I-theories of bare demonstratives take the semantic referent of a demonstrative to be determined by an inner state of the utterer. These states are typically taken to be states that constitute having certain referential intentions. E-theories take the referent to be determined by factors external to the utterer.1 These are typically taken to be criteria like salience, conversational relevance and the like. The issue has recently flared up again in an exchange between Gauker (2008), who defends an E-theory, and Åkerman (2009; 2010), who defends an I-theory.
Semantic theorising generally takes one of two roughly distinguishable forms. The first is the so-called ‘method of cases.’ Here the theorist starts by considering our intuitive judgments about the truth-conditions of utterances and tries to develop a theory that accounts for such judgments. The second method we may call ‘the method of theoretical constraints.’ Here the theorist starts with some basic constraint that [End Page 43] all semantic theories must meet, and tries to show that some class of theories is ruled out in virtue of not meeting this constraint. The two methods are, of course, not completely independent, and both have distinctive advantages and disadvantages. This is typical of case-driven and theory-driven methods of enquiry in general.
Semantic theory at present is mostly case-driven. Theories are developed in order to account for cases previously thought to be problematic, these theories are then measured by how well they account for such cases, and objections to such theories typically take the form of putative counter-examples. It is, by contrast, a distinctive feature of the Gauker- Åkerman dispute that it is largely theory-driven. The core of the dispute is Gauker’s basic claim that I-theories of bare demonstratives provide a hearer with no effective method of determining the referent of a bare demonstrative (Gauker, 2008, 362). Hence such theories cannot be true. Åkerman (2009, 159), in turn, defends the I-theory by denying that it provides the hearer with no resources to determine the referent of a bare demonstrative.
My primary aim in this paper is not to adjudicate between Gauker and Åkerman, but to argue that both I-theories and E-theories violate an even more basic constraint on semantic theorizing. I will argue that, if we accept the most popular and intuitive view of communication, sometimes referred to as the Standard view of communication, we are forced to conclude that both I-theories and E-theories are fundamentally flawed. Instead we are led to a third view, distinct from both I-theories and E-theories, on which bare demonstratives are not devices of semantic reference at all.
II The Standard View of Communication
On the Standard view, communication is a matter of speakers trying to make evident, and hearers trying to ascertain, what a speaker has in mind. Interpretation, in other words, is a matter of ‘mind-reading’ (Carston, 2002, 42). What happens is that speakers perform some communicative action in order to communicate some contentful entity that they have in mind. The hearer then uses the communicative action in order to determine what the speaker wished to communicate, effectively treating the communicative action as a clue that can be used to ascertain what the speaker tried to communicate. The Standard view is not without its critics,2 and it is far from clear how various details about this [End Page 44] process should be fleshed out. It is, however, widely accepted among linguists and semanticists.3 These background assumptions about communication allow us to define the following notions:
Speaker’s referent: The speaker’s referent of a term is the object that the utterer, upon an occasion of use, is trying to draw to the attention of the hearer by uttering the term (and wishes to be understood, by the hearer, as drawing to his attention by uttering the term).
Conventional rule: The convention governing the use of a term. This is what a speaker and hearer need to know in order to be competent users of a term and may also be termed the ‘linguistic...