Translation as icosis as negentropy at the edge of chaos

Kobus Marais’s monograph Translation Theory and Development Studies: A Complexity Theory Approach carves out new territory in translation studies, namely what might be called translational development studies – but it also seeks to fuse that new subdiscipline with an invigorated complex-ity-theoretical framework. This article seeks to promote and advance Marais’s project by offering correctives to two areas where his own theoretical framework remains somewhat blurry – in fact, undeveloped – namely the translator’s agency and social constructivism. The article explores an emergentist theory of “icosis” (somatic plausibilization) as a solution that, like Marais’s own approach, is steeped in Peircean semeiotic.


1.
Introduction I propose in this article to stage an engagement with a single book: Kobus Marais's 2014 Routledge monograph Translation Theory and Development Studies: A Complexity Theory Approach. I do this not because I want either to attack or to promote the book, let alone to offer a fair and balanced review of it, but because I find it an important statement that should be transformative for the field of Translation Studies -and also because I find it quite problematic in one key area, and I want to dedicate the article to the sketching of a possible solution.
I want, to put that differently, to participate in the transformative effect that Marais's book should have on the field. I want to enter into a transformative dialogue.
In a sense Marais's book is three books. The first book, coterminous with Part I, is a study of complexity theory: "Toward a Philosophy of Complexity" (Chapter 1), "Emergent Semiotics" (Chapter 2), and "Developing Translation Studies" (Chapter 3). The second book consists of a single chapter, the first chapter in Part II, Chapter 4: "Translation and Development". It contains some wide-ranging summaries of the tensions within development studies, but mainly takes potshots at Western translation theories as "constructivist" -which he takes to mean that the descendants of European colonizers believe they have the power to create and shape reality. The third book, covering the rest of the monograph, Chapters 5-7, consists of a series of empirical research reports conducted by Marais and his students in South Africa, mostly mappings of what got translated by whom, and in what language pairs. Specifically, the problem I find in the book is an excluded-middle problem -a persistent, though not ubiquitous, binarization of options that both (a) takes the classic form binaries usually take in hegemonic Western thought, namely ME vs. NOT-ME, a.k.a. "the right way" (mine) vs. "the wrong way" (all those other translation scholars out there), and (b) employs the dread non distributio medii, the shunting over into opposite poles of everything messy in the middle. One might be inclined to shrug this off as par for the course, what one would normally expect in an academic treatise of this sort -except that Marais's core complexity-theoretical message in the book is that binaries are (to overdramatize slightly) enemy territory: In my view, the paradigm of simplicity is the cause of the binary thinking that dominates the reductionist paradigm. As Morin (2008, 39) argues, this paradigm can see the one and the many, but it cannot see that the one is simultaneously the many. It can see phenomena, but it cannot see, or at least it cannot theorize, the interrelatedness of all phenomena (Morin 2008, 84). Put differently, it can see parts and it can see wholes, but it cannot see the interrelationships between parts and parts and between parts and wholes. The simplicity paradigm cannot see that difference is similarity and that the universal is the particular. In short, it cannot deal with complexity, or paradox. (Marais 2014, 20) The simplicity paradigm vs. the complexity paradigm: not perhaps the best logical format to impose on an argument against binary thinking! In one sense, of course, it is clear that Marais faces an argumentative impasse: he sees all the ways in which scholars have failed to address the complexity of translation, and sees why -their/our unthinking adherence to the reductionist models that have dominated Western thought since Plato -and wants to correct the errors. This is of course a standard impulse in academic discourse: there is all this wrongness, but fortunately, finally, I am here to rescue us. How else does one justify the writing and publication of books and articles? What else is there to say, in the end, what else is worth saying, besides "they're wrong and I'm right"? And yet, awkwardly, what Marais is right about is the clutch of oversimplifications and overstabilizations attendant upon the impulse to binarize argumentation into "they're wrong and I'm right".
The fact is, I agree with both sides in this impasse: I agree that Marais is right about the need for complexity theory, and I agree that binary polemics are both a very bad way to argue for complexity theory and, in the end, utterly unavoidable.
What are avoidable, however, I want to suggest, are the specific excluded-middle pitfalls into which Marais keeps stumbling in formulating his polemics. He does not need to caricature his opponents to the extent that he repeatedly does. He does not need to reduce their methodological and theoretical preferences -in particular, their/ our constructivism and analyses of translatorial agency -to straw men in order to make his case.
What typically happens, in fact, when he resorts to these extremist strategies, is that the position he attacks comes to seem remarkably reminiscent of the position he defends. On the face of it, this is quite astonishing: his extremist caricatures of constructivism and translation scholars' explorations of the translator's agency, designed to render those orientations not just indefensible but ludicrous, seem nonetheless to apply unconsciously to his own preferred model.
The crux of the problem, as I see it, is that Marais believes he needs to accuse the other side of exaggerating human control of reality -indeed, needs to escalate those supposed exaggerations to mythic proportions. His opponents' conception of agency, he thinks, projects onto translators not only fully conscious decision-making but Heroic Agency, Super-Agency, and that is not only wrong but an expression of Western power and privilege. What drives translation, and all other forms and flows of sociality, he insists, is not sovereign rationality but the complex structure of open systems. A tidy binary: not Lockean liberal agency -King Reason at the helm of the Free White European Male -but the nonequilibrium of nonlinear dissipative systems. And his opponents' constructivism, he thinks, similarly entails a belief in the human power to create reality ex nihilo, through the sheer force of rationalist Will, and that is a phantasm born of Western colonialism: the delusion conjured up in and by "powerful societies where people have the power to construct their reality. In a postcolonial context, " he adds, "it is an open question whether people have that power" (Marais 2014, 66). Only a European or North American would harbor such delusions. Africans, never having had that power, are humbler, and so less susceptible to such ignes fatui.
On one page (144) in the book, in fact, early in Part II (the empirical study of development and translation in South Africa), while Marais continues to hammer away at the colonial delusions of the West, he also tentatively sketches out a middle ground between elitist Western agency/constructivism and an utter lack of agency/constructivism -a middle ground that remains agentive but without the kind of Western colonial elitism and activism that Marais finds objectionable in most TS work: 1 Studying development from a translation studies perspective will require more thinking on agency. The type of activist agency currently advocated in translation studies will have to be revisited. As (a part of) translation studies frees itself from its bondage to critical theory and academic activism, it will be able to see that perhaps there are many other ways of being agents than being activists. (Marais 2014, 144) That is still quite denunciatory ("frees itself from its bondage"), but at least now there is room for a middle-ground theory of agency that is not instantly shunted over into the colonial delusion of omnipotence. Note, however, that Marais does not know what that middle ground might consist of: it "will require more thinking, " he says. "The type of activist agency currently advocated in translation studies will have to be revisited. " There is a potential content to be filled in later -perhaps. Lower down on 1 In some places Marais finds that "the intricate relationship [between agent and system] has been lacking in translation studies" (2014, 44; emphasis added); in others, however, he finds that an impressive number of major translation scholars are "interested in the relationship between translatorial action and social systems, and their interest goes both ways, that is, how the agent influences the system and how the system influences the agent" (Marais 2014, 90 the same page he offers a fuller working out of this two-pronged approach (attack the wrong way, hope for some future exploration of the right way): I thus contend that the focus on agency in translation studies is part of a Western analysis of reality. You can only contribute if you are actively for or against something. It also rests on a very strong belief that your actions matter and that you are in control of history and nature, that is, humanism. Nonlinear systems theory relativizes the importance of human agency. The outcome of your input cannot necessarily be predicted.
I am not arguing that one should forego the notion of agency. What I suggest is that we look for other modes of agency, that is, translation that serves or translation that builds. These forms of motivation for action are also agentive in nature. What I am trying to say is that agency in the critical theory definition of the word is not necessarily the only kind of agency contributing to the construction of social reality. The typical anonymous, voiceless, invisible translator slaving away in a stuffy little office, translating boring municipal regulation after regulation, is contributing as much if not more to the construction of social reality than the verbose literary translator who performs an aggressive feminine translation of a literary classic. Western notions of high visibility, branding, and status should not be the only ones defining the agency of translators. (Marais 2014, 144) That "typical anonymous, voiceless, invisible translator slaving away in a stuffy little office, translating boring municipal regulation after regulation", is here tentatively assigned not only agency but the kind of constructivist agency that makes her or him a contributor to "the construction of social reality" -precisely what Marais elsewhere in the book, by attacking constructivism tout court as an egregious colonial fantasy of agentive omnipotence, seems to deny. About the middle ground that he seems to be speculatively theorizing, or at least positing, he says only: "The kind of agency involved in these actions needs to be thought about, and to refer to Latour (2007), agency is the one thing we know virtually nothing about" (Marais 2014, 144).
Offering a tentative suggestion for that middle ground is thus my purpose in this article. My brief is that, because Marais does not know how to fill in the gap between his binary extremes, he sets a marker for that gap, with every intent of coming back to fill it later, but in the meantime keeps falling, as if lured in by some Western fatal attractor, into the right-wrong/up-down rhetoric of hierarchical binarization -and, therefore, that he could use some help.
One way of thematizing Marais's polemic, as I have suggested, is as the (former) empire writing back to the center (see Ashcroft et al., 2002). Another way might be as an historical mythomachy between the Enlightenment and Romanticism/Idealism, say, between Locke's Aristotle and Hegel's Aristotle. What Marais really needs in Part I of his book, I suggest, is a post-Kantian complexity-theoretical explanation of the human agency that constructs reality anywhere and everywhere in the world, so that he does not need to rely rhetorically on the heaping of abuse on the Enlightenment/colonial Heroic Super-Agency model that he problematically attributes to his Translation Studies opponents. Here and there he seems to come close to stumbling upon such an explanation -this is the sense in which the positions he attacks uncannily adumbrate the positions he defends -but he pointedly, and to my mind self-defeatingly, sidelines the social and affective neuroscience that might help him theorize that approach in useful ways.
My version of that neuroscience is what I call icosis, built hegelizingly out of Aristotle's observation that, given a choice between a story that is true but implausible and a story that is plausible but untrue, we tend to prefer the latter, because plausibility is a sign that the story has been vetted by the community. 2 "Things that are plausible" in Aristotle's Attic Greek are ta eikota; by Latinizing eikos "plausible" as icos-, I derive the coinage "icosis" for the communal process of "plausibilization". Icosis in my theorization is the fourth stage of somatic theory, following (1) Antonio Damasio's somatic markers, which stabilize what we have learned from experience by reminding us of those lessons somatically (Descartes' Error); (2) what I call the somatic or somatomimetic transfer, also borrowed from Damasio (Feeling), where the stabilizing force of somatic markers is circulated from body to body through the mimetic power of the mirror neurons; and (3) the somatic exchange, in which the dyadic somatic transfer is almost simultaneously reticulated throughout an entire group, so that everyone feels, say, the "same" collective approval or disapproval of a given action or attitude. Through that (1-2-3) group somaticization process, then, (4) group norms are socio-affectively stabilized and plausibilized ("icotized") as truths, realities, stable identities. Such icotic stabilizations and plausibilizations are channels of human agency, and they have the effect of constructing reality -but they are complex dissipative systems that are out of equilibrium and therefore always susceptible to symmetry-breaking events, and they mostly operate so far below the level of conscious awareness that they usually seem mysterious to us, like the operation of some nonhuman "force" like God or Truth that simply imposes objective reality on us. 3 I offer Marais icosis, which maps the middle he excludes, as a solution to his binarization problem: it is a human constructivist agency that is also an Aristotelian/post-Kantian complex adaptive system characterized, as Marais would say, summarizing John Holland, by (1) aggregation ("the complex, large-scale behavior that emerges from the interactions of less complex agents" [33]), (2) nonlinearity, (3) flow ("the resource thus flows from node to node via a connector with the nodes acting as agents and the connectors as possible interactions" [33]), and (4) diversity ("one finds parts of different nature or agents of different nature" [34]). I will return in the Conclusion to consider the Deleuzian implications of this model. Note, however, what I am not saying: my brief is not that Marais is wrong and I am right. My brief is rather that Marais is right about almost everything, and wrong about this one tiny detail. My correction is dwarfed by his rightness, but also participates admiringly in his project by helping him move past this one self-limiting argumentative strategy. That arguably does not quite rescue me from the binarizing implications of the "you're wrong and I'm right" rhetoric of academic discourse -I'm still correcting Marais -but my correction serves to bring Marais's complexity theory of translation into better and stronger alignment with complexity theory, and so strengthens his argument. It is corrective support for an argument that is itself a complex adaptive system that by definition is out of equilibrium. My support is an attempt to buttress that equilibrium. Because it is an intensification of the argumentative system's negentropy at the edge of chaos, however, and because negentropy is not a state but a homeostatic sorting conduit that does not just export chaos and import order to keep its own entropy low but also imports what it guesses might be the right quantity of chaos in order to keep its entropy hopping, the equilibrizing/organizing/structuring effort of my support for Marais's argument is not proof against chaotic collapse. So far from 3 Another word for those complex systems of human nonconscious agency in the West would be Hegel's Geist, usually translated as "spirit" or "mind, " which is actually the byproduct of human action in the aggregate. In ancient Chinese thought, those systems are typically called 天 tian "heaven" (Confucians) and 道 dao "way-making" (Daoists). Both 天 tian and 道 dao are often mistakenly deified in the popular imagination, but in the ancient texts they actually mean something like mysterious doings/forces that we do not understand and cannot control. See my Dao and Deep Ecology for discussion. seeking to establish a stable right-or-wrong binary, in other words, I am pursuing diversity along the nonlinear flows of Marais's argument, seeking to aggregate the interactions of his less complex arguments into enhanced equilibrium on a higher level.

Agency
The two points on which I offer correctives to Marais's argument here, then, are agency and constructivism. The common ground undergirding both correctives is my sense that Marais wants to assign too much Enlightenment/empiricist/reason-based agency to his opponents and to claim too little Romantic/Idealist/affect-based agency for his own argument.

Systems without human agency
Let me begin by noting that his defense of complexity theory tends to be based on examples of pre-social -physical, chemical, biological -systems without human agency. This, I admit, is a somewhat tendentious claim, since Marais does everywhere stress the importance of "the mind/brain/individual as the basic level from which social phenomena emerge, itself emerging from physical, chemical, and biological substrata" (Marais 2014, 110, and see section 2.2, below). Somehow, however, whenever he describes such emergences in translation and other social phenomena, the negentropic organizing effects of human social complex adaptive systems tend to be presented rhetorically as if they proceeded non-agentively: Open systems are governed not by the second law of thermodynamics, that is, entropy, but by negentropy. This means that they do not decay into chaos but maintain their organization by interacting with their environment. Whereas entropy tends toward dissipating the differences on which structure and order are built, negentropy tends to lead to the maintenance of difference (M. Taylor, 2001, 119-21).
Negentropy is the reason why two people cannot produce the same translation. Human beings are not closed systems; thus, their thoughts and interpretations cannot be predicated based on initial conditions. Being open systems, the same stimuli, such as a text, could give rise to widely differing interpretations and thus translations because the initial conditions in two brains can never be the same. Translation is thus not a process of which one can predict the outcome; translational action can only produce probable outcomes. One cannot predict how two transla-tors will translate or what effects a translation would have in a society. The laws of prediction have to be replaced by laws of probability. In this respect, a field such as translation studies has much to learn from the conceptualization of complexity theorists. (Marais 2014, 32-3) The problem there, I suggest, is that in Marais's account the "initial conditions" out of which translations emerge are not agentive minds but the physical, chemical, and biological substrata of minds, namely brains: "the same stimuli, such as a text, could give rise to widely differing interpretations and thus translations because the initial conditions in two brains can never be the same. " Human "thoughts and interpretations" arise unpredictably out of those cerebral initial conditions; and translations, rather than emerging unpredictably out of divergent mental agencies as initial conditions, would appear to be among those "thoughts and interpretations". In other words, the differences between translations have to do not with the different translators' divergent kinesthetic-becoming-affective-becoming-conative(-becoming-cognitive) experiences and experientially guided inclinations, but with the different initial neural conditions in their brains. Another binary.
Not only that: despite what Marais notes about the power of mind to effect downward causation on reality (Marais 2014, 67), presumably including the physical reality of the brain, there is apparently no downward causation here. All causation is upward, from the brain to thoughts/interpretations/translations. This is rather surprising in light of the "fact" -or rather, Marais's rhetorical framing -that brain-based "negentropy is the reason why two people cannot produce the same translation": negentropy, after all, the organizational sorting that systems do "at the edge of chaos" (Stuart Kauffman's pithy phrase from At Home in the Universe), is the maximization of stability and equilibrium in a system. By rights it should be the reason why two people can produce similar translations despite divergent brain structures! If it is because of negentropy that the linguistic, cultural, and professional skill-sets of translators "do not decay into chaos but maintain their organization by interacting with their environment", presumably those negentropic interactions are in large part mental interactions with the linguistic, cultural, and professional environment. The professional negentropy of translators obviously has a lot to do with language-learning, training, mastery of marketplace norms, and so on, which, as Daniel Simeoni noted back in the late 1990s, shapes/structures/stabilizes (exerts partial/imperfect downward causation on) both the linguistic, cultural, and professional environment and the brain shaped by that environment: "Translators govern norms as much as their behavior is governed by them" (Simeoni 1998, 24). The fact that what is stabilized, organized, equilibrized in these interactions tends to be partly idiosyncratic -or what back in The Translator's Turn I dubbed "idiosomatic" -does justify Marais's observation that "negentropy tends to lead to the maintenance of difference"; but that tendency is the unstable and unpredictable result of interactions not just with stabilizing forces in the professional environment, but with entropic forces as well, including divergent brain structures, no doubt, but also transient states of body and mind such as alertness and exhaustion, hunger and thirst and a full bladder, and distracting neural excitations from loud jarring noises, physical commotions, and unresolved emotional issues. The symmetry-breaking distractions of a new love affair are very different from the symmetry-breaking distractions of a crushingly bitter break-up. The translator's negentropic "sorting" of entropic forces "at the edge of chaos" is obviously an agentive effort to master the chaos, to impose sufficient "structure and order" to get the job not only done but done well -and just as obviously that effort only ever succeeds in part. Translation, like any social activity, is a dissipative system not because it is roiled with chaos but because it can never perfectly banish chaos. And while the translator, like any other social actor, is not always aware of working to banish chaos -to impose order on the internal and external environment -it would be inaccurate to deny that work "agency" on the grounds of insufficient awareness.
Marais's rhetorical inclination to deemphasize translatorial agency is also reflected, it seems to me, in his insistence that "the laws of prediction have to be replaced by laws of probability". Probability is a mathematical concept designed to measure blind chance. In the world of "laws of probability", agency is a human intervention that skews measurement -like saying: "Watch me toss this coin ten times and catch it so that I get heads at least eight tosses out of the ten. " In icotic theory, you will recall, the counterpart term is "plausibility": is it plausible that someone could catch a tossed coin accurately enough to raise the probability of heads from .5 to .8? Plausibility is a measure not of mathematical likelihood but of group normativization -how well the members of a group have been conditioned to accept collectively normativized opinions as truths and realities.
But let us think through the difference between probability and plausibility with a story -a thought-experiment.

A mini-novel
At the individual level, the probability that a subject will perform a certain action in a certain context can certainly be calculated. Imagine a married couple, which I will anonymize for gender with "spouse" and ze/zir pronouns. And so for several years now Spouse A has been keeping a log of such "negative enhancements. " At first, however, showing the count to Spouse B led to scoffing: "That's just a few isolated incidents. " So Spouse A began counting not only "negative enhancements" but "positive enhancements" and "neutral renderings" as well. To zir mind this expanded log proves zir right: out of a total count of several thousand textual passages, ze can show mathematically that the probability of a negative enhancement is .68, the probability of a positive enhancement is .13, and the probability of a neutral rendering is .19. "See?" ze presses Spouse B. "You're imposing the imprint of your own personal style on my novels!" Unfortunately, Spouse B continues to scoff: "You're just cherry-picking passages that confirm your paranoia!" But then one day, writing a paragraph about a female character who is sexually attracted to men in uniform, Spouse A unplugs zir laptop and carries it over to where Spouse B is working.
Spouse B complies. "So?" "So I'm predicting, " Spouse A says, "that you will call this character a 'slut. '" Spouse B reads through the paragraph again. "But she is a slut!" "That's your stereotyped interpretation, " Spouse A says. "Do I call her a slut?" "Not in so many words, " Spouse B admits. "You're too good a writer to be so obvious about it. " "And you're too good a translator to be so obvious about it as well, " Spouse A retorts. "But I'm still predicting you'll call her a slut. " "We'll see, " Spouse B says.
Well, you've guessed it. Just to prove zir spouse wrong, feeling irritated and aggrieved at being so predictable, Spouse B does not call the character a slut -not in that paragraph, nor anywhere else in the translation, either. That female character is portrayed throughout without attributive aggressions. Spouse B has, Spouse A feels, bent over backwards to render zir descriptions and attributions accurately, faithfully, neutrally. Everything else in the translation conforms to the usual -mathematically probabilized -pattern, within what Spouse A regards as a standard deviation. But, as an experiment, Spouse A decides to praise Spouse B only for rendering that one character without added negativity, and to leave zir probabilistic assessment of the rest of the translation unspoken.
So what do we conclude from this story? Like Spouse B, Kobus Marais too may want to conclude that I have cherry-picked the textual passages in his book that confirm my critique. I have not even compiled a quantitative log of such cases, and so could not throw statistical probabilities at him. He may even want to compile such a log to prove me wrong.
Setting that aside for a future discussion, however, the obvious conclusion of my little story is that if the "law of probability" based on a purely quantitative log of past events is a dissipative system -and of course it is, despite the closed-system implications of the word "law" -Spouse B's irritation at being thought predictable is the symmetry-breaking event that causes the system to tip. Well, causes it to tip this one time: Spouse A can of course continue to believe that the calculated probability of Spouse B's translatorial negativity will remain high, and that expectation may well be borne out in practice.
But now suppose Spouse B decides to take it further: decides not only to break the pattern and always translate passages about sexually active female characters "faithfully", without intensifying zir attributions and descriptions in ways that Spouse A calls "negative", but to extend the new pattern to other characters as well. To Spouse A's surprise -and, to be honest, mixed feelings, zir public pleasure at the more "faithful" renderings mixed with private dismay at the breaking of zir probabilistic patterns -the probability of negativity plunges alarmingly below .5, and then below .3, and seems headed for 0.
And since in this mini-novel I am the omniscient narrator, I know and can report that, despite Spouse B's repeated protests that ze is still translating exactly the same way as before, ze is actually not only changing the way ze translates but learning to like the change. A new translatorial style is emerging. Spouse A's "law of probability", which to Spouse B felt gratingly like a "law of predictability", has given rise to a new plausibilization, and thus, in my retelling, a new "law of plausibility" -or rather, perhaps, a new "epistemology of plausibility". By predicting a probability to an agent capable of affecting the probabilistic outcome, Spouse A skewed the mathematical system. Ostensibly non-agentive probability flipped over into undeniably agentive plausibility.
Spouse B decided -out of spite, at first, but then, gradually, out of more complex motivations -to prove Spouse A wrong.
Of course, despite the complex emergence of Spouse B's new stylistic plausibilizationthe creation of a new pattern ("system") out of symmetry-breaking deviation from the old -the quantitative social scientist's (or say Spouse A's) inclination to impose a new numerical "law of probability" on it could still be (at least arguably) predicated on an objectivizing empiricism, one that wanted to base probability judgments on non-agentive data.
The same would be equally true if the quantitative social scientist set out to study a population sample of (say) a thousand translators, with the same deagentizing methodology. Non-agentive probability is of course easy enough to posit even in studies of human agency: all the quant has to do is exclude affective interactions in groups from consideration. But the guidance wielded by affect could also be incorporated into this sort of quantitative "law of probability" study as well, by aggregating, say, the social groupings (stable categories) of age, gender, and social class as independent variables, and reducing "affective guidance" (as dependent variable) to stable stereotypical "mechanisms" activated variably in the different social groups: • women will tend to translate novels with more empathy, men with more hierarchical aggression (despite my Spouse A/Spouse B/ze/zir anonymizing, you probably mentally made Spouse A female and Spouse B male); • young men will tend to translate novels with more empathy, old men with more hierarchical aggression; • old upper-class men will tend to translate novels with more empathy, old working-class men with more hierarchical aggression, and so on.
"Empathy" and "aggression" as statistical artifacts -which is to say, as mechanistic reductions.
But now if we start exploring the agency that is always operative in actual human decision-making, the reductionism of a "law of probability" becomes completely inadequate -both because probability cannot account for agency, and because a law cannot account for decisions (let alone whimsical or resentful impulses) to act against pattern. At the dyadic level, those of us who have lived in committed relationships for decades know that our significant others are always capable of surprising us in radical ways. Long-term relationships, supposedly so stable as to be boring, are actually dissipative systems out of equilibrium. Symmetry-breaking events are always possible. But even in the aggregate, studying behavioral and attitudinal trends in thousands (or even dozens) of people, we know that the rough "accuracy" of stereotypical attributions to large groups is only probable because it is plausible, and only plausible because the attitudes and behaviors in question have been "plausibilized" (icotized) as group norms, and those group norms have been icotized as ontologized belief structures, all of which has a behaviorally and attitudinally aggregating effect on group members. Nonconscious or preconscious affective-becoming-conative beliefs and inclinations are shaped by the groups to which we belong -which is to say, our individual agency is shaped by collective agency, almost always without our conscious awareness, and never perfectly. Surprise is always possible, and surprise as the leading edge of emergence is also possible.

Translators with "Little Intention"/"No Particular Intent"
And Marais mostly seems to know this. The theory of translatorial agency he proffers is adapted slightly from the "Introduction" Ralph Stacey and Douglas Griffin wrote to the essay collection they edited, Complexity and the Experience of Leading Organizations, according to which "the system does not exist prior to symbolic interaction between individuals. A system is an emergent phenomenon that emerges out of the relationships between individuals" (Marais 2014, 94). Interactive agency, in other words, is the lower level out of which social systems emerge, and it should go without saying, of course, that social systems also exert downward causation on interactive agency (shaping individual identity, personality, and so on). Even more promisingly for icotic theory, Stacey and Griffin draw on George Herbert Mead's Mind, Self, and Society to argue that systems are not "real", but only take on the feeling of being real by emerging specifically "out of the bodily interactions or relationships between human beings" (Marais 2014, 94; emphasis added).
Given the deep grounding of icotic theory in shared and circulated somatic response, this resonates immediately. But is the reticulation of socioaffective impulses through groups the kind of "bodily interactions or relationships" Marais is borrowing from Stacey and Griffin? He never says. His theorization of translatorial agency remains at precisely this fairly cursory level throughout. He never quotes directly from Stacey and Griffin's "Introduction", where we do find clarification: "All we have are vast numbers of continually iterated interactions between human bodies and these are local in the sense that each of us can only interact with a limited number of others" (Stacey and Griffin 2005, 9). In other words, what Stacey and Griffin mean by "bodily interactions" is physically situated interactions: the fact that existing in corporeal form restricts our mobility. We can't be physically present in more than one place, and thus one situation, at a time.
The first problem this arouses for a reader like me is that it does not give us much to go on. How do social systems emerge out of situated human interactions? What kinds of interactions give rise to such social emergences? Is conversation enough, for example? If so, does the interaction have to be explicitly verbalized, or is body language enough? And, once we move past the 7% of human communication that is channeled through the disembodied words that appear in corpus transcriptions of those conversations, how does the embodied remaining 93% work to generate sociality? 4 My icotic theory can answer those questions. In Feeling Extended (Robinson 2013b, 16-24), in fact, I develop an early formulation of that theory out of the work of George Herbert Mead. None of that is in evidence in Stacey and Griffin, however, and Marais gives us a rather minimalist version of Stacey and Griffin.
The second problem is that I do not see much evidence that Marais was particularly interested in developing the workability and applicability of this stripped-down model of translatorial agency through his empirical study of translation and development in South Africa, in Part II. He insists in the Conclusion to Part I, in his bulleted list of "advantages of thinking about translation from a complexity perspective" (Marais 2014, 114), that his complexity theory of translation "provides a theory of agency, explaining the relationship between agents and society" (Marais 2014, 114), and he returns to reiterate that much as a theoretical foundation for Part II: "If it is true that societies emerge from the complex interactions and links between individuals (Chapters 1 and 2), and if it is true that these interactions are of a semiotic nature (Chapter 2), and if it is true that in multilingual contexts these interactions need to be facilitated by means of translation (Chapter 3), it follows that translation has a role to play in the way in which societies emerge" (Marais 2014, 120).
And I agree, this does indeed all follow. But (a) what happened to agency in all that? Whenever the issue comes up, Marais seems content to state the importance of "explaining the relationship between agents and society", but nowhere theorizes that relationship, except to say that "when you talk about agency, you are asking how individual 4 I refer here to Albert Mehrabian's so-called 7%-38%-55% rule, according to which, as his research for Silent Messages showed, three different communicative channels contribute differentially to our "liking" for a person: the words themselves contribute 7% of the effect, tone of voice 38%, and other body language (facial expression, gestures, posture, body positioning, and so on) 55%. For example, one might imagine a local official in the Free State, South Africa, where Marais lives and conducted his research for this book, interpreting a visitor's English speech into Sesotho or Setswana, with a high degree of verbal accuracy (7%), but with a tight tonality (38%) that is arguably ambiguous but at the very least seems to be signalling some degree of distancing, combined with frequently rolled eyes (55%), turned away from the speaker, so that (a) only the audience gets the strongly negative message and (b) the interpreting official has plausible deniability if someone accuses him or her of prejudicing the audience against the speaker's words. actions cause other individual or social actions, which is a question concerning the influence of the agent on social reality, that is, on other agents" (Marais 2014, 89). Again, a relationship -a causal relationship, this time, though not necessarily an intentional one -without an accompanying theory. And (b) where is situated embodiment? The jump to semiosis is important, I agree again; but semiosis also lends itself, if one is not careful, to abstraction, disembodiment, and I do think that Marais could have done a lot more to remind us of the affective, conative, and kinesthetic aspects of semiosisof Peirce's emotional, energetic, and logical interpretants, for example.
Even more important, (c) how does this minimalist theory of translatorial agency as translation playing a semiotic role in the emergence of societies explain Marais's specific South African case studies of translation in development contexts in Part II? And above all, (d) how do the case studies complicate and develop the bare-bones theory of translatorial agency stated quickly on two pages in Part I? I think it is not an overly harsh assessment of Marais's book that he does not do much with this problematic. The project that I tentatively sketch out in those two previous paragraphs is a massive one -one that I am contributing to, in a minuscule way, in this article, and may contribute to more significantly in the near future, but one that will ultimately require the efforts of many more translation scholars than Marais and myself.
For now, though, note that Marais repeatedly tends to encapsulate his empirical findings in Part II with vague talk of a single emblematic type of translator -and that his remarks on that emblem do not reflect well on his theoretical engagement with his qualitative data. There is, apparently, a whole class of translators -the majority worldwide, Marais claims -who have no conscious desire to leave their personal imprint on their translations, and so, apparently (though he hedges on this), have no agency: 1. "This does not solve the question as to how to account for the large number of texts translated everyday [sic] by anonymous 'agents of translation' who may have very little intention concerning agency beyond earning a living. " (Marais 2014, 90) 2. "Current theories of agency in translation … cannot theorize the majority of translation activity in which the translator has no particular intent, other than making money of [sic: or?] having to do a job. " (Marais 2014, 95) 3. "The typical anonymous, voiceless, invisible translator slaving away in a stuffy little office, translating boring municipal regulation after regulation…" (Marais 2014, 144) "Little intention concerning agency beyond earning a living" in 1 and "no particular intent, other than making money" in 2 seem to be roughly synonymous, and I may be reading Marais's implicatures incorrectly, but he seems to be suggesting that the minimal "intent[ion]" aimed exclusively at "earning a living"/"making money" signals a more or less complete lack of translatorial agency. "Agency in translation" seems to be defined as the translator's intent(ion) to impose a personal style on her or his translations; Marais seems to be implicating here that most translators have no such intent(ion), and are in it just for the money.
Indeed, in 1 the scare quotes around "agents of translation" would appear to suggest that these translators are not really translatorial agents at all. They are at most economic agents, working human parts in the biocapitalist machine whose translatorial performances are perfectly deagentized, mechanized. They are also "anonymous" in both 1 and 3, and "voiceless" and "invisible" in 3, all of which, combined with "intentless" in 1 and 2, seems to be a jab at (a) Western theories of the translator's heroic visibility, which Marais explicitly equates with misguided colonial theories of translatorial agency, and (therefore?) (b) a reversion to the binary opposite of that Heroic Agency visibility, namely no (translatorial) agency at all.
"Majority" in 2 also sounds very quantitative, more strongly redolent of a depersonalized/disembodied "law of probability" than of social plausibilizations effected through affective-becoming-conative "bodily interactions between humans" (Marais 2014, 94) or "the bodily ways in which the anthropos interacts with both other anthropoi and the environment" (Marais 2014, 110). And while this is more a suspicion than anything else, it also seems to me that the dismissal of (1/2) "intent[ion]" in cases where translators are just mechanically doing a job for money would appear to deny the relevance of socially constitutive "bodily interactions between humans" altogether.
But one wonders: a. How exactly does Marais know when translators have "little intention concerning agency beyond earning a living" or lack "a particular intent, other than making money [or] having to do a job"? Is he asking them, and believing them when they say they do not care about the text, are just doing it for the money? Is he seeking confirmation for his inclination to believe those replies by reading body language? Or is he just imposing easy reductionist stereotypes?
b. If he means that there is a kind of economic quasi-agency or sub-agency that drives a translator to "do a job" in order to "make money, ", which is somehow categorically (stably, non-complexly) different from the translatorial Heroic Agency that drives a translator to leave a personal interpretive/stylistic imprint on a target text, how does he know (i) whether that binary distinction is actually at work in specific translators doing specific jobs, (ii) whether the distinction is ever airtight, and (iii) whether the distinction is air-tight in "the majority of translation activity" worldwide?
c. Is the Enlightenment/colonial Heroic Agency that drives a translator to leave a personal imprint on a target text necessarily so conscious and deliberate, so definitively grounded in an explicit decision to leave an imprint, that it would be not only possible but unproblematic for each individual translator to recognize the operation of that translatorial Super-Agency in specific translation jobs, or in specific decisions (this word or that, this phrase or that, this register or that, etc.) in a given translation job, and thus, as in (b-ii), to distinguish it from the shadowy mercantile sub-agency?
i. And, conversely, does a given translator's or translator class's lack of awareness of the translatorial choices that leave a stylistic imprint therefore univocally signal a lack of Heroic Agency, or even a lack of any agency at all?
ii. Is it possible that the agency that drives a translator to leave a personal imprint on a target text is always nonconscious, and therefore only very rarely even vaguely heroic? Is it possible that all translators, even the ones who say they do not care about the job and only do it for the money, nonconsciously leave their own stylistic imprint on their translations?
My guess is that Marais might answer Question c-ii in the affirmative: yes, it is possible that translators nonconsciously leave personal imprints on their translations. After all, he notes that "a translation performed forward and backward, that is, from source to target and from target to source, will not yield a copy of the first source, because of the unidirectionality of history" (Marais 2014, 39) -and, we might unpack that last clause, because the differences between the forward-translation and the back-translation are driven not by conscious translatorial decision but by "history", which is to say, by the open-system complexity of translation.
Whether he would agree to identify those differences as "the translators leaving their personal imprints on their translations", however, is not clear. As he writes elsewhere: "In closed systems, were the initial conditions identical, that is, were two identical brains to tackle the same translation job, with the same brief, at the same time, under the same conditions, one could imagine having identical translations. However, in open systems, with the slightest difference in initial conditions, one cannot predict the outcome; that is, one could not have identical translations" (Marais 2014, 10). No need for agency: the translations differ not because they are performed by different translatorial agents, who bring different experiences of language and other people and the world to the task, but because the "brains" perform the translations "in open systems, with the slightest difference in initial conditions".
But then what would he do with translatorial agency? If he would continue to insist on answering Question c-i in the affirmative -yes, agency requires a conscious decision -his solution to the apparent discrepancy between his answers to Questions c-i and c-ii would have to be the complete depersonalization of the translator's decision-making: the irreversibility of translation is driven not by the open-system complexity of translator agency but by "the unidirectionality of history". The unpredictability of translation is driven not by the open-system complexity of translator agency but by "the slightest difference in initial conditions" in two or more open nervous systems.
My guess, in fact, is that if pushed Marais would back off from the extremism of his attack on translatorial agency, and admit that his anticolonial resistance was not so much to the possibility (indeed the omnipresent reality) of nonconscious affective-becoming-conative agency in every translatorial decision (indeed every human action), as it was to only a single egregiously exaggerated version of that agency -the one I have capitalized as Heroic Agency, a.k.a. the translator's visibility, narratoriality, and so on. If I am right, his invocation of "the majority of translation activity in which the translator has no particular intent" is not intended to deny translatorial agency, just to minimize its Enlightenment heroism -and only gets rhetorically transformed into a noncomplex binary out of Marais's understandable vexation with Western colonial privilege. It is, in other words, what we might call "backlash binarism".

The negentropic movement of information
Marais's complexity theory of translation is based on the idea, very similar to Juri Lotman's translational "two-language" model of cultural semiotics, that negentropic "sorting" (organizing, structuring) takes place at the boundaries between systems -"at the edge of chaos" not just on the periphery of a single system, but intersystemically. This "inter-ness" or "inter-ing", as Marais dubs this negentropic interfacing, for him is translation writ large (see Marais 2014, 42-5, 96-105). He writes: Open systems allow for the flow of information of various types; that is, it allows for life (Morin 2008, 10). This means that these systems do not operate according to the laws of thermodynamics, a position that assumes equilibrium in systems. Systems theory has realized that equilibrium means death for any system. It is the apparent complexity or chaos that, together with simultaneous structure, makes life possible. Both structure and change are thus paradoxically a precondition for life.
… This interaction takes place by means of a movement of information, be that symbolical, chemical, biological, or any other kind of information. This information is organized within a system so that noise is diminished and negentropy is achieved. The interesting point that I wish to highlight here, and that I expand on later, is that this "inter-ness" or "inter-ing, " this need for exchanging information between systems in order to keep them alive is the philosophical underpinning of translation. All systems need some kind of "inter-action. " … According to Latour (2007), the social, which is connected to the natural -if one has to make such a distinction -refers to links that change relationships continuously. It is a sociology of connections, but not static connections, rather connections that translate, that is, carry over or transfer, all the time. The social refers to moving relationships, in which carryings over, that is, translations, of various natures take place. Of these, linguistic carryings over are but one category of inter-ings or inter-actions. (Marais 2014, 38-9) I like that idea a lot. My only problem with it is that in social systems it is not just the "movement of information" or "exchanging information", it is the interpretation and organization of information as knowledge. Again, Marais knows this -"Scholars point out", he notes in Chapter 6, "that ICT has made it possible to turn knowledge into information in order to store and/or disseminate it" (Marais 2014, 174) -but he often forgets it. 5 It is not the "connections that translate, that is, carry over or transfer, all the time", it is the people who make the connections that translate. Information can move, and can be exchanged, without human agency. And yes, that kind of movement can be troped as translation, a term that has been used over the millennia to mean many different kinds of movement, not all of which have involved human interpretive agency. But what exactly do we gain by reducing translation to the non-agentive exchange of information between systems? 5 Chapter 6, from which I take this quotation about converting knowledge into information, also theorizes the human embodied kind of knowledge that circulates icotically, and manages normativity as well as the normative conversion of opinion into fact, as "tacit knowledge". This is an extremely important point, and one that could -and, I would argue, should -have been brought out of its confines in Chapter 6 and mobilized for the theorization of translatorial agency.
My guess, again, is that Marais did not really intend that reduction -that it was a kind of rhetorical accident, born of his excitement about complexity theory, the modern scientific version of which was originally developed for nonhuman systems. 6 Or would we want to say that submental human systems, like the synaptic movement of biochemically stored information from neuron to neuron in a nervous system, are purposive, and therefore agentive? If human agency is (thought to be) at work in the movement of information through human nervous systems, should we take that agency to inhere in the physical, chemical, and biological strata of the brain, or to be produced through the downward causation effected by mind on brain?
Border disputes like that remain to be explored. It should be clear, in any case, that the complex purposivity of nonhuman systems operates at a different evolutionary level from the normativity of human social systems -what I've called icosis. People in groups are not only working to impose negentropy, to "import" it and "store" it, but they're also organizing that work around group norms, and ontologizing those norms as "realities". It is not just probability; it is plausibility. It is not just negentropic; it is icotic.
Obviously, the normativity imposed on their members by groups of humans and other social animals is an emergent system -it emerges "upward" out of that lower-level purposivity, and partakes of some degree of similarity with the systems out of which it emerges. But "negentropy tends to lead [not only] to the maintenance of difference" (Marais 2014, 32), but also to the mobilization and management of new differencesand the question becomes, when something "new" like social normativity emerges out of the old, whether we should draw a semantic or other semiotic line through that new difference. Would we want to say that broad-based semiotic "translation" consists now only of normative inter-pretations of information as knowledge and no longer of merely purposive inter-exchanges of information? Or would we prefer, with Marais, to leave "translation" definitionally open to all forms of "inter-ing", including, say, the semiocapitalist movement of banking information through computer systems? 6 The origins of modern complex systems theory have been traced back to the political economic theory of the Scottish Enlightenment, especially perhaps Adam Smith's "invisible hand of the market, " which is ultimately "emergent" (spontaneous) in the sense that it arises out of human economic interaction but is neither carried out rationally by human plan or design nor imposed on human affairs by a deity. For discussion, see my Who Translates? (Chapter 6) and Andy Chan's (2016) article on "pushing hands" and "the invisible hand. " Because the transhuman "agency" of the market is not controlled rationally by any individual or group, however, it is sometimes supernaturalized -as it was by Adam Smith, disapprovingly -as a mystical force ("invisible hand"); the later applications of complexity theory to physical, chemical, and biological systems have tended to naturalize/scientize it as a random force. I am suggesting that Marais is drawn to the latter. (And please note that I am not remanding it to the former, namely mystical/supernatural forces.) We could go either way, it seems to me; see for example my discussion of Jon Solomon's (2014, 172-73) critique of semiocapitalist translation in Critical Translation Studies: The corporation as "source author" translates its "source code" both for the consumer as "target reader" and into the consumer as "target code"and the target code recodes the source text so that it becomes better able to address the target reader. The primal scene of translation as capitalist "growth" (reciprocal learning as a revenue-generator). It is in this sense at least (perhaps in some others as well) that Solomon charges translation with complicity in the corporate-state: "In relation to translation I would argue, in other words, that it must be considered in light of the reproduction of stateness (which is a way of producing and managing 'anthropological difference' for the sake of capital accumulation), and that it (translation) plays a crucial role in the management of the transition to a new type of world order based on the 'corporate-state'". (Robinson 2017, 148) But perhaps those neural-net computer systems ("reciprocal learning as a revenue-generator") have already been normativized and thus human-agentized by their programmers for the single all-encompassing norm of profit, as "the reproduction of ['corporate'-]stateness"?

Constructivism
Kobus Marais is not a fan of constructivism. He only ever defines it in short subordinate clauses that sound more like broad stereotypes, or caricatures, than like definitions; and he never cites, let alone critically engages, even one actual source defining or discussing constructivism. His attacks thus seem to be based on his best guess at what "constructivism" might mean, based on the fact that it has the verb "to construct" in it.
Constructivism of course is core Kantianism -the Idealist belief that we have no reliable access to the "thing in itself ", and that therefore what we take to be reality is a social construct. We do of course experience the material reality of the thing itself: yesterday I slipped and fell and bunged up my right knee. Nothing dangerous, but the lump that formed was painful for a few hours, and today I can still feel some soreness when I press on it. The rock I fell on, and the lingering pain in my knee, are "the thing in itself ". But we cannot know that material reality reliably. Our nervous systems interpret sense-data for us in coherent ways -and those ways are rendered coherent by the groups we belong to.
In Marais, by contrast, this is the sort of critique we get: 1. "I shall argue that extreme forms of constructivism philosophically negate the ecological model of reality. This view, which conceives of humanity as dominating over nature, in the fashion of fundamentalist Christianity or, for that matter, fundamentalist constructivism, is part of what causes the destruction of the universe. " (Marais 2014, 10) 2. "Reality is not ours to control, to command. It is something we cannot control and something of which we stand in awe. This is contrary to the humanistic, constructivist position. Also, in translation studies, which is currently dominated by constructivist views, one has to reconsider the notion of human control over reality. " (Marais 2014, 29) 3. "Strong versions of constructivism represent an epistemological position that is not only unecological but suited to Western conceptualizations because it is related to powerful societies where people have the power to construct their reality. In a postcolonial context, it is an open question whether people have that power. " (Marais 2014, 66) "Extreme forms" in 1, "strong forms" in 3: the implication would appear to be that there might exist "moderate" or "weak" forms of constructivism that Marais would applaud, but he mentions something like a weaker form only once in the book, and that very sketchily, in a place we have seen twice before: "The typical anonymous, voiceless, invisible translator slaving away in a stuffy little office, translating boring municipal regulation after regulation, is contributing as much if not more to the construction of social reality than the verbose literary translator who performs an aggressive feminine translation of a literary classic" (Marais 2014, 144). Put aside the aggressive anti-feminism of "aggressive feminine translation", and the aggressive attack on literary translation (or High Literary Translation Theory) in "verbose": 7 the big question is, What is that translator contributing, and how? What might the "complex" constructivism adumbrated in that sentence be like? Marais does not provide enough detail for us even to venture a guess. As a result, "constructivism" as a kind of blanket term for colonial delusions of omnipotence remains a much-abused whipping-boy throughout his argument.
And yet, on the very next page following 3, he gives us this: 4. "In other words, not only does physical reality give rise to semiosis, through the biology of the brain from which mind emerges, but through mind, semiosis is also able to exert downward causative power on reality, changing reality, creating new forms of reality. " (Marais 2014, 67) This claim, coming as it does hard on the heels of the damning association of constructivism with Western colonialism, makes me wonder: is semiosis "able to exert downward causative power on reality, changing reality, creating new forms of reality" only in the colonial centers? Or does it have that power even in postcolonial contexts?
The problem here is, on the one hand, the middle that Marais excludes between (2) "ours to command" (white Western privilege wielding conscious, deliberate, godlike agency) and (4) "semiosis is also able to exert downward causative power on reality, changing reality, creating new forms of reality" (semiosis as a mysteriously outsourced agency beyond human control). This is a stable, noncomplex binary that he is only able to construct through his extremist anticolonial middle-excluding caricature of constructivism. That would be 2. That mysterious outsourcing of semiosis that he broaches in 4 sounds more like what I know as constructivism, but Marais deagentizes it by middle-excludingly depersonalizing "semiosis". He does hint at the excluded middle between those two extremes in the adverbial phrase "through mind", which hints at what "we" do without conscious godlike control; but again, because he defers (perhaps indefinitely?) discussion of the neuroscience behind "through mind", he is in no position to complexity-theorize the simultaneously (or alternatingly?) upward and downward causation of white Western privilege and mystical semiosis -or rather, perhaps, the downward causation that belief in the constructivist positing of white Western privilege exerts on the actual constructivist operation of semiosis.
On the other hand, however, the problem is also that Marais seems to ontologize semiosis as "real" -as the complex "intering" structure of reality. Semiotics, he says, "is rooted in the brain, one can even say in the psychological, which emerges from the brain, and it is simultaneously, paradoxically, part of the social where more than one physical brain interacts" (Marais 2014, 71). No, "semiotics" is not "rooted" in the brain. It is not rooted anywhere. It is not "part" of anything. Semiotics is in fact the study of semiosis, which is not "rooted" anywhere either. 8 Semiosis is a situated production of meaning that is not one thing "simultaneously" or "paradoxically" in several places or systems at once. Semiosis is not a reality-structure but a reality-structuring activity.
Nor does the "simultaneity" of different levels or activity-domains of semiosis (what Lotman calls "semiospheres") render them parallel, or equivalent, let alone consubstantial -any more than the "simultaneity" of gestural communication in primates and verbal communication in humans makes them the same thing. 9 Semiosis as social meaning-production emerges out of semiosis as individual psychological meaning-production, which emerges out of semiosis as neural pattern-production, storage, and recovery. Each semiotic system operates in and as its own semiosphere, emergently. Each works in its own way. Not one is reducible to the lower level out of which it emerged. It is misleading, therefore, to say that "one does not have to pose a typical constructivist divide between the first, nature, and the second, culture. Nature and culture are one because semiosis is both physical-chemical-biological-psychological and social" (Marais 2014, 71). No, nature and culture are not one. That is a grossly reductionist claim that undermines Marais's entire complexity-theoretical project.
The "typical constructivist divide" that he attacks here is in fact one of the core insights of the Peircean semeiotic: that signs are not things in themselves, but are constructed as signs by interpretants working on objects. What Marais calls "emergent semiotics" is in its original Peircean sense explicitly constructivist complexity theory. The functioning of semiosis in, say, slime molds ("Man's Glassy Essence") is not "real" -it is a constructivist semeiotic, developed by Peircean interpretants to explain the emergence of meaning-production in single-celled organisms. The recurring triadic patterns that Peirce finds in slime molds, the practice of scientific inquiry, and cosmic evolution ("Evolutionary Love") are similar because Peircean triadic interpretants are constructing them as similar. 10 The Peircean semeiotic is post-Kantian Idealism, not Enlightenment empiricism. It is a mental structuring procedure, not an ontological blueprint. It is constructivism, not objectivism. Or rather -to ease out of the binaries in those three previous sentences -in the Idealist frame, it is mental constructivism that gives the impression of ontological objectivism. What makes it a conduit for complexity theory is its radical situated perspectivism: semiosis operates in many different systems, in ways that seem similar when reduced to abstract triadic patterns but that also generate throughout the universe astonishing emergent (irreducible) diversity.

Conclusion
By way of wrapping up this constructive (and constructivist) intervention into Marais's complexity theory of translation, let me give a final Deleuzian thought to his summary of John Holland's list of the characteristics of complex adaptive systems: aggregation as "the complex, large-scale behavior that emerges from the interactions of less complex agents", nonlinear flows "from node to node via a connector with the nodes acting as agents and the connectors as possible interactions", and diversity as "parts of different nature or agents of different nature" (Marais 2014, 33-34).
In the terms Deleuze and Guattari develop for this complexity thinking in A Thousand Plateaus, translation would be a line of flight out of the source-textual/cultural territory that not only deterritorializes the source text and reterritorializes it as the 10 "Man's Glassy Essence" and "Evolutionary Love" were the fourth and fifth instalments in what is known as Peirce's "metaphysical" or "cosmological" series written in 1890 and published in The Monist in the early 1890s. They can be found in The Essential Peirce, vol Marais expresses other Peircean ideas throughout the book, citing only other sources, suggesting that perhaps he has not read Peirce. For example, he argues that "while the logic of difference has been made clear, at times one needs to draw boundaries, though contingent and temporary, to these deferral processes because you have to act (Cilliers 2005, 263-64)" (80). This principle derives from Peirce's solution, late in life (between 1903 and1907), to the problem he posed in the late 1860s of "endless semiosis". The poststructuralists borrowed that early notion from him as a defining trope, but never found (or never wanted) his late solution, namely, that habit stops semiosis in order to facilitate action. target text, but also deterritorializes the target language/culture and reterritorializes it in the image of the source text. But what interests me most here is not so much translation as it is the translator's agency -or what Deleuze and Guattari might well call translatorial becoming-agent.
What, after all, in translation terms, might be the "interactions of less complex agents" out of which aggregations emerge? They would apparently be the "nodes" by and through which the flows move -but what might that mean in the interactivity of translation, or of translation studies? And what would the "agents of different nature" be, and what would constitute their "difference" or "diversity"? Is it possible to map this complex adaptive system onto "translation" with human beings as agents? Should we say, as in skopos theory, that translation commissioners, project managers, researchers, translators, editors, and end-users are the diverse nodes by and through which a translation flows into being? That might be a Taylorized becoming-translation; in what way would each of those "nodes" also be a becoming-agent? Should we imagine, for example, that each "node" is a professional role with a job description, and individual human beings become those node-agents not only by being hired to perform each task, and then performing it, but by forming a more or less stable professional unit that processes a text from source to target in an industrial production line?
No. That would not be Deleuze and Guattari's conception of "becoming-agent", and I very much doubt it would be Kobus Marais's either (though it somewhat suspiciously resembles those intentless "[non-]agents of translation" who are actually just cogs in the biocapitalist machine). Deleuze and Guattari's "becoming-" construction is never about assimilating oneself to stable ontological categories, or, as they would say, "imitating" the "molar" states of assigned subjectivity that we associate with specific body shapes ("man", "woman", "child", "animal", etc.) or roles ("project manager", "researcher", "translator", "editor", "end-user"). Becoming for them is always "molecular". As Louise Burchill explains, this means that it is "a process of desire opening us to creative exploration of modes of individuation, intensities and affects (relatively) untrammelled by the forms, functions and modes of subjectivity society imposes upon us" (Burchill 2010, 88). In Burchill's account, for example, their infamous concept of "becoming-woman" consists not "in imitating women but in producing in ourselves the relations of speeds and slownesses-the spatio-temporal determinations-and correlative affective intensities that are proper to the girl in her identity of a 'molecular woman' or 'microfemininity'" (275-6)" (Burchill 2010, 88). 11 In the translation marketplace, for example, "molar" subjectivity -the imprisoning territory of a static role in geographical space and linear time -would be what Marais calls "the typical anonymous, voiceless, invisible translator slaving away in a stuffy little office, translating boring municipal regulation after regulation". But to clarify: it would not be that person. For that matter, in Marais's demeaning description "that person" is not "that person" either: what Marais describes is a stereotype, a category, a molar territory. The molecularity or becoming-molecular of the person who sits there in that "stuffy little office, translating boring municipal regulation after regulation", would be the unfolding of possibility as a fractalized becoming-agent, an infinite series of bifurcating moments or "micro-agencies" pulling him or her simultaneously back into a binding past and a potentially unbinding future. For Deleuze and Guattari becoming is a participation in the relationalities and pressures mobilized by what they call "the time of the event", which they define as "the floating line that knows only speeds and continually divides that which transpires into an already-there that is at the same time not-yet-here, a simultaneous too-late and too-early, a something that is both going to happen and has just happened" (Deleuze and Guattari 1987, 262). 12 To put that in icotic terms, the molecularity of the Deleuzian translatorial becoming-agent would be the bound openness of that icotic complex adaptive system called "the translator": the affective-becoming-conative experience of past normativities, which bind and constrain, intershot with the inevitable partial failure of such bindings and constraints and the potential opened by that failure for creative deviation. What is depressing about Marais's description of "the typical anonymous, voiceless, invisible translator slaving away in a stuffy little office, translating boring municipal regulation after regulation" is the sense we have of the hopelessness of that quasi-agency, the icotic pressurization of that sub-agency into debilitating role-paralysis. Not only is this translator "anonymous, voiceless, invisible"; not only is s/he "slaving away in a stuffy little office": s/he is typical. S/he is a type, a stereotype: a molarity, trapped by past and present icoses in molar subjectivity.
The becoming-molecular of this becoming-agent, by contrast, is shot through with the "not-yet-here", the complex becoming of an open system that is never perfectly closed off to the future. Emergence is always possible -even if it is only the occasional indulgence in parodic out-loud readings of excruciating source-textual formulations to one's humorously commiserating colleagues across the room, or in silly bad translations called out and not written down. Or even if it is only a slightly off translation that, written down and admitted into print, no one else will notice as even slightly problematic, but still gives the "typical anonymous, voiceless, invisible translator slaving away in a stuffy little office" a tiny smile of secret satisfaction. Or even if it is only a parodic translation written during a break and shown to colleagues, which somehow, as if by accident, finds its way into print, and launches a new career more promisingly laced with creativity and possibility. And so on.
But would we want to cluck our tongues at this Deleuzian vision as the mere phantasmatic flowering of the Romantic imagination in the colonial West? Would Marais claim that, while emergence is always possible for open systems, the work of a municipal translator in South Africa, say, is a closed system? Is the openness of becoming-agent something imaginable only by "powerful societies where people have the power to construct their reality", while "in a postcolonial context it is an open question whether people have that power" (Marais 2014, 66)?
Blown up into a utopian globalization, perhaps, yes, translatorial becoming-agent might be dismissed as typical colonial grandiosity. But Deleuze and Guattari are working with molecularity: tiny fragments of micro-agency. At that level, surely everyone on Earth is engaged in becoming, becoming-agent, self-agentizing, at every moment of every day? If that were not true, surely there would be no hope at all anywhere, not only in the formerly colonized world, but in the penthouses and shiny offices of the rich and powerful? Surely in a world, or a country, or a region where becoming-agent was perfectly impossible, even development in the top-down neoliberal model (bringing the poor outwardly up to the standards of the "developed" "first world") would be a mere sham -and development in the bottom-up human-centered mode ("opportunities for increased humanness", "people finding or constructing meaning for their lives", "the experience of the lifeworld", "the entire universe of participants", "the beneficiaries of development are not conceived of as recipients but as contributors" [Marais 2014, 130-31]) would be simply unthinkable?
As I say, Marais only mostly seems to foreclose on the openness of translatorial becoming-agency, in "the majority of translation activity in which the translator has no particular intent, other than making money of [sic] having to do a job" (Marais 2014, 95). At least once in the book -on that magical page 144 -that same intentless translator "is contributing as much if not more to the construction of social reality than the verbose literary translator who performs an aggressive feminine translation of a literary classic". It remains unclear just what s/he is contributing, but I would have to assume that it partakes at some level of the becoming-molecular excess that I have theorized as translatorial becoming-agent.