Nitsuh Abebe writes in the NY Times (archived) about a kerfuffle that had hitherto escaped me but is obviously in my wheelhouse, “Whose Punctuation Is More Human: Yours or A.I.’s?”:
There are countless signals you might look for to determine whether a piece of writing was generated by A.I., but earlier this year the world seemed to fixate on one in particular: the em dash. ChatGPT was using it constantly — like so, and even if you begged it not to.
As this observation traveled the internet, a weird consensus congealed: that humans do not use dashes. Posters on tech forums called them a “GPT-ism,” a robotic artifact that “does not match modern day communication.” Someone on an OpenAI forum complained that the dashes made it harder to use ChatGPT for customer service without customers catching on. All sorts of people seemed mystifyingly confident that no flesh-and-bone human had any use for this punctuation, and that any deviant who did would henceforth be mistaken for a computer.
Those deviants were appalled, obviously. I am one; I am, even worse, a former proofreader who could speak at length and with passion about the uses of the narrower en dash. I understand very well that this dash-happy lifestyle is maybe atypical, but I had not expected to see its whole existence questioned. The dash is a time-honored and exceedingly normal tool for constructing sentences! Dickens, Dickinson, Nietzsche, Stephen King novels, this magazine — all strewn with dashes. Part of what makes them popular, in fact, is that they can feel more casually human, more like natural speech, than colons, semicolons and parentheses. Humans do not think or speak in sentences; we think and speak in thoughts, which interrupt and introduce and complicate one another in a neat little dance that creates larger, more complex ideas. (Or, sometimes, doesn’t: The copious dashing in J.D. Salinger dialogue is a great illustration of all the thoughts we leave unfinished.) This is the whole thing punctuation is for. […]
I am not writing this to defend dashes. I am writing this because I want to suggest that the phrase “everyday use-cases” signals a genuinely epochal shift in our perception of what writing even is.
Consider that, for a good stretch of recent history, most of the written material that people spent time with — the stuff beyond signs and menus — was full-on writing-writing: text that somebody sat down and composed, maybe revised or edited, maybe even had professionally printed. And this kind of communication was different from our daily interaction with our peers: You talked to your peers, mostly. Even after the internet arrived, this basic psychic arrangement persisted.
And now it does not — like, at all. “Emails or text messages,” posts and chats, DMs and comments, DoorDashers telling you the restaurant is out of coleslaw: Oceans of communication that used to be handled by speech are now left to lone individuals typing into the internet. Even if you remain a dedicated reader, you may still end up spending more of your time dealing in on-the-fly typings, because that has become the everyday use-case of writing.
The whole thing is worth reading, not least for Mr. Abebe’s splendid style (varying between polite dyspepsia and genial enthusiasm); I am in thorough agreement with him on all points, and I am infuriated by the common tendency among the terminally online to assume and proclaim that whatever is not personally familiar to them does not and must not exist. Thanks, Eric and cuchuflete!
To be fair, once upon a time there was a world of notes sent by messenger or by mail, serving much the same purposes as texting: note to the grocer to be delivered by the domestic, or notes to a friend or a business across town, sent by a messenger boy. Admittedly their volume was much smaller.
(I once found inside a Parisian 19th century binding a contemporary laundry list. That was exciting. Unfortunately it was not mine to keep.)
Presumably the reason Y’s once-upon-a-time did not endure is that the rise of the telephone undercut the functional need for quite a lot of written communication, probably including a lot of the quickly-dashed-off sort. To be fair Abebe specifies “recent history,” presumably after the rise of telephonic communication.
Just last week, a couple of my colleagues presented our AI workshop aimed at both understanding the threat (deepfakes initiating crisis response, mostly) and potential uses for people in our field. And someone raised her hand and said “I can always tell AI-generated writing by em-dashes.” And I was like “crap, everyone’s gonna think I’m an AI.”
[But having now read the remainder of the post, it’s a good deal more interesting than the narrow issue of em-dashes.]
I — am not — a robot.
It’s easy to write en–dashes and em—dashes here, and, ‘pon my word, there are fewer glitches than with “smart quotes”.
As this observation traveled the internet, a weird consensus congealed: that humans do not use dashes.. That suggests—to me at least—that those doing the congealing are not accustomed to reading non-interweb printed matter.
I learned to set type around 1957, at age ten or thereabouts. The job case had particular boxes assigned to en quads, em quads, and dashes of various lengths. I’ve been using em quads ever since, and can’t see any reason to drop them from my punctuation palette.
@cuchuflete, perhaps you meant using em-dashes ever since? Otherwise I would think you might be typing like this. Alas, I have but one font of Caslon 12pt that has matching em-dashes with it. For every other typeface you can tell their em-dashes are zweibelfisches.
On the topic of quads, were you trained in the tradition of thins, mids, thicks or 5-to-the-em, 4-to-the-em, 3-to-the-em, etc? I have one case that has 7/em, 8/em, 9/em demarcated, which I never use for fear of not being able to distinguish them again when dissing.
Em dash named after Emily Dickinson — via LLog [scroll down]. Elle Cordova is particularly — but entirely justifiably — scathing.
@chuchuflete, I can’t see any reason to drop em from my punctuation palette either.
and, ‘pon my word, there are fewer glitches than with “smart quotes”
I see what you did there…
@pc, Thanks. You caught my error. I meant em dashes, not em quads. I am not familiar with variable quads as you describe them, having had only California job cases full of such faces as Century Schoolbook, Park Avenue, Baskerville, Copperplate Gothic,Palatino, Cheltenham and my best beloved Goudy Old Style. Sizes ran from 6pt up to 48pt and even 72pt for 19th c. wood type display faces. The first press was a 6” x 9” Sigwalt with multi-colored decals like an old Singer sewing machine on the cast iron body.
https://www.briarpress.org/36220
I was not a letterpress-typesetting child or teenager but an old-fashioned-manual-typewriter one, meaning that I could produce either a hyphen or two-hyphens-together as a simulation of a handwritten dash. When I learned probably some point in my twenties that there was a supposedly important distinction between an em dash and an en dash and that a hyphen was supposedly another thing altogether (and possibly a minus sign yet a fourth thing?), I was paralyzed by this new-to-me complexity and reacted by eschewing dashes thereafter rather than try to learn the subtle distinctions that were claimed to exist by those who cared about such things.
I am, even worse, a former proofreader who could speak at length and with passion about the uses of the narrower en dash.
As could I; indeed, as I did when I was the main agent in an epic reformation of dash provisions at Wikipedia’s Manual of Style, back in the Classic era. See the current provisions here; they have stood up reasonably well, but show a more pronounced AmE prejudice than when I left the scene.
In my experience very few academics use Word with good understanding of hyphens (standard, optional, or hard), dashes (en or em), minus signs (different from both hyphens and en dashes), or hard spaces. How could we expect AI to glean wisdom from the leavings of such writers?
A final gratuitous gripe: Let no one ever call me or any other editor a “proofreader”, whether former or active.
(Op, I didn’t notice that all my extra spacing in “Otherwise I would think you might be typing like this.” had gotten removed. An em quad is a large amount of space.)
@cuchuflete – all classics! Alas, I have only Park Avenue in that list.
@pc,
You—and others here—might well enjoy this:
<
source: https://words.fromoldbooks.org/Jacobi-PrintersVocabulary/
The curse of the typewriter, where dashes are impossible, has prevented them from being included on computer keyboards. The Mac has solved the problem with Cmd + hyphen or something; MS Word has solved it with Ctrl + minus (on the numeric block!) for the en dash and Ctrl + Alt + minus for the em dash (…paradoxically easier on the German layout, where there’s an AltGr key; the em dash is not used in German…). But few people seem to know that. Let alone Alt + 0150 (en), Alt + 0151 (em), which work pretty much everywhere. Most people seem to believe that if something isn’t shown on their keyboard, they can’t write it.
Scientific journals often have rules about dash usage that authors have to follow before their manuscripts can be accepted – they’re not left to the copyediting process. My current manuscript is full of em dashes that I’d never use on my own.
I vaguely remember a mid-century lament that the correspondence between scientists, once conducted in properly archivable letters, was now lost to the telephone. It is back to writing…
“which work pretty much everywhere” – not everywhere:(
But I know the combinations… and don’t know those for Word because I don’t use it.
I think a more practical solution (if don’t change machines often) would be creating combinations by your taste.
When using TeX or LaTeX, you get an em-dash by typing — (i.e., three hyphens) and and en-dash by typing — (i.e., two hyphens). Very easy. I use both quite frequently.
The software here seems to convert both to an em-dash when displaying them, however. No en-dashes for me…
George, one hyphen between spaces ” – ” is en-dash here.
(between words:
one-hyphen
two–hyphens
three—hyphens)
P.S.
four—-hyphens
five—–hyphens
one- hyphen
two– hyphens
Aha, more ” smart ” quotes
I think it’s interesting that, even when the topic of different dashes and their lengths does come up, people never seem to talk about the height at which the various dashes are set. In a lot of fonts, the standard appearance of the N-dash (when it is used to indicate a range of numbers, dates, etc.) has it set lower than a hyphen or M-dash. This is often the most visible difference between the N-dash and a minus sign (which is typically very similar in length, but set significantly higher).
When using TeX or LaTeX, you get an em-dash by typing — (i.e., three hyphens) and and en-dash by typing — (i.e., two hyphens). Very easy.
And it goes without saying that you get a minus sign by typing a hyphen in math mode.
@Hat, thanks for informing me that I—like many here, it appears—have been failing the Turing test since high school.
@George Grady: Thanks for informing me that I can type — here more easily than with &mdash.
I […] have been failing the Turing test since high school
Fascinating! Turing test normally exists for machines, whether they good enough at emulating people or not. Not human beings good or not in differentiating themselves from machines. Are we that far gone already?
With the compose key — a thing you can enable on Linux out of the box, and on Windows with WinCompose — you can type compose, -, -, – for em-dash and compose, -, -, . for en-dash. I replaced caps lock with it years ago. On cellphone keyboards it’s also fairly simple to type an em-dash: hold on a regular one. However, in a chat you’re generally more likely to use send or delete instead.
While the eagle-eyed reader may have noticed that I surrounded them by hair spaces, unfortunately there’s no (default) entry for that. Unicode entry with Ctrl+Shift+u, 200a, space / enter is one possibility.
For many of these items I override Word shortcuts with my own Windows-wide shortcuts, coded using the brilliant and easy AutoHotkey freeware. Strongly recommended.
Yes, in theory I’m with Noetica.
And with gaming mice.
In practice I’m lazy:)
Gaming mice? Euw …
Hat, did you ever end up reading Because Internet? This excerpt here seems like a much shallower retelling of McCulloch’s thesis.
No, I’m afraid I didn’t. So many good intentions, so little follow-through…
So can I get more solecism points by adding spaces around m-dashes, or by omitting them? Inquiring minds want to know! (We have one [1} German[?] vote for not adding them being correct, by a rough count, and the rest of you lot in opposition. WP says it’s “Old British Style” to omit them).
ObDanish: tankestreg, and a hyphen is a bindestreg. I wuzn’t larnt about the uses of N-anythings in Danish typography. (Number ranges and the like use hyphen or m-dash according to the length of the connected words/phrases). And it seems that lots of people print tankestreger as not much longer than hyphens, or identical to them. With preceding and following space.
But a minus is a minus, and belongs in maths.
Sometimes I use ChatGPT for auto translation. It’s a good tool but of course a few edits are needed, and during editing I replace all em dashes (and all guillemets in Russian). They are simply antithetical to the computer communication styles.
my use of dashes (generally en) is, i believe, consistent – but not based on proper inculcation in correct usage. it’s all in the dickinson/rukeyser/olson territory of rhythm/breath indications to me, which in my mind are thoroughly entangled with indications of sense (like cantillation-marks). so i like my dashes to not be hyphens, but almost always think an em-dash is more of a pause than i want.
and to Y and JWB’s points: part of what first persuaded me that we were having XIX Century II: Even Longer was noticing the way that email and text messages have recreated the texual communication systems of the pre-telephone middle/upper classes – multiple daily mail deliveries, telegrams, hand-carried notes, etc.
I learned embarrassingly late that current Hebrew typography uses spaced en-dashes instead of em-dashes.
Em dashes are seldom surrounded by spaces in English, but I’ve seen that in edited works nonetheless; evidently there’s a style guide or two that want it that way.
Gedankenstrich, Bindestrich*; the former is always surrounded by spaces, even for ranges. (Ideally perhaps hair spaces for ranges, but always spaces.)
* Rarely Trennstrich – “separation stroke”, because the hyphen arguably has two opposite functions.
IIRC the convention in the U.S. is no spaces, in the U.K. some variety of thin spaces around the dash.
Wot, no xkcd?
Although now people will realize three-per-em space that all this time I’ve been using weird medium mathematical space whitespace characters in my hair space hair space hair space speech dot dot dot…
(I note that it is “dot dot dot” and not “ellipsis”)
(I actually dislike the squished look of the Unicode ellipsis character that is often automagically substituted for three dots in a row, and often use dot space dot space dot)
Hm.
dot dot dot: …
dot space dot space dot: . . .
dot hairspace dot hairspace dot: . . .
dot zerowidthspace dot zerowidthspace dot: ...
dot emspace dot emspace dot: . . .
(It’s an elliptical workout)
Four. Four spaces, four dots.
YOU HAVE NO CHANCE TO SURVIVE MAKE YOUR TIME.
HA HA HA HA . . . .
They do something in Spanish typography that I find very logical, though I’ve not seen it in English or French. I don’t know if it’s universal in Spanish, or just happened to be used in Spanish texts that I’ve read. When em dashes are used to enclose a parenthesis —as here, for example— they put a space before but not after the opening dash, and after but not before the closing dash.
frans said
Good idea, but I haven’t succeeded in replacing caps-lock with anything except do-nothing (I’m the sort of typist who only ever presses caps-lock accidentally). Ukelele allows me to replace almost any key with up to 20 characters—for example, alt-d gives me access-date = —but it doesn’t offer a way of defining caps-lock.
I have no strong preference for whether there should or should not be spaces around M-dashes. I don’t use spaces, myself, but either one looks fine. However, when there is a space just at one end, it looks just egregiously wrong.
I prefer m-dashes to have spaces on both sides. I find m-dashes with no spaces somewhat uncomfortable. And just on one side just looks strange, but I don’t know what to think about it?
Consider the complete set of sentence-dividing punctuation marks used in standard English, as abstract entities with less-abstract realisations:
The unspaced em dash is unique among these, since it is the only mark that fixes the distance between the words adjacent to it. This makes the horizontal disposition of some full-justified lines uneven, and no remedy is available. For that reason I reject it whenever the client’s preferences and any prevailing style rules permit.
The spaced em dash is no better, because its width combined with two stretchable spaces will often make an awkward hiatus in a full-justified line. I reject it whenever I can.
The spaced en dash shares that fault with the spaced em dash – but its width makes it far less of a problem. I strongly prefer the en dash.
Editing AmE texts (and others) that require the em dash, I grit my teeth and comply. If I also have to use the non-sentence-dividing en dash in AmE ways, which differ sharply from BrE and OzE ways, I console myself with the fact that it is less likely to be confused with the sentence-dividing dashes that are in use; but I grumble at it also, and at the arbitrary inventedness and inflexibility of Chicago’s rules for it. In small ways they’re softening under pressure from rational argument, but it’s taking ages.
In all of the above I mean the ordinary space that we get with a click of the spacebar. Roughly standard-width fixed-width spaces (or wider ones like em spaces) both complicate the story and relieve some of the difficulties. Thin, hair, and other narrow spaces (always of fixed width, in standard realisations) are yet another matter; I don’t normally regard them as separate items in the punctuation inventory, but as rigidly applied or ad hoc adjuncts to punctuation marks. For example, if a text needs to have (unspaced) em dashes I’ll sometimes attach hair spaces at each end of every one of them, and conceptualise them as unspaced em dashes with a slight typographical adjustment.
@Athel
Apparently it’s possible to copy the compose key logic more or less directly to macOS, but of course it’s quite possible that some other approach is generally preferable on Mac.
https://web.archive.org/web/20200206084455/http://lolengine.net/blog/2012/06/17/compose-key-on-os-x
Talking of “AI”, I’ve now twice received this spam from academia.edu:
Sadly, as I have not clicked on any links in this exciting message, I can provide no further details, but I would imagine that the source is George Soros, continuing his nefarious campaign to undermine Western values by any means necessary.
It’s probably the Nigerian Royal Academy of Sciences. You just need to send them an initial contribution to become a member and give them the PIN of your credit card.
Though I yield to none in my admiration for the enterprise and creativity of the West African internet fraud community, I have a horrible suspicion that this is actually genuine (in the technical sense of being actually from academia.edu.)
The inanity is not inconsistent with the degree of intelligence displayed by a system which regularly sends me emails enquiring whether I might be the David Eddyshaw mentioned in papers by a certain David Eddyshaw.
Oh. I get the “are you the” emails, too, but I just think “I bloody well hope so” and don’t try to get the rest of the subject line displayed, so I didn’t know…!
I have stopped telling them NO I BLOODY AIN’T when they ask me if I’m that human metabolism researcher somewhere in the US who seems to get mentioned a lot. Because they promised me to take note of the fact, but did the emails stop? (And I’m not getting “Premium” just to be able to see what the other me is writing. I’m sure the emails would continue anyway).
On the other hand their system has figured out that I like historical linguistics, and I get links to some interesting papers I would never have found otherwise.
I stopped subscribing to premium because I stopped doing the things I wanted premium for. Now I’m getting the “Are you…?” mails without actually being able to check which papers I’m mentioned in. I couldn’t care less.
The juxtaposition of clauses with an em dash is the thing that suggests AI usage to me. “It’s not—it’s…” is the problem, not the use of an em dash in itself, and I think that it’s poor style even if it’s purely composed by a human without machine intervention.
I also find it amusing, because sure, people either don’t know the rules, they’ve overridden their machines and the software on them, or they don’t know how to use them, but you’re getting a hyphen or an em dash with ease (since two hyphens becomes an em dash in a lot of programs). As mentioned it’s trivial to get an en dash on a Mac, including on an iPad with a keyboard, and even on phones one can have more or less decent typography. So that it’s a tell-tale sign of AI is off, and that’s in addition to the phenomenon of “if I don’t know about it, then it doesn’t exist”. I imperfectly share your reaction (imperfectly, because I’m sure I’m guilty of it sometimes).
As to other languages and their conventions, you’re “supposed” to use, for French texts, em dashes for some things but a lot of editors use en dashes (like signaling dialogue; I think that I’ve got that right: that it’s supposed to be an em dash, but en dashes have supplanted the former, despite being a bit harder to type). Then mid-sentence, the spaces are inserted despite the problems that this causes with the line, and they use dashes in places where I find them hard to read, like for values in a range. In a project where I’m using Latin and where I get to pick and choose the conventions, I used Latin names or abbreviations of the book (Arabic numbers instead of I and II), commas instead of colons after the chapter, but an en dash surrounded by a space with multiple verses instead of a hyphen. Colons separate non-sequential verses. I find it elegant if a bit unusual.
Anyway, back to AI: I find it bizarre, because interrupting sentences with parentheses and em dashes is not purely an Anglo convention, but I think that it’s done more fluidly in English texts. So that means people use them in their real documents, as AI is getting it from somewhere! In contrast, it’s something that I have to watch when I write in French. I don’t see nearly as much of this interruption, which is always dangerous, since it’s about gut feeling, but that’s sometimes how it goes with your second, etc. language and usage.
Often it’s three, and two gives you an en dash.