Andrew Kidd, University of Minnesota
KENNETH BURKE CRITICALLY RESPONDED to numerous academic trends throughout his long career and one of his earliest such intellectual engagements was with the Logical Positivist movement in philosophy, which enjoyed an influential, albeit brief, period of prominence in the 1930s. Although positivism was breathing its last gasps as a major philosophical force by the time The Grammar of Motives was published in 1945, its influence remained strong enough that Burke felt obliged to respond to the scientism which still permeated the academic hallways of the time. Burke tried to present an alternative to a model of empirical thought which he felt was excessively reductionist and overly rash in its rejection of metaphysical thought or linguistic alternatives to the positivist model of inquiry. In its place, he advocated a “critical realism” which tried to accommodate the notion of an objective reality existing independently of the individual, with an equal acknowledgment of the role language plays in our own specific interpretations of reality (Brock, 1999; Heath, 1986).
Ironically, while Burke was putting forth his logocentric alternative to empiricism of the Logical Positivists, philosophers of science themselves were also coming up with their own alternatives to a philosophy which left too many gaping holes, and came to similar conclusions with their own models of inquiry. “Scientific realism,” the most prominent of these alternatives to logical positivism, came about from philosopher’s attempts to explain how a scientific theory is able to “work” even when it is based in non-observables. Accordingly, scientific realism postulates that our theories in science work because they are literally true, and that the acceptance of theories as true hinges on how well they are able to explain and provide a generalized description of external reality through the fulfillment of the required axiomatic propositions, and not on empirical claims (Psillos, 2002; Van Fraassen, 1980).
Although one would expect the Burkean scholar to welcome this move away from the extreme empiricism of the Positivists, the rationalist alternative of scientific realism poses its own problems to a symbolic study of rhetoric. For one, scientific realism proposes that language and other forms of symbolic interaction are a form of “mental toolkit”, which enables us to provide literal representations of reality while Burkean models, as well as most schools of the rhetoric of science, necessitate that we view our symbolic models not as literal but as figurative representations that can be interpreted as true. This is also a similarity that logical postivism has with the Burkean model of rhetoric as well as other interpretative models, although the similarities end there; logical postivism insists that the use of linguistic terms or statements, including the axioms of geometry and mathematics, are only valid to the extent that they correctly correspond to an empirically verified proof.
In an attempt at reconciliation between the positivist and realist schools, Van Fraassen (1980) has tried to return the philosophy of science to empiricism, while at the same time retaining those aspects which the scientific realists had done right, beginning a new conversation in the philosophy of science which has continued for some twenty-five years now. Van Fraassen’s notion of “constructive empiricism” hinges on a notion similar to that of scientific realism, that theories are sets of models and statements which serve as a useful description of physical reality when they are shown to “work” properly. However, according to Van Fraassen, a good theory need not necessarily be true, nor is truthfulness arbitrarily assigned when a theory is found to fulfill certain axiomatic propositions. Rather, the truthfulness of a theory is confirmed if and only if it is found to fulfill the tests of empirical adequacy, but it is not necessarily “inadequate” if it does not meet these tests. Moreover, Van Fraassen rejects the rules of linguistic determinacy which both logical positivism and scientific realism insist upon; instead, any form of symbolic representation, regardless of whether or not it meets propositional rules, may constitute a valid basis for a scientific model so long as it is stands as an accurate description (Van Fraassen, 1980). Here, we find ourselves edging closer to the Burkean conception of science, with a theory’s validity dependant on well it uses language to explain - or more accurately, persuade - the audience of its truthfulness.
What we are interested in here is less an attempt to draw analogies between Burke and contemporary philosophy of science than a desire to examine what is the relevance of this modern philosophy to a Burkean model for studying the rhetoric of science. It shall be demonstrated that Van Fraassen’s model of constructive empiricism is useful in explaining how the four master tropes described by Burke in A Grammar of Motives. To undertake this task, Tietge’s application the four master tropes to a study of scientific discourse shall be used as the theoretical basis for a synthesis of Burke and Van Fraassen.
We begin by examining Tietge’s description of how Burke’s four master tropes correspond to scientific practice. According to Tietge, the general statements Burke makes about science as an extension of the human symbol-making capacity in the Grammar of Motives necessarily lends itself to a discussion of how we relate the role of the four master tropes in communicating knowledge and meaning in scientific discourse. Accordingly, the key rhetorical features of scientific expression have their corresponding equivalents in Burke’s tropes. “Reduction”, the representation of features and phenomena in mathematical form, has its counterpart in the trope of metonymy. “Perspective” has its counterpart in the trope of metaphor, as it involves the attribution of features to an object of phenomenon which it actually does not have, but is useful in providing a simplified understanding (as when physicists describe quarks as having “color”: naturally, it is impossible for them to actually have colors as we know them but the metaphor has proved highly useful in explaining how the particles interact). “Synecdoche’s” counterpart is found in the use of representation of science, the inductive use of specific examples to establish universal principles; “dialectic” has its counterpart in irony, using examples to describe how outcomes differ from predictions or expectations (Tietge, 1998).
In using Burke’s four master tropes as a means of explaining and situating scientific practice, Tietge arrives at conclusions which parallel those which Van Fraassen derived from his notion of constructive empiricism. Both Van Fraassen and Tietge describe their individual interpretations as moves away from scientific realism, although the definitions of ‘scientific realism’ used by both scholars differ slightly and should be clarified before proceeding further. The scientific realism Van Fraassen is responding to specifically maintains that scientific theories, while capable of being either true or false, are to be considered literally true when their predictive value in explaining the world existing outside the mind has been validated and is also a rationalist response to the extreme empiricism of the logical positivists who focused on the use of observed evidence and argumentative structures in theory (Van Fraassen 1980). Van Fraassen is similarly critical of logical postivism but also maintains that weaknesses exist in the precepts of scientific realism, which must be further examined as well if we are to come up with an accurate model of how science proceeds.
On the other hand, the philosophy of scientific realism which Burke described and responded to, in a conversation continued by Tietge, encompasses the key assumptions of the philosophies of both scientific realism and positivism. The Burkean reply to scientific realism, that theories and observations are as much functions of symbolic action as other forms of rhetoric are, is further extended by Tietge to describe science as an ordering of observations and predictions through the use of language in an understandable manner (Burke, 1945, Tietge, 1998). As such, scientific theories are not literal descriptions or interpretations of objective reality as scientific realism insists, nor are they mere linguistic or social constructs as some of the more extreme constructivists would have us believe. They are, instead, an attempt to place in order our observations and interpretations in a manner which is clear, coherent, and understandable to their audiences, be them other scientists or the lay audience as a whole. If they fall short of a complete or accurate description of external reality it is less a failing of the theories themselves than of the very nature of language itself, subject as it is to ambiguity and imprecision, to say nothing of individual preference and competence in the selection of tropes (Tietge 1998).
It is at this point that we are able to proceed to a synthesis of Burke’s four master tropes with Van Fraassen’s constructive empiricism. Van Fraassen opposes logical positivism on the basis that a theory’s validity need not hinge on observables, be they direct observations of phenomena or observations of their effects. In turn, he also opposes realism on epistemological grounds, arguing that claims about what is and is not observable are not to be accepted as literal representations of reality but as linguistic constructs. Theories and models are as much “inventions” as they are explanations and may be defined and defended through any conceivable application of language (Van Fraassen, 1980). This is in part the same approach to science which Heath identifies in Burke’s approach: viewing theories as perspectives on reality instead of representations of reality itself and holding that language is as real as the theories it defines (Heath, 1986).
There is, however, one area where Burke and Van Fraassen are very much at odds and it is at this point that the four master tropes come into play. Van Fraassen holds that scientific theories and models are still primarily mental constructions and, as such, can be defined through any language by any means or manner we choose. We need not even define models in terms of any form of regimented or formal language, and can merely consider them to be abstract objects so long as some form of linguistic representation is utilized. Although a theory may not necessarily be literally true or empirically verifiable in order to be valid, a theory which both satisfies the given axioms and is empirically verifiable may be viewed as literally true (Van Fraassen, 1980).
For the Burkean scholar, however, “any old language” simply won’t do, nor can the theories and models be conceived or interpreted without the use of some systematic application of language and symbols. The Burkean perspective also encounters friction with the notion of constructive empiricism in that it denies that we can satisfactorily maintain that any theory may be viewed as literally true given that the limitations of language necessarily mean that we cannot provide a full description (Heath, 1986). Here is where the four tropes become useful: we may use them to more precisely define the use of language in the construction of scientific theories and models. This has a precedent in Schiappa’s (1993) synthesis of Burke with Kuhn’s philosophy of science, where it was demonstrated that the linguistic tropes were consistent with the epistemological pragmatism in scientific theories put forth by Kuhn. As noted by Schiappa, Burke provides the sort of social philosophy of language which Kuhn viewed as necessary for the advancement of the philosophy of science. This, according to Schiappa, leads to a pragmatic approach in which paradigms are selected on the basis of their utility and language becomes the basis of shared claims and concepts through which said paradigms are both developed and selected. The following will attempt to do for scientific realism and constructive empiricism what Schiappa did for Kuhn: demonstrate that Burke’s approach to rhetoric, embodied in the notion of the four tropes, provides a vehicle for linguistic analysis in Van Fraassen’s philosophy of science. The key difference is that whereas Schiappa argues that that combining Burke and Kuhn leads to a social constructionist viewpoint, the synthesis of Burke with Van Fraassen leads to an approach where hypotheses and models are more appropriately regarded as mental constructs with social interactions nonetheless playing an important role in the evolution and maturity as theories.
Using examples found in Van Fraassen’s text as well as those who have responded to him in the unending conversation of the philosophy of science, we can demonstrate how the four master tropes become the source of description in the empirical construction of scientific theories. The first of these tropes we shall try to use in this manner is metonymy, and as this corresponds the most closely to the quantitative models and representations used in the sciences, it is probably the most important of the four tropes in this regard. Theories and models themselves, when viewed according to constructive empiricism, become a form of metonymy, as they serve as a reduction of observable phenomena to a statement which can be adequately be expressed linguistically. Whereas the other tropes rely on the imprecise definitions of formal language, the use of mathematics to provide for a metonymic reduction of objects allows for a more accurate model, although not a perfect one. Giere (1985) has responded negatively to both Van Fraassen and the more hardcore realists’ claims of isomorphic closure in scientific modeling; that is to say, the notion that any quantitative model which satisfies the required axioms may be regarded as a complete and accurate representation of object or phenomenon. The notion of “constructive realism” put forth by Giere argues against this claim in a manner similar to Heath’s interpretation of Burke (1986), although arrived at independently. According to Giere, the precise geometric models used by Van Fraassen are not valid analogies to scientific theories, as no quantitative model can ever provide an exact and complete representation of an actual object. Although Tietge maintains that metonymy is somehow separate from the other tropes in its use of a more precise language in the invention of theories and models, when we take Giere’s arguments into account we find that it too is subject to the same linguistic limitations which restrict the other tropes. Metonymy is still a trope apart, however, in the sense that it is used a form of persuasion almost exclusively within the rhetorical community of scientists where it occurs, and it also serves as the “root trope” from which the use of metaphor, dialectic, and synecdoche all spring forth. Whenever any one of the other three tropes is used in constructive empiricism, it is as usually as a means of either re-interpreting a metonymic model or to supplement it (Tietge, 1998).
Giere further argues that instead of relying on certitudes, a model’s validity can be evaluated on the basis of how similar it is to an actual object and on certain specified degrees of realism. Accordingly, just about any form of representation which meets the tests of empirical adequacy maybe considered valid. Here, we have the basis for an understanding of the use of synecdoche in the use of models which serve as the representations of nature which a theory aims to describe. Since we cannot ever have a complete representation of reality, we rely on synecdochal models which derive general principles from representations of specific objects. The validity of the use of synecdoche in a particular model is therefore dependant on the tests of empirical adequacy such as those insisted upon by Giere or Van Fraassen. Moreover, since theories in scientific realism are viewed as endpoints instead of reductions, they rely largely upon a linguistic approach; thus, theories serve as linguistic representations of physical reality and models, in turn, serve as representations of theories. If metonymy is the trope which corresponds to the understanding of theory in constructive empiricism, then synecdoche corresponds in parallel to theory in scientific realism.
Trying to locate the role of metaphor and especially irony in either scientific realism or constructive empiricism proves to be rather more problematic. As Burke himself noted in A Grammar of Motives, these tropes tend to shade into one another in a manner which makes them hard to define; this, added to the fact that contemporary philosophy of science has still only made a tentative investigation of the roles of metaphor and irony in science, means that the status of these two tropes in probing the validity of scientific theories and models remains up in the air. Both Tietge and Schiappa, however, have attempted to demonstrate how the Burkean definitions of metaphor and irony have a role in scientific understanding. In trying to explain the use of both metaphor and irony in scientific discourse, Tietge uses the examples of Charles Darwin’s (1959, cited in Tietge, 1998) use of the term “community of descent” to both provide an understanding of the deeper implications of natural selection through the use of metaphor, as well as an example of irony in the way that it inverts an accurate interpretation of what the theory of evolution truly means. Metaphors such as “community of descent” and “natural selection,” as noted by Tietge, have a certain degree of cognitive power as they serve as a means by which representations of nature can be done so linguistically in a manner which is not necessarily exact but which is sufficiently pragmatic to provide a reasonable conceptualization of a working model.
It has been further noted by Giere (1999) that we cannot establish perfect analogical models in a theory because the similarities can never be fully exact and therefore definitions of the individual objects can never be fully precise. Metaphor especially embodies this double-edged usefulness of analogy in science; but just as metaphors are essential to conceptual understanding in language, so too are models essential to the foundations of scientific theories. As noted by Psillos (2002), scientific realism views model construction in science as being guided by an analogical approach based upon substantial similarities which are either formal (derived from mathematical descriptions of a system) or material (descriptions of purely physical properties), and can then be tested against the phenomena they intend to describe. Although models are on one level synecdochal in the perspective of scientific realism just as they are in constructive empiricism, given that they are representations of larger structures which eventually lead to the development of mature theories, their dependence on analogies based on substantial similarities makes them metaphoric as well.
As for irony, it is perhaps best reinterpreted as a form of dialectic, which is also how it was interpreted by Schiappa (1993). Although a dialectic interpretation fits well with the Kuhnian view of science as a series of competing paradigm shifts, it is less comfortable with either scientific realism or constructive empiricism, both of which conform with the traditional view of science as eventually moving towards an accurate vision of physical reality differing primarily in how that vision is represented mentally and linguistically. Irony is then perhaps best defined, in the context of both scientific realism and constructive empiricism, in terms of models being adequate, but not exact, representations, so that the observed outcomes are not always as expected. Needless to say, irony is more suited to constructive empiricism and other pragmatic theories than to scientific realism, where, as Psillos maintains, “belief in truth is better” (1999, p. 204). The history of science is replete with examples of irony where the outcomes of observables did not match those of predictions, the most famous being perhaps the null effect of the Michelson-Morley experiment and the nonexistence of the hypothetical planet Vulcan, the puzzling outcomes of both of which were later resolved by, respectively, Einstein’s special and general theories of relativity. In constructive empiricism, an ironic outcome does not necessarily constitute a failure of a theory if it continues to meet certain other tests for empirical adequacy. Both Maxwell’s electromagnetic theory and Newton’s gravitational theory continued to meet those tests and continue to be considered adequate even though they have been supplemented by new theories necessitated by new knowledge borne by new observations.
Hopefully, the reader has brought back from the preceding article two key impressions. The first of these is that Kenneth Burke’s four master tropes have a renewed importance in social and rhetorical studies of science given the emphasis on language and symbolic systems that preoccupies much contemporary philosophy of science. The second impression is that there is room for a dialogue between both the rhetoricians and philosophers of science that will allow both sets of scholars to work towards a greater understanding of how scientific theories are developed. If this dialogue is to be successful, however, both sets of scholars must try to go beyond the tropes to other discourse rules and structures which are used in science, as well extending the scope of their investigations beyond the hard sciences of physics and biology typically used as subjects, to encompass theories in sociology, economics and psychology as well.
In addition to asking what rhetorical theory can provide to the philosophy of science, we should also be asking ourselves what the philosophy of science can provide for rhetoric. The preceding paper provided only one side of a conversation; I happily invite the other side to join in.