Is God's power limited by logic? Is our power of thought limited by the fact that we use language, or that we are acculturated? A family of questions about limits – the limits of being, the world, thought, the mind, life, freedom, and meaning – are philosophically troubling. What lies beyond those limits? Can we cross them? Can sense even be made of the idea that there is a limit to the universe, to thought, to the meanings language can grasp? I propose to offer a partial investigation of the usefulness to moral philosophy of the idea of limit.

In characterizing language we sometimes say that it has limits. But we may mean two different things by this: we may mean that it has boundaries or, alternatively, that it has contours. Boundaries or barriers face outward. They are the kind of limits that fence things out. Identifying or erecting the barriers of language is discovering the restrictions on, or actively restricting – defining – what language can in principle grasp or be. Usually, this involves abstraction from actual linguistic behavior, and is expressed by ‘impossibility’ claims, for example when saying that words cannot express certain things, or when formulating conditions that expressions must meet in order to be truth candidates. In opposition, contours or outlines face inward. They don't enable definition, but characterization – a description of the reality of that whose contours they are. Contours thus give a real-life portrayal of something. Drawing the contours of a linguistic activity involves a certain attention to reality: to detail and particularity. A particular greeting, for instance, may be characterized as mechanical and contrived, or alternatively as warm and genuine, or formal, or awkward, or sarcastic… This would involve attention to the greeting's contours.

Contours and barriers are internal to different philosophical methods. What contours and barriers are is clarified by how we draw contours and identify or erect barriers. Sometimes, I shall argue, we draw contours by using remarks I call ‘helpers’. I shall contrast these with another kind of remarks, which I call ‘theses’, often used in the identification and erection of linguistic barriers.

Within the barrier category, we may make a sub-distinction between contingent barriers that certain educations or political realities may create, and necessary barriers that are embedded in the very nature of language. I shall be mostly interested here in necessary barriers. We can further make a sub-distinction between two kinds of necessary barriers: all-encompassing barriers that delimit all sense, and circumscribing barriers that fence out some possible sense. Sections I–II concentrate on all-encompassing barriers, whereas Sections V–VI focus on circumscribing barriers. The latter can be further sub-divided into mobile and immobile barriers. I touch on this distinction in Section VI.

The attempt to draw linguistic boundaries can sometimes be confused. Sections I–II discuss the possibility that a confusion is involved in Sabina Lovibond's application of the idea that language has necessary all-encompassing barriers. Sections III–IV explain the alternative of drawing linguistic contours. Sections V–VI examine Richard Rorty's thesis, according to which the fact that we can shape our language indicates that its barriers are not all-encompassing. I compare his discussion with Sarah Bachelard's discussion of euthanasia. My claim here is that trying to improve on the language we have might be part of drawing its contours, rather than moving or replacing its boundaries. The discussion will reveal a difference between two kinds of contour drawing, and thus between two kinds of helpers: ones that help to draw the contours of the actuality of linguistic activities, and ones that help to draw the contours of their potentiality. Section VII argues that the value of drawing contours instead of barriers is that the former better reveal the fact that we care about our ways of making sense.

In many of my arguments, I follow in the footsteps of Cora Diamond. I shall try to make that explicit, although I shall not examine any of her works in any detail. This paper is a tribute to Diamond's philosophy.

I

John McDowell advises us to investigate ethics ‘from the inside out’.Footnote ¹ Our conception of right conduct, for example, is graspable only via the knowledge we have as members of our moral form of life. According to this so-called quietism,Footnote ² gaining moral knowledge is a matter of assimilating a form of life.Footnote ³ We can easily envision an issue of limits here: the barriers between us, who are on the inside, and some imaginary others, who were not acculturated the way we were – barriers that seem to restrict where we and they can travel in thought.

McDowell recommends a course of investigation, and I would like to examine a tempting but confused way of taking McDowell's quietist advice, signs for which can be found in Sabina Lovibond's writing on moral philosophy.Footnote ⁴ After assessing the confusion, I outline how I think the advice should be taken.

‘[T]here is something we cannot do’, Lovibond argues, namely, ‘codify’ moral thought and practice. The forms of knowledge deployed in practical reasoning ‘outrun anything we could “reduce to a set of (mechanically applicable) rules”’.Footnote ⁵ We cannot say things of moral significance in such a way that ‘their correct uptake owes nothing to any substratum of tacit likemindedness on the part of our audience’.Footnote ⁶ One cannot reason fully explicitly, Lovibond maintains, but not because it is too difficult. We are concerned here not with a contingent barrier and an empirical difficulty, but with a necessary barrier, and the conceptual impossibility of going outside the space of moral reasons while retaining the meanings of our sentences – going outside these barriers while keeping the grammatical identity of our words. Any attempt to reason explicitly in this sense would amount to not saying what we intended, or indeed anything at all. There is no moral sense outside the boundaries of morality.

‘[T]he uncodifiability of what is apparent to [morally acculturated people]’ Lovibond argues, ‘is offset […] by the endless availability of real-life material from which the spirit of their thinking can be reconstructed’.Footnote ⁷ Involved in moral thinking there is, accordingly, a kind of compensation – a kind of indirection, to use Cora Diamond's termFootnote ⁸ – and a necessary one: the kind of account of ethics we might want to supply is impossible for us, logically impossible. However, we have another, logically valid, even if a bit of a roundabout option. We can account for ethics from within.

II

It seems that the way Lovibond conceptualizes the barriers of language exposes her to a philosophical criticism. The criticism was formulated by Wittgenstein: ‘In the actual use of expressions’, he writes, ‘we make detours, we go by side roads. We see the straight highway before us, but of course we cannot use it, because it is permanently closed’.Footnote ⁹ Here, Wittgenstein gives voice to a kind of philosophical fantasy, and Lovibond seems to share a fantasy of that kind. Wittgenstein's remark is applicable to various issues: e.g. the existence of infinite numbers, past facts, and other people's minds. We may think that there are infinite numbers, there are past facts, or there are thoughts in other people's minds. We want our minds and language to capture these. But, alas, they are out of reach. We cannot count to infinity, go back in time, or enter other people's minds. The highways are blocked and sealed, and we have to compensate, make detours, and approach all these things by indirections.

The idea that a necessary indirection is present in thought may naturally involve the idea that there are some thoughts we logically cannot think, or some things we logically cannot do: count to infinity, go back in time, think other peoples' thoughts, and so on. Here we risk confusion: As in many other cases, ‘The great difficulty’, Wittgenstein says, ‘is not to represent matters as if there were something one could not do’.Footnote ¹⁰

Lovibond seems to be confused in this way. She appears to claim that the barriers of moral language fence some sense out: a kind of reasons and a kind of activity of reason-giving we cannot offer or engage in. Now, presumably, to conceive of such an activity would be to conceive of its reality: for instance, of what would constitute trying to engage in it, or learning, or encouraging, or forbidding it; and to conceive of such reasons would likewise be to conceive of their actual reality: the ways they justify actions, for example, and the sort of thinking that goes into estimating how good they are. In the case at hand, however, all this is irrelevant. Lovibond seems to rule all this out, and to maintain that the boundaries of moral sense encompass all moral sense and leave nothing outside. Yet, if one could not clarify the reality of those activity and reasons, but still wanted to say that something was impossible, it would very much seem that one would not be saying of anything that it cannot be offered or done.Footnote ^11, Footnote ¹²

III

Confusions like the one Lovibond's mode of presentation creates can often be traced back to a certain conception of limits: that in which the limits of sense are barriers and the drawing of those limits is a kind of fencing something out. In this section I explain how to conceive of the idea that language has limits differently: instead of barriers, we can use contours.

We saw that Lovibond's apparent conception of that impossible kind of moral reasoning disintegrates when we try to clarify what she has in mind: the reality of that moral reasoning. What is it to learn how to give reasons of the impossible kind? How can one estimate their soundness? The absence of answers to such questions indicates that we have nothing clear in mind. When we appreciate that, we can also learn to see that attempt at clarification, and the kind of attention it invites, as an attempt to draw the limits of sense – to characterize language – that is different in kind than Lovibond's. This sort of clarification is what I call drawing the contours of sense, which does not imply that some thing is out of reach, but that something is unclear.Footnote ^13, Footnote ¹⁴

We can draw the contours of a particular activity: a certain greeting, a certain description, a certain request. But we can also draw the contours of kinds of activities. Let me step back for a moment from the characterization of moral language, and give an example in which we would naturally want to encourage contour drawing. Imagine your friend tells you of a strange facial expression an acquaintance of her made in response to something she said. You ask her to describe the expression, and in response she gives you a geometrical description. If you didn't think she was joking, you might naturally tell her: ‘Look, I don't want to know the geometry of the expression, I asked you to describe the expression’. This would be a sort of grammatical remark. It is meant to function methodologically – to foster sensitivity to a sort of activity: that of describing facial expressions. We express happiness or malice or whatever, not by bending our mouths in such and such angles etc., but, e.g., by smiling, wincing, snarling, frowning, or gapping.

To emphasize, I am not now interested in the contrast between your friend's geometrical description and the kind of description you asked for. I am interested in the contrast between erecting barriers and drawing contours; the difference between the natural and the geometrical description is not such a contrast. There is nothing in your friend's description, which should lead one to believe she was trying to erect barriers to some activity. I am interested in the contrast between your grammatical remark, which was uttered in response to the geometrical description – the remark that contrasts the geometrical description with an ordinary description of a facial expression, and thus helps to draw the contours of both kinds of descriptions – I am interested in the contrast between that and a possible barrier – erecting remark, like the claim that facial expressions could not be studied geometrically.

If someone were to claim that facial expressions cannot be studied geometrically, she might thereby define that she will not consider geometrical descriptions – whatever they may involve – as descriptions of facial expressions. In philosophy, however, we sometimes try to go farther. We look for absolute generalizations. We want to claim not merely that an action of a certain kind (e.g. geometrical description) will not be considered as manifesting a certain activity (e.g. description of facial expressions), but to argue that certain activities are logical impossibilities – for instance, a geometrical description of facial expressions, or the codification of moral thought and conduct. That would be an attempt to draw a barrier – to define and fence something out. Making such claims may commit us, even if we do not intend it, to the confused idea that this fenced out logical impossibility is at least conceivable. After all, we want to claim something about it: to say it is impossible. It would seem there is some thing we have in mind: an impossible practice. As we've seen, though, in attempting to characterize the practice by drawing its contours, we find that we have not fenced anything out. That is, we have no activity in view, whose contours we may wish to draw, and thus that we have not really erected any fence. The attempt to define, to erect a barrier, has not fully materialized as it were.

My claim is not that we cannot have theses, barriers, or definitions in philosophy. The claim is that such attempts are often confused. Barriers have two sides. And so, erecting one might not be a useful – might actually be a confusing – philosophical method.Footnote ¹⁵ Alternatively, in drawing the contours of a practice we may invite a kind of attention to the practice as we know it, attempt to give a true sense of its reality, and foster a realistic view of its outline and shape. In ethics, for example, we may mention the fact that we do not normally appeal to guidebooks, or experts, to tell if something we did was just, brave, considerate, or cruel. We may say that we do not, normally, codify moral practice, and that would give a sense of what thinking and acting morally is like.

The characterization I gave to the philosophical activity of drawing contours is thus far partial. What contours are is clarified by how contours are drawn in different cases, and section VI below distinguishes between two kinds of contour drawing. In the next section I would like to deepen the distinction between contours and barriers.

IV

Contours and barriers are internal to different philosophical methods. Let me examine the distinction between the two kinds of methods.

I call remarks that are designed to educe sensitivity to different dimensions of our activities and their reality ‘helpers’. Consider these two sentences: 1) ‘A rule is something that is followed’, 2) ‘Unlike cubes, spheres do not have sides’. Think of situations in which you would use them. Sometimes they can be appropriate responses to philosophical claims. (Claim: ‘Moral rules are algorithms for people's behavior’. Response: ‘A rule is not something with which you calculate behavior, but something you follow’.) They can also be responses to grammatical confusions. (Confusion: ‘Can you please count the sides of that ball?’ Response: ‘What are you talking about? Balls don't have sides.’) This would be to use them as helpers, as if to help someone regain linguistic balance. However, such sentences can be used differently: as theses – theses that define what rules and balls are.

Theses and helpers are thus two kinds of philosophical uses of sentences. To characterize the difference, I would first like to make some distinctions. Later, I shall argue that we cannot identify theses or helpers by the words they utilize, and that the very same sentence can be used as a thesis or as a helper.

First, then, the point of using sentences as helpers is to help draw the contours of activities. They lead our gaze toward dimensions of a practice, but are not meant themselves to arrest our gaze. Theses are intended not just to characterize, but to define – often they formulate conditions: necessary, sufficient, or both. They thus fix our gaze on certain dimensions of a practice, and are designed to arrest it. Using a sentence as a thesis involves fencing something out: When one states in philosophy that something is necessarily this way or the other, and is thereby putting forward a thesis, one implies that there is a way in which things cannot be, or that if something is different, it is excluded by the definition. Theses might sometimes be involved in drawing contours and in giving a sense of the reality of some practice, but it is not their function, and there are philosophical theses, as we've seen, that do not exclude anything – theses that have not materialized completely as theses.

Second, used as theses, there will normally be a question about sentences' truth: We can talk about counterevidence against them, and it is at least intelligible to say they are erroneous. Alternatively, used as helpers, there will normally not be a question regarding their truth, but about their usefulness. If, for instance, one came up with a reason or a purpose for talking about spheres as having sides, this would not be counterevidence against the thought that spheres do not have sides, but a reevaluation of its usefulness.Footnote ¹⁶

Third, used as a thesis, different people might understand the same sentence differently. A moral realist and a moral anti-realist, for example, may both hold that there is truth in ethics. For both it is a true thesis. They thus make the same sort of use of the idea about truth in ethics. At the same time, the idea means something different for the two kinds of philosophers: While the realist may think it indicates the existence of an irreducible moral reality, the anti-realist may hold no such reality exists. Theses can thus support different theories. In contrast, sentences can be used as helpers only in so far as people do not in fact understand them differently – that is, in so far as they have the power to elicit a true sense of the reality of some practice.

‘Theses’ and ‘helpers’ are classifications by use. It is not the words, but what one does with them that matters. This first means that we cannot tell whether a sentence is a thesis or a helper by the words included in it, or by its syntax. We have no alternative but to examine how it is used. The word ‘necessary’ does not prove that a sentence is a thesis, and neither does a conditional syntax – if-then. Even the words ‘logically impossible’ do not prove that a certain sentence is a thesis. Not every claim that some activity is a logical impossibility is confused. The claim that we cannot do something might be an attempt to draw contours, rather then erect boundaries. To settle that, one must clarify its use: to look at what is done with it.

Second, to say that ‘theses’ and ‘helpers’ are categories of use means that often the very same philosophical text can be used in those two different ways: not just mean something different (as when realists and anti-realists support different theories on the idea that there is moral truth), but be meant differently.Footnote ¹⁷

What I suggest is exemplified by an examination of Wittgenstein's remarks about language-games: ‘If you follow other rules than those of chess you are playing another game; and if you follow grammatical rules other than such-and-such ones, that does not mean you say something wrong, no, you are speaking of something else’.Footnote ¹⁸ Such remarks have lead commentators to the idea that, according to Wittgenstein, language is a many-faceted rule-governed activity or set of activities.Footnote ¹⁹ This is taken as a definition of language, a thesis about what sort of thing a linguistic activity – language game – is in general. Here is another ‘Wittgenstein’ remark about how games can help us when we investigate language:

The activities Wittgenstein mentions here are unlike chess in various ways. They do not have sharp contours. It is not always clear when one game stops and another begins, when a game has come into being, or when it has been altered. The games he considers are also not all rule governed. Not all involve rules, and those that do, do not all involve rules in the same way.

Rather than giving a theoretical characterization of language, then, Wittgenstein's remarks – both of them – can, and are in fact designed to, call attention to and help us to see the reality of differences between linguistic activities. The chess remark helps to see the reality of some games and some linguistic activities. It implies that chess is largely rule-governed (this is a characterization of chess – of its contours) and that some grammatical remarks are rules (this characterizes some grammatical remarks – their contours). But the chess example was not meant to be generalized over the whole of language. In other words, such a generalization would arrest our gaze, and would not allow us to discern other dimensions of our linguistic behaviors. The ball examples help us to see and characterize some of those other dimensions.

‘There is not a philosophical method, though there are indeed methods, like different therapies’.Footnote ²¹ This, I believe, is a Wittgensteinian philosophical method: to demonstrate that one can use theoretical remarks otherwise. One can, that is, argue against philosophical theses not by denying them, finding counterexamples, contradictions, or contrasting theses. One can use them differently: as helpers.

In particular, we can use McDowell's quietist advice, to investigate ethics from the inside out, (and I suspect it is intended) not as a thesis, but as a helper. To go back to the facial-expression example from the previous section, we could use the quietist advice to say that your friend's geometrical description of that facial expression was given from ‘sideways-on’, as if she was foreign to the activity of describing facial expressions, and to say that your remark tries to lead her into inspecting this activity from ‘within’. The point of the quietist advice, understood this way, is not to reject a kind of descriptions, (e.g. geometrical descriptions of facial expressions), but to help us to attend to the grammar of certain descriptions. Since it helps to examine the contours of different kinds of descriptions, it is also a useful tool for diagnosing cases in which what looks like a description is really a confused use of language.

I should, in conclusion of this part, also point out that it is perfectly possible to use Lovibond's apparently theoretical claims as helpers. We may for example take Lovibond's claims about that substratum of tacit likemindedness that is found in moral communication, not as claims about explicitness that moral communication is logically prevented from attaining, but as reminding us of how moral communication looks like, and of the kind of attunement it involves. Insofar as she does indeed use her seemingly theoretical claims in such a way, she is drawing contours, not erecting barriers.

V

The discussion thus far raises a worry about a certain type of attempts to state theses and erect barriers in philosophy. I suggested that we can alternatively attempt to clarify the contours of our linguistic activities, often with the use of remarks as what I called helpers. However, one might think that this implies that we can only investigate where the limits of language are, but not where they should be. So the question is this: Do my claims thus far pull in the direction of linguistic conservatism: namely, towards the idea that language is a given, and has a fixed shape which our linguistic investigations cannot affect?

In what follows I argue that it does not. We can investigate language with a view to improving it. This is the point and essence of some linguistic investigations, and in this sense, it is possible to change the limits of sense. I'll compare two sorts of attempts to do so. This section examines the suggestion that the boundaries of language are not all-encompassing, but exclude some sense – which can be included in the realm of meaning by a redefinition of the boundaries of language. Here I use Richard Rorty's discussion in his Contingency, Irony and Solidarity.Footnote ²² Section VI examines the idea that when we characterize linguistic activities we might be interested not in their established contours, but in the contours of their potentiality. Here I use Sarah Bachelard's discussion of euthanasia.Footnote ²³

Rorty writes with a view to improve on the language we have. We make transitions between vocabularies, he argues, each involves passing through an interregnum during which we witness a contest ‘between an entrenched vocabulary which has become a nuisance and a half-formed new vocabulary which vaguely promises great things’.Footnote ²⁴ During those interregnums, Rorty argues, we are invited to try new ways of making sense, tempted to ask new questions, and encouraged to leave old difficulties behind, until new patterns of thought and behavior are formed and settle as the new regime.Footnote ²⁵

There is thus room for Rorty for a kind of discussion of language that does not participate in already given forms of thought and talk, but attempts to create them. For example, metaphors, according to Rorty, are like slaps and kisses: they are at bottom semantically meaningless sounds we emit or signs we create on paper, which have effects on our audience. Nevertheless, metaphors might be involved in changes of vocabulary. A metaphor might, he writes, ‘become a truth-value candidate […] it will gradually acquire a habitual use, a familiar place in the language game. It will thereby have ceased to be a metaphor – or, if you like, it will have become what most sentences of our language are, a dead metaphor. It will be just one more, literally true or literally false, sentence of the language’.Footnote ²⁶

The process of vocabulary change, Rorty thinks, is not aimed at discovery of some independent truth, which the new vocabulary is better suited to capture. He insists that such investigations achieve their goal ‘not by inquiry but by imagination’.Footnote ²⁷ He thus contrasts the kind of investigations he favors – investigations that involve imaginative creation and invention, the investigations of ‘the poet’, as he calls them – with what he calls discovery or inquiry of truth, which characterize the investigations of ‘the physicist’.Footnote ²⁸ The poet, Rorty says, ‘is typically unable to make clear exactly what it is that he wants to do before developing the language in which he succeeds in doing it. His new vocabulary’ Rorty continues, ‘makes possible, for the first time, a formulation of its own purpose. It is a tool for doing something which could not have been envisaged prior to the development of a particular set of descriptions, those which it itself helps to provide’.Footnote ²⁹

The Rortian physicist is interested in the actual boundaries of our language games. The poet is interested in its potential boundaries. To investigate language in an attempt to improve on it, Rorty holds, requires that we leave behind the vocabulary we have, and cross a barrier of sense. There are meanings on the far side. The poet can sense them, but she cannot completely grasp them from where she is. Metaphors are the poet's vehicle to an indirection she must make.

To examine Rorty's view, I would like first to recall my claim above that the idea that a necessary indirection is present in thought may naturally involve the notion that there are some thoughts we logically cannot think, or some things we logically cannot do, and may thus be confused. Rorty's claims seem to structurally resemble Lovibond's, and his views seem to be confused in much the same way: they seem to involve an attempt to fence something out, state a thesis, and erect boundaries. Yet, this attempt seems not to have completely materialized as such. I do not wish to dwell on this point. I do wish to note that, as in Lovibond's case, despite the confusion, there is no reason to assume that Rorty's theoretical arguments cannot be used as helpers with which to draw the contours of some linguistic activities.

The difference between Rorty and Lovibond is in the kind of boundaries they erect or discover. For Rorty, the bounds of sense are not all-encompassing. They only circumscribe what we can do now. What is now fenced out of sense is not fatally non-sense. It may be brought inside the bounds of sense in a new language game. To examine this conception of the nature of linguistic boundaries, let me contrast Rorty's views with those of Sarah Bachelard.

VI

Bachelard, like Rorty, attempts to improve on the language we have. However, she does not think that in order to do that, we must cross a barrier of sense.Footnote ³⁰ Shaping language, to fit our changing needs for example, might involve drawing a particular kind of linguistic contours – not of established language usage, but of its potentiality.

Bachelard argues that in the discussion of euthanasia we sometimes use a concept of a human being, which deprives us of the possibility of making sense of certain experiences. This happens when we draw a parallel between humans and animals, and this sometimes happens when we think of the terminally ill as creatures that can be put out of their misery. Bachelard thinks it is wrong to draw this parallel, because it serves to conceal our ‘capacity to make or to fail to make something of our suffering in our lives’ and thus makes us ignore ‘the requirements of a proper responsiveness to our sufferings’.Footnote ³¹ For Bachelard, those capacities and requirements go to characterize our concept of a human being. ‘[T]o act in the spirit of putting a human being out of his misery is radically demeaning. It is to treat a human being as someone incapable of responding to the claims of the meaning of his suffering, and hence as less than fully our fellow, as less than fully human’.Footnote ³² To justify this claim Bachelard does not examine what human beings are, and whether it would or would not make sense to say of them that they can be put out of their misery. She rather asks: ‘is there an argument for thinking that the concept of a human being as a being who can be put out of his misery is the ‘right’ concept of a human being to have?'Footnote ³³

As I understand Bachelard, people are in fact asking that very question when, in the face of terminal illness and great suffering, they find themselves raising questions like ‘What if she were a dog? – What would I do then?’Footnote ³⁴ They connect animals to humans, and thus make sense of such suffering. The connection does not force them to accept a way of thinking about humans, for although it may make them think about the similarity between animals and humans, it may equally make them think about the difference.Footnote ³⁵ The important point for my purposes, though, is that by exploring those different possible connections – their naturalness or rightness – Bachelard is guided by her sense of what it is to be human. That is, she is guided by her sense of the contours of this concept. This goes to characterize the sort of investigation she conducts, as well as the sort of interest she has in language, which is neither wholly conservative nor completely revolutionary.

Bachelard's investigation exemplifies a kind of ‘non-theoretical’ interest in language. She is not attempting to erect new linguistic boundaries, or move existing ones. She is interested in the contours of language. At the same time, she wants to shape language, and so she is interested in a special kind of contours. We already saw that exploring the contours of linguistic practices may involve using helpers to point out their established reality – saying, for example, that we do not talk of spheres as having sides, or that we express emotions by making facial expressions. But sometimes, when exploring the contours of concepts and linguistic activities, we may not be interested only in their established reality. Sometimes, there is no single established way of using a concept – as when we ask what would be to treat a terminally ill person as fully human. In such cases, we may note, some ways of using the concept are more natural than others. That would not be a helper that helps to characterize an actual reality, but that helps to examine that reality and perhaps create a new one. And this indicates the existence of a different kind of helpers: helpers that explore the contours of the potentiality of concepts. Bachelard, I suggest, is using a helper of this special kind when she talks about the ‘right’ use of a concept. She is, as it were, exploring a linguistic horizon.

To contrast Bachelard's investigation with Rorty's, let us ask: What if we read Bachelard's question with Rortian eyes? If we did that, it would seem that in asking her question, Bachelard is adopting a point of view outside our – outside her – concept of a human being in an attempt to contrast it with a competing conception. Bachelard would seem to have a choice: either she is a ‘poet’ (in Rorty's sense) in the process of creating new boundaries to the concept of a human being, which would help us to make sense of certain experiences that we currently cannot make sense of, or she is asking a question within the bounds of a concept we already have. This Rortian reading (mis)identifies Bachelard's contour drawing as one kind or another of erection of barriers of sense – either as of the poet or of the physicist kind. What she cannot do, according to this view, is assume that we have a concept of a human being which does not allow us to make sense of certain experiences, and at the same time think that by investigating its boundaries she can come to a conclusion regarding how those experiences could or should be made sense of. Yet, this is exactly what she does.

The Rortian view assumes a theoretical interest in language, and accepts various theses about it: notably, that language games have immobile boundaries and cannot be changed but only replaced. Now, we do not have to accept Rorty's theory. We might suggest a contrasting one, in which the boundaries of sense are mobile and flexible, and language changes and develops shrinks and expands constantly, e.g. in response to fluctuating needs.Footnote ³⁶ I am here less interested in the differences between these two kinds of theories, and more in their similarities: Both share the idea of characterizing language as clarifying what in principle can and cannot be done in language, both are putting forward theses about that, and both are looking to identify the nature of linguistic boundaries.

My claim, and what I think Bachelard's investigation demonstrates, is that we can, alternatively, have an altogether different, ‘non-theoretical’, interest in language even when we attempt to shape it: an interest in drawing the contours of the potentiality of linguistic activities. Bachelard asks about the ‘rightness’ of a concept. Beyond ‘rightness’, we can and sometimes do ask about the natural, intelligent, interesting, funny, useful, needed, and even compelling, extensions of the use of concepts. Investigating all that would be investigating the contours of concepts by exploring their horizons.

In fact, we can use both theoretical ideas I mentioned above – that of linguistic activities as having immobile, and that of linguistic activities as having mobile boundaries – as helpers to shed light with on different dimensions of the horizons of language and of how it changes in different cases: how it takes new shapes, develops, degenerates, is replaced, and so on, in the face of formulations of new scientific paradigms, discoveries, technological inventions, political transformations, developments of new styles of art, and so on.

VII

In conclusion, I would like to ask: what is the value of examining the contours of language – both actual and potential? The discussion thus far shows, I think, that examining all this allows us to examine what Cora Diamond called ‘the good of concepts’.Footnote ³⁷ I do not mean the goods that we can instrumentally gain by using some concepts, but the good intrinsic to concepts, which cannot be understood without them: the good of making sense of the reality we live in-the political situation in which we find ourselves, the fact that we were born to certain parents, that we fall in love, that we are going to die. In other words, drawing the contours of language allows us to explore the ways we care about how we make sense – think and act.

We care about how we think. Patterns of talk reflect patterns of care. Investigations of linguistic barriers are partly meant to reveal that, but they tend to conceal some of those patterns. For example, claiming that facial expressions cannot be studied geometrically (or alternatively that they must), unnecessarily limits the ways we are interested in facial expressions. Similarly, claiming like Lovibond that we cannot codify moral practices (or adopting another theory, according to which we must), unnecessarily limits the ways we are interested in moral practices. It thus misses the fact that we do not always care about things, and make sense of them, in the same ways.

Similar unnecessary limits might also be placed on how language can change. Part of Rorty's motivation in investigating vocabulary changes is to show that we might not always care about things in the same ways: vocabulary changes signify transformations in our patterns of interest. He himself wants to bring about such a change. However, his interest in linguistic barriers leads him to make a sharp distinction between invention and discovery, which undermines his ability to distinguish between kinds of inventions and kinds of discoveries. He allows no room for the idea of re-inventing our vocabulary, or re-discovering it, or of claiming traditional vocabularies as our own. In other words, Rorty uses the idea that patterns of talk reflect patterns of care as a thesis, which informs his sharp distinction between two sorts of ways of caring about how we think – conservative and revolutionary. Either we keep our concepts, he holds, or we throw them away and get new ones. Being interested in our established linguistic patterns, he holds, must be different from being interested in improving on them. You are either a poet or a physicist. You cannot be both. It thus becomes very hard for him to see how patterns of care, and how changes in those patterns, are actually reflected in language.

Let us look at some of the relevant data. Our care about moral practice is reflected, for instance, in our seldom codifying it, and in that we seldom require scientific kinds of accuracy from descriptions of moral phenomena. Our care is also reflected in our interest in how others think. What it is to care about our concepts, about the possibilities of making sense, the contours of this, will be clarified in part by an examination of different cases in which we appeal to foreign practices to shed light on our own: Political-theologians, for instance, employ old theological ideas to explain modern politics. Christians today identify Buddhists as having religious thoughts. Science fiction novelists warn us about the future possibility of losing touch with ways of making sense. Philosophers promote views by trying to persuade us to adopt new distinctions.

Some, but not all, differences between ways of making sense indicate cultural distances. Others are differences between ways in which our thoughts might be brought to bear on each other. We make connections between ideas, and we sometimes have a choice: we can connect ideas in different ways. Sometimes we are forced to make such connections, for instance when we feel that our established ways of making sense do not equip us with the means of making sense of certain experiences, or that they equip us with contrasting ways. Bachelard, as we saw, is investigating such a case. By exploring different possible connections between the concept of a human being and the concept of an animal, she enables an investigation of the former – the good of that concept, and the patterns of care it can, and can be made to, express.

We can thus employ the idea that patterns of talk are tokens of patterns of care not as a thesis, defining for us what language is or what it can become. We can use it as a helper with which to draw attention to the contours of linguistic activities. For example, we can use it to help someone stay sensitive to the significance of slight differences in wording; or we can, like Bachelard, look at different linguistic patterns to investigate what we care about. Saying, with Diamond, that concepts are goods can be used as a tool for looking at the contours of our life with words, and of the different types of interests we might have in how we think.

One last remark. Bachelard's investigation shows that concepts may allow us to bridge distances between what we experience and what we can express. This is part of their good. However, this indicates our vulnerability to situations in which language does not keep up with experience. We may sense in such cases that language puts a barrier in front of us. This, however, does not entail that the very nature of language restricts thought. That is, it does not indicate the existence of necessary linguistic barriers. Rather, different contingent occurrences, including the interventions of other people and political regimes, might erect contingent barriers that interfere with the language we are taught or allowed to have. Blindness to this is another price that inattention to the good of concepts – to the contours of their actuality and of their potentiality – may have to pay.Footnote ³⁸

VIII

It might seem that the upshot of my arguments is that no sense and no use can be made of the idea of a necessary barrier logically restricting what language can grasp. ‘The great difficulty’, says Wittgenstein, ‘is not to represent matters as if there were something one could not do’.Footnote ³⁹ But can we make sense of the idea of a logically necessary linguistic barrier? Can we do more with the idea of a limit? I leave that for another occasion.Footnote ⁴⁰