The question "What is literature?" is not, like "What is hematite?" asked out of ignorance. It is a question of interest only to those who already have a sense of the extension of the concept and who want, for whatever reason, to think about the defining or differential qualities of the phenomena to which, as they know perfectly well, the term is generally applied. In attempting to respond, one can talk about what literature does, how it functions in this or that society or institutional context, or one can inquire whether there are properties that literary works share and features that distinguish literature from other cultural objects or activities. The first approach can generate much interesting discussion of the role of literature in establishing or contesting a national culture, in giving concrete, vivid expression to moral, ethical, and developmental scenarios, in teaching disinterested appreciation, in establishing bourgeois hegemony, and so on. Literature has been given diametrically opposed functions—a set of stories that seduce readers into accepting the hierarchical structures of society, and a practice where ideology is challenged or subverted—but unless the functioning of literature is described in rather vacuous terms, there is not likely to be a single function that all literary works perform, and as soon as the functions or effects are described with enough specificity to become pertinent and interesting, one finds that each of these functions (constituting a nation, contesting ideology) can also be performed by nonliterary discourses.
Adopting the second approach and trying to identify the defining features of works deemed literary leads to discussion of important characteristics of literary works, such as their fictionality, their noninstrumental use of language, their high degree of organization that extends to levels and to linguistic features usually regarded as transparent, their dependent yet transformative relation to other texts regarded as literary; but, again, each of these qualities is likely also to be shared with works not usually regarded as literature. One of the major lessons of theory has been that literariness is not confined to literature but can be studied in historical narratives, philosophical texts, and rhetorical and cultural practices of very different sorts. Moreover, for many works, it does not seem to be [End Page 229] objective properties that make them literature but rather the fact that they are read in certain ways, placed in the cultural framework of literature, subject to particular sorts of attention. And once one begins to think of the cultural or interpretive frame as crucial to literature and widens the historical scope of one's inquiry beyond a particular society or historical period, it is hard not to reach the deeply unsatisfying conclusion that literature is whatever is treated as literature by a given society. Just as weeds are not defined by objective properties but by a culturally and historically variable framework—weeds are plants that are not wanted in the lawn or garden—so literature may be the name of a variable cultural function rather than a class defined by distinctive properties of language. Though at one level this may be true, it is not at all what one wants as an answer to the question, "What is literature?"
In his 1973 article for New Literary History, "The Notion of Literature," Tzvetan Todorov speaks of functional and structural definitions of literature and, running through some of the possibilities, concludes that "whether or not the functional notion of literature is legitimate, the structural notion definitely is not."1 There are no defining features that distinguish literature from other discourses, and theorists' failure to identify the "specific difference" (12) that characterizes literature leads him to wonder, in conclusion, whether it could be "that literature does not exist" (12).
For any reader of Wittgenstein, such a conclusion seems both naïve and premature. Garry Hagberg, pursuing the Wittgensteinian notion of family resemblance, overlapping features that make a class recognizable, even though there is no one feature shared by all members of the class, takes up the challenge. Refusing to accept Todorov's implied conclusion, that literature does not exist if there are no essential features that distinguish works of literature from other works, Hagberg asks what are "some of...