One of OAC's most active recent threads has built on the notion of a web of explanations. Introducing this notion, M. Izabel observes that no single theory ever seems adequate as an explanation of the phenomenon to which the theory is being applied. Thus, "A web of theories is what is needed to get to the center of any matter that is problematic."


But what does a web of theories look like? This is not an idle question.  The image with which we begin, when we start to address any topic, our "root metaphor" or "world hypothesis" affects the directions of our thought. What if, instead of starting with a web, we started with a three-dimensional space, in which different types of explanations move in different directions, the approach advocated by Andrew Abbott in Methods of Discovery: Heuristics for the Social Sciences and illustrated by the following diagram.


Concerning this diagram, Abbott writes,


The three dimensions are the three types of explanations. for each of these, the origin stands for explanations focused on everyday particulars, on commonsense events. These are an anchor for each explanatory program, rooting it in the everyday world. From this base, "universalizing" moves reach from the origin toward abstraction along each of the principal axe of explanation. The syntactic program explains the social world by more and more abstractly modeling its particular action and interrelationships. The semantic program explains the world of social particulars by assimilating it to more and more general patterns, searching for regularities over time or across space. Finally, the purely pragmatic  program tries to separate more and more clearly the effects of different potential interventions or causes from one another.


What is of particular interest in this scheme is, I would argue, a tacit fourth dimension that runs from the center to the extremes of all three axes, from informal, verbal accounts to formal, mathematical descriptions. Commonsense understanding expressed in everyday tropes is the origin for all three dimensions, but all three also move toward greater abstraction and formalization. On what is here the horizontal dimension, ethnography is more abstract than commonsense understanding; but pattern search, using, for example, data mining, is more abstract than ethnography. On the vertical dimension, historical narration becomes fully formalized in complex computational models. On the third dimension, commonsensical understanding is clarified through experimentation and abstracted in standard causal accounts (SCA), typically based on statistical data and techniques for statistical analysis, most commonly linear regression. 

But even if you have no interest in these methodological subtleties, the critical point to note is that the goals of explanation  diverge radically along the three dimensions. The semantic program's goal is translation, rendering the alien and at first incomprehensible in familiar terms. The syntatic program's goal is narrative, a story about how and why a certain sequence of events led to what we observe. The pragmatic program's goal is leverage, detecting key points at which intervention will be effective. 

Translation, story-telling, or intervention? Historically anthropologists have been involved in all three. What remains unclear is how one could effect the other. It is all too easy to see why translation, which emphasizes cultural difference, comes to be seen as pointless obfuscation by policy-makers looking for leverage for interventions. Or why both translators and policy-makers have looked askance at pseudo-historical just-so stories. Returning to our starting point, I wonder what sort of fabric can be woven of threads pointing in two, let alone, three directions.


My jet-lagged brain begins to blur. Dear colleagues, what do you see?

Views: 763


You need to be a member of Open Anthropology Cooperative to add comments!

Comment by John McCreery on September 13, 2011 at 4:50am

M. With all due respect, what "I am more at ease with" is not a convincing argument. It is, if anything, evidence that you are (like most of us, I suspect) so brainwashed by what Abbott calls standard causal analysis (SCA) that you have trouble getting your head around what Abbott has done in Methods of Discovery. Instead of assuming a that the goal is a covering law explanation of the type you mention w=f(x,y,z), he begins by considering what scholars actually do when they do different types of research. He observes that,considered as practices, the explanations constructed by ethnographers/data miners, historians/computational modelers, and social scientists/advocates of standard causal analysis, turn out to be very different, both in what is actually done and in how it is evaluated by the research communities in question. 

When he asks, for example, what makes ethnography, conceived as cultural translation, a.k.a., interpretation, compelling, the answer is the way in which details of what is being described cease to appear strange and seem to make sense in the language in which the ethnography is written. The other's assumptions may be different from the reader's but, viva la difference, they no longer seem absurd. As an example of this domestication of the alien, he cites a remark by Evans-Pritchard in Witchcraft, Oracle and Magic among the Azande, in which E-P says that, while living among he Azande, he, too, used the poison oracle to organize his schedule and found it as useful a way as any other he had tried. Here, in one brilliant stroke, E-P establishes his own "I was there" credibility as a witness and invites the reader to consider the decidedly creepy sounding "poison oracle" as an everyday mechanism for sorting out what to do next. 


A good example of the difference between ethnography and history can be found in the introduction to James Watson and Evelyn Rawski (ed.) Death Ritual in Late Imperial and Modern China, where Rawski observes how differently the anthropologists and historians invited to the conference that produced this volume treated their data. While the anthropologists debated the significance of ritual details in the moment in which they were observed, the historians treated the same details as variables whose changes over time could be traced in documentary evidence, often over several centuries.

The pertinent point here is, however, that neither ethnographers nor historians were treating these details as outcomes predicted by covering laws conceived as mathematical functions. Neither worried, moreover, about the hypothesis construction and testing demanded by standard causal analysis. The fact of the matter is that, in these research communities, standard causal analysis is regarded as overly simplistic.  The credibility of the explanations, i.e., interpretations or narratives, is judged by coherence and range of detail covered instead of hypothesis-testing based on quasi-experimental designs. 

Thus, returning to Keith's point, it turns out on Abbott's closer reflection that "x therefore y" covers a range of practices: commonsense association, rhetorical trope, and logical implication, as well as cause and result strictly construed and properly demonstrated by use of experimental or quasi-experimental methods. Thus, to imagine that "x therefore y" necessarily implies demonstrable cause and effect, is evidence of the brainwashing mentioned above.



Comment by M Izabel on September 12, 2011 at 5:50pm
Nice observation, Keith.  The problem with the model, John, is its structure that is also selective. I am more at ease with  the notion that in culture, it is w because of x, y, and  z or even a, b, and c  and more.
Comment by John McCreery on September 12, 2011 at 3:55pm

[continued from the previous comment from which the last line should be deleted]  The lack of covering laws that work in all times and places is a problem for pragmatic decision makers. And they are not likely to find what they need in ethnographic descriptions or historical accounts that focus on the what and why of the phenomenon in question. They must look, instead, at causal relationships in the environment surrounding the phenomenon. A familiar marketing parable underlines this point.

It concerns a company that devotes itself to making the world's finest drill bits. Its secret alloys and patented manufacturing techniques make its bits far superior to any competitor's product. This company is put out of business by a company that makes lasers. The customers, you see, weren't interested in bits. They were interested in holes.

Comment by John McCreery on September 12, 2011 at 3:39pm


I may not have explained Abbott's diagram with sufficient clarity. His position is that the three axes diverge in the ways indicated from a common origin in commonsense understanding. In all three directions analysis moves toward greater abstraction. That said, neither translation (the goal of semantic understanding) nor storytelling/ sequential modeling (the goal of syntactic understanding) are the same as standard causal analysis/identifying levers (the goal of pragmatic understanding). 


Consider, for example, tatami. From a translation perspective, explanation consists of understanding that, for example, according to Wikipedia:

tatami (?) is a type of mat used as a flooring material in traditional Japanese-style rooms. Traditionally made of rice straw to form the core (though nowadays sometimes the core is composed of compressed wood chip boards or polystyrene foam), with a covering of woven soft rush (igusastraw, tatami are made in uniform sizes. Usually, on the long sides, they have edging (heri) of brocade or plain cloth, although some tatami have no edging.[1]

The translator's problem is finding more felicitous language to communicate the features of tatami important for the text and scene in question.

History demands a different type of explanation, e.g., the first tatami were small, one person woven rice-straw mats, taken out on special occasions and offered to guests, who would otherwise sit on bare wooden floors. For a series of reasons [not spelled out here], these mats became larger and more permanent floor coverings, produced in standard sizes that vary by region, with the coverings and edgings typical of modern tatami. Note that in this type of explanation, the elements are not necessary or sufficient conditions simultaneously present in the phenomenon being described. They are, instead a series of contingencies, events in an on-going story that has led to the phenomenon whose history is being examined.

Standard causal analysis aims at writing a function y=f(x1, x2,....xn), such that a phenomenon y is the predictable outcome of factors x1....xn. The desired function is assumed to be a covering law, equally valid in all times and places. Contingencies may effect the law's effects, but the law itself remains valid. To the best of my knowledge, there is no covering law that accounts for tatami in this way. There are only translations/descriptions and reasonably reliable historical accounts of how tatami came to be. 

One implication is that pragmatic decision makers looking, for example, for levers to implem

Comment by Keith Hart on September 12, 2011 at 2:40pm

Thanks very much for putting this up, John. It is a really helpful take on the problem raised by M. I would like to approach it through a quote from Steve Jobs:

“Creativity is just connecting things. When you ask creative people how they did something, they feel a little guilty because they didn’t really do it, they just saw something. It seemed obvious to them after a while. That’s because they were able to connect experiences they’ve had and synthesize new things. And the reason they were able to do that was that they’ve had more experiences or they have thought more about their experiences than other people.

“Unfortunately, that’s too rare a commodity. A lot of people in our industry haven’t had very diverse experiences. So they don’t have enough dots to connect, and they end up with very linear solutions without a broad perspective on the problem. The broader one’s understanding of the human experience, the better design we will have." [Wired, February 1996]

I would suggest that the model you present, for all its 3- and even 4-dimensional attributes, is essentially linear, since explanation as a goal of inquiry is linked to causal analysis. If x, then y. Connecting the dots is what used to be called intuition. Now I can see why someone interested in industrial production or the efficiency of policy might go for a method such as the one you outline. But I believe that an approach to connection as a web or network points in another direction -- to creativity. I am not sure that explanation is even our main goal as anthropologists. A good description is often an explanation in itself. And to some extent, a web of theories is an oxymoron, being the aspiration to make discovery hinge on theory, when it is often the other way round.

I think that I write with a web of theories and experiences somewhere in my mind. Sometimes, if I am lucky, they light each other up in a configuration that comes out as living prose. Memory is at least as much a source of experience as it is of theory. It doesn't always pay to formalize the relationship as a systematic method.

This is not to rubbish what I have learned from you and the other discussion, just to say that we can start from different assumptions about the goal of our writing. Mine has to to with helping people to discover their own thoughts by juxtaposing theory and experience in a personally accessible form.


OAC Press



© 2019   Created by Keith Hart.   Powered by

Badges  |  Report an Issue  |  Terms of Service