Writing Is Decision-Making

It’s true, even though I said that I wouldn’t worry too much about students using ChatGPT, a few doubtful cases have made me wonder what to do about it. I’ve seen a lot of good advice and discussion already (see e.g. this piece by Matthew Noah Smith among further discussions and resources), but nothing has quite convinced me for my own endeavours and settings. What I am particularly worried about is that some students might stop entirely with working through crucial hardships of writing: trying out formulations, thinking carefully about structure and terminology, setting goals, failing, revising, refining and trying again. Obsessing (especially in how we grade) about the quality of the product (the exam or essay), we might forget about the point of teaching writing. After all, it’s not the odd successful exam or essay but reflecting on shortcomings and setting priorities that will foster learning. As Irina Dumitrescu aptly puts it: “But the goal of school writing isn’t to produce goods for a market. We do not ask students to write a ten-page essay on the Peace of Westphalia because there’s a worldwide shortage of such essays. Writing is an invaluable part of how students learn. And much of what they learn begins with the hard, messy work of getting the first words down.” The main reason for emphasising such tasks, then, is not to torture students, but to teach them thinking successfully and affording control over the process of thinking. So before I set out my ideas for examination, let me briefly motivate my approach.

Two phases of writing. – As I see it, writing is a kind of decision-making. While (1) tacitly articulating things in one’s head and attempting to write them down might count as thinking, (2) coming down on a certain way of phrasing means to decide or commit oneself to a particular mode of expression. It’s crucial to see that these are two very different phases and the way from phase one to phase two might be very long and disparate. As a student, I simply could not get to phase two without very torturous and long processes of trying things out. And sometimes I would never even reach phase two. Other times, I would need to write down two to three pages in order to end up writing and committing to a phrase that I had initially formulated in my head. It felt like slowly working towards finally writing down a phrase legitimately that I had idly considered in the beginning of writing my text. (Even if you’re different from me and do everything in your head before writing down a single line, you need to practise weeding out bad formulations before.) When you do this in handwriting, the constant crossing out and revising remains visible. Today, computers and formatting allow even the most hapless scribbles to look like parts of a finished book manuscript. The perfection of layout suggests a perfection of presentation that leaves the traces of desperate revisions invisible. Coming to phase two, then, means to have ruled out plenty of unsatisfactory formulations and alternative modes of structuring. Arguably, shortening this process of phase one by jumping on the next best phrase or sidestepping it completely by leaving it to ChatGPT means sidestepping thinking altogether and ending up with at text that no-one ever decided on.

Accordingly, I want to discourage students generally from unreflectively holding on to the first form of words that passes through their minds. Rather, I’m looking for tasks that make students ponder on their work and encourage second thoughts. So I hope to design something that works even for students who are not resorting to ChatGPT or other forms of cheating.

What I want students to go through. – Is this a fitting expression, and what is left out in using it? Does this structure work, given the content? What would change if I presented things in a different order? What is the main point I need to get across? How did I come to think of this as the main point? Should I rather focus on a seeming side-issue? Etc. Between the blank page and a successful piece, there are so many things and versions and other potential pieces that might be equally successful. Despairing over such choices is a crucial part of the process of writing. Leaving it to ChatGPT means learning nothing, nothing at all about writing and about yourself, let alone about ways to find your voice. Drawing out the gloomy consequences of leaving thought-processes to machines, Maarten Steenhagen sees us heading “towards a de-skilled society. More and more, thinking itself is being turned into a service, a product that is offered by some company or other. When people look for answers or want to understand something, they turn to Google, Bing, or to social media. There, they are likely to find easily digestible, byte-sized snippets that will do for most practical purposes.”

So what are the tasks I’m going to try out in my courses? – How can I see and evaluate whether students thought about the presentation of their ideas? I guess by asking to do so explicitly. So in future exams and essays I will add two kinds of tasks to the standardly requested answers (or papers).

1. On the level of content: Instead of having students just write down answers to exam questions, I will ask them to motivate their answer in relation to an insight they had. Ideally, this insight should relate to a previous discussion in class. It could take the form of “I think this or that in the light of the following idea, premise, assumption, argument (where the specific item relates to a discussion in class)”. If one wants to extend this procedure, one could add further steps to the motivation, such as an objection to the answer given and a tentative response to the objection. (This idea builds on my teaching of structured questions.) So whereas the actual answer is the item to be graded, the additional items (motivation, objection, response) ensure a relation to the previous action in class (or whatever you ask it to be related to). Obviously, the addition items allow for a fine-tuning of the grade, too, but the main point is to encourage reflection, ideally by means of relating to actual discussions in class so as to introduce elements that cannot be achieved by ChatGPT.
2. On the level of articulation: Here, I would ask students to add a reflection on their formulations or terminology. Either positively, by explaining why they have chosen a certain form of words, or negatively, by explaining why they have decided against a certain form of words. The precise term or phrase is for the students to pick. What they need to do is say something like “I used the term necessary because this excludes the possibility of exceptions.” Or, “I first thought about using the term thing but then I realised that what I meant could also include processes.” Again, the point is not to turn this into a demand for whater-tight arguments for certain modes of expression, but rather to encourage and monitor some level of reflection on one’s own language. It goes without saying that this also could be done or requested in relation to discussions in class. (Ideally, this exercise will also train the grasp of “operational concepts”, i.e. the means through which we express certain contents. See on this my conversation with Daniel-Pascal Zorn.)

While these tasks are thought of in relation to exam questions, they could also be introduced in essays and other assignments. Here, they could easily be requested in the form of footnotes offering some self-reflection.

I don’t know if these and related tasks will prevent ChatGPT from being used and abused, but at least the request to invoke discussions that happened in class will be difficult to mimic for such a device. In any case, they would take some reflection for making the relation, ensuring at least some reflection on part of the student.

At this point, I’m just beginning to experiment with tasks that encourage reflecting on one’s texts. I’m pretty sure, there are many people who have already thought of this and related issues more thoroughly. (* I am particularly grateful to Sara Uckelman for sharing her reflections. You can follow up on these on FB.) Please feel free to add ideas or, as always, comment on the ones presented.

***

As it happens, this blog is now up and running for five years. So I’d like to thank you all for your continuous reading, encouragement, and discussion.

Published August 18, 2023September 1, 2023