This elderly and baggy-fleshed word has caused no end of trouble. Originally, it meant grammar. Litteratura was the translation into Latin of the Greek grammatice. According to Curtius, the original culprit was Quintilian (II.1.4), who divided the study of grammar into two parts: “correct speech and the interpretation of the poets.”
On this conception, the study of literature belonged to the trivium. Most English professors are annoyed when a new acquaintance, learning what they do for a living, says, “Oops, I’d better watch my grammar!” “I teach literature,” the professors growl. The chuckling and self-effacing acquaintance has the better instinct for the history of their discipline.
In traditional study of the bible, the grammatical sense is the first level—the entry level—of its fourfold interpretation. (The commonplace that the bible contained “four senses” of meaning was the starting point of biblical interpretation from Augustine through the Middle Ages.) Since the Masoretic text is not punctuated, and since Hebrew has no capital letters, the syntax had to be reconstructed in even the clearest cases.
Even now, the effort to construe the text’s grammar remains a profitable place to begin the study of the bible, especially since the received text has become so familiar to most students that they unconsciously “read” its traditional interpretation rather than its literal meaning. Literalism—not the religious doctrine, but the unsophisticated insistence upon bare grammar—is defamiliarizing.
In the medieval trivium, however, grammar did not include the study of morphology and syntax; it was what would now be called prescriptive grammar. Correct speech, in Quintilian’s phrase. Poets were interpreted and valued, then, as ethical influences. Hence the term author, since the authors were considered sources of moral authority. Medieval literary study, the province of grammarians, included the formal devices of poetry, but did not stop there.
These were the assumptions behind literary study for a very long time, as neatly expressed by Sidney:
Good form and good influence were what students of literature were taught to look for in literary texts. (“Political correctness” is a reversion to an earlier educational theory.) It was not until the Victorians that bonnes lettres yielded the field to belles lettres.
The very idea of literature once entailed literary value, and still does (if only because you and I have not world enough and time to read it all). Mischief was introduced, though, when beauty replaced good as the value of values. The problem is not with the concept of literary value, but with restricting value to beauty, which restricts the kind of writing that can exhibit it.
Literature is just the writing that arouses the impulse to preserve it and pass it on. (I call that the “canonical impulse.” Canons are inseparable from literature. To call something literature is to start a canon.) “When an inability to stay interested in Sappho lasted longer than the parchment she was copied on,” Hugh Kenner says, “the poems of Sappho were lost.” There are many reasons to keep something from being lost, however.
These many reasons cannot be contained by a list of genres, no matter how long it is extended; nor by distinguishing fiction from non-fiction (because there are whole literatures, of which Jewish literature is only one, to which this distinction is an utter stranger); nor by “privileged criteria” like sublimity or irony or artistry or “stylistic range” or “bravura performance” or anything else that can be humanly imagined (because exceptions to the rule will immediately suggest themselves).
Literature is simply good writing—where “good” has, by definition, no fixed definition.
 E. R. Curtius, European Literature and the Latin Middle Ages , trans. Willard R. Trask (Princeton: Princeton University Press, 1973), p. 42.