One person's approach to writing.
27 Jul 2020This post covers my process for writing academic papers such that it is as enjoyable, stress-free, and efficient as possible for me. A key feature of my process is to break writing into two parts: generation and editing. I will include some suggestions for those who have difficulties with the generation part, in other words, starting a first draft.
My lab is not typical; I’m a PI, I supervise one other person who is a machinist, and then I collaborate with others, some of whom are, and are not, at the same location. Some of these wonderful collaborators are students. I have done solo and partner writing. Given that my methodology has developed over time and is tailored to my life and personality, I hope that you develop your own approach that works for you.
Roadmap
Philosophy: get words on the page first, edit later.
My philosophy is that editing is easier than generating text. So instead of trying to write the perfect paper from the beginning, I continuously edit drafts. But how do I get that first text, the generation part?
At some point in high school, I was introduced to writing in a stream-of-consciousness style. The essence of the technique is to get ideas on the page (ok, screen) as quickly as you can for a specified period of time, such as 15 minutes, without worrying about details. For instance, if I don’t have the energy to flesh out an idea, I leave a note and move on. If I am not feeling inspired to write on a given day, I set a timer for a minimum time to write. In the early days, this was 15 minutes. Usually, once I started writing, I would feel that I had more to say and would keep writing. I have written in my notebook, or in a text file for a first draft. This blog post first started as a kanban task, and you can take a look here (.txt file) for that first draft. The latex conference style file is not necessary when writing early drafts.
I have known some students who have a really hard time writing any first draft, a very common occurrence informally called “Blank Page Syndrome.” It may be helpful to consider that an advisor can give feedback on a draft. An advisor cannot give feedback on no draft.
Here are some techniques that I have suggested to people and also done myself. Again, the important thing is to find a method that works for you, not necessarily to follow anyone else’s method:
- record themselves with a sound recorder, use text-to-speech to transcribe.
- start with an outline.
- take email or direct messages about the work and dump them into a latex file, convert to a document by reorganizing the ideas.
- write on paper, later transfer to the computer.
- write in native language and translate using automated translation tools.
Since editing is easier than generating text, I have also sent a partial draft to a student, knowing that certain elements were not quite correct. The student then enthusiastically edited out almost all of my content and finished the paper. Mission accomplished!
I used to be very particular about where I could write, the workspace, etc., but with the addition of kids and multiple projects, now I write whenever, wherever I can and sometimes with interruption. Because of this, my focus is very much on doing what I can in the moment, and abandoning any hope of generating a great stream of text the first time. For me, I love listening to music and am more relaxed listening to music, so if I am writing, I am usually listening to music at the same time.
Timing and Feedback.
I like to spread the writing process over a long time period if you consider it on a calendar, but the actual time spent writing per day is small. For instance, I tend to work in 1-1.5 hour blocks writing or editing. When I was at work, I put a “please don’t interrupt” sign on my door for these times. After a writing time block, I would break for tea, other tasks, or a walk. This approach allows time for me to think about how to organize the content and what other items may be needed. My process has some similarities to those in Jeffrey J. McDonnell’s article, where he talks about writing at a specified hour per day as a way to consistently move work to publication. When I consider myself in “paper-writing mode,” I am writing or editing for two blocks a day. Figure creation, email, code, and other tasks fill in the rest of the day.
The next sections will give an overview of the different draft stages, and the role of these draft stages when working with collaborators. I work with collaborators with multiple professional and personal commitments, and I recognize that the first step of feedback – reading a draft – takes time. Consequently, I plan to send my collaborators some kind of draft at least one month before the deadline. Usually this is an in-between draft with notes for sections that may not be complete. By the way, I ask for a similar courtesy from my collaborators when I am not leading the paper.
Early drafts.
I start with a 2-page document that is written mainly for myself. With this document, I try to answer the questions of: “what is this work about?,” “what are the contributions?,” and “how does it move the community forward?”. This draft is an overview of the work. Usually I include a rough figure or two.
Sometimes, I send this document to my collaborators to give them an idea of where the work is heading and how their data will be used.
In-between drafts.
Figure 1. An early version of this post, printed and then I wrote comments on the printout. Those with children in the USA may recognize the Melissa and Doug-branded puzzle board – they make excellent portable work surfaces!
As mentioned in the first section, I use a method of getting the ideas into a document, and restructuring later. Often, by writing, I get a better sense of how to restructure the ideas to make them more clear. I revise the draft often. I don’t have a rule, but usually after I add a new section, I will read the whole draft and start revising the parts that now don’t make sense. I leave latex comments for items that may need to revised in future versions. I like to print out the document and mark it up as in Figure 1. Marking up a physical copy also gives me a welcome break from the screen.
When I am restructuring and rearranging the text, I like to structure paragraphs so that there is one main idea per paragraph. I indicate this main idea to myself by a commented note so that when I am scanning the document, I can easily see the progression of the ideas. Since I am writing this post in Markdown / HTML, I am using HTML comments. If I were writing in latex, I would use a latex comment. To illustrate, the comment for this post, this paragraph, is as follows:
<!--restructuring/ rearranging ideas, practically how to do that with comments. -->
When I am restructuring and rearranging the text, I like to structure paragraphs so that there is one main idea per paragraph. I indicate this ...
When submitting a completed paper to arXiv, you can quickly remove all of these comments in a submission version with the arxiv-latex-cleaner tool.
I will send in-between drafts to collaborators as well. Usually at this stage, I want feedback for technical issues, but I may also request feedback about writing and expressing the ideas. When the paper is near completion, we work on writing more, and I will start to ask questions like, “surely there is a more elegant way to express this concept, can you help here?” I use a different text color for notes to co-authors. In latex, you can do this by creating commands for each co-author as follows:
\usepackage{tikz}
%%% or \usepackage{xcolor}
...
\definecolor{blue}{rgb}{0, 0, 1}
\newcommand{\amy} [1] {\textcolor{blue}{\textbf{Amy:} #1}}
...
\amy{Co-author X: what do you think about this argument, does it make sense to you?}
Final-final-final drafts.
When the paper is starting to look complete, I pretend to read the paper like I am a new reader. Is the flow of information logical? Can I understand certain paragraphs only using the information from the preceding paragraphs? Can any ideas be refined so that they are more clear? If you have time, you can send your paper to friends or colleagues and ask them to read with these questions in mind, and to mark any regions where it was difficult for them to understand the text.
I also read the paper out loud. This will often help me catch little editing mistakes, like missing words, additional words, or copy-paste errors.
Finally, I hand the paper over to my husband. He is my proof-reader extraordinare. I find typos in others’ writing, but in my own writing I tend to just skim over them.
Then, I submit!
Check-as-you-write style and spell checkers.
There are new and old tools that check your text as you write, for a variety of issues, such as spelling and grammar. If you find that you are altering your text for the tool, and especially if you have trouble starting drafts, my view is that it is better to run these tools after you have done some writing and editing to organize your ideas. I will reiterate that as an advisor or co-author, I can work with words that have grammar issues. I can edit grammar problems. I cannot work with no words.
Other resources:
Thanks to Dmytro Mishkin for feedback on an early version of this post.