In a chapter in The Social in Question, Bruno Latour summarizes the work of 19^th-century sociology Gabriel Tarde. Latour is particularly interested in the anti-structuralist import of Tarde’s social metaphysics and anthropology. Tarde writes, for instance, “In general, there is more logic in a sentence than in a talk, in a talk than in a sequence or group of talks; there is more logic in a special ritual than in a whole credo; in an article of law than in a whole code of laws, in a specific scientific theory than in the whole body of a science; there is more logic in each piece of work executed by an artisan than in the totality of his behaviour.” Applied to language, this undoes the fundamental structuralist duality of parole and langue. Tarde makes this point by observing that “People who speak, all with different accents, intonations, pitches, voices, gestures: here is the social element, the true chaos of discordant heterogeineities. But on the long run, from this confusing Babel, a few general habits will be outlined which can be formulated in grammatical laws.” There is, Latour points out, no “structure beyond or beneath speech acts.” Rather, “Tarde imagines a kind of sociolinguistics, of pragmatics absolutely opposite, in which the structure is only one of the simplified, routinized, repetitive element of one of the locutors who has managed to include his or her local tradition into the general idiom.” Tarde doesn’t demean standardization, but the effect of standardization is simply to enable speakers to generate new complexities. Tarde again: “In turn, those [grammatical laws] since they allow many more locutors to speak together, will help them to find a specific turn of phrase: yet another kind of discordances. And those laws will succeed all the more so in diversifying the minds that they will have been better fixed and more uniform.” As Latour exclaims, “Macro features are so provisional and have so little ability to rule over the occurrences that they only manage to serve as an occasion for more differences to be generated! Instead of a structure of language acting through our speech acts, the more structural elements float around in the shape of grammars, dictionary, exemplars, the more they will allow speech acts to differ from one another!” Further, for Tarde, the standardized level isn’t bigger or broader than the specific locutions: “there are no levels,” Latour says. Rather, the standardization is a simplified version of one element of the group. In Tarde’s words, “Let’s insist on this crucial truth: we are led to it when we remark that, in each of those vast regular mechanisms —the social, the vital, the stellar, the molecular— all the internal revolts that succeed in breaking them are provoked by an analogous condition: their components, soldiers of those various regiments, provisional incarnations of their laws, pertain to them by one side only, but through the other sides, they escape from the world they constitute. This world would not exist without them; but they would subsist without it. The attributes each element owes to its incorporation in its regiment do not form its entire nature; it has other leanings, other instincts coming from previous enrolments; and some which are coming from its own store, from its own proper substance, to fight against the collective power, of which it is a part, but which is only an artificial being, made only of sides and facades of beings.” Latour glosses this with another exclamation: “Extraordinary picture of a social order constantly threatened by immediate decomposition because no component is fully part of it. Every monad overspills the artificial being of any ‘superior’ order, having lent for allowing its existence only a tiny part, a facade of itself! You can enrol some sides of the monads, but you can never dominate them. Revolt, resistance, break down, conspiracy, alternative is everywhere. . . . The social is not the whole, but a part, and a fragile one at that!”