Cognitive models in language and thought: Ideologies, metaphors, and meanings 199–227. Berlin: Mouton de Gruyter. Sebeok, Thomas A., Jesper Hoffmeyer, and Claus Emmeche, eds. 1999. Biosemiotica. Special issue of Semiotica 127, 1/4. Silverstein, Michael. 1979. Language structure and linguistic ideology. Chicago Linguistic Society 15, vol. 2 (parasession): 193–247. Simpson, Paul. 1993. Language, ideology and point of view. London: Routledge. Skutnabb-Kangas, Tove. 2000. Linguistic genocide in education: Or worldwide diversity and human rights? Mahwah, NJ: Lawrence Erlbaum. Stockwell, Peter. 1999. Towards a critical cognitive linguistics? In Annette Combrink and Ina Biermann, eds. Poetics, linguistics and history: Discourses of war and conflict 510–28. Proceedings of PALA XIX, Potchefstroom University, South Africa, March 29– April 1. CD edited and produced by Peter Stockwell and Martin Wynne (available at http://www.pala.ac.uk/confs/potch/CD/home.htm). Published for PALA by the Hu- manities Computing Unit, Oxford University. (Available at http://www.nottingham .ac.uk/~aezps/research/papers/CRITCOG.PDF) Toolan, Michael, ed. 2002. Critical discourse analysis: Critical concepts in linguistics. 4 vols. London: Routledge. Urban, Nancy Y. 1999. The school business: Rethinking educational reform. PhD disser- tation, University of California, Berkeley. van Dijk, Teun. 1993. Principles of critical discourse analysis. Discourse and Society 4: 249–83. van Dijk, Teun. 1997. Discourse as structure and process. London: Sage. van Dijk, Teun. 1998. Ideology: A multidisciplinary approach. London: Sage. van Dijk, Teun. 2002. Discourse, knowledge and ideology: Reformulating old questions. LAUD Paper no. 546. Essen, Germany: Linguistic Agency of the University of Duisburg-Essen. Weingart, Peter. 1995. ‘Struggle for existence’: Selection and retention of a metaphor. In Sabine Maasen, Everett Mendelsohn, and Peter Weingart, eds., Biology as society, society as biology: Metaphors 127–51. Dordrecht, Netherlands: Kluwer Academic Pub- lishers. White, Michael, and Honesto Herrera. 2003. Metaphor and ideology in the press coverage of business consolidations. In Rene ´ Dirven, Roslyn Frank, and Martin Pu ¨ tz, eds., Cognitive models in language and thought: Ideologies, metaphors, and meanings 277–323. Berlin: Mouton de Gruyter. Whorf, Benjamin L. 1956. Language, thought, and reality: Selected writings of Benjamin Lee Whorf. Ed. John B. Caroll. Cambridge, MA: MIT Press. Winter, Steven L. 2001. A clearing in the forest. Chicago: University of Chicago Press. Wodak, Ruth, ed. 1989. Language, power and ideology. Amsterdam: John Benjamins. Wodak, Ruth, and Paul Chilton, eds. 2005. A new agenda in (critical) discourse analysis: Theory, methodology and interdisciplinarity. Amsterdam: John Benjamins. Wolf, Hans-Georg, and Frank Polzenhagen. 2003. Conceptual metaphor as ideological stylistic means: An exemplary analysis. In Rene ´ Dirven, Roslyn Frank, and Martin Pu ¨ tz, eds., Cognitive models in language and thought: Ideologies, metaphors, and meanings 247–77. Berlin: Mouton de Gruyter. 1240 rene ´ dirven, frank polzenhagen, and hans-georg wolf chapter 48 COGNITIVE LINGUISTICS AND PHILOSOPHY peter harder 1. Introduction How one sees the relationship between Cognitive Linguistics and philosophy de- pends on what one takes to be the role and nature of philosophy. The approach followed below is the one presented by J. L. Austin (see Austin 1970: 232): it views philosophy as constituting the overarching arena for discussions about the nature of the world and our knowledge about it, within which independent disciplines have gradually crystallized into domains of their own. The decisive factor in the process that creates such independent areas is the rise of a descriptive practice that is generally recognized as adequate for a particular area of inquiry. When such a method becomes established, work within the field ‘‘graduates’’ from the stage of philosophical argument to systematic investigation in well-defined terms. To some extent, such areas then point the way for the rest of philosophy to follow; already in antiquity, the systematicity of mathematics endowed it with the status of a para- digmatic form of knowledge that was a source of inspiration for the rest of phi- losophy. In more modern times, the most epochal instance of this is the process that gave rise to natural science with physics as the paradigm discipline. This process of ‘‘cultural modularization of knowledge’’ means that the terri- tory of philosophy is gradually diminishing. There is, for instance, no longer any- thing corresponding to ‘‘philosophy of nature,’’ in which the constitution of matter is discussed. In contrast, there is still an area of ‘‘philosophy of mind’’ because discussions about the nature of the mind have not in the general opinion reached a stage of solidity in which they can be discussed in isolation from the overall framework. Austin envisaged a period in which also the area of language might come to constitute an independent domain—and this is a natural perspective for viewing the relationship between Cognitive Linguistics and philosophy. Even for areas that have reached the stage of independence, however, philos- ophy retains the function of keeping order in the universe from an overall perspec- tive. The question of how different areas with different approaches relate to each other cannot be answered by any single discipline. In this view, philosophy is an arena of inquiry, not a body of definitive results. To say that something is a phil- osophical question may occasionally suggest that you can say whatever you like about it. The reason for that impression is that philosophical questions only arise in cases where there are no straightforward answers to a question, but where we must go back to more basic and general principles. A fair number of such questions can still be found in Cognitive Linguistics. Taking Austin’s approach means that the chief aim below is to address some foundational issues that I view as important for the theory and practice (goals and methodology) of Cognitive Linguistics. I will make no attempt to express the phil- osophical position of Cognitive Linguistics (there is none) or to follow up all ref- erences to philosophical issues in the cognitive linguistic literature (e.g., to Aris- totle’s views on the nature of concepts or Merleau-Ponty’s on embodiment). The points I will take up are those that call for a continuing clarification process that would also be useful to Cognitive Linguistics. The main issue to be addressed is the nature of the object(s) of investigation. No consensus has been achieved, either inside or outside Cognitive Linguistics, on the precise status and properties of mental entities, including their relation both to the human body that generates them and to the outside cultural and physical en- vironment. That issue and its implications will be a leitmotif below. For obvious reasons, the status of mental entities is crucial also to philosophy— and since Cognitive Linguistics makes claims about the mind that bear upon what philosophers have said, this aspect of the problem constitutes an issue in its own right. In order to discuss that problem, it will be necessary to touch on some of the basic concerns of philosophy. These include the relationship between ontology (i.e., the nature of reality, including the objects under description) and epistemology (the question of how we can acquire true knowledge of reality). The latter is con- cretely manifested in the form of methodology (what are the appropriate scientific, descriptive procedures that reflect our epistemological stance?). Thinking about the mind is historically bound up with dilemmas that span all these levels, with a rough polarity between, on the one hand, idealism and rationalism, which share a com- mitment to mental foundations of understanding, and on the other hand, em- piricism, which takes actual experience, entering the mind via the senses, as the foundation of knowledge. 1242 peter harder The aim of both positions, however, is to arrive at true knowledge of the world—that is, whatever may be the nature of mental representations, there are criteria which these representations have to live up to in order to work ‘‘properly’’ from a philosophical point of view. These criteria involve epistemology as well as ontology: our thoughts must work reliably and tend toward valid conclusions about reality. Below, the word validity will be used with reference to both dimen- sions. The criteria of validity, which are central to philosophy, constitute a major difference of perspective between philosophy in its metaposition, viewing all areas from above, and Cognitive Linguistics as a discipline in the process of carving out its own domain. The implications of that difference constitute another main theme below. Thinking about mental entities has had a turbulent history in the twentieth century. An account of the developments that led up to the emergence of Cognitive Linguistics may therefore be illuminating in order to pave the way for a discussion of the principles involved. Especially in relation to the ‘‘first cognitive revolution’’ (see Sinha, this volume, chapter 49), including Generative Grammar, which con- stitutes the immediate predecessor to Cognitive Linguistics, the process of devel- opment is an important part of the present picture. The account below will there- fore begin with a historical account. 2. Linguistics and the Philosophy of Science in the Twentieth Century The history of linguistics in the twentieth century is profoundly influenced by de- velopments in the overall philosophical perspective. Other disciplines, for example, biology, have to a greater extent developed as a result of increasing domain-specific knowledge leading to the rise of, for example, a molecular dimension of the subject. Developments in linguistics, in contrast, have been marked by revolutions from above. Already in Classical Antiquity, linguistics (or ‘‘grammar’’ as it was then called) experienced a partial liberation from its philosophical beginnings with the mor- phological school of the Alexandrian grammarians. The tradition they founded was extremely durable, continuing uninterruptedly until the twentieth century—even today, most people who have had any experience with school grammar probably think of grammar in their terms, as classes of words with different sets of inflections. But the liberation from philosophy was only partial; the semantic foundations of this pattern of description remained bound up with classical ontological assump- tions going back to Plato and Aristotle. These included an assumption that linguistic categories directly mirror mental categories, which again mirror the categories of cognitive linguistics and philosophy 1243 the world. The ‘‘speculative’’ approach (from Latin speculum ‘mirror’) was applied through intuitive ‘‘hermeneutic’’ (interpretive) description (with unclear empirical control, which is why the word speculative gradually acquired a new sense and a doubtful reputation) and, throughout that entire period, did not undergo much change. Therefore, it was vulnerable to radical twentieth-century changes in philo- sophical assumptions about the world and about principles of scientific description. The underlying cause of these changes had to do with the paradigmatic status of natural science understood as carrying the beacon of progress—not only intellec- tually but also technically and socially. From the Renaissance onwards, the emer- gence of a world picture based on blind, objective laws, supported by demonstra- tions of the technical possibilities created by insight in these laws, gradually changed the picture of human knowledge. A central trend was the erosion of faith in sys- tems of intuition-based ‘‘ideas’’ in favor of faith in scientific method, both inside philosophy itself and in the general intellectual climate. From the beginnings of philosophy, epistemology and ontology had been intimately connected, while the relations between them had varied. The triumphs of scientific method tipped the balance decisively in favor of a greater weight attached to epistemology—that is, in favor of the question of ‘‘how’’ rather than ‘‘what.’’ The result was that world knowledge became increasingly subservient to the dictates of methodology, what- ever the consequences might be for the familiar picture of the world. Being intu- itively obvious was no longer the same as being taken for granted. Although this development had been at work since the sixteenth and seven- teenth centuries, some of the potential consequences only became apparent in the twentieth. Behaviorism, which defined psychology in terms of overt behavior only, was a radical consequence of this way of thinking, eliminating the mind entirely as part of the scientific world picture. Since scientifically reliable insight could only be achieved by observations of overt behavior (methodology), it followed in practice, if not in theory, that responsible psychologists could not accommodate such a thing as the mind in their world picture (ontology). From a philosophical perspective, this conclusion was not by any means obvious (see Russell 1956: 293); and jokes like that of the two behaviorists who met and said, ‘‘How am I?—You’re fine!’’ showed a continuing awareness of the apparent absurdity of the position. Within psychology, it had all the destructive influence that only a denial of the chief object of de- scription can have on a scientific discipline (see Sinha, this volume, chapter 49). The fact that psychologists were willing to pay this price shows how powerful the epis- temological and methodological considerations were. In linguistics, Bloomfield re- vised his linguistic system to make it fit with the assumptions of behaviorist psy- chology, excising all reference to mental objects. Psychologists and their linguistic followers in America, however, were the only scholars who went so far as to deny the existence of the mind. Within philoso- phy, the development away from the familiar world picture toward principles of scientific method took another form, called ‘‘the linguistic turn.’’ The idea was, roughly speaking, that scientific status is bound up with the formulation of sci- entific problems and solutions, rather than with the actual substance that they dealt 1244 peter harder with (see Carnap 1934). By fixing the properties of the scientific metalanguage, in practice a version of formal logic, the philosophy of science could be a guardian of the reliability of knowledge without getting involved with the actual world. The inspiration came from the use of mathematics as a tool for the precise formulation of physical laws (the model of true scientific knowledge noted above). One of the aims of tightening up criteria for what counted as tenable forms of thought was precisely to get rid of the whole inherited but prescientific ontology, as expressed in volumes of what was seen as mere verbiage (since it was not expressed in the language of formal logic). Within linguistics, there were domain-internal tendencies that could derive inspiration from this view of what counted as reliable knowledge. The structural revolution occurring in the wake of Saussure was also motivated by a desire to get rid of the ontological ballast from antiquity. Saussure insisted that words as lin- guistic entities were severed from meanings understood as preexistent entities in the mind or in the world. The classical, in the etymological sense, speculative grammar, in which language mirrored immutable ideal reality (such that substantives de- noted substances, etc.), was thus rejected in favor of a picture that focused on structural relations between linguistic forms. This transfer was further promoted by a movement toward structure within philosophy itself. Carnap (1928) set up the concept of ‘‘logical syntax’’ to capture the logical analysis of the structure of the world and suggested that science can essen- tially only describe structures. It is not surprising that in European structuralism this parallel was a major source of inspiration. The abstract and formal nature of structural relations, as opposed to intuitive content in the form of ideas or concepts, had the same aura of rising above inherited muddles in linguistics as it had in phi- losophy. It is perhaps not inappropriate to point to modernist developments in art (see Ortega y Gasset [1925] 1976) to give an idea of the pervasive drift in the early twentieth century away from a baggage of assumed, presupposed substance, toward imposing a new abstract order on experience. In this movement, the alien flavor is a virtue, precisely because it demonstrates a successful break with accepted (hence suspect) criteria of naturalness. Although this movement was by no means restricted to philosophy, its most powerful manifestation remained the alliance between physics as a model science, on the one hand, and formal modeling as the paradigm for scientific description, on the other. The reason Generative Grammar achieved its dominant position in modern linguistics is that it presented an application of this model of description to language and launched it at the point when the behaviorist barriers against theo- rizing about the mind were ready to be dismantled. The rise of cognitive science, the ‘‘first cognitive revolution’’ around 1960, came about not because mental entities as we intuitively know them suddenly became scientifically respectable, but because the restrictions imposed by behaviorism were becoming obsolete. Chomsky (1959) achieved his most spectacular breakthrough with his review of Verbal Behavior by B. F. Skinner, the main exponent of behaviorism as applied to the psychology of language. The force of his argument had two sources: it cognitive linguistics and philosophy 1245 introduced more advanced descriptive techniques, and it was directed against a position that was intuitively unattractive because of its reduction of human powers to reflexes. The complexities that Chomsky introduced thus combined the thrust of increased scientific sophistication with a defense of human dignity. Moreover, in being a founding member of the newborn discipline of cognitive science, Gener- ative Grammar was part of a movement that appeared to mark a wholly new epoch in all sciences dealing with human nature, combining exciting new vistas with rig- orous scientific precision. Only much later did the limitations of the formal appa- ratus that was at the core of the first cognitive revolution become an issue. The sudden scientific acceptability of complex theories of cognition can pri- marily be attributed to the invention of the computer. If a machine could handle complex information processing in ways that could not be handled by simple stimulus-control relations between events, the credibility of trying to understand human information processing in those simplistic terms was gone. The first step in Generative Grammar was the demonstration that the formal model that could be distilled out of behaviorist principles of description was simply too impoverished to account for the known structure of linguistic facts. The scientific legitimacy that was already accorded to formal modeling in physics was thus simply transferred to formal modeling in linguistics. It is therefore natural that Generative Gram- mar should understand itself as carrying the beacon of true scientific principles of description. In the generative context, the frequently invoked term empirical refers to the conception in terms of building a formal model, deriving empirical predictions from it, and testing them against the data, which had now been finally extended to linguistics. Previous forms of linguistics were assigned the status of alchemy in relation to chemistry, with Bloomfieldian linguistics as the primitive precursor to the true science of language. The identification with hard science was confirmed when Generative Grammar allied itself with a semantics based on logical, truth- conditional foundations (i.e., the level of ‘‘logical form’’). Philosophically, however, the situation became rather more complex when Chomsky reinterpreted his own philosophical position in rationalist terms, claim- ing that the nucleus of the formal system must be innate in the human mind. As pointed out by Itkonen (1992: 73), this is a wholly external maneuver in relation to the anatomy of Generative Grammar per se, which (in good accordance with the model science of physics) was originally predicated on the ‘‘physical properties of utterances’’ (Chomsky 1975: 127), just as in the thinking of his Bloomfieldian predecessors. It is sort of odd for a science to invent a new and more attractive object of description while retaining the descriptive apparatus, and this is an important clue to Chomsky’s way of thinking. Apart from the rejection of behaviorist oversim- plification, Chomsky’s early writings were not initially motivated by new insights in language, but by the idea of putting linguistics on a new scientific footing. From that point of view, it is natural that the exact nature of language, or rather 1246 peter harder ‘‘grammar,’’ as an object would have to depend on what fitted best into the new scientific footing. A grammar that is inherent in the basic wiring of the human mind is much more congenial to his formal system than one viewed as a property of actual, messy empirical talk. Chomsky’s repeated insistence on internal language as the only proper object of linguistic description (for a recent reaffirmation, see Chomsky 2000) reflects a basic commitment to formal, metalinguistic precision as the criterion of scientific status. This claim is maintained in spite of the fact that the hard-nosed attitude of precision coexisted with a rather soft intuitive underbelly when it came to relations with the data. His postulated underlying structures are prevented from having empirical status in terms of the criteria of hard sciences because of the problem of experimental access to mental as opposed to physical entities. When it became apparent that there were no compelling reasons to prefer Chomskyan intuitions over intuitions less regimented by rather idiosyncratic prin- ciples of formalization, the time was ripe for a new Second Generation Cognitive Science in which Cognitive Linguistics had a central role (see also Sinha, this vol- ume, chapter 49). 3. Cognitive Linguistics and the Rebellion against Methodological Imperialism Cognitive Linguistics represents a radical break with the earlier twentieth-century trend of shaping linguistics in ways that reflect philosophical views of what counts as scientific description. Cognitive Linguistics focuses squarely on mental, con- ceptual entities as legitimate objects of description in their own right. Historically, this can be understood as reflecting increasing permissiveness in the philosophical climate. The first generation of cognitive scientists remained careful in their outline of a science of the mind because the fear of the science police was still present: their way of mapping out the science of the mind was designed to prove that such a science would not revert to prescientific forms of description. When discussions about the nature of mental entities had become generally accepted, the need to stay within established ‘‘formalist’’ bounds of scientific respectability was no longer acutely present and diminished in favor of a desire to bring new kinds of phe- nomena to light. From that perspective, an important mission of Cognitive Linguistics, as part of the new generation of cognitive science, was to show that established barriers prevented a full investigation of the field. The most uncontroversial achievement of Cognitive Linguistics is probably that it opened language description to a rich cognitive linguistics and philosophy 1247 new landscape of conceptual phenomena and mechanisms interrelated in multiple ways with the whole of human experience and shaped in accordance with patterns of human imagination. In addition to these positive qualities, Cognitive Linguistics was also very much aware of being everything that Generative Grammar was not: nonformal, nonmodular, non-truth-conditional, and so on. Cognitive Linguistics saw itself as a liberation from misguided earlier philosophical positions. This is true especially when it comes to methodology. Langacker ( 1987: 31) has gone on record as being wary of devoting too much importance to methodology as opposed to the actual object of description. Both Langacker (1999: 26) and Lakoff and Johnson (1999: 79) espouse the open-minded doctrine of ‘‘converging evi- dence,’’ that is, of letting all methodologies be used and see what they collectively point to. When it comes to ontology as opposed to methodology, the position of Cognitive Linguistics is more well-defined and follows from the core area of the approach. The central domain of Cognitive Linguistics is cognitive, conceptual structures defined by their content rather than their formal structure alone. In accordance with this, Langacker (1987: 97) has made explicit his commitment to the conceptualist approach to meaning. This commitment is made in direct con- trast to the views expressed by Lyons ( 1977: 113) as a representative of first gen- eration cognitive science. Generative Grammar and Truth-conditional Semantics saw no independent role for a purely conceptual-semantic level, because it would be an expendable ‘‘middleman’’ between the two central levels of formal-syntactic structure and logical form. However, the conceptualism adopted by Langacker and other cognitive lin- guists should not be understood as aggressively oriented against other ontological commitments—rather, its function is specifically to defend the legitimacy of the central elements in the theory, regardless of what other elements they may share the universe with. This ontology goes naturally with attributing a central methodo- logical role to intuition: since mental content is only accessible to human subjects, there is no way of getting at the central data without them. The basically optimistic feeling about the sturdiness of its object of description and of the intuitive descriptive practice that goes with it has served Cognitive Linguistics well in its investigation of the central phenomena of its field. These include construal, imagery of different kinds, mental spaces (this volume, chapters 3, 14, and 17), and the way these are associated with different linguistic categories. But the problems that worried most of modern psychology and philosophy about the nature of mental entities have not evaporated entirely. Cognitive Linguis- tics would not have been possible without a certain consensus about the value of intuitions. Still, as had been noted at earlier points in the history of psychology, intuitions do not always provide clear-cut answers. One of the points where conceptual dis- agreement is frequent in scholarly circles is when it comes to the question of how much to lump together under one concept. Where this is usually a problem at the 1248 peter harder metalevel (involving the scientific terminology), in Cognitive Semantics it involves the object of description, linguistic meanings. The problem was brought to the forefront by Sandra and Rice (1995) in relation to the issue of polysemous networks of meaning, as they have become familiar from the analysis of the preposition over in Brugman (1981) and Lakoff (1987). Sandra and Rice found that it was very difficult to confirm the existence of cohesive semantic links in polysemous networks of meaning by experimental means. The question this raises, as expressed in the title of their article, is whether the mental constructs posited in network analyses reflect any other mental object than the linguist’s mind (see also Rice, Sandra, and Vanrespaille 1999). The article thus directly challenges the scientific status of intuitions. The discussion continued in a series of articles in Cognitive Linguistics (Croft 1998; Sandra 1998; Tuggy 1999; see also Sanders 1997). Whereas Croft and Sandra stress the limitations of what can be postulated on the basis of intuition-based linguistic evidence, Tuggy defends the possibility of using linguists’ intuitions as a source of information about mental entities. This is not the place to enter into the discussion, but it shows that there is a methodological, hence philosophical, di- lemma for Cognitive Linguistics here: while convergent evidence makes it possible to relax constraints on methodology, divergent evidence brings back the issue. In particular, the latter calls for a clarification of criteria of what exactly is meant by ‘‘psychologically real.’’ The question of what the relation is between mental content as it manifests itself intuitively and mental content as manifested in experimental findings is a philosophical question, which cannot be answered simply by looking at the data. At this point, the foundational question of the nature of the object of description raises itself as part of the road to further progress, bringing us back to the central issues raised in the introduction. 4. Conceptual Structures and the Philosophy of Mind: Internalist and Externalist Perspectives The question of the nature of mental entities is also becoming more salient in relation to the issue of how the domain of cognitive phenomena is to be defined in relation to phenomena inside as well as outside the individual subject. The general tendency is to use the word cognition to cover as broad a range of phenomena as possible, rejecting narrow compartmentalization and stressing the essential con- tinuity of all cognitively imbued domains. This is motivated both by opposition to cognitive linguistics and philosophy 1249 . important part of the present picture. The account below will there- fore begin with a historical account. 2. Linguistics and the Philosophy of Science in the Twentieth Century The history of linguistics. in the philosophical climate. The first generation of cognitive scientists remained careful in their outline of a science of the mind because the fear of the science police was still present: their way. looking at the data. At this point, the foundational question of the nature of the object of description raises itself as part of the road to further progress, bringing us back to the central