LING484: Discourse Analysis: Conversation J. McFadden, Spring 2007
Some guidelines for writing short discourse analysis papers
1. Remember that your analysis must be data driven. This means you can't decide what your point is before you turn to your transcript - your data should drive your analysis. Don't start writing until you've read through your transcript carefully and made notes on anything and everything that seems interesting. Then choose one point and go with it, excluding any other interesting observations that don't fit.
2. Make sure your paper has a single clear point. In other words, your reader should be able to finish reading your paper and summarize its content briefly and concisely, preferably in a single sentence. So, for example: "Jen's argument was that Goffman's discussion of move might usefully be broken down into the following categories: X, Y, Z." Or: "Goffman's discussion of move seems not to account for circumstances in which blah blah blah." But not : "Jen demonstrates two examples of moves " - because that's not really analysis. And not: "Jen identifies moves, turns, floors, transition relevance places, speaker overlap, frames and footing while simultaneously demonstrating the inherent sexism of male interactants and their intent to dominate female discourse" - because that would be (obviously) biting off more than anyone could possibly chew in 3 pages; because it would be mixing lots of terminology without leaving adequate room to consider their connections; because it would be making gross generalizations; and because it would be attributing intention/internal states, which is not typically something that discourse analysts can access.
3. Do more than just identify structures or apply frameworks - try to build on them productively. There are countless ways you could go about this - but two strategies that seem to work especially well are to (1) make connections between a couple of readings from different weeks (for example, can your data help you relate floor to repair? intonation units to turn structures?); or (2) critique a particular analytical application as it applies to your data.
4. No matter how interesting a given excerpt or observation might be, don't include it in your paper unless it contributes directly to the one clear point you're making.
5. Be rigorous and careful in your use of terms and definitions. So if you're talking about "floor" as Edelsky defined it, you should cite Edelsky - this is the cue that you're not using terms in a layman's sense. (We're all used to talking about 'floor' in everyday talk, but that's different from how we use it analytically - and the difference is an important one.) You should also make sure that your paraphrases/usages of the definitions are accurate - it's okay to quote from the readings if that will help you!
6. Don't assume your reader automatically knows what you're doing. In other words, make sure your introduction identifies the fact that you're doing conversation analysis. This means you wouldn't want to start with something like "Sacks, Schegloff and Jefferson claim that turns are organized systematically" - because you could be talking about turns at chess or cards, for all your reader knows. Simply changing this to "turns at talk" makes a big difference. Also avoid using phrases like "in my conversation", because that assumes your reader knows you and knows what you mean by "my conversation". Instead, you could say "in a conversation between three friends at dinner", which requires no prior knowledge of you or 'your conversation.'
7. Make every step of your analysis explicit. Don't just say "line 15 shows X" - point out why it shows this. If you are including counts of some phenomenon (and especially if you are using percentages), say how you got these numbers. If you're examining "interruptions", be clear about what you're calling an interruption and why.
8. In general, try to preview excerpts before they appear. So you can say something like "in excerpt 1 below, Bonnie and Clyde are reminiscing about their recent adventures robbing gas stations in Missouri" and then go on to show excerpt 1. This works a little better then giving the same information after the excerpt.
9. Make sure you attach a marked-up transcript. This is where you can make annotations about the transcript that you didn't have room to include in the paper. It's okay to use markers, highlighters, different fonts, and to scrawl stuff in the margins. (Just remember that your observations should also be relevant to the point of your paper.)
10. Proofread!