Textual analysis vs. discourse analysis
This began as a Facebook exchange, but I think it's pedagogically interesting enough to reproduce here and add one more comment of my own, then see where the ensuing discussion takes us.
Efe Sevin: this might be one of the best pieces I have read recently....
An Array of Qualitative Data Analysis Tools: A Call for Data Analysis Triangulation.
Patrick Thaddeus Jackson: As though "rigor" and "replicability" (and "reliability," in the fairly narrow sense of "observer-independent") were synonyms. That, I would say, is a very contestable proposition indeed.
Efe Sevin: I think it all boils down to 'data analysis is a systematic search for meaning' understanding. So, there is actually a meaning hidden somewhere in the qualitative data and it is possible to extract it through rigorous analyses. Especially domain analysis, taxonomic analysis, and componential analysis are, I would say, promising tools to explain meaning in a context (in a replicable way). The words are not synonymous but they are closely linked to each other.
Patrick Thaddeus Jackson: The key move here is to assume that "there is actually a meaning hidden somewhere in the qualitative data." And that is the leap of faith, I think: that meaning is there to be found, instead of co-produced by the act of looking for it.
Efe Sevin: 'Meaning is to be found' is an assumption that I do not entirely disagree.
Efe Sevin: let me try to rephrase that - I do not think the co-production, or production of meaning in a specific context, necessarily means that meaning cannot be analyzed systematically.
Patrick Thaddeus Jackson: Ah, but your two comments are quite different, since systematicity and externality are two very different claims. Systematic co-production and systematic discovery are not, after all, the same thing.
Efe Sevin: Right- but there is no prerequisite relation among them. I am trying to claim there can be a systematic discovery of co-production. A rigorous/systematic analysis does not need to assume that there is an external reality.
Patrick Thaddeus Jackson: Granted. But "discovering meaning in the text" does presume that there is an external reality, in a way that "co-producing meaning with the text" does not.
Craig Hayden: If I may wade in here... cannot the researcher simply just foreground his/her value commitments or other reflexive predispositions before entering into any systematic analysis? You don't have to be a rigid "realist" to acknowledge that there are structures (patterns, institutions, etc.) that exist externally - in so far as they are identifiably and contingently external. Sure, this is a double hermeneutic problem, but that doesn't mean that you have to buy scientific realism wholesale. Simply that you have to acknowledge the limitations of our own systematically. Once we speculate on the enduring externality of structures, we wade into the "magical thinking" of realism.
Patrick Thaddeus Jackson: "External" to what?
Craig Hayden: You know, I keep struggling to comprehend my own fundamental inclinations toward realism, while recognizing that it's not really compatible with co-production. If qualitative data exists in some form somewhere (say, a corpus of textual artifacts) do they still have meaning *external* to the researcher? Or, for example, can someone really be a structuralist at all? I suppose it's about the ambition (or pretension) you bring to your claims.
Patrick Thaddeus Jackson: I think there's a difference between "external to the researcher as an individual" and "external to processes of knowledge-production." Accepting the former doesn't make one a realist in any significant way; it just makes one not a solipsist. Accepting the latter -- and saying that meaning inheres "in the text" would, I think, be accepting the latter, since it would place meaning in the realm of "brute facts" that are definitionally the case whether one apprehends them or not -- makes one a realist of some sort. Denying that there are any brute facts makes one a subjective idealist; denying that it is sensible to talk about brute facts or other mind-independent externalities makes one some variety of social constructionist.
Patrick Thaddeus Jackson: And none of this has anything at all to do with systematicity. I'm a huge fan of systematic procedures of data collection and data analysis; I'm just very skeptical about "triangulation" and the collection of ontological baggage that it brings with it.
Craig Hayden: Yes, that makes sense, and is what you say (repeatedly) in your book. And I'm on board with systematic procedures, if only to make for the better argument.
Patrick Thaddeus Jackson: Of course, we might have to ask ourselves *why* systematic procedures make for the better argument ... but that's a conversation for another time, and probably not for Efe's wall ;-)
Efe Sevin: Well, I actually did enjoy the conversation :) I have two things to add - first what I believe the systematicity in one sense includes an apprehension of social reality. The hidden assumption is that reality can be 'categorized' prior to da...ta gathering or analysis. Secondly, the main argument of the article comes back to what Dr. Hayden mentioned. There is indeed a body of qualitative data waiting to be analyzed. The methods offered in the article just enable the researchers to get the best out of the data set. (However, then again 'getting the best out of the data set' inherently includes a conceptualization of reality external to the researcher, and argues that it is possible to extract contextual meaning from a text - and I am okay with both statements.)
Craig Hayden: Well... a lot to say here. Probably best over coffee.
So my final comment would be that this discussion provides a good account of the difference between the content analysis of a text and a discourse analysis that uses text. Content analysis attempts to isolate the meaning of a text so as to use that meaning in an argument, perhaps correlating that meaning to some outcome, perhaps explaining that meaning in terms of the interests it serves or the relations it obscures. The point is that the meaning is an attribute among other attributes. But a discourse analysis of a piece of text would try to use the text to get at the practical context surrounding it -- in Shotter's terms, to move from dead already-spoken words to living words in their speaking, with the emphasis here firmly on "in their speaking" as a practical context. So what's of interest in a discourse analysis is not the text (which is why, parenthetically, Foucault is rather dismissive of "interpretation" as an activity, and this formed part of the argument between Foucault and Derrida to the extent that Foucault accused Derrida of diving too far into the text), but the effect that the formulations in the text have, how they contribute to an over-ongoing flow of events and happenings. Killing the text by subjecting it to automated techniques of content analysis (or to doctrinaire decodings in terms of some given notion of ideology) is, in this respect, the polar opposite of a discourse analysis. Epstein should help us grapple with this distinction in the context of a concrete analysis.
Efe Sevin: this might be one of the best pieces I have read recently....
An Array of Qualitative Data Analysis Tools: A Call for Data Analysis Triangulation.
Patrick Thaddeus Jackson: As though "rigor" and "replicability" (and "reliability," in the fairly narrow sense of "observer-independent") were synonyms. That, I would say, is a very contestable proposition indeed.
Efe Sevin: I think it all boils down to 'data analysis is a systematic search for meaning' understanding. So, there is actually a meaning hidden somewhere in the qualitative data and it is possible to extract it through rigorous analyses. Especially domain analysis, taxonomic analysis, and componential analysis are, I would say, promising tools to explain meaning in a context (in a replicable way). The words are not synonymous but they are closely linked to each other.
Patrick Thaddeus Jackson: The key move here is to assume that "there is actually a meaning hidden somewhere in the qualitative data." And that is the leap of faith, I think: that meaning is there to be found, instead of co-produced by the act of looking for it.
Efe Sevin: 'Meaning is to be found' is an assumption that I do not entirely disagree.
Efe Sevin: let me try to rephrase that - I do not think the co-production, or production of meaning in a specific context, necessarily means that meaning cannot be analyzed systematically.
Patrick Thaddeus Jackson: Ah, but your two comments are quite different, since systematicity and externality are two very different claims. Systematic co-production and systematic discovery are not, after all, the same thing.
Efe Sevin: Right- but there is no prerequisite relation among them. I am trying to claim there can be a systematic discovery of co-production. A rigorous/systematic analysis does not need to assume that there is an external reality.
Patrick Thaddeus Jackson: Granted. But "discovering meaning in the text" does presume that there is an external reality, in a way that "co-producing meaning with the text" does not.
Craig Hayden: If I may wade in here... cannot the researcher simply just foreground his/her value commitments or other reflexive predispositions before entering into any systematic analysis? You don't have to be a rigid "realist" to acknowledge that there are structures (patterns, institutions, etc.) that exist externally - in so far as they are identifiably and contingently external. Sure, this is a double hermeneutic problem, but that doesn't mean that you have to buy scientific realism wholesale. Simply that you have to acknowledge the limitations of our own systematically. Once we speculate on the enduring externality of structures, we wade into the "magical thinking" of realism.
Patrick Thaddeus Jackson: "External" to what?
Craig Hayden: You know, I keep struggling to comprehend my own fundamental inclinations toward realism, while recognizing that it's not really compatible with co-production. If qualitative data exists in some form somewhere (say, a corpus of textual artifacts) do they still have meaning *external* to the researcher? Or, for example, can someone really be a structuralist at all? I suppose it's about the ambition (or pretension) you bring to your claims.
Patrick Thaddeus Jackson: I think there's a difference between "external to the researcher as an individual" and "external to processes of knowledge-production." Accepting the former doesn't make one a realist in any significant way; it just makes one not a solipsist. Accepting the latter -- and saying that meaning inheres "in the text" would, I think, be accepting the latter, since it would place meaning in the realm of "brute facts" that are definitionally the case whether one apprehends them or not -- makes one a realist of some sort. Denying that there are any brute facts makes one a subjective idealist; denying that it is sensible to talk about brute facts or other mind-independent externalities makes one some variety of social constructionist.
Patrick Thaddeus Jackson: And none of this has anything at all to do with systematicity. I'm a huge fan of systematic procedures of data collection and data analysis; I'm just very skeptical about "triangulation" and the collection of ontological baggage that it brings with it.
Craig Hayden: Yes, that makes sense, and is what you say (repeatedly) in your book. And I'm on board with systematic procedures, if only to make for the better argument.
Patrick Thaddeus Jackson: Of course, we might have to ask ourselves *why* systematic procedures make for the better argument ... but that's a conversation for another time, and probably not for Efe's wall ;-)
Efe Sevin: Well, I actually did enjoy the conversation :) I have two things to add - first what I believe the systematicity in one sense includes an apprehension of social reality. The hidden assumption is that reality can be 'categorized' prior to da...ta gathering or analysis. Secondly, the main argument of the article comes back to what Dr. Hayden mentioned. There is indeed a body of qualitative data waiting to be analyzed. The methods offered in the article just enable the researchers to get the best out of the data set. (However, then again 'getting the best out of the data set' inherently includes a conceptualization of reality external to the researcher, and argues that it is possible to extract contextual meaning from a text - and I am okay with both statements.)
Craig Hayden: Well... a lot to say here. Probably best over coffee.
So my final comment would be that this discussion provides a good account of the difference between the content analysis of a text and a discourse analysis that uses text. Content analysis attempts to isolate the meaning of a text so as to use that meaning in an argument, perhaps correlating that meaning to some outcome, perhaps explaining that meaning in terms of the interests it serves or the relations it obscures. The point is that the meaning is an attribute among other attributes. But a discourse analysis of a piece of text would try to use the text to get at the practical context surrounding it -- in Shotter's terms, to move from dead already-spoken words to living words in their speaking, with the emphasis here firmly on "in their speaking" as a practical context. So what's of interest in a discourse analysis is not the text (which is why, parenthetically, Foucault is rather dismissive of "interpretation" as an activity, and this formed part of the argument between Foucault and Derrida to the extent that Foucault accused Derrida of diving too far into the text), but the effect that the formulations in the text have, how they contribute to an over-ongoing flow of events and happenings. Killing the text by subjecting it to automated techniques of content analysis (or to doctrinaire decodings in terms of some given notion of ideology) is, in this respect, the polar opposite of a discourse analysis. Epstein should help us grapple with this distinction in the context of a concrete analysis.
posted by ProfPTJ at 11:14 AM
1 Comments:
I was planning to use this correspondence in my article entitled "The Implications of Web 2.0 on Academic Learning: The Power of Procrastination" but since we started to discuss it, let me jump in as well.
What I don't understand about discourse analysis is...well 'discourse'. Discourse is not an uber-complicated social phenomenon that can be only 'felt' (This is my interpretation of Foucault's work. In order to explain discourse/power/knowledge etc. he takes such a metaview, if I may call it so. He tries to explain all the events related to what he believes to be the dominant discourse.)
A content analysis, on the other hand, looks at discourse as a concrete thing (not necessarily dead words in this sense, but let's say outside Shotter's dynamic time-space understanding). You pause the time, take a chunk of text-based data, and look at what people are saying. This article shows several ways to get the best out of the qualitative data sets in terms of reliability and validity without losing their verbal richness.
Discourse is, in communication, how you talk about an issue. So, what is a better way to understand the discourse than looking at what you actually say?
Post a Comment
<< Home