Theories of Meaning What is the kind of data that theories of meaning shall be able to handle and what is the point with systematizing these data? In short what is semantics? Theories and Theories of Meaning There are different meaning theories connected with logic, linguistics, cognitive science, philosophy etc. It is arguable that we shouldn t ask for a general purpose of a theory of meaning but instead look for the specific role of the concept determined by every theoretical context. We could claim that outside a specific theoretical context there is no point to ask for the nature of the meaning relation or to ask for the stuff and structure constituting the carriers of meaning. Cummins (1989) has an analogous argument for treating the question about the nature of mental representation as a question within philosophy of science. We should pick a theoretical framework as computational theories or connectionist theories, or whatever, and ask what explanatory role mental representations plays in that framework and what the representation relation must be if that explanatory role is going to be well grounded. 1 If we accept Cummins strategy we should take each theory exploiting any semantic concept for it self and ask for the role of that notion in the theory and then try to se if any explicit definition of the concept can fulfill that role. I will evaluate this philosophical method and set it up against a more commonsense oriented view. A Basic Epistemic Level There is an epistemically basic level at which we understand such phenomena as thought, perception, action, language, speaker, speech, desire, happiness, pleasure, pain etc. For me it is natural to conceive of all these phenomena as somehow related to the human mind. I do not want to claim that animals minds are not related to the same or similar categories, though. Further we are able to sometimes recognize what we say think or in general signal. We are also capable to judge when things go wrong in different ways: we misperceive; make invalid inferences; infer false conclusions; get false reports; hallucinate etc. Perhaps these abilities are part of the human nature, perhaps they are products of social evolution, but that does not matter for the point I want make here. I would claim that semantics was born at the very day someone started to systematize those aspects of mind that concerns correctness in thought, speech and perception. Then some concepts filling roles similar to those filled by truth and meaning in current theories must have obtained. The recognition of these basic phenomena may seem very distant from the construction of any formal theory of semantics or of a naturalized meaning theory fitting a computational or functional theory of cognition or whatever. In formal semantics are almost all features 1 Cummins (1989) p13
abstracted that were counted at the basic level. The modern dissociation of psychology from semantics states that psychology is uninformative about, e.g., the concepts of logical consequence and valid inference. The separation of psychology from philosophical semantics involves the silence of psychology about the nature of truth and semantic content of thought. Semantics Versus Psychology There is still a demarcation problem concerning the semantic domain versus the psychological or perhaps the cognitive domain. Take the issue of descriptivism versus direct reference theories e.g.. The classical program of descriptivism comprised a broad set of issues of both, prima facie, cognitive and semantical character, such as: the problem of informativeness of identity statements, the issue of substitutivity of coreferential terms in contexts of broadly intensional character and in quotational contexts, the semantics of propositional attitude reports, the problem of assigning truth values to negative existentials and the problem of explaining how reference is secured. Frege especially attended to the two first issues mentioned above. His solution in terms of the distinction between sense and reference is more than well known.. Russell took issue of negative existentials to be a crucial one for semantics. He distinguished between genuine names and ordinary proper names that were considered concealed definite descriptions. By that move, Russell went into position to treat all of the issues mentioned above. I do not intend to dwell upon details in the development of descriptivism and its historical antagonist, the direct reference approach (with major proponents such as Kripke, Donnellan, Putnam and Marcus). I restrict myself to mention the most salient feature of the new approach to semantics. It is expressed by Putnam s device that intension does not determine extension. 2 This move is partly motivated by the demand that semantics shall be able to handle ordinary language. Kripke showed that a general problem with strict descriptivism is that it does not seem to work for ordinary proper names. Take Clinton for example. What description secure the reference of this name? Do all successful uses of Clinton consist of a hidden description uniquely picking out the same individual? Another motive is part of a naturalistic movement in the philosophy of mind and language. Here it says that the concept of meaning shall not refer to a nonphysical property of its own kind. There is an underlying ambition to produce a semantics that is implementable in brains and perhaps in other mechanical systems. From this perspective the domains of semantics and psychology must get new demarcations and both in turn must be related to the domain of physics. At the basic conceptual level we take for granted, e.g., that meaning is in thought and that we know the meaning of our thoughts when thinking them. We also consider twins at twin-worlds 2 Putnam "the meaning of meaning" har jag för mig.
to be thinking thoughts with the same meaningful content, as we do our selves. All this is inconsistent with a purely externalistic account of thought content. The list of clashes, between our basic intuitions regarding meaning and the externalistic and naturalistic approaches to meaning, can be made longer, no doubt. The point I want to make is that one cannot escape from performing some kind of analysis of the basic data. These must either be explained or explained away, by a comprehensive semantics. Perhaps there is no comprehensive semantics and perhaps that is no problem. To find out if we have that problem or not I will try to model the relation between the basic level and the theories. The Role of The Basic Level in Theorizing REALITY? Are the theoretical categories equivalent with real categories? Different theoretical categories Are data at the basic level data for theories? Commonly recognized categories Are the basic categories equivalent with real categories? REALITY?
Let us take an example of a kind of theory so general that it perhaps by some would be considered to cover all cases involving meaning. What I think of is a general theory of representation. Such a theory should be able to handle at least the following categories: (i) individual representations (ii) relations among individual representations (iii) the production of representations and their relations (iv) the use of representations by (v) the device or devices responsible for production and use of representations (vi) the communication of representations (vii) the system hosting i,ii,v and responsible for iii,iv,vi Here comes another example of a general theory formulating what can be meant by direct refernce theory I shall give a general characterization of, the so called, New theory of reference in (0)-(2) below. (0) The meaning of a sentence is a function of the meaning of its parts. (1a) The reference relation between referendum and referent is not equal to, determined, secured or established by a mediating entity such as a concept, description or sense of any kind. In short, no semantic property is involved in securing reference. (1b)The reference relation is secured by an appropriate external relation, between referent and referendum. (2) A referring item contributes to sentence meaning with its reference only. No associated concept, description or sense of any kind is contributed to the sentence meaning by referendum. The conditions could be combined all three or embraced separately. Sometimes does the epithet "direct reference theory" stand for combinations of all conditions. Sometimes (1) & (2) are alluded to. Often (1b) is narrowed to cover only proper causal relations. Therefore "causalism" is an often used label of direct reference theories. What about the following story about what we do when we construct theories as those sketched above? By assumptions a model is built containing categories (e.g. those above) and relations among them. We then construct a theory about our model. That is, we formulate basic truths about it and draw conclusions. Next step is test our theory. Of course we had some phenomena in thought while we were constructing our model and the theory about it. Those things we had in mind were known from the basic level or from another theory. The step we now take is to identify certain phenomena with the categories defined in the model. From our theory about the model we know how certain kinds of phenomena behave so we
expect the world to behave the same, or approximately the same or the same if there are no disturbing causes etc.. If the world behave otherwise the theory is false about the world or the model shouldn t be identified with the phenomena at issue but perhaps with other existing phenomena. The word model here is used with at least two distinguishable meanings. First there is the construction giving meaning to our theory. Secondly, if the world contains phenomena behaving as the theory truly says that the model does, then the world is modeled as well. What has this to do with the relation between the categories known from the basic level and theories? It has to do with the construction of models. We build them from pre-theoretically known categories. Even if taken from another theory they are known independently of the theory we shall construct. Without these categories we could not begin theorizing. What I have said so far does not prevent any one from claiming two things. First she could argue that the pre-theoretical categories have nothing in common with the categories known from the basic epistemic level. Secondly she could argue that the pre-theoretical meaning is radically transformed within the theory. Both these statements, especially the last, could be used to argue for Cummins strategy. One could claim that when if there is an empirical theory with a notion of representation supposed to fulfill a certain theoretical role but missing an explicit definition, then we cannot go to the pre-theoretical meaning but to the theory. A theory using a concept not defined is no theory. So my suspicion is that so far as the theory works there is enough conceptual power built into it. So when is spoken of a need of definition of a notion which is already part of an empirical theory it is not really clear what is meant. When Cummins speak of empirical theories he mentions, e.g., folk psychology, orthodox computationalism, connectionism, and neuroscience. If some partially defined notion of representation figures in these sketches of theories I would say that we have gained from other theories and from our basic knowledge about thoughts etc. So against Cummins view that we must enter the theoretical framework I would claim that we must enter some framework, pre-theoretical or theoretical to start our theorizing. NU SKITER JAG I DETTA. Mycket är nerplitat mer eller mindre nu.