Tallis thinks that one of the basic confusions of post-Saussurean criticism is a confusion of the levels of parole and langue . Signifiers and signifieds are, for Saussure, purely differential; but words are not signifiers or signifieds, but types of signs, which are combinations of the two. Signifier and signified does not, as Terry Eagleton implies in a badly garbled statement, equivalent to signs and referents.

Words, actual verbal tokens in a language, don’t belong to langue but to parole , and as such they has a positive content that the components (which only exist notionally) do not have. Tallis says, “It is language that is a pure form without substance; while speech includes at the very least a particular phonic substance.”

The post-structuralist conclusion that Saussure’s differential/negative account of signifiers and signifieds implies there is no reference relies on “the identification of signifier and signified with, respectively, sign and referent.” But “Saussure himself contrasts the negativity of the signifier and the signified with the positivity that results from their fusion in the realized linguistic sign – the actual verbal token as it appears in discourse: ‘the statement that everything in language is negative is true only if the signifier and the signified are considered separately; when we consider the sign in its totality, we have something that is positive in its own class.” Signifier and signified are “purely relational.” They are like the two sides of a sheet of paper, so distinguishing them is “purely notional” anyway. On the other hand, “the sign as a whole, realized and put to use, is none of these things.”