Beki Grinter

Writing Papers

In computer science, discipline, research on April 24, 2011 at 4:02 pm

Matt Welsh has written a nice post about how to write acceptable papers. As he notes in his post, some of this is out of your hands because Program Committee meetings have their own Psychology. But his points are well taken, and here I’m going to use them to riff.

Spelling and Grammar. Yes, yes, yes. I don’t mind a few typos, and frankly especially when left to write myself I usually produce a few typos too (thank you to the reviewers for catching them) but it does get very difficult to read a paper with lots of mistakes. It screams of carelessness. In that category I would also add formatting. Each conference has a required paper format. It is always tedious to reformat a paper, but I also think that choosing the right format is a sign of a persons interest in having the paper published. I’ve read things that are single column (most ACM archival tracks are two column) in a variety of fonts that are not Helvetica/Times. Just get the formatting right, or at least plausible. He’s also right on about good figures, and I also add tables.

English. This is really complicated. Native English speakers are at a huge advantage when it comes to writing papers. I include myself in this group even though I am not from the United States. I just have to remember to replace s’s with z’s and remove extra u’s. Or I just leave them in. If you are British/Canadian/Australian etc… and writing a paper I do recommend not switching between U.S. and English however. Pick a version of the language and aim for consistency in it. One other thing, a hallmark of U.S. English is active voice. I know many American academics that loathe passive voice. I think when I came to the U.S. it was either me, or a cultural thing, that I tended to write long, passive voice, sentences. I was taught not to. I actually like active voice now because it forces the writer to be specific about who did what to whom. There’s nothing that magically happens, and in a world of human-computer interaction we should be able to identify the action. Also I was taught that organization is a verb and not a noun, there are many types of organization so organization as a noun is not specific enough.

But, the dilemma I have as a reviewer of non-native English speaker written papers is to balance being inclusive and being comprehensible. I want to inhabit a world of conference authorship that includes as many voices as possible. In HCI, there’s a particular reason for this. Some of the work that we do is inherently framed in cultural terms. To the extent that culture overlaps with language, I feel that it’s incumbent on the community to be inclusive, to understand alternate perspectives that inform the global science of HCI. But I have read papers where that science, the contribution, eluded me from start to finish, and it is those papers that I do tend to reject. The others I try to work on the grammar at least in places, but usually I encourage the authors to share it with a native English speaker to see whether it can be improved.

All of this, of course, is written from the perspective of people submitting papers to conferences where English is the published language of the conference.

Related work. I agree with everything he says. Related work is not a list, it’s a synthesis, and it should read like one. I tend to use two strategies to organize my related work sections. First, chronology. That’s charting shifts not listing individual papers in order of when they were published. Second, I use thematic clustering. Take the related work, divide it into groups, decide what each groups central contributions are, write them up.

I have another pet peeve. It’s less common in HCI but every now and again I run into papers that have the related work at the end. After everything else has been presented. I have never understood why this was a good practice. To me the primary work that related work does is to argue the case for your research and its contribution. Related work frames the question you picked, how you approached its answer, and why the contribution is made (because it stands in relief to the other work already published). So, why leave it to the end to tell the reviewer how this is different? Why suspend the framing to tell the reviewer about a system or empirical study that they may or may not believe is a novel contribution because they haven’t read the related work yet. OK, stronger statement from me, related work at the end frequently feels less substantial than a related work section that appears after the introduction because the former is not put to work in the ways that the latter is. Late related work, last refuge of a scoundrel?

Tackling the Introduction and Getting to the Point together. Agreed, agreed. Engaging introduction and making sure that the reviewer swiftly knows why they are reading the paper. All good stuff. I was taught that you should have your research question out by the end of the first paragraph. I frequently do it in the form of the “gap” set up (I think this makes my students laugh, but I’ll just say that it was my favourite Thompson Twins album, so that’s really helping isn’t it). I argue that research is happening in a particular space (that’s a first shot at saying what the related work will likely be, at least some of it) and then argue that there’s a gap or omission in the area because something specific has not yet been addressed. The paper that I’m writing will address that gap. I think I have a reputation for being gap oriented, perhaps I should get a different type of set up. Anyone want to share theirs?

State your contributions. I’m almost embarrassed to admit this, but well there we are, no-one is reading anyway. I’m a late comer to the whole state them explicitly. I hope that problem set up, hard fought related work, and discussion helps people know what they are, but I’ve not typically put them explicitly into the introduction. But I can see the appeal. I think it’s a good way to get things out front and center.

And that leads to the final comment of Matt’s. No bullshit. Again I agree. I’ll add that what is true of the results, is also true of the methods. Being clear about what you did and what its limitations are is essential. Of course you want the good to shine in your work, but never try to hide the problems, especially if they are minor. Never allow a reviewer to figure something minor out, it encourages the hunt for more hidden secrets. Not worth it.

Slightly off topic, but the no bullshit also applies to CV’s. I can’t tell you how much it irks me to see non-archival track material show up in the archival track section of the resume. Never, ever, under any circumstance (thus ensuring one is found) list, for example, something that appears in the Extended Abstracts of the CHI proceedings as archival i.e. under the conference papers section. They’re not, nor should they be. Again it’s encouraging the reader/reviewer to look for more. Never wise.

  1. re: related work at end. it’s the default pattern for a lot of communities including systems. such readers (1) almost certainly know the related work because the communities are a lot more inbred (hence the much higher citation counts compared to HCI – there are always 30+ very closely related papers on any topic) and (2) expect to get to the technical details quickly and find it jarring to have a lengthy section between intro (which should have an accurate but soundbite-length contribution statement anyway) and the meat. note that (2) assumes that an accurate but soundbite-length contribution statement is even possible, which it is in many communities but is not in social science. in other words the strategy that sounds more substantial to you (related work up front) sounds like BS to them, as well as vice versa.

    i agree with you that if you’re not writing a paper where the main contribution has to do with a clear technical differentiator (and there are plenty of HCI systems and interaction papers where the latter is the case), moving related work up front makes more sense since the intro probably didn’t do enough for the critical reader.

    • thanks for the explanation, given your experience, I am sure that you are right. I’ve never gotten used to it, and when confronted with it I find it very jarring.

  2. i generally agree with you about being unclear about the non-archival thing on CVs and web pages (i.e., it annoys me to have to parse them out), but i have to say that the conference committees do NOT make it very easy for the non-CHI-mafia by adding a bunch of tracks where your paper is reviewed, you probably have a presentation, the paper appears in the ACM DL, and people want to be sure that the (technically) non-archival contributions are “valued” (since they are supposedly for some community like practitioners that is otherwise oppressed by the academics).

    the thing that seems reasonable to me to expect is that everyone will be accurate and complete. if it’s a paper in the proceedings, fine, list it as such. if it’s extended abstracts, fine, list it as such. if it’s a demo paper that happened to be in the proceedings, fine, but please mark it as a demo. the paper categories in everybody’s CV are all different anyway. but actually marking an extended abstracts paper as “in proceedings of” – NO NO NO.

    • I’ve been experimenting with a category on my vita called “lightly reviewed” which seems to be a reasonable catch all for workshops, magazines, extended abstracts, and so forth. I agree that this section of the vita is most difficult and the categories are not well standardized, and you want something that’s useful enough to catch a reasonable amount of it, so you don’t then have to have another section or too. I think my main beef is to keep the archival stuff well and clearly separated from the others. But it’s non-trivial as you say. And oh yes, to call them proceedings… terrible.

      Another area that used to catch similar ire was when CHI proceedings were also letters and people would put their conference papers into the journals section of the vita using the letters reference. I knew people who were furious about that.

  3. A question: Why most conferences have 2 column format? Is it because, this utilizes space in more efficient way (because we waste lesser space, when we break a paragraph) and this makes the packaging more compact?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: