Shortly I'm about to print another essay on one of the critical works of Northrop Frye, which I'm sure will be as welcome to the assembled hordes (mini-hordes?) of comicdom as the flowers that bloom in spring. But because Frye uses a term I don't accept, I'm leading into the longer essay with a shorter one on that term, and why I think there's a better one.
The word is "convention," used in the sense of "a literary convention." American Heritage online gives this definition of the word in this sense:
"A widely used and accepted device or technique, as in drama, literature and painting."
In itself this sounds innocuous enough. However, the same source tells us that the adjectival form "conventional" can sound innocuous:
"Based in or in accordance with general agreement; customary."
Or not so innocuous:
"Devoted to or bound by conventions to the point of artificiality; ceremonious."
I've argued elsewhere that not only does this negative application of the adjectival form taint the word "convention," it also makes little sense in terms of application to the first version of a convention, when it would be, rather, an "invention."
For instance, Poe's detective C. Auguste Dupin is usually taken to be the first fictional amateur detective. Dupin did not arise out of nothing, of course, but if it were agreed that his only influences were, say, Poe's knowledge of then-contemporary crime detection, then the character of Dupin could not be a literary convention because (after American Heritage) his type was neither widely-used nor accepted at the time within the corpus of literature. He would be, rather, an "invention."
This then has the effect of putting a descendant of Dupin like Doyle's Sherlock Holmes behind the evaluative eight-ball: no matter how popular Holmes may be, no matter how many readers may think him a superior creation compared to Dupin, poor Sherlock will always be a mere "convention," though few would really call Doyle's creation "conventional."
For me it's plain the word "convention" is a malapropism: whoever first applied it to literature was not thinking of how literary works are made-- that is, that they are always derivative of something in their symbolic universe-- but how they are received, as to how their readers think of them as being conventions that they either enjoy in a conventional manner or reject because they find said works too conventional.
The word I prefer I take from Vladimir Propp, one of the influences on structuralism: in place of "convention" I would generally write "function." There is, I think, no automatic onus to its adjectival form "functional:" if one wants to use it with opprobrium one has to add something, as in "drably functional."
Thus a character who is acutely aware of hidden plots does not have to be confined to the "convention" of the detective: as a literary function such a character crosses generic boundaries and can be found as much in MACBETH (cf. Edgar, who disguises himself to ferret out the crafty plans of evil brother Edmund) as in the more strictly generic tales of Poe and Doyle.
I should add that when Propp formulated his term he applied it only to plot-functions, for his single influential work, MORPHOLOGY OF THE FOLK-TALE, was devoted to folk/fairy tales alone, and not to literature as a whole. However, one can usually find functional congruities in character-types, setting-types, and types of theme in all walks of literature, even in literary works that endeavor to reject any connection with genre and/or convention.
Back in this essay I spoke of functions without any great associative complexity as "simple variables," akin to narrative "furniture" that an author had to move about. Somewhat later I used "null-myth" as a term of evaluating such simple variables in terms of their lack of mythicity. "Function" is meant to be more inclusive. Say that I consider Sherlock Holmes mythic while I deem August Derleth's imitation-Holmes "Solar Pons" to be null-mythic. That does not mean that I might not be amused in some way by a Solar Pons tale, depending on how well the author presents his material on the purely kinetic level. But I would not expect the level of associative complexity that makes the Sherlock stories generally more appealing.
Both Holmes and Pons stories share functions that their respective authors did not "invent." The Holmes stories, because of their added associative qualities, may be said to be "super-functional" in that author Doyle forges more felicitous associative connections within the literary elements of his tales than Derleth does. But Doyle doesn't escape the need for narrative functionality.
Indeed, even the most avant-garde artists can't do without some representational functionality. If James Joyce wants a character to go through a door, and doesn't have it in mind to discourse on the doorness of doors everywhere, then the door through which the Joyce character steps will be no less functional-- or even "conventional"-- than the one through which a Poe or Holmes character steps.
And that's why the convention of misapplying the word "convention" is functionally wrong.