a lot of issues that we normally have with the survey method. Right. So for example, one worry that was discussed quite a lot in the early days of experimental philosophy of language was, okay, so how do we know that people are really using their linguistic intuitions in these cases, right? So that they are just telling us how they would intuitively understand that expression rather than telling us what their pet theory is about this or what their intuitive theory is. So how do we know that we get this really basic linguistic response out of their linguistic competence rather than some sort of higher level thing which we are maybe not so much interested in? And eye tracking is great in this respect because it just gives you that sort of immediate response. And that is not as far as at least that paradigm thinks is the case, is not sort of tainted by sort of our theoretical background theory, right?
- [58:55]
B
Unless perception is cognitively penetrated, then it becomes a bitter problem.
- [59:01]
C
Yeah, exactly. But this paradigm holds that that's not the case and that that's more robust.
- [59:12]
B
We're out of time, unfortunately. We've already run for the entire period. But I had a lot of more questions about reference change and indeterminacy and conceptual analysis and conceptual engineering. But unfortunately, or fortunately, other listeners will have to actually read the book to get some of those details. So in the meantime, to close up, are you working on some other further collaboration at this moment, or what are your next steps?
- [59:58]
A
At this point, we're not really working on, like, concretely working on any joint paper or let alone another book at the moment. But I think we're both still working on very much the same questions. And also, for example, co organizing a symposium at the ecaf, the European Congress for Analytic Philosophy, in Madrid this summer. Myself, I'm working both on experimental projects and on theoretical projects that are still very much connected to the work that went into this book. So I'm involved in experimental projects where primarily the aim is to develop better survey methods that are not subject to the shortcomings of past and present survey methods. That's very much an ongoing project. And at the same time, I'm engaged in a couple of article projects of a theoretical nature where I'm picking up loose ends from the book, things that we mentioned just in passing in the book, but didn't have the opportunity to go into in any detail in the book, such as what does our view. What, what does. How you say about deflationist theories of reference and related questions? So I'm very much working on, on the same, same sets of, of, of issues also after the book came out.
- [61:37]
C
Yeah, for me it's similar. So the, so I'm so in the, in the book, at the end of the book, we say just a few words about, okay, so how would all that maybe help us with understanding artificial intelligence? And maybe. And how can we make sense of what seem to be linguistic utterances of large language models and so forth? So how to apply that stuff to these things? So I tried to work on a paper on that topic. But then I also tried to currently try to think about another somewhat larger project back in the day, 10 years ago, in my dissertation on thought experiments in philosophy, where you try to look at the question, what's the thought experiments in philosophy? A good, trustworthy method. I gave the very disappointing, boring answer that it depends that the. And then in this book, what we do in this book to some extent is we zoom in on a specific case and really show what it depends on. Like that we say, okay, so look, if you have all these other background assumptions, and if you make these, and these and these and these assumptions about speakers and dispositions and what intuitions are and what dispositions and how they are connected and so forth, then you can give that story about the role of thought experiments in philosophy of language in that discussion. And what I would like to do next is to zoom out of this again and say, okay, so how can one nevertheless give a positive, interesting story about the relationship between philosophical problems and questions on the one hand, and the appropriate methods on the other. Given that you find yourself with a certain philosophical question, how do you find the appropriate method for that question in a more general way, and taking into account that there is not one story that can be told for all of philosophy, but that it depends. But to really clearly say systematically what it depends on and how to find yourself on that path from, let's say, problems to methods.
- [64:05]
B
Yeah, no, that's a great topic there. Well, I want to thank you both for taking the time to talk with us at New Books and Philosophy.
- [64:20]
A
Thanks for having us.
- [64:21]
B
Yeah, I wish you both of you luck with your individual projects and anything you might do together. So thanks again.
- [64:30]
A
Thank you.
- [64:30]
C
Thank you.
- [64:31]
B
You've been listening to an interview with Daniel Koenitz, professor of Philosophy at Utrecht University, and Jussi Haukioja, professor of Philosophy at the Norwegian University of Science and Technology. We've been talking about their new book, Foundations for Metasemantics, which is just out from Oxford University Press. I'm Carrie Figdor. This is New Books and Philosophy, a podcast with the New Books Network. I hope you enjoyed the podcast and thank you for listening.
- [65:03]
A
Thank you for listening to this episode
- [65:05]
B
of the New Books Network. We are an academic podcast network with the mission of public education. If you liked this episode, please share it with a friend and rate us on your preferred podcast platform. You can browse all of our episodes on our website, newbooksnetwork.com Connect with us on Instagram and Bluesky with the handle
- [65:23]
A
eBooksNetwork, and subscribe to our weekly Substack newsletter at newbooksnetwork.substack.com to get episode recommendations
- [65:31]
C
straight to your inbox.