Linguistic units are signs, a view which suggests that linguistics is closely related to semiotics, the philosophical theory of signs. In this tradition, there are many definitions of the term `sign'; a useful source of background information is Innis (cf. [14]). The linguistic tradition starts with de Saussure (a dyadic sign concept), continues through Bühler (with signs related to three components of the context of use) to Jakobson (with signs related to six constitutive factors in the situation of utterance), and beyond.
In the logical tradition, which started with Peirce and continued through Morris (see Innis, [14]), a basic three-way distinction was made. Carnap introduced a variety of three-way distinction with a terminology which has become widely used: syntax investigates the relation between signs and other signs, semantics the relation between signs and the world, and pragmatics the relation between signs and their users. A later development, starting from Carnap's work, is logical Model Theory, in which sentences are constructed from elementary items in a vocabulary, and interpreted compositionally in terms of a model.
It is presumably no accident that the distinctions made by Bühler and Carnap were relatively similar; both were professors in the Faculty of Philosophy at Vienna University around 1930. Jakobson's theory builds on and extends Bühler's: both were active in the Prague School context around 1930.
About ten years ago, the sign theory of de Saussure was revived by Pollard and Sag (cf. [31]), and formulated in terms of a feature-based grammar. Their sign model makes no reference to `the world', or `the user', like the sign model of de Saussure, but is, also like de Saussure's model, conceptualis or mentalist.
Unlike de Saussure, Pollard and Sag put the compositionality of signs in the centre of their approach, rather like the syntactic aspect of the logical approach to semiotics. For Pollard and Sag, signs consist in general of smaller signs, and the properties are a function of the properties of their constituents. Signs have phonological shapes, and meanings, which are also a function of the properties of their constituents.
Suggestions for discussion: