The motto of the Institute of Chartered Accountants of Scotland suggests that the discovery of truth is a central concern of its members. This is a surprising motto for a technocratic organisation, seeming to express moral ambitions more general than those associated with accurate bookkeeping and conscientious vouching. In the academic accounting literature that addresses this issue, truth, in various guises, has also been a central concern. This literature has however, tended to be either dogmatic or despairing—either stipulating unconvincing, and possibly interest driven rules for the discovery of truth, or arguing that all attempts at truth discovery or truth telling in accounting are, in some sense, spurious. This controversy reflects a similar controversy in the wider academic community. The social sciences (in a debate which, in the Anglo-American tradition probably takes Winch, 1958, as its starting point), the study of history (the range here is enormous—classically St. Augustine, Kant, Hegel, in the 20th century analytic tradition Collingwood, 1994, Hempel, 1965 and Popper, 1972; and many more), are prone to regular and disruptive methodological debates, and the physical sciences are far from immune.

The methodological issues raised with respect to research in accounting, in the social sciences generally, and in the physical sciences, are already well aired. Also, somewhat different (although clearly related) problems arise in business and professional practice to those that arise in relation to academic research. The approach taken in this paper has been inspired by the particular difficulties associated with business language use, and it may shed some light on the other more traditional, and perhaps somewhat stale, debates.

If we are to seek out (and tell) the truth we should, presumably, be clear about what the truth is. Many disciplines (some more convincingly than others) attempt to define, or give some general theoretical account of, their fundamental principles and concepts. Would a fundamental theory of truth help us in the way that these theoretical accounts are meant to help the disciplines they apply to?

The answer to this question is that whether or not it would, no such theory or definition can be forthcoming. We can establish this, without examining the strengths or weaknesses of any individual theories of truth, by considering the following paradox:

Let us suppose that there was a general theory (‘Theory X’), which told us how to establish whether or not a statement was true. How would we go about establishing the truth of theory X? We cannot refer to theory X for a test, because the question is exactly whether the tests implied by theory X are reliable. But if theory X is the general theory of how to establish the truth of statements, then no other test of its truth is possible.

This paradox (one of a group of ‘self-referential’ paradoxes, which cause fundamental problems for theories of knowledge, of ethics, of meaning, and of mathematics as well as for theories of truth) is generated by any theory of truth which (a) attempts to give criteria for what counts as a true statement and (b) claims to be completely general. It should lead us to reflect on whether the debating preliminary of requiring the definition of fundamental terms is really as universally valid as it pretends. If we needed to define truth before attributing it, we would be unable to speak at all.

Note that theories of truth for some restricted set of statements—say the statements within a particular disciplinary area—do not generate this paradox so long as the theory of truth is not one of the statements within the practice. This is because such a theory would not, directly or indirectly, refer to itself. We could, for example, have a theory of truth in accounting so long as that theory was not a theory that depended upon accounting principles, or was not a theory within the accounting discipline itself. One way of expressing this is to say that we can have ‘meta-theories’ of truth—theories of truth in one area of discourse that are expressed in another discourse. Meta theorising, generally, is what philosophy does—it attempts to ask questions at the greatest level of generality, questions that are relevant to all rational disciplines (and some less rational ones). Unfortunately, a discipline which is essentially characterised as meta-theorising cannot have a general theory of its own principles of validity which is expressed in another discipline since it, itself, encompasses all meta-theories of this type, and so philosophy necessarily falls foul of the paradox. There can be no general theory of philosophical truth that incorporates or determines criteria for truth telling.

A concrete example of a theory of truth for accounting would be a set of accounting rules and principles which gave comprehensive guidance on how accounting language should be used, and on how the various measurements and tests which accounting reports depend upon should be carried out. This sounds, of course, very much like a ‘Conceptual Framework’ project (notoriously FASB, as first reported in FASB, 1978). If concepts such as ‘rule following’ and ‘measurement’ were sufficiently well defined in ordinary speech for their application in this specific context to be unambiguous, such an approach to defining accounting truth might succeed. Unfortunately, rule-following is an insufficiently transparent process to do the work that is required here, and measurement processes can rarely be well defined outside of their context of application. Conceptual framework projects cannot work as theories of truth if either (a) the external principles upon which they depend are ambiguous or (b) they depend essentially on principles that, in turn, can only be made sense of within the language whose theory of truth is being considered. In the first instance, they fail because of their ambiguity, and the in the second, because they depend upon a circular definition.

Note, also, that the self-reference problem is not solved by having a multiplicity of truth theories, since a conjunction of these (or some appropriately conditionalised conjunction) would still be a general theory of truth, and so would give rise to the paradox.

Setting aside this difficulty for the moment (it will be relevant to the discussion again later), let us turn to some of the main themes that appear in philosophical discussions of truth. In the twentieth century, within the analytic philosophical tradition, these discussions generally refer to Tarski’s schema for theories of truth. Possibly recognising the paradox referred to above, Tarski considered only theories of truth expressed in one language (the language in use, or the meta-language) that apply to another language (the ‘object’ language). According to the schema, any theory of ‘truth in L’ (where L is the object language) would have to give the correct truth value to every statement of the type ‘P is true in L if and only if S’ (often referred to as ‘T-statements’), where S is a statement in the language in use, and P is a statement in the object language L. In other words, a theory of truth for the object language, given in the meta-language, would have to tell us whether any proposed T-statement was true or false. An example of a statement of this type is:

‘il n’y a riens hors du texte’ is true in French if and only if there is nothing beyond the text.

Clearly the simplest kind of theory of truth-in-L that would pass such a test is just a list of all of the statements of this type that are true—a complete translation manual for the object language L, in other words. Since such a list would be indefinitely long, a grammar and a dictionary (i.e. a set of rules for generating such a list, or testing whether or not any given statement was in the list) would normally do.

Breaking the ‘meta-language’ rule results in T-statements that are disturbingly uninformative, but oddly suggestive of one kind of approach to empirical truth telling. For example:

‘Snow is white’ in English is true if and only if snow is white.

Statements like this look a little bit like statements about the relationship between language (the bit in the single quotes) and the world (as described in the ‘language in use’, English). They seem to point to a relationship, a correspondence, between true statements and the way the world actually is. Tarski himself was (arguably) a ‘correspondence’ theorist—that is a theorist for whom true empirical statements are just those which correspond to the way the world is. ‘Snow is black’ does not correspond to the way the world is, because snow is not black. Most people, asked to give a common sense account of empirical truth, would probably come up with something that looked like a correspondence theory. Among the Oxford English Dictionary’s definitions of ‘true’, we find “3. a. Of a statement or belief: Consistent with fact; agreeing with the reality; representing the thing as it is.”