Interpretive Anthropology

Information

Interpretive Anthropology

A forum for conversations about the concepts and contributions of Clifford Geertz to anthropology.

Members: 70
Latest Activity: Jan 3

Perspectives in Anthropology

New article on Geertz and Foucault.

http://perspectivesinanthropology.wordpress.com/2011/06/13/foucault-and-g…considerations/

Discussion Forum

Ethnography: The Stranger Effect and Gift Giving

Anyone who has conducted ethnographic research understands the importance of the participant-observer methodology. When gathering information, it is essential to have an informant that is able to…Continue

Started by Neil Turner Apr 24, 2014.

In what ways do you think Geertz’s work differs from the concept that observers must stand outside the cultures they are observing? 13 Replies

The concept of the interpretation of cultures in anthropology is associated with the work of Clifford Geertz during what has been termed the “interpretive” or “literary’ turn in contemporary…Continue

Started by Neil Turner. Last reply by Neil Turner Apr 19, 2010.

Is anthropology experiencing another crisis in representation?

Traditionally, the practice of anthropology required leaving one’s home country for long periods of time, learning a new language, immersing oneself deeply in another culture for the purpose of…Continue

Started by Neil Turner Dec 30, 2009.

Interpretative Anthropology - is it still a valid approach to cultural analysis?

For the better part of this week, I have been reading posts from many of the groups and discussions that attract my interests and have found many of the forums interesting and exciting. It is against…Continue

Started by Neil Turner Dec 5, 2009.

Comment Wall

Comment by Ranjan Lekhy on December 6, 2009 at 3:49am
Dear Neil, no theory is ever complete and no theory is wrong. I am seeing the 'thick description' is still usefull in research methodology.
Comment by Arnab Sen on December 15, 2009 at 2:16pm
In my field of "consumer anthropology" I suspect 'thick description' is still a valid method and a counterpoint to mechanistic, transactional models so often used to understand how people behave with each other in consuming products, services, ideas and images.
Comment by Jacob Lee on December 15, 2009 at 9:49pm
Ranjan Lekhy said:
"...and no theory is wrong"

Seriously?
Comment by John McCreery on December 16, 2009 at 6:32am
Perhaps we should reconsider what we mean by "valid" and "wrong." To me both words imply a simplistic two-valued logic, A or not-A. I prefer a model suggested in one of the early works of Noam Chomsky. That is, we replace the idea that scientific method is a decision procedure (input theory and data; output decision, right or wrong) with the notion that scientific method is an evaluation procedure (input two theories and data; output a ranking that indicates that one theory is better than the other, given the data in question).

The most fully formalized example of an evaluation procedures is statistical hypothesis testing, which always involves two theories (the hypothesis being tested and the null hypothesis) and output that indicates how likely it is that one hypothesis is better than the other given the data used in the test.

The problem for anthropologists and others involved in case-based research is that we rarely have data that conforms to the sampling conditions that statistical hypothesis testing requires. But the basic evaluation model still works pretty well.

The classic examples can be found in classic British murder mysteries. The reader is confronted with a number of possible murderers. Bits of relevant data are scattered through the story. The detective's task is to assemble them into a story that proves the murderer's identity beyond a reasonable doubt, There are always several possible stories, but the one that provides the solution accounts for more of the data and the sequence in which it appears in the process leading to the murder than any of the others.

Similar considerations apply, I would argue, to competing thick descriptions. Which is thicker (accounts for more details in the data) and more persuasive (accounts for the sequence of events described? These are the questions that drive the evaluation that one thick description is better than the other — given the data in question.

"Given the data in question" is important, since even the currently better account may fail when new data becomes available.
Comment by Jacob Lee on December 16, 2009 at 9:51am
Nice thoughts John.

Two comments really, regarding:
(1) hypothesis testing
(2) the scientific method as a procedure

First (1), I wonder if we ought to consider Bayesian learning as an alternative to traditional hypothesis testing. I claim no personal expertise, but Bayesian methods are one of the dominant paradigms used in machine learning today.

Re: (2) First, I basically agree with you. Different hypotheses will accord with one's data to different degrees and we can construct a partial ordering of these hypotheses.

I am reminded of a different but similar idea in logical concept learning in machine learning. Logical concept learning involves learning a target concept using positive and negative instances in an example set. Examples are given in a logical 'language of examples' or LoE. Hypotheses are given in a logical ';anguage of hypotheses' or LoH. The two languages needn't be the same. A concept is considered learned if it can correctly classify test examples to some acceptable degree. The principal question is how to move from examples to hypotheses. In general there will be many hypotheses consistent with a single example. Each additional example eliminates some of these possible hypotheses (assuming no noise).

A too-general hypothesis will incorrectly classify some negative instances as positive instances. A too-specific hypothesis will classify some positive instances negatively. We could start with some current best hypothesis, and then proceed to generalize or specialize that hypothesis as needed. But there are problems with this: first, it may not be obvious what a good candidate for a first best candidate hypothesis would be. Second, in some cases we might find that we have to backtrack because our hypothesis revisions have made our hypothesis either too general or too specific.

In logical concept learning, an alternate representation is possible. Instead of directly representing some preferred working hypothesis, one represents a subspace called a version space of the total hypothesis space. It turns out that under certain conditions one can efficiently represent a vast version space of hypotheses by simply noting the set of most general and most specific hypotheses consistent with some current set of examples. As new examples are added one contracts the hypothesis space by either by specializing the most general hypotheses, or generalizing the most specific hypotheses (through pruning). If the concept can be captured using the attributes in the example set, and given sufficient variety and number of examples, the hypothesis space will converge to some minimal set of hypotheses (usually the ideal is one). With noiseless data, nothing can slip through the cracks, because the version space has the structure of a lattice and is a partial ordering.

Note that at any given time, those hypotheses outside the version space are *are* wrong, and those still inside the version space *might* be right, but there is a definite sense in which some hypotheses are *more* right than others.
Comment by Ranjan Lekhy on December 17, 2009 at 2:09am
Arnab, it is great to see you here! I am too busy, but let me leave a short reply here. I confess that I have no great theoretical knowledge on interpretative anthropology. At TU (Kathmandu) I studied few chapters of Interpretation of Culture and some works of Sherry Ortner. Here, at the University of Hyderabad, I had to write an assignment on interpretive anthropology (Dan Sperber and C. Geertz) during my MPhil. Though, I was more interested in Structural Marxism, I did not pay more attention on interpretative anthropology.

Perhaps, you are true in your own case, but I realize that it is still useful to me while I am trying to understand the perception of ecology, poverty, and livelihood among the Tharus. I think, people (always) do not speak same what they think, people do not do same what they speak, and people do not mean same what they think, speak and do! Meanings of their thinking, speaking, and doing change from time to time and from situation to situation (caution: not all the actions). Thus, I still think that text should be in context! That’s why, we anthropologists do more than general interview and questionnaire works. However, I cannot even imagine that human beings behave just mechanistically as you say! I hope my saying is not ambiguous!

Dear Jacob, yes I think that ‘no theory is wrong’ but in the sense that ‘theory’ is itself not a truth! A theory is a window to see the truth which is constructed by human mind, and thus no matter how intensive is in its objectivity, is just a part of subjectivity! So, I see the meanings of subjectivity and objectivity blur here. We can just say that a certain theory is much explanatory or closer to truth than others. Now, theory of Evolution is not so much popular in anthropology, does it mean that theory of evolution is wrong?

I hope veteran professors would like to enlighten us!
Comment by Jacob Lee on December 17, 2009 at 4:30am
@Ranjan
I appreciate the clarification.

I'm most comfortable talking about whether some proposition is true. A theory or hypothesis is not truth, but a sentence expressing a proposition, something that can be true or not true. Meaning is subjective of course, but that doesn't mean, IMHO, that the truth of some assertion, once a meaning is fixed, is itself subjective. Either some state of affairs obtains, or it does not, though I must profess my ignorance of sophisticated physics which may have a somewhat different metaphysics.

I agree with you that something can be closer or further from the truth. And this is something that is very interesting in itself. Suppose that in a given room there are exactly one hundred candles at some moment in time. If I asserted that there were one hundred candles my assertion would be true. If asserted that there were ninety-nine candles then my assertion would be false, but in some sense my assertion is closer to the truth, closer than if I had asserted that there were fifty candles in the room. So we might say that we have a discrete notion of truth, and a continuous notion of truth that coexist in some way.

Now we can let things get interesting. What if I had asserted that there were at least fifty candles in the room? This assertion is true! But is it more or less true than my false assertion that there were ninety-nine candles in the room? Something does not seem to fit.

I think that this example shows that the truth and falsity, both as a discrete value or as a matter of degree, are not the only relevant dimension by which we are to judge the correctness of a statement/hypothesis/theory. Specificity is also required. Following Luciano Floridi's notion of strong semantic information, I would suggest that together these somehow define a statement's informativeness.

It seems to me then that a thickly descriptive ethnography is therefore a highly informative ethnography.
Comment by John McCreery on December 17, 2009 at 3:21pm
I think, people (always) do not speak same what they think, people do not do same what they speak, and people do not mean same what they think, speak and do! Meanings of their thinking, speaking, and doing change from time to time and from situation to situation (caution: not all the actions).

Yes, oh yes, indeed. There are interesting discussions of the problems this poses in Victor Turner's work on the Ndembu in Central Africa (the essays in The Forest of Symbols are a good place to start. Also in Robert Weller (1994) Resistance, Chaos And Control In China- Taiping Rebels, Taiwanese Ghosts And Tiananmen .

Re Arnab's mention of consumer anthropology: It is important to remember, however, that the utility of thick description is different in an applied, e.g., business, context from its value in a scholarly context. In the latter, it provides a richer, deeper, more detailed understanding of the case in question. In the former, it can lead to fresh insights that identify the hot buttons that marketers can use to sell their products.
Comment by Arnab Sen on December 17, 2009 at 5:10pm
@John McCreery
Agreed that the objective of thick description is different in in business ethnography and in scholarly enquiry. In the former yes, what it gives us is a rich and nuanced understanding of the meanings, values, motivations that operate in a cultural space. In the former, no, it doesn't give the brand owner or marketer consumer hot buttons, what it does give is a rich and nuanced understanding of that set of interactions, transactions, values and meanings which have some relation with the brand or its category. Typically it is thick description through a filter. A flaw, if judged by academic benchmarks but a necessary hazard of the occupation that comes with the objective of brand outward research. But in some broad category researches, especially when trying to understand some technology and finance categories, a full-bodied academic thick description is always a better idea than a filtered one. Take for example a study on mobile telephony in a complex market like India. Can we describe communication without reference to kinship, hierarchy, gender, sexuality, religion, the works. Like a good anthropologist one would have to attempt a thick description of every respondent in terms of every (eurocentric) anthropological category. What most of us end up doing is going to the field with a hypothesis or rather set of hypotheses and see how well they feed into a description. I've personally done outrageous things - gone to field to study group dynamics among teens (for a soda, LOL) and tested four models of sharing food from evolutionary ecology to figure out how well these models work in the cultural space of students sharing knowledge, gossip, movies, music, pictures, software, transacting in access, licence, sex. It always looks like at least two conflicting models can comfortably coexist, the thicker the description, the more such linear models (and potentially conflicting) models can coexist. It's quite like you've pointed out with a reference to Forest of Symbols that meanings are contextual. Yes, and they morph through the ritual cycles of structure, antistructure and liminality in life. Think of journeys across cultural spaces as diverse as a face to face community, a hierarchical and patriarchal home, a campus where experimenting with alternatives offers the liminal zone between the order, control, comfort of home and the anonymity and freedom of a threatening world outside...
Comment by Arnab Sen on December 17, 2009 at 5:12pm
Oops sorry correction - in line 2 for 'former' read 'latter'

Comment

You need to be a member of Interpretive Anthropology to add comments!

 

Members (70)

 
 
 

Translate

OAC Press

@OpenAnthCoop

© 2015   Created by Keith Hart.   Powered by

Badges  |  Report an Issue  |  Terms of Service