A few weeks ago, a pal sent me a satirical article about the use of the em-dash. It was written in a response to a more serious article from Inside Higher Ed. In short, this punctuation mark has come to prominence of late because Large Language Models often use them when asked to write creative passages.

Before we dive into that, however, what is an em-dash?

At its simplest, it’s a wide horizontal line that was historically the same width as the letter ‘M’ on a typewriter. In the age of the desktop computer, the size doesn’t necessarily scale directly. It always runs up against the neighbouring words:

It seemed the problem—at least, we considered it a problem—was that everyone was using the em-dash again.

It lends a disjointed pace to the writing, either introducing related thoughts or changing the subject.

This should not be confused with its sister mark, the en-dash. As its name implies, this was once the width of the letter ‘N’. This is a more flexible mark:

The Duckworth–Lewis Method – and its later variant with Stern – has been adopted by the International Cricket Council.

Here, the first dash is used to join the words Duckworth and Lewis without spaces either side. The others two do have spaces and are used like brackets to add extra information. This punctuation mark is not typically used to imply the same pace or a disjointedness as its sister.

Merriam-Webster has a more detailed guide to such punctuation, although I notice the name of the dictionary itself uses a hyphen, not an en-dash.

While reading up on why Large Language Models favour the em-dash, I found no clear answer. My theory is that many of them have been trained on older works, when this mark was more prevalent, and the software is picking up on this. The satirical article from earlier specifically names Mary Shelley and Emily Dickinson as heavy users.

I’m such a heavy user of the en-dash that I have a Windows keyboard shortcut especially for it. As I write, however, I think back to the one and only time I’ve used the em-dash extensively in a written work: my university dissertation.

Throughout the course, I was determined not to have my work pigeonholed into one category or another. This worked well until I needed to put together a cohesive dissertation based on what I’d written. The solution was to borrow a somewhat scatterbrained character from one of my stories as a connecting thread. I would make her use the em-dash to talk through the pieces she had supposedly written in character.

I graduated with an MLitt Writing Practice & Study in 2016, long after automated plagairism software was introduced, but a few years before Large Language Models were available. Had I handed in that essay today, I’m convinced I would have been questioned about it even if the plagiarism test had come back clear. But this old-school punctuation mark really was the best choice for the story at that time.

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.