Culler points out that there are both paradigmatic and syntagmatic relations at every level of language. Nouns combine with prefixes and suffixes, and the possible syntagmatic combinations help to define the noun: “A noun is partly defined by the combinations into which it can enter with prefixes and suffixes.” Friend can take be- and -ly but we don’t find “disfriend, friender, friendation, subfriend, overfriend, defriendize.”

What is the force of Culler’s “can”? Why “can’t” we say “I disfriended him” or “he’s more a subfriend than an overfriend”? The answer is, we can, and the sentences are sensible.

Culler, though, following Saussure is talking about the “can” of langue rather than parole . In the linguistic system of English, we don’t add de - or sub- to friend . The fact that when we do add such prefixes we get arrestingly odd words is a sign that this is not part of the English sign-system.

Culler makes the same point about syntactical relations. There are pre-ordained syntags that might go with “he frightened,” which might include “George, the man standing on the corner, thirty-one fieldmice” but not “the stone, sincerity, purple, in.” Again, Culler uses the word “can”: “Our knowledge of the syntagmatic relations enables us to define for he frightened a paradigmatic class of items that can follow it.”

But we can say “he frightened the stone” or “he frightened sincerity.” In isolation these sentences make little sense, but put them in a narrative or a poem and they might well be enriching. Inventing utterances that violate the syntagmatic expectations, in fact, is stock-in-trade for many poets, especially modern(ist) ones.

This is not a refutation of Saussure or Culler. But it does suggest the limitations (which Saussure and Culler are quite well aware of) of the structuralist focus on synchronic system. Saussure and Culler may be entirely correct about langue , but that says relatively little about the possibilities of the actual exercise of language or the interpretation of literature.