"Man is a creature who makes pictures of himself, and then comes to resemble that picture" - Iris Murdoch

Sunday, 25 February 2018

The meaning of meaning

Path among Pines by Emily Carr, c. 1930
Path among Pines, c.1930, by Emily Carr (Vancouver Art Gallery) [Public domain], via Wikimedia Commons
I once had to write an essay about the meaning of meaning. More precisely, we were asked whether the meaning of a sentence was the same as the conditions in which it would be true. But I thought of it as the meaning of meaning, and decided that I had had just about enough of philosophy, and of studying in general. This "philosophy of language" nonsense was the last straw. I don't think this any more—in fact, I think all social scientists should know some, just as they should know some maths. But that isn't what this post is about. Not yet anyway.

I'm going to write instead about the other meaning of "meaning"—the wishy-washy one we apply to events, actions, situations etc. when we ask whether they are meaningful to us, and how. I'm going to be summarising part of the Canadian philosopher Charles Taylor's 1971 article "Interpretation and the Sciences of Man". Taylor is explaining why we need the right-hand column in the matrix I included in my last post:

interested in causes)
interested in meanings)
e.g. classical and Marxian political economy
e.g. Wittgensteinian language games
e.g. rational choice theory, game theory
e.g. dramaturgical models

That matrix was philosopher of science Martin Hollis' window, a set of four approaches in social science he sees as jointly needed to understand society, but difficult to put together.

I hope it's fairly intuitive why the left-hand column matters: one thing we'd like from social scientists is good explanations of what makes things happen the way they do. But it might not be clear what exactly the right-hand column is about, and why it should be roughly equal in importance (or even more fundamental, as Hollis suggests). I think Taylor offers a good starting-point for thinking about this, and I'd be keen to hear from readers whether what I'm presenting here makes sense, what might be missing, and where might be a good place to find it.

Taylor starts by discussing why we would think the human sciences would have an interpretative, or hermeneutical, component. (For reasons of space, I'll have to skip some of the detail of how interpretation in the social sciences overlaps with hermeneutic approaches in literature, theology and elsewhere, but it's a nice article if you have time and access. It also includes some highly relevant discussion of politics and social conflict.) For Taylor, interpreting an object of study is trying to make it clear, to make sense of it.

He identifies three features of an interpretative science:
  1. We must be able to apply this idea of coherence or incoherence to the thing we are studying. It has to be something that can make sense, or fail to. 

  2. There has to be some difference between the sense made by the thing—the meaning—and the way we express that sense. Otherwise there couldn't be such a thing as clarifying the sense: we must be able to say, this was unclear before, but now we've expressed it more clearly.

  3. There must be someone (or a group) who this sense is for. A cloud formation might have a "coherent" pattern, it might fit together in some way, but it doesn't have a meaning, unless it has a sense for someone. 
Why would we be skeptical of such a science? Taylor thinks this comes down to the question of how we judge whether or not a claim is true, or correct. Suppose we are interpreting a text. We do this correctly when we make sense of it. But how do we know that we have made sense of it? We rely on our understanding of the language the text is expressed in, which tells us that we have moved from an unclear expression to a clearer one.

But what if someone does not share this assessment? At this point, we must refer her to interpretations of other related texts, and show that they fit with this one. We might appeal to a part-whole relationship: the part we are interpreting now fits into the whole that we are interpreting overall. But what if they don't share any of these interpretations at all, because they don't share our understanding of the language?

At this point, we seem to be stuck, and at this point, a skeptic might say: this isn't a science at all. Here's a real science. There are data, which we can all agree on and don't need interpreting. Taylor calls these "brute data". To these we apply the laws of logical and mathematical inference. This is clear and solid stuff. This is science.

But, argues Taylor, we need to keep hold of interpretation to study people and society. "Meaning" is necessary to characterise human behaviour adequately, and we know this from our experience and from the way we talk. We can't really convey how Lata is acting without some reference (at least implicit) to what the relevant situation, action, demand or prospect means to her. These meanings help explain her purposes, desires, feelings and emotions.

This kind of "meaning" has three features:
  1. The meaning is for a subject: here, Lata. (This matches feature (c) of an interpretative science.)

  2. The meaning is of something, e.g. Lata's situation, and we can distinguish that thing and its meaning. That means we can characterise the thing without reference to its meaning for the subject, and with such reference. We can talk about something that happens to Lata, and what it means to her, and discuss the connection. (This corresponds to (a) and (b) above.)

  3. Things have meaning in a field of meanings, i.e. in relation to other things' meanings. 
This last feature is crucial to the logic of an interpretative science. Suppose Lata meets someone for a coffee and finds the experience refreshing. The meaning "refreshing" can't exist in isolation. She must have some grasp of other, related meanings: both contrasting kinds of meetings (say, tedious ones), and analogous experiences (like refreshing drinks).

Underlying all this are social practices. Obviously, Lata can't find her coffee meeting refreshing unless she can participate in a practice of meeting for coffee. Less obviously, the coffee-meeting practice involves distinctions that can shape how participants act and relate to each other, distinctions which ideas like "refreshing" and "tedious" point to. In fact, Lata needs quite a rich field of meanings indeed to find her away around the task of being a good coffee partner—some of which aren't specific to the coffee-meeting practice at all, such as those meanings which apply more generally to being a good conversational partner.

Lata does not necessarily have language to articulate this all. This is why there is work to be done here, of the sort that Taylor is talking about. Art and literature can help us see these things more clearly, but there are also scientific ways of going about these things, as exercised by anthropologists and sociologists.

Wait, cries our hard-nosed skeptic. There may be work to be done here, but this isn't science. All this—what we are investigating, and the way we are investigating it—is subjective. The best we can do is record publicly observable behaviour, which can include whether people agree to certain statements about that behaviour: why they act as they do whilst drinking coffee, or why they voted in that unpredictable manner, or what they want from the economy. But we can’t actually get inside their heads.

What Taylor is warning us, as Wittgenstein does before him, is that this distinction between inside and outside is false. The meanings are not just inside people's heads. They are in the social practices. Getting at those social practices most certainly presents challenges. Taylor has more to say about this, as, of course, do anthropologists, sociologists and others. So do philosophers across a range of fields and traditions, including philosophy of language. These are all things I hope to dig into in coming posts. But I'd welcome directions!

Finally, I'll just raise one query about Taylor's discussion. As we've seen, he talks about meaning being for a subject. This seems to be the central idea. Very quickly though, we have shifted to talking about that meaning in terms of language and the social practices they're entangled with. Here, the subject is a social group, and individuals qua members of that social group. But it might be quite important to us what meanings are for individuals as individuals, whether we are social scientists interested in variation within a group, or power imbalances, or as policymakers trying to incorporate this all into objectives and policy tools. So we need to stick a question mark over the thought that language, that very social thing, is as central here as we might have first thought.


  1. [This comment was too long for the form, so I have broken it down into two parts. Part 1.]

    I really enjoyed this post! I should preface this by saying that I haven't read the Taylor paper properly, so it may be that he addresses this point directly.

    My question concerns, I think, the point which you raise right at the end, about the relationship between meaning for an individual vs. meaning for a group. I take this to be an instance of a puzzle which emerges quite generally in social thought, but I will try to phrase it in terms of the framework which you describe here.

    You write that the requirement that "things have meaning in a field of meanings" is "crucial to the logic of an interpretive science." I take it that part of the point here is that by situating a given interpretable phenomenon within a wider field of interpretable phenomena with which it has non-trivial links, we provide interpretation with a kind of 'grit' which is necessary for it to constitute something like a 'science' as opposed to mere free-association. (The latter option is a bit of a caricature, of course.) Relating the phenomenon to the others in the field allows us to apply tests of coherence, etc.

    But another requirement of something’s being an interpretive science as a study of _society_, as opposed to an interpretive study of the individual (perhaps like some forms of psychoanalysis?), is that its results are at least somewhat generalisable across individual subjects. It seems to me that these two requirements generate some tension with the claim that “meaning is _for a subject_.”

    All participants in a given practice P at least have that practice in common. But if we are committed to situating a practice within a wider field of practices in order to interpret it, we have to ask what gets into that field. As soon as we do this, we notice that the participants in P will all belong wider sets of other practices which only partially overlap with each other. If we were to take the whole set of all practices which members of P are involved in, the meaning that P has within _that_ field might not be a meaning it has for anyone in particular, because there may be no-one who belongs to all of those practices. (Indeed, that larger set might contain various forms of contradiction or antagonism.) On the other hand, if we interpret P within the field of practices which a given member of P is engaged with, the meaning we obtain might be a meaning _only_ for that member. So there is a question of how this kind of interpretive holism can generate a meaning which is _for_ anyone in particular, without being totally idiosyncratic.

    Anthropologists of the early 20th Century are sometimes described as holding that cultures are self-contained, fully coherent unities, and that enculturation involves each individual wholly internalising the culture in which she is raised. I don’t know if any of them ever held exactly that view, but either way it does not seem to me to be viable. There is a weaker view, however, which might solve the problem of producing generalisable meanings which are nevertheless still meanings _for_ particular individuals.

  2. [Part 2]

    This view might be something like the following. Take a group C which is intuitively something like a ‘self-contained culture’. People within that culture might belong to various subsets of all of the practices found in that culture, and there may be no points which are shared by everyone. Nevertheless, we might think that the fact that practice P has survived in that cultural context must mean that it is in some sense compatible with all of the other practices found there, such that whatever subset of the practices in C a person belongs to, if the belongs to P then the meaning of P for her will be the same as it is for a member of C who belongs to a different set of practices including P. In this way, meaning could still be set holistically, at the level of a culture. And there may also be room for a kind of holism at the level of the individual – perhaps you cannot belong _only_ to a single practice, but whatever subset of practices you do belong to, their meaning for you will be fixed by their place in the cultural whole.

    This picture does seem to me to place weaker requirements both on the homogeneity of ‘cultures’, and on the effectiveness of ‘enculturation’. Nevertheless, it still seems quite strong in both respects. As I mentioned at the beginning, I take it that something like this kind of tension between interpretive holism, generalisability, and ‘meaning for’, is quite ubiquitous in much social thought. I’ve just tried to sketch how it might manifest in the vocabulary you have provided, by describing a couple of seemingly unattractive extremes that it might drive us towards. I would be very interested to hear your thoughts on whether it is a real problem, and if so whether there is a way out.

    1. Thanks so much for this Pete (and sorry for late reply). You've introduced a lot of dimensions here I hadn't considered. I think you've pretty much convinced me about your conclusion the holism of meaning, but I also think I'm looking for a different "way out". I'll respond here by: (1) restating the outline of what you're saying (let me know whether I'm getting it right!), (2) bringing in a bit more of what Taylor himself says, (3) responding to a few substantive points and (4) briefly indicating where my "way out" is heading...which will ultimately require another post.

    2. This comment has been removed by the author.

    3. [earlier commented deleted because I missed a key word!]

      (1) It seems to me you've raised three considerations which could be separated: (a) what a study of meanings has to be to count as *science*, (b) what it has to be to count as *social* and (c) what meanings might have to be like *anyway* (e.g. because they are, in fact, produced by practices). You've offered two sketches of what practices and meanings might be like if they satisfied all three considerations, where how you're thinking about (c) is constrained by how you've set out (a) and (b). Let's call these sketches Problematic Scenario and Acceptable Scenario.

      If that's right, I want first to sound a note of caution, and then suggest a distinction about (b) ("social").

      First: I'm wary about saying too much, in advance, about what *kinds of things* meanings/practices must be (a metaphysical question, if you like), if our study of them is to be scientific, and indeed socially relevant. This might block us from seeing something relevant. But I do agree that *epistemologically* the *way we think* should be scientific, and socially relevant. (I realise this metaphysics-epistemology distinction might not be strict.)

      Now the suggestion. You've said that for our study of meanings to count as being a study of society, our results have to be generalisable. Generalisability is an important issue which is worth discussing in detail. But I think there is a difference between our study being *of society* and our study being *socially relevant*. For instance, for the purpose of defining economic aims for macroeconomic policy, it might be useful to understand, through detailed case study, how particular individuals think about meaning. Such study might really be *of an individual*. Psychoanalysis is (usually) different, because the application is genuinely individual (helping the person live more effectively, without inner conflict, etc).

      Therefore: I'm wary of saying meanings *should* turn out to be set holistically. (However...)

    4. (2) This is also a nice opportunity to bring in some more Taylor from the paper which I had to leave out.

      I really like your point about relating phenomena to provide "grit", making connections, enabling testing, etc. Lots to think about here.

      But Taylor's point is more basic, and again, more metaphysical: it's about what meanings *must be like*. He draws a very close analogy to concepts. (I'm going to spell this out those unfamiliar with phil of language, but also please excuse my own rustiness!)

      A colour concept would not function as a concept at all if it existed on its own. There couldn't just be "blue" - it wouldn't pick out anything in the world, because it wouldn't have boundaries set for what it shouldn't refer to. Those boundaries are set by the other colour concepts.

      Likewise in my coffee meeting example. Lata can't pick out what a "refreshing" meeting is if there is nothing else a meeting could mean to her.

      Second thing I should mention is that Taylor actually develops this framework in his 1989 Sources of the Self (opening chapters), especially on the individual vs. practice question. One key innovation there is the idea of *expressing* meanings to other listeners/interlocutors.

      Having said that, in this 1971 paper there’s a key idea which might affect the analysis in your sketches. This is the idea of intersubjective meanings, valid for everyone. For Taylor, it’s not possible to identify what a bit of social behaviour is at all without using the distinctions provided through intersubjective meanings. They set up a common reference world.

      It is possible for an individual to have different meanings in their head to everyone around them, but this has to be because they have got intersubjective meanings from some other set of social practices (e.g. they’ve been educated in a missionary school). Even social cleavage and rebellion usually happens in the context of these intersubjective meanings: in the first case, people disagree in terms they still share, and in the second, they try to break away from terms without quite knowing how.

    5. (3) So a quick response to some of specific points now. Your Acceptable Scenario seems to operate on a different (though quite possibly compatible) logic to Taylor’s. The meanings seem to emerge out of a sort of equilibrium-finding process. Once we reach a set of compatible practices, the meanings will be fixed accordingly.

      I like this. But as in economics: how long do we take to reach equilibrium, and do we ever get there? E.g., I think on Taylor’s logic, it would be possible for certain *local* practices to be mutually incomprehensible to each other, as long as that didn’t stop people being able to interact in more general social settings. So there’s space for some fluidity there, in the extent to which the set of practices as a whole really fix the set of meanings.

      A second point: I think it’s true, as a psychological (and I guess epistemological) point about human beings and their meaning use, that it would be *difficult* (and costly) for an individual to sustain her own set of meanings in conflict to everyone around her. But I don’t think this comes down to your point about the generalisability of our *study* of meanings: I think this is a matter of (c), i.e. what meanings and practices really have to be like.

      I’ve not really answered your (a), actually, about scientific study. I do think, having said all this, that we do need to consider that question at length, and separately, because I’m suspicious about “generalisability”. I have a post I want to write soon about Dilthey. But I’m happy to explore this in further comments now too.

    6. (4) I meant to respond to you within 25 minutes, and ~4 paragraphs, and I’ve failed miserably at that! So I’ll just say: there’s a lot more to say here, but your arguments have pushed me further towards accepting that *meanings* must emerge out of, and must ultimately be governed by, a social process. So I don’t think by “meaning for an individual”, I am ultimately talking about meanings. I am talking about something that is phenomenologically more fundamental.

      So that will be the next post (in this series). But it would be great to have your thoughts on any of this. Thank you again! – it’s been a huge help thinking all this through (and knowing how much more there is to look into).