Andrew Bowie
What do philosophers think is meant if, for example, somebody says: “I don’t understand Brahms”? Approaches to such an assertion can range from attempts to specify how semantic content can be conferred on a string of noises or marks, to investigations of the historical and cultural power relations that might influence and inform a statement. In the former case, the assumption is that there is something which can be identically conveyed by another string of noises than the original string, and establishing what this is constitutes the prime concern of the “philosopher of language.” In the latter case, such an assumption seems to miss the point, insofar as the content of the statement could make manifest a perspective which connects Brahms to the history of music, psychology, German history, gender relations, and so on. The former approach has the apparent advantage of seeking something circumscribed to analyze; the latter seems to have the disadvantage of leading almost anywhere. The latter seems, moreover, to depend on some answer of the kind offered by the former: surely, unless we grasp the literal meaning of the terms, we cannot even begin to understand what is at stake in the latter? Michael Dummett therefore maintains that a “full-blooded theory of meaning” “must give an explicit account, not only of what anyone must know in order to know the meaning of any given expression, but of what constitutes having such knowledge” (Dummett 1993, 22). But how does one get from this contested kind of understanding to what the utterance may concretely mean if stated, say, by Richard Wagner? What, for example, does one mean by “Brahms”? Importantly, it is not that one cannot give an answer to this: on the contrary, the answers can arguably go on indefinitely, and they can all be true, depending on the context of the answer.
At this juncture, the question becomes the very point of a theory of meaning of the kind suggested by Dummett, if the knowledge it seems to require is indeterminately large. Such a theory would not enable us to engage with issues that we already acknowledge as being germane to our cultural, social, and political life, because one never gets past the first base of specifying in philosophical terms what is being talked about. Is the “knowledge” in question “knowing that” or “knowing how,” and can this distinction be made in a definitive manner, given that the “knowledge” must clearly be “tacit” for most everyday speakers of a language, as opposed to philosophers with what would have to be some kind of privileged theoretical insight (see Gascoigne and Thornton 2013)? Lee Braver sums up the underlying issue here in an account of Wittgenstein: “The trouble with such words as “understanding” comes through thinking of a few cases and trying to carry over their analogy to all other cases. For example, conscious mental acts do play a great role in understanding, but we should not try to make every case of understanding look like these cases” (Braver 2013, 23). The same can be applied to the word “meaning”: what dictates that it refers to certain kinds of verbal expression, when the very capacity of such expressions to be expressions at all is what is in question? Dummett follows precisely the model Braver finds problematic, and the implications of the resulting difficulties go beyond the particular case of Dummett’s theory.
What I am sketching here is a version of the kind of dispute which goes on between the “analytical” and “European” traditions of philosophy (see, e.g., Braver 2007; Taylor 1985; Wellmer 2004), and what is at issue can be suggested as follows. While people have a remarkable facility for understanding often hugely complex forms of communication and interaction in everyday cultural and social contexts, from the structures of symphonic works to what their partner means when they say “I don’t understand Brahms,” philosophical analysis seeks to isolate one form of understanding as if it were the key to all others. Having done this, however, such philosophy signally fails to come up with agreed accounts of that form of understanding. This means that the superior forms of insight regarded by some as the preserve of philosophy here actually lead to an impoverished conception of the domain of meaning, which is generally inferior to that articulated in other domains, because it fails to make more sense of the world. If the task of philosophy is to make sense of things, such philosophy seems to fail at the most basic level, being unable to account for the sense we must always already be able to make if an analysis of linguistic sense is to be possible at all.
Much of the project of which Dummett is part, which relies on the existence of meanings as “objective representations” or “senses,” can be seen as based on a misapprehension of key aspects of the nature of meaning, language, and understanding, whatever else it may contribute to a philosophical account of language. This is not to question the advances in logic made by the analytical tradition in the wake of Frege, but rather to question the scope of such approaches with respect to what actually happens when human beings understand or fail to understand each other and the world. Heidegger sees what is at issue here in terms of how we understand the ground of logic, rather than seeing logic as the ultimate ground. My claim will be that understanding is itself better understood if it is seen as an art, rather than as something that is to be exhaustively explained by a philosophical theory.
In this context, maintaining that understanding is a kind of art does not commit one to defining the meaning of “art.” Defining art actually gets in the way of understanding the key point here by seeking to establish the scope of a classifying term that, as the history of modern art should teach us, is constitutively resistant to being defined in such a way. The reiterated question in relation to any new artifact in the art world of “Is it art?” admittedly seems to demand an answer based on the scope of a concept. However, as Heidegger suggests, this misses the essential point about art, which is that it is the occurrence of something which is significant precisely because it is not covered by the scope of a classifying term and can itself change the scope of such terms.
F. D. E. Schleiermacher’s remark that “we call art … every compound product in which we are aware of general rules, whose application cannot in the particular case be again brought under rules” (in Rössler 1990, 232–233) helps initiate a tradition of philosophical approaches to language which offer a more productive way of seeing how understanding relates to art, and so inherently resists attempts to objectify it by reducing it to being the result of the application of rules. This resistance should not be seen as meaning that understanding is therefore something wholly mysterious. The very idea that language is mysterious only makes sense if language in some other way is not mysterious, otherwise we would not understand what it is for something to be mysterious at all. It is a fact of our being that we understand most of the time, even though we may be only aware of a tiny part of what could be understood. Given that we could not even ask questions about how to explain understanding if it were not already unquestionably part of what we are, there must be domains of prior sense on which any attempt to explain understanding always relies.
One way in which these domains became apparent in modern philosophy is via an issue in Kantian philosophy which relates to questions of regress in grounding knowledge that are connected to the skeptical tradition. David Bell has pointed out that the analytical tradition has rarely had much time for the “Schematism Chapter” of the first Critique (Bell 1987). In contrast to the analytical tradition, Heidegger claims that this chapter leads to the “core of the whole problematic of the Critique of Pure Reason” (Heidegger 1973, 109). Vital questions in modern hermeneutics can be shown to emerge from the issues raised here by Kant. The problem which schematism is supposed to answer has in it all the ingredients of the core problems for a “full-blooded” theory of meaning, and Kant’s attempts to overcome this problem have been repeated in various forms in the subsequent history of philosophy (see, for example, Bowie 1996).
Kant’s aim is to explain the subsuming of intuitions under concepts. The latter must remain independent of the vicissitudes of the empirical world for knowledge to be possible at all, whereas the former are endlessly diverse, because we never receive exactly the same intuitions at any two moments of our lives. For Kant, mediation between the spontaneity of the understanding and the reception of intuitions is the job of the “imagination,” whose location, as Heidegger shows, moves from the side of spontaneity, in the first edition, to the side of receptivity in the second. What is required is something, the schema, which can establish identities even though what it identifies is, qua what is given to receptivity, not strictly the same at all.
Judgment for Kant consists in the subsuming of intuitions under concepts, as rules for identifying what is given in receptivity. Kant, though, suggests there is a problem in any philosophical account of judgment, and thereby suggests a crucial link to art:
If judgement wanted to show universally how one is to subsume under these rules, i.e. distinguish whether something belongs under the rule or not, this could only happen via a further rule. But because this is a rule it requires once more an instruction by judgement, and thus it is shown to be the case that the understanding is admittedly capable of being instructed and equipped by rules, but that judgement is a particular talent which cannot be given by instruction but can only be practiced.
(Kant, Krv, B, 172, A, 133, my translation)
This “talent” is required to avoid the regress which results from the need for rules for the application of rules. What is at issue is the ability to grasp immediately what something is to be “seen as,” an ability which would become inexplicable if cognition were purely rule-based. Schleiermacher talks of the schema as a “shiftable” image (Schleiermacher 1988, 145). As Schelling would later point out, language functions by a related creation of identity from difference (Schelling 1856–61 I/3, 509). The ramifications of this methodological point for questions of meaning and understanding are considerable, as Schleiermacher will make clear.
The basic problem here has been central to recent developments in semantics. In Making It Explicit, a book whose title echoes a central term, “auslegen,” meaning “explicate,” in hermeneutics, Robert Brandom refers, in relation to the question of interpretation, to “Wittgenstein’s Regress Argument.” The argument shows that a “rule says how to do one thing correctly only on the assumption that one can do something else correctly, namely apply the rule” (Brandom 1994, 21). Brandom sees the argument as the “master argument for the appropriateness of the pragmatist, rather than the regulist-intellectualist, order of explanation” in semantics (ibid., 23). The argument prevents a regress of rules for the application of rules, by grounding rules in pre-theoretically constituted practices that are inherent in our “being in the world.” Brandom wants to use his approach to establish a pragmatist version of analytical philosophy. However, he thereby excludes some of what can be learned about language from the hermeneutic tradition’s relation to aesthetic issues (see Bowie 2007).
Language consists of noises or other perceptible articulatory marks, of the kind which are also present in the world in a way which is not linguistic. In order to become language, noises and marks have to be in a manner in which non-linguistic things are not. Heidegger puts it like this: “Words accrue to meanings. But word-things are not furnished with meanings” (Heidegger 1967, 161). For a noise or mark to be a word, meanings in the world must come to be articulated via the noise or mark: these thereby cease to be objects describable in terms of the physical sciences, not least because they are themselves part of what makes such sciences possible. Albrecht Wellmer suggests the link to art when he says that, as an object of scientific investigation, the work of art “loses all the qualities that make it a work of art” (Wellmer 2009, 214). Without the object that is investigated by the sciences, the work could not exist, but the object is not what constitutes the work: that depends on its relations to a world. Something analogous applies to language: only if it renders aspects of the world intelligible is the thing which makes meaning possible linguistic. In turn, the thing which is a work of art is only a work if it discloses a world in which things make sense.
An example of what is at issue here is the fact that the incorporation of more and more kinds of sound, from ever greater dissonances to everyday noise, has been a key characteristic of the development of modern music. What makes no sense beyond its manifestation as objective phenomenon in the everyday world can come to make new sense in the context of musical practice. In the terms of Heidegger’s essay “Origin of the Work of Art,” qua “earth” the word or the sound in music is a thing that closes off sense, while at the same time being necessary for sense to emerge, and it is only as what makes “world” manifest that it is a word or musical sound. The tension between earth and world is the space in which sense is made in terms of what Heidegger characterizes as “unconcealment”: the essential nature of being is to be hidden, so the primary nature of language lies in its capacity to reveal, to bring earth to unconcealment as world (see Wrathall 2012).
The issue here is the scope of what can be considered to have sense, which is where the divide between the hermeneutic tradition and the analytical tradition is most evident. The aspect of language which is the object of semantics may be characterized in terms of the ability of assertions to pick out aspects of things, and by Brandom’s idea that meaning entails inferential commitments, such that by claiming “x is copper,” for example, I can fulfill my commitment to the claim that something is copper by knowing it is a metal, not iron, platinum, etc. However, such conferring of semantic content is parasitic on the ways in which things already make sense in a world. In contrast to hermeneutics, semantic approaches rarely regard poetic usage as germane to how linguistic sense is made. Wittgenstein refers in the context of literary and poetic usage to “these words in these positions,” meaning that the replacement of the specific words in those positions would lose the sense of the poem, even if the new version were semantically equivalent to what the poem says. Cashing out this sense in rule-bound terms is impossible, because the positioning of the words is significant precisely because it is specific to the particular poem. Similarly, an inferentialist approach to the content of the poem could not articulate the sense which arises from the formal arrangement of the words. Even though claims can be made about it, such sense is not exhausted by what can be justifiably claimed: if it were, the poem could be replaced by its paraphrase. Both language and art are, as objects in the world, as “entities” in Heidegger’s sense, open to analysis in terms of a whole variety of disciplines, from psychology, to linguistics, to anthropology, etc. The fact that they convey sense at all, and the fact that they demand investigation, cannot, though, be fully explained in terms of the cumulative knowledge provided by such disciplines.
Wellmer (2009) points out that in language the empirical manifestation of a sign can remain essentially identical, at the same time as the meanings expressed by the sign continually change. This fact is most evident precisely in the case of texts which are understood to be art. Explaining such changes in terms of what has already happened, in terms of concepts as rules for identifying and explaining recurrent phenomena, is impossible, because it is precisely the emergence of new sense via the reconfiguring of preexisting entities that is the decisive point. Meaning is constituted holistically, emerging from interrelations between the ways the world is manifest, which change how iterable signs signify. This idea is better understood in terms of how the parts of a work of art signify in relation to the whole and to the contexts of the work than by the attempt to isolate the meaning of discrete utterances.
These ideas put the notion of philosophical analysis as a grounded enterprise in question. If one has not always already understood, how could what one is doing in understanding be grounded by an explanation of understanding? Only if one gives priority to that which is rule-bound, as though everything about understanding is in principle like a phenomenon to be explained in a science, could explanation, as certain versions of semantics require, take priority. Just how problematic this approach is will be apparent in a moment.
In the German Romantic tradition to which Schleiermacher belongs, the most significant factor in language is precisely its capacity to disclose the world in new ways. This idea is linked to changes in the status of wordless music, which, because it is the least representational form of art, becomes the most significant form of art for some theorists (see Bowie 1997; 2007). The rise of “absolute” wordless music, and the associated move of the new subject of aesthetics away from theories of mimesis toward theories of art as what renders things graspable in new ways, opens the way to conceptions of language which do not see the prior function of language as the representational fixing of a world of ready-made entities, and which insist that the “world-disclosing,” “unconcealing” aspect of language is both ontologically and logically prior to its semantically determinable representational aspects (see Taylor 1985).
The tension here between conceptions of language as “world-disclosive” and conceptions of formal semantics is apparent, for example, in the very different ways in which Wittgenstein is understood in hermeneutic and analytical approaches. Following his remarks on poetry, Wittgenstein claims, for example, that “Understanding a sentence in language is much more related to understanding a theme in music than one thinks” (Wittgenstein 1971, 227. See Bowie 2007). By rejecting a strict division between verbal and other sense, he points to dimensions of sense which play almost no role in analytical approaches to language.
At much the same time as Schleiermacher develops his hermeneutics, Bernard Bolzano lays the foundations for the semantic tradition. The basis of that tradition is “the separation of meaning from psychological processes,” whereby “the objective representation associated with the word “table” (i.e., the meaning of “table”) should not be confused with tables, the objects of that representation” (Coffa 1991, 30) or with the subjective representations of anyone thinking about a table. There are as many subjective representations as there are acts of thought on the part of persons thinking about something, but Bolzano’s claim is that the “objective representation designated by any word is, as long as this word is not ambiguous, single” (Bolzano 1963, 66). When Dummett asserts that “for Frege a word simply has a sense … he does not think that its bearing that sense in the mouth of a speaker depends upon his performing any mental act of endowing it with that sense” (Dummett 1991, 276), he is carrying on the tradition already established by Bolzano. The real question, though, is what role “objective representations” play in actual interpretation of concrete utterances.
In the Critique of Judgment, Kant distinguishes between “determining judgment,” judgment of particulars based on a preexisting general rule, and “reflective” judgment, establishing a general rule in relation to particulars. Reflective judgment, which is also the basis of apprehending things aesthetically in terms of the interrelations of their parts, seeks to arrive at new rules. Its function is apparent in relation to Wittgenstein’s idea of understanding “something which only these words in these positions express.” In an analogous manner, Schleiermacher distinguishes “grammatical” interpretation, in which “the person … disappears and only appears as organ of language,” from “technical/psychological interpretation” in which “language with its determining power disappears and only appears as the organ of the person” (Schleiermacher 1977, 171). Hermeneutics requires both kinds of interpretation, but “to carry out grammatical explication on its own is a mere fiction” (ibid., 164). While there is no doubt that one can identify lexical, pragmatic, syntactical, and grammatical rules in language, which form the focus of linguistics and of analytical approaches, these aspects of language are, Schleiermacher insists, “not positive means of explanation, but negative ones, because what contradicts them cannot be understood at all” (ibid., 171–172). Completed technical interpretation, which relies upon reflective judgment, is a regulative idea concerning how we seek the new sense made by individuals using established, rule-based linguistic material. Schleiermacher does not, as is often claimed, employ the psychologistic notion of “empathy” in interpretation, but he does think that understanding what an individual concretely means in a situation must also depend on an understanding of that individual and the world they inhabit, and this understanding cannot be wholly rule based.
Schleiermacher also proposes a version of the regress argument Brandom attributes to Wittgenstein, which includes his claim about interpretation as art:
The complete understanding of speech or writing is an artistic achievement and demands a doctrine (Kunstlehre) or technique to which we give the name hermeneutics. We call art … every compound product in which we are aware of general rules, whose application cannot in the particular case be again brought under rules.
(Rössler 1990, 232–233)
Interpretation “only bears the character of art because the application is not also given with the rules,” and there are no “rules … that would carry the certainty of their application within them” (Schleiermacher 1977, 81). Donald Davidson thinks much the same, employing the questionable notion of a “passing theory” for the “art” in question, maintaining that semantics in real communication inherently relies on reflective rather than determining judgment: “For there are no rules for arriving at passing theories, no rules in any strict sense, as opposed to rough maxims and methodological generalities” (Lepore 1986, 446).
The hermeneutic tradition does not, then, rely on the assumption that a theory of “meaning,” qua definable “sense” of a word, can be the foundation of an account of understanding. That the basic structures in question here remain constitutive for the hermeneutic tradition’s approach to the question of meaning is apparent in Heidegger’s 1928 lectures on Metaphysical Foundations of Logic:
Thinking and the use of rules might be unavoidable for the carrying out of all thought, thus also for the foundation of metaphysics itself, but from this it does not follow that this foundation lies in the use of rules itself. On the contrary, from this it only follows that this use of rules itself requires grounding, and it further follows from this that this apparently plausible argumentation is not at all capable of carrying out a foundation.
(Heidegger 1990, 130)
The realization of the impossibility of grounding meaning and truth by explanation in terms of rules is, then, common to the most significant hermeneutic theories.
The disintegration of any kind of unified project for a theory of meaning in analytical philosophy—and thus in some senses of any clearly characterizable notion of “analytical philosophy”—is arguably a result of the hermeneutic insights at issue here becoming more widespread. Dummett, on the other hand, insists that “the conviction that a philosophical explanation of thought can be achieved by a philosophical analysis of language, and … that a complete explanation can only be achieved in this way and in no other” defines the analytical project (Dummett 1988, 11). He claims in relation to Davidson’s equation of truth and meaning that
if we want to maintain that what we learn, as we learn the language, is, primarily, what it is for each of the sentences that we understand to be true, then we must be able, for any given sentence, to give an account of what it is to know this which does not depend upon a presumed prior understanding of the sentence; otherwise our theory of meaning is circular and explains nothing.
(Dummett 1993, 43)
The hermeneutic tradition can, in contrast, be defined by its acceptance of an inherent circularity in understanding, because there is no way of escaping the need to have already understood something before attempting to explain understanding. The analytical approach exemplified by Dummett works on the assumption that understanding is inherently cognitive. As we saw, he demands a theory which “must give an explicit account, not only of what anyone must know in order to know the meaning of any given expression, but of what constitutes having such knowledge” (Dummett 1993, op. cit., 22). However, this seems inevitably to lead to the further question of what constitutes having knowledge of such knowledge, etc., and so into another regress.
The essential divide in contemporary approaches to these issues lies between those, like Dummett, who think that a theory of meaning must also give us an account of the notion of truth without presupposing an understanding of truth, and those, like Heidegger and Davidson, who think truth must in some way be presupposed. In Heidegger’s terms, as Mark Wrathall (2012) has shown, propositional truth is dependent on prior “unconcealment,” which is exemplified by what happens in art. This approach avoids a predominantly cognitive focus, where understanding is essentially a form of propositionally articulable knowledge. The focus is instead on the idea that what happens in new understanding, when the world becomes unconcealed in new ways, is the key to understanding. Dummett, in contrast, thinks that knowledge of the meaning of a sentence is also linked to the rules for verifying its truth conditions: but this just repeats another version of the same problem. How do we decide which rules are the right rules for the verification of a particular utterance, without again falling into a regress of judgments? Dummett’s view requires either a thoroughly metaphysical assumption about the status of senses, or a final arbiter of what a sense actually is and what behavior exhibits the grasping of it. But who is this arbiter if not just another language user like the one being observed, and what shows that she really understands? The position Dummett advocates ends up having to assume a kind of God’s eye view of language which gives a metaphysical status to Fregean senses.
It is here that the hermeneutic tradition allows one to consider one of the least adequately addressed issues in modern philosophy. While the modern period is characterized by an unprecedented growth of well-warranted explanation of natural phenomena and a concomitant technological command of nature, modern philosophy since Descartes has signally failed to come up with a convincing epistemological account of that success. The fact is that the regresses that result from reliance on a conception of meaning based on an account of understanding as something essentially cognitive echo the regresses that have characterized the history of foundationalist epistemology. By assuming that our primary relation to the world is cognitive, such philosophy generates dilemmas with respect to meaning whose significance is hard to convey to those outside professional philosophy. At the same time vital cultural, social, and political issues are largely ignored.
This is not to say language and meaning are not in certain respects mysterious: the very fact that analysis demands an objectification of language in language itself suggests why. Heidegger talks in this respect of the idea that we are always “on the way to language,” rather than masters of it. The question is how we conceive of the mysteries: the analytical tradition’s objectifying assumptions lead to the same dead ends as much of the history of epistemology, where even well-warranted knowledge becomes mysterious. In contrast, the hermeneutic tradition no longer seeks a definitive philosophical perspective on meanings, in the name of an openness to new understanding, which, though it still acknowledges the role of rules, is not grounded on rules of analysis. The following historical constellation is informative here. The emergence of the modern hermeneutic concern with language and interpretation in the second half of the eighteenth century goes hand in hand with the emergence of the modern aesthetic concern with forms of sense that are not primarily cognitive and are significant precisely because they resist definitive interpretation and reveal new sense. The latter is in part a revolt against the reifying consequences of the natural sciences’ seeing nature solely as a mechanical system of laws. Analogously, modern philosophy in the analytical mode too often seeks to ape the natural sciences with respect to its approaches to language, and thereby loses sight of too many dimensions of meaning which are crucial in making sense of the world, dimensions which are best understood through what happens in art.