Writing an academic article isn’t as difficult as it might seem from the outset — as long as you have enough evidence, a clear sense of the existing literature, and a good model to work from. In this first of several guides to writing a journal article, I want to work through this last element: a good example from an appropriate journal. (This is intended for anthropologists, but it might apply more generally.)
The first thing to do is identify a journal you’re interested in publishing in. If this is your first article, target a journal that focuses on your subfield or geographical region of interest. Generally speaking, these smaller journals have word lengths of approximately 6,000-8,000 words. These short articles tend to focus on one key idea from your research, and mobilize 2-3 cases to support it. Once you have a significant number of dissertation chapters written, it should be relatively easy to weave together a first draft. But putting something together and making it relevant to the journal you’re interested in publishing in are two different projects. So the best thing to do is to find a good article published in the journal you’re targeting and work from that to get a sense of what the journal is looking for.
A good model isn’t one that comes from a senior academic. Instead, find one published in the last 12-18 months by an assistant professor or someone recently graduated from their Ph.D. program — someone roughly like you. Your model author will also be working from dissertation material, which is significantly different than the kinds of evidence later-career academics work from. Moreover, because junior people are working to position themselves in the field, the burden of their articles is significantly different.
If the editorship of the journal has changed hands since the publication of the article you’re working from, be sure to look at any introductions to the journal that the new editor(s) has written. If the journal is changing its focus or generic form, the new editor(s) will generally make that known early in their leadership.
Once you have the model in hand, read it once all the way through. Then, circle back with a highlighter and read it again. You’re going to need a few different colored highlighters for what’s about to come, and each time you read the article, it will get shorter and shorter…
On this first pass, highlight all of the primary evidence in the article — all of the actual empirical content generated by the author’s research. So for most anthropology articles, this means descriptions of spaces, people and events; it also includes quotes from interviews and other qualitative (and sometimes quantitative) data.
On the second pass, highlight (in another color) all of the secondary evidence in the article, but focus your attention on the content outside of the literature review. This often includes paraphrasing other authors, as well as historical or other anthropological work on the same topic. This does not include theoretical citations.
On the third pass, highlight (in yet another color) all of the argumentative content — the thesis, the topic sentences, and wherever else claims are being made (there shouldn’t be too many in the introduction or conclusion). This may include theoretical citations, especially if the author’s purpose is to argue with a set of theories or theorists.
On the fourth and final pass, highlight (in still another color) all of citations in the literature review section of the paper. Most journal articles should have 1-2 pages where the author is positioning the research alongside other work in the same subfield, other approaches to the same topic, and other research on the topic and subfield in the same geographic region. In some cases these sections can be quite long, but in most subfield and area journals, they’re relatively short.
Once you have this set of tasks completed, you should have a well marked up document. It will provide you with a few things: 1) a sense of how much evidence you need for an article of the same length, 2) a feeling for how much secondary literature you need to engage with, and 3) a scaffold of an argument and its relation to the empirical content that supports it.
With this evidence in hand, you should begin to think about your work and what might be successful at the same scale. Short articles in the 6-8,000 word range usually only have one substantial argument, and use a few cases to argue it. (For an example, you can look at an early article of mine, here.) Generally, your argument can’t get too complicated — you need a well defined problem and interpretation of it for a short article — so it will often be less than a dissertation chapter, or might borrow content from a number of chapters.
Like I discuss in relation to developing a publishing strategy for your early career, often when you’re writing for subfield or area journals, you’re making an argument with the existing literature in that field. So, what does your dissertation research add to dominant approaches to your subfield, area or topic? Just tackling that question is enough for a first or second article, because, in the beginning, you’re trying to do two things: First, you need to get people to pay attention to you, and, second, you need to start putting out articles that you can cite to support later, more complicated arguments in longer, more complex articles… But early on you need tight, short articles that make it clear to your readers what your interests are and what debates you’re contributing to at this stage in your career.
And here is Step 2.
Matthew Wolf-Meyer 
I was excited to try this new method for analyzing an article but had trouble executing the model because some of the categories you describe appear to blur into each other. Most empirical content was fairly easy to pick out. You are performing close reading of pharmaceutical industry texts and analyzing interviews, so I spotted these fairly quickly. However, in some spots in the paper, it seems that argument and description are intertwined — this happens in many good ethnographic papers — which it makes it difficult to distinguish between primary evidence, secondary evidence, and argumentative content.
I generally don’t use highlighters because my eye tends to skip over those elements. I suspect that that’s the point, and the highlighting is supposed to give you the big picture, to help you drown out noise. In this case it might be helpful to explain not only that the article gets shorter with more and more highlights, but also that the colors in relation to each other is what matters. You hint at it in the description above, but I think the description could be more explicit. For example, how is argumentation situated in relation to evidence? How much argumentation is foregrounded before you even introduce your evidence? Is there a specific sequence or order into which colors fall? Should everything be highlighted at the end or will you end up with just an intro and a conclusion mostly unmarked? All questions but the last are questions that the author can pose to him/herself as s/he marks up the article of choice.
It could also be helpful if you marked up your own article and showed us some of these relationships to make your points.
I liked this exercise. While it initially confirmed what I got from the first read, by the time I had completed the exercise, it offered a visual guide to article structure. I’m finding this visual guide very useful as I turn to my own material because it strips the argument structure of its content. I also found it useful to outline the template article, noting the number of paragraphs in each section and, based the coding suggested here, the dominant rhetorical functions of each section. Putting together the coding and the outline, I have a handy map of articles for this journal.
I found the primary/secondary research distinction fine, though outside of the lit review this particular article did not have much of the latter. I found the exercise of highlighting the firsthand material in the template article especially useful, in part because I realized that I had chosen an article that draws on a different kind of source material than I do — the template is entirely from firsthand observations, my own work is also document analysis. Though I think it will be fine in this case, in the future I will (also) seek template articles using the same types of primary sources that I use.