Punctualization: Law and Greimas

There's a great little summary of Actor Network Theory by John Law published in a journal called Systems Practice (which henceforth will be my name for anything I'm up to at the moment: ordering a pizza = systems practice), titled "Notes on the Theory of the Actor Network." I want to turn to it not only because we are turning to Reassembling the Social, but also because he gives us another way to think of the black box, which links things to certain problems the SR and OOP people are trying to figure out. Latour in Science in Action basically has two operations for the black box--that of opening it and closing it. But Law says that this box--or actant--can also be described as a "punctualization" of the network which supports it, or of those transformations or translations of the actant that opening the box makes it possible to trace and retrace. The terminological shift here cues up a host of basic issues that Latour often seems to want to pass by--which is understandable, because bringing them up distracts us from tracing and retracing--but which for me are quite crucial. Law sums these issues up explicitly in a question: "Why is it that the networks which make up the actor come to be deleted, or concealed from view?" (385). In other words, the issue is one of manifestation--and I use this word because it is that of Greimas, who we will turn to in a moment. The box suddenly is less functional, all of a sudden, and becomes more or less apparent. Obviously in an open box you see more--and this is why Latour uses this figure. But your focus is then on what's inside (the other actants), and not on the status, as it were, of that box and how it has altered. Thus Law's question seems to me important.

We sketched out punctualization above, but here's Law's full description of it:

All phenomena are the effect or the product of heterogeneous networks. But in practice we do not cope with endless network ramification. Indeed, much of the time we are not even in a position to detect network complexities. So what is happening? The answer is that if a network acts as a single block, then it disappears, to be replaced by the action itself and the seemingly simple author of that action. At the same time, the way in which the effect is generated is also effaced: for the time being it is neither visible, nor relevant. So it is that something much simpler--a working television, a well-managed bank or a healthy body--comes, for a time, to mask the networks that produce it.


Actor network theorists sometimes talk of such precarious simplificatory effects as punctualizations, and they certainly index an important feature of the networks of the social. Thus, I noted earlier that I refuse an analytical distinction between the macro- and the microsocial. On the other hand, I also noted that some network patterns run wide and deep--that they are much more generally performed than others. Here is the connection: network patterns that are widely performed are often those that can be punctualized. This is because they are network packages--routines--that can, if precariously, be more or less taken for granted in the process of heterogeneous engineering (385).

This is a lot more sociological in orientation, perhaps, than anything in Latour: we don't find much mention of "routines" in quite this way, and we certainly don't get much deconcealing, revealing in Latour, though there is some detective work in Aramis (tellingly, though, it ends up as a detective story without a murderer).

Why? Latour would rather focus on what this deconcealing gives you, to which you are, after you have opened the box or de-punctualized the punctuated, immanent.

This goes back to particular problems having to deal with power and force that I raised last time: Latour sometimes traces actants on the level of power, and other times decomposes them into forces--which makes them actants proper. If there is a bit of confusion, it is for the reason that Law here gives us in these sociological terms: sometimes actants wield a lot of resources (other actants), and so can mobilize them all quicker, such that we see them as punctuations. His example is a corporation or a government: we will suddenly be referring to these actants, rather than to the doers that, according to another view, make these things up.

Now, what is crucial is that for Law, we are right to refer to the actant in this way, or at least are somewhat right. We are recognizing, through our punctuation of the network, that there is a network of such a size behind the actant.

Now, this is for the reasons Latour also gives--and we note how he couches this less in the sphere of routines: the actant is shaping reality itself by recruiting allies. But this takes on a whole new meaning when aligned to what Law is saying: when we talk about powers, or about closed boxes, or about punctuated networks--all three here mean the same thing--we are talking about an actant that has so shaped reality that in referring to it that it seems natural it would be sustained by so many forces, so many of what is inside the box, or the wide and deep network itself. Or rather, that seeming naturalness which allows us to skip over the network in referring to the punctuated is precisely the amount of reality it has achieved through the network.

So, when Latour is talking about powers, or punctuated networks, he is referring to reality as any of us do. That is, it isn't as if he has to switch gears and enter the level of forces only in order to refer to the networks, or actually go about depunctualizing the punctuated in order to refer to that network that is punctuated--though he would encourage us all the time to do so.

But what Latour always stresses is that the situation is the same in reverse: when he refers to forces, he's referring to reality as any of us do, as well. And this is what, I think, confuses me occasionally, because it skips over this whole level where the problem of manifestation is actually opposed to the immanence to the network that ANT gives you.

And it is opposed. We're only making clear here that this opposition is not total. The question then becomes: what is the precise relationship of that manifestation of the network to its immanence with respect to the theorist?

This, however, was long ago answered by A.J. Greimas--who I turn to because, of course, he was fundamental in the development of ANT though not (as far as I've seen) very much discussed (Latour adapts and cites often, but, like Law, never quite explicates or interprets--either because it is all to patently Greimassian, or because we don't want to rehearse the differences)--whose opposition this is, in effect. I say "in effect" because the notion of immanence in this sense was originally formulated by Louis Hjelmslev, in his dense Prolegomena to a Theory of Language, though it has since acquired many valences which may be relevant (mostly through Deleuze). Hjelmslev was saying essentially that an immanent linguistics (glossematics) would not define itself with respect to transcendent elements like physiology or psychology: it would disregard the possibility that the only phonemes produced, say, could be those the human body was shaped to make, and would therefore remain founded only on its own properly linguistic presuppositions, remaining, always, close to them. It's important to note that while this is a structuralist move, this produces a slightly different sense of structure than that of Saussure and (in a certain sense) Lévi-Strauss: structure is here not a word for something like a synchronic logic governing certain elements, but is rather a system that generates these elements out of itself--something more self-consistent. It's no mistake that Hejlmslev was thus able to posit different and more subtle layers of signification with which structuralist analysis then could proceed (indeed, without Hjelmslev, we don't get anything like the discourse of Benveniste): his notion of structure forced him to stop applying the same old categories over and over, and move on to the generation of new terms, out of the old.

Now, Greimas takes over this term and interestingly opposes it to manifestation by saying each is the other's contrary (in the logical sense: that is, they are not in a relationship of contradiction). But what is immanent and what manifests? For Greimas, the structure itself: the immanent level is the one where the deep structures of language lie, and remain self-defined (like the semiotic square itself, which inter-defines its elements), and the manifest is the surface layer, where language moves towards its articulation and encounters the constraints of medium. Obviously, this is a bit weird, given what Hjelmslev meant by "immanence." But this is how Greimas believes we should take the term: "the terms manifestation vs. immanence are borrowed from Hjelmslev, but they can be compared profitably with the categories surface vs. deep in linguistics, manifest vs. latent in psychoanalysis, phenomenal vs. noumenal in philosophy, etc." as he says in "Towards a Theory of Modalities," (in On Meaning, 125) but also many times more elsewhere. One sees that transcendence, to which Hjelmslev opposed immanence, has to become something more like expression itself: thus, manifestation involves not moving beyond what is self-defined, but something more like the subjection of the "structural trunk" to the "specific requirements of the linguistic substance through which it [the trunk] is expressed" ("Elements of a Narrative Grammar," in On Meaning, 64), or the production of closure (as Greimas calls it), the lopping off of the pure possibilities of the deep structure.

So when one moves backwards, from the manifest to the immanent, as we are doing in ANT, what is one doing? The short answer returns us to square (or, rather, box) one, since obviously to move in the other direction than closure would be... opening. But because, as we saw, the manifest and the immanent are contrary, and not contradictory, this move backward can be seen as a more gradual return to the deep structure, the elementary units, out of which the more complicated manifested units are made.

Except in Latour this move back is not towards a deeper structure--and this problematizes everything once more, and confuses the stability of our opposition of the deep and fundamental and the simple qua immanent to the transcending that manifests. Latour tries to clarify all this at the end of We Have Never Been Modern, where he talks about immanent transcendence--one can think of this as his rewriting of Greimas. I'll try and work out this situation in another post. Nevertheless, I wanted to show that the old Greimassian distinction has some relevance, in that it provides the framework in which to think of this punctualization of the network: if we conceive of the relation between punctualized and the network as relation of the manifest to the immanent, or punctualization as the manifestation as the immanent, and conceive this relation correctly, we end up with a less outright contradictory opposition which needs to be overcome. This means that when we become immanent with the network, or de-punctualize the punctualization, we are engaged in an operation that isn't overcoming a massive gap so much as bringing the two levels into a relationship of contrariness. This rehearses, then, on a more technical level, what we were getting at before when we said talking about powers is, in a way, also acknowledging the size of their forces without specifying them.

One way to think about what Latour then does with this Greimassian situation--if I can just sketch out what happens through the massive reversal the desire to open produces--is to look at all of this coming to bear on the famous narrative grammar, and say that Latour narrativizes the deep structures, rather than the other way around. That is, Greimas thought the simplicity and minimal character of the deep structure tended towards narrativization (or the generation of combinations that had to be cut off in closure), and that we always could see, as it were, these simpler structures behind the manifestation (which is what makes possible his amazing narrative grammar). Thus, in Greimas, the actant itself is a simplification or concentration of a complicated textual surface, full of various actors. Latour--and Harman--reverse this: we see the actors (which can also be thought of as accidents or properties of objects) narrativize actants, which are much more complicated, or which do not become more simple. However, we are still, in this maneuver, becoming immanent--the network is self-articulating and self-consistent in these particular ways.

One way to describe many of my objections to Latour is not that manifestation has to work the other way (I like this about both Latour and, even more, about Harman), but that this reversal should turn the relationship of contrariness into a relationship of contradiction (or rather, there should be some similar sort of closure for his type of opening up). This, however, goes against (or maybe it merely refines? I think it refines) my point saying that the great thing about Latour is moving away from a logic of otherness and towards a logic of more.

(On that last point, I will show in my next post on "Morality and Technology: The End of the Means" that Latour himself isn't always so interested in this move, and is quite happy to talk about otherness--further complicating all of this! I think Harman will be clarifying it a bit though for all of us.)