Tips on Scientific Writing

At the moment, I am providing a lot of feedback on graduate student writing and it feels like I see the same types of errors that I (and I know many others) made when they started scientific/academic writing. As giving the same advice over and over falls into the “three times makes a blog post” rule, it’s time to collate the advice I often put forward into a single resource I can point people to.

The whole post is as assortment of thoughts in no particular order. I am initially publishing this post in January of 2023, but I think this will become a #living-post where I will continue to add as time goes on.

Active Voice

Decide on your voice before you start writing. As in, explicitly tell yourself who is doing the writing and who is doing the reading.

For me, this often means reminding myself explicitly that I am writing in the first person and will maintain an active, not passive, voice.

I am writing this blog post. A blog was not written (by me).

We collected data from participants. Data was not collected via an online subject pool.

You can of course swap between these two types when appropriate, but this should be a conscious choice.

Your Audience

I also find it helps to hit the right tone by creating a persona of my intended reader to make writing feel more manageable.

I tell myself…

To really make this come alive, I will often write a name of one of my colleagues at the top of the page to remind myself a real person will read this. This, to me, makes it real and allows me to better understand what I can and cannot expect my reader to know. It also allows me think think more clearly about what they might get out of reading what I have wrote.

For this post, there are three graduate students at the UvA whose work I have read and edited recently that I think might benefit from reading what’s on this post.

They all are very smart. They are just getting started in writing about music cognition. They want to improve their writing and be taken seriously. Not all of them speak English as a first language. Each sentence I write here, I write with them in mind.

Empowering Your Reader

One of the goals I have as a writer (and educator) is I want people to feel smarter after they read something I wrote. I am not sure exactly how I can turn this into actionable advice, but I feel it fits in with deciding not only who your audience is, but what you want them to get out of you writing. I think you have to be empathetic in order to be a good writer.

Not everyone shares this view, but I don’t want someone to feel dumb after reading what I have wrote. I don’t believe this has to be the case when it comes to writing about stats or anything technical.

Tense

It’s important to also spend a moment deciding when things happened before you write. This is again a writing choice that maybe (?) can change within a document, but needs to be a conscious choice in my opinion.

Scope of Claims

One problem I often see a lot in the student writing I edit is over-the-top, grandiose, bold, declamatory verbiage. Most writing guides aptly point out that these adjectives really don’t add much to the narrative and distract from the main point.

It’s really not up to you as a scientific writer to convince your reader of how important or relevant or impactful your research is. Your reader is smart and can decide for themselves.

I really don’t think we can know what is going to be impactful and not for the most part (see: the entire history of science). Save this kind of writing for when you’re trying to get funding. Only write what you would be comfortable saying in front of a big room of all your favorite scientists.

Motivating Your Writing

One rhetorical technique I wish would die is motivating a subject’s discussion by saying no one has done it before.

I, of course, have used this. We all have. And if you work in research it is so important to have an element of novelty to your work. But often this goes too far.
Just because no one has done it does not mean that it’s a good idea to do.

But what makes something a good idea to do? This really depends on you and your reader’s scientific values.

In my own little, crazy world, I don’t think you should have to justify. I very much subscribe to anything goes ala Feyerabend. It’s my view that scientific funding and opportunities to present at conferences would be more fair if we adopted more of a lottery scheme. But I know that is fringe and I have upset some senior academics when I have suggested this approach be adopted by certain exclusive conferences…. And I want to have a career in this, so I will just leave it at that for now.

The more sensible option (in my opinion) is to take the “standing of shoulders of those that came before you” approach. Other researchers have 100% thought about what you have thought about before. It is nearly impossible to have totally new ideas, especially when you are just starting out. If someone in your field has not done it, someone in a related, more mature field probably has. Just look at everything music cognition borrows from computational linguistics. And other than being better than saying no one has done it before, it also sets up your Discussion.

In practice, this approach looks like…

  1. Write a series of small, short, factual sentences of what other people have said on a topic.
  2. Only write statements that you can support with a reference.
  1. Continue this conversation by then adding in what you can contribute.

Now whatever you do is connected to such a richer discussion. You can talk about the findings of your study in reference to so many other claims. Ideally your results help disentangle what other theories might predict.

This also helps you move away from narratives that read like the following:

  1. People have studied a general population doing task A.
  2. But what about this special population doing task A?
  3. Well we think that the special population will be different from the general population (but we won’t posit a mechanism as to why this might be!)
  4. Collect data, find a significant difference, claim there is a difference (but you still can’t say why they would differ, which is also a problem considering that with enough data there will always be a statistically significant difference.)

This type of narrative, in my opinion, really stems from our statistical methods/training where we think that it’s acceptable to say that “we’ll its not nothing! p <.05”. I also thought this was what science and research was about for some time, but the more I read and talked with people who were way better than I was at thinking about research methods, the more I realized how uninformative this way of thinking is.

As scientists, we want to know about the much larger network of causal relations that make the world what it is. Really engaging with literature that links theories to hypotheses improved my thinking a lot on this. Also thinking about the causal mechanisms responsible (aka read Statistical Rethinking).

Tools to Get Going

Simple Language Tool

When writing scientifically, one of the pieces of advice I often give is to tell a story using short, factual claims that you can support empirically. Ideally this follows some sort of funnel metaphor where you start with the bigger claims, then work your way down to your specific hypothesis.

Said another way:

  1. You claim a lot of other people have talked about a topic.
  2. You talk about a subset of that larger discussion.
  3. You say what some of the problems are with that discussion.
  4. You then arrive at your insight.

But in constructing this funnel over and over again, people often trip on their own feet. They want to write big, beautiful, complex sentences. What ends up happening is that the first draft is just very confusing and wordy.

One way to get around this is to write the general narrative you want using something like the XKCD Simple Writer Tool. Take ten minutes to write in as simple language possible the narrative you want to convey. Then once you have this skeleton, you can build up that structure in order to suit your taste.

Multiple Pass Method

The other way I find helpful to write is sort of like a Frankenstein method (aka building a monster).

It goes like this.

  1. Write down in short bullet points every single thing you want to talk about on a page.
  2. Order those bullet points into a narrative that makes sense (adding more bullet points as needed!).
  3. Turn the ordered list into an outline (add titles and headings and transition points).
  4. Add references to each part of the tree when you know you’re going to need to support what you say.
  5. Write your paper with this outline as your guide (several passes and all the critical information will be there).

This is the method that I use primarily, also because it allows me to use all the vim + emacs key bindings I like to use. Though not part of the writing process, this makes writing more enjoyable for me. The process of writing should be fun.

Further Reading

In order to write well, you also need to read a lot of good writing and books about writing good. The goal for writers should be to develop their own taste and preferences and be able to justify these.

In music science, we have a lot of great writers. At some later point I will add a list of my favorite writers.

Books About Writing

I also enjoy reading books about writing.

Below I list a few that I have read and some notes about them.

Potpourri

David Huron Writing Advice

This link here is a list of advice on writing empirical papers in music and science from David Huron.

(contributed and hosted by Daniel Shanahan)

Commmiting to signposting

It’s OK, important, and makes it easier for your reader if commit to some sort of signposting. What this means in practice is that if you say you are going to talk about three possible explanations for you data, you do that, in the order you brought them up. This allows your reader (reviewer!) to just ensure that you did what you what you say you were going to do.

(paraphrased, suggested by Dominique Vuvan)

Shifting Levels of Focus

Be aware of what level of focus you are currently talking about and don’t dive too deep or ascend too fast (it gives the reader the bends!). For example, at the start of your Discussion, you can talk globally about what you set out to accomplish and if you did that. You might stay more global for another paragraph or two to say how this links to other work that has been done before. Then one you have said all you need to say at that top level, you can dive down and talk about more specific ideas.

(paraphrased, suggested by Leigh Van Handel)

The Sportsbroadcaster

I absolutely love this next one from Leigh Van Handel.

When watching sports, there’s often one person whose job it is to describe what is happening in a game. The other person’s job is to explain why that is interesting or important. As a writer, you want to be more like the second person.

(nearly copied exactly, suggested by Leigh Van Handel)

The IKEA Flatpack

When writing the Materials and Procedure section of your paper, have an IKEA model in mind. In Materials you list everything you have and need to assemble this experiment. The Procedure is the step by step assembly instructions.

(paraphrased, suggested by Dominique Vuvan)

Write as a real person doing real things with other people

The scientific literature does not have its own agency. Research is done by real people doing real things in the real world. Stay away from phrases like “the literature suggests” whenever you can attribute these ideas to the people and events that actually lead to their creation.

Avoid language that will get people’s backs up

This one goes in line with needing to respect what other people have read and thought. It’s important to avoid language and words that will get people’s backs up. In practice, I think this translates to avoiding expressions that have explicit value judgments.

For example, one idea that often crops up in the music cognition student writing that I often see is how to talk about, what some people might refer to as, “real music”.

Whenever I see this, I have a general idea what the writer wants to convey. They often want to delineate the sounds that people might actively experience in everyday listening from these bleeps and bloops we play for people in experiments in the name of control.

But to get to the actual issue, it almost follows a meme like template:

Does the existance of “real” music imply the exitance of “unreal” or “fake” or “not real” music?

For native English speakers, unreal, fake and not real all carry a lot of (different) baggage when it would come to describing music.

If you are writing and find yourself using a term like this, know it’s not correct, the best thing to do is just flag it and ask a trusted colleague to help find the right technical term. And it really is a technical term in this case, which again is fine, writing for academic audiences can be technical. In this case we might call it “naturalistic” (possibly problematic!) or “ecologically valid”. It would really depend on the context here, but I think a rule of thumb of avoiding value laden terms explains a lot of this type of problem.

This happens all the time in interdisciplinary research since everyone has slightly different terms for everything. When writing scientifically, there are a lot of different people from different countries, with different levels of English, schools of thought, and assumptions reading your work.

Post History

Have a piece of advice you this others would benefit from reading?

Comment on this Post on Twitter

#writing #science #gradschool #living-post #tips #reference #faq #email-faq