Although the previous two chapters seek to define the statement, they actually describe a certain kind of function that governs statements: the enunciative function. The statement itself has no consistent linguistic unit, but varies according to its place in the enunciative field. What does it mean, then, to 'describe' a statement, and how does such description fit with the analysis of discursive formations (outlined in Part II)?
Three 'tasks' are involved in the description of statements. First, the vocabulary used in the description must be fixed in accordance with the discussions of the last few chapters. We can call any group of signs that is produced on the basis of language a 'linguistic performance,' and the actual act that produces that group of signs materially a 'formulation.' We can call the units of meaning that grammar recognizes in a series of signs a 'sentence' or 'proposition.' And finally, distinct from these, we can use the term 'statement' to designate the 'modality of existence' of that series of signs. Discourse, as Foucault will show, is comprised of 'a group of sequences of signs, insofar as they are statements.' Thus, a given discourse can now be provisionally defined as 'the group of statements that belong to a single system of formation.'
Second, the statement cannot be described in and of itself (like a sentence or proposition), apart from its associated field; it is always 'an element in a field of coexistence.' But, although the statement is not immediately visible in all its defining relations, neither is it 'hidden' in the way a secret or unspoken meaning might be hidden in a verbal performance. 'The statement is neither visible nor hidden.' The description of statements is less concerned with the (more or less hidden) signifying meaning of statements than with their conditions of existence, perseverance, and disappearance. There may, for any statement, be something 'unsaid,' a 'lack,' but this unsaid is always defined by 'exclusions, limits, or gaps' that are part of the conditions for the emergence of a given statement rather than a silent meaning somehow built into the statement itself.
Third, the statement, though it excludes any inherent, hidden meaning, is also not entirely visible. This is partly because the level of the statement is simply the level of existence that enables sentences and propositions to be analyzed; thus, it is almost too basic too see (it has the 'quasi-invisibility of the “there is”'). The statement is also hard to see because it entails a very specific and unusual approach to language. Whereas language, which always exists for something outside itself, is always missing something, always 'hollowed by absence,' the statement is language at the level of 'existence itself.' Studying a statement, we don't follow its linguistic pointing to something else, but rather the condition of its particular existence, its given-ness. The statement is quasi-invisible because analytic methods like grammar depend on it as their basic element, while their conclusions lie on a level apart from the statement. Traditional methods never see the level of the statement because it 'constitutes the element of their possibility.'
How does this description of the statement fit into the larger analysis of discursive formations? Foucault offers the argument that he is not constructing a true 'theory,' with various elements deduced from more basic ones, but rather seeking to establish the possibility of a certain mode of description. Statements and discursive formations fit together as various levels of this descriptive analysis, rather than as logically deduced elements in a unified theory of language. Discursive formations are groups of verbal performances linked together strictly at the level of the statement (not the level of the grammatical sentence, the logical proposition, or the psychology of the formulator). Since statements relate to each other through the enunciative function, the discursive formation can be thought of as 'the general enunciative system that governs a group of verbal performances.' The four kinds of analysis in the description of a discursive formation (formation of objects, formation of subject-position, formation of concepts, and formation of strategic choices) each mark one of the ways in which the enunciative function operates.
Four propositions sum up this mode of description. First, the level of the statement and that of the discursive formation are correlative; for now, one leads to the other. A formal theory, however, would eventually have to establish an order of analysis. Second, the statement is not an element in a discursive formation in the sense that the formation is constructed of interchangeable units that form a whole. The laws that govern a statement are its existence in the discursive formation; the identity of the statement is inseparable from the laws that govern the discursive formation. Third, a discourse does not simply exist as a unity at one point in an otherwise chaotic span of historical time. A discourse is itself historical 'from beginning to end,' and is 'a unity and discontinuity in history itself.' Fourth, 'discursive practice' can now be defined more precisely, not as the expression of an individual or as a set of logical laws, but rather as the specific 'conditions of operation of the enunciative function.'