Idiomdrottning’s homepage

Re: Smart quotes are a skeuomorphism

November writes:

Smart quotes are a skeuomorphism. And they’re one that I’m not particularly fond of. […] A “skeuomorph” is when the design of something deliberately mimics an older element that’s no longer needed or in use. So like the faux wood paneling on newer cars, or save buttons in modern programs being shaped like floppy disks, or the whole concept of electric candles, or Winamp skins that look like jukeboxes.

He’s right.

I’ve often been bugged by the same thing; you copy and paste something and it’s borked because it is in fancy quotes.

(I use “smart quotes” to refer to the process of computationally trying to figure out which fancy quote is meant by a particular fancy quote. That’s not to say that my use of “smart” vs “fancy” is universally correct or better, just wanna try to get more consistent in how I write about them.)

Even worse, I go back and forth between using fancy quotes in the source text vs using straight quotes there but exporting them as fancy.

To make things even more, hmm, curious:

Straight quotes are also a skeumorphism! The original skeumorphism. In calligraphy and in block type setting, fancy quotes were the norm. Straight quotes were invented as a stop gap thing for typewriters. The use of straight quotes in ASCII is a skeumorphism from that era.

Straight quotes don’t have any semantic meaning of their own, they’re just placeholders for whatever quote is appropriate.

For example, there are a ton of similar-looking single quotes.

Probably more!

Good old ASCII 27 a.k.a. UTF-8 0027 is nothing on its own but is used a placeholder for any of these.

(Yeah, yeah, descriptivism and all that says that straight quotes now do actually have a meaning since they are in use in the wild.)

(Also, in the Lojban language, it’s a character. For a while, when typing Lojban, I would take care to write gi'e instead of gi’e even in the face of fancifying programs.)

So if we’re getting rid of onions in the varnish, of skeumorphisms, maybe it’s straight quotes that need to go.

He also writes:

If they weren’t separate characters, if they just DISPLAYED as smart quotes but were treated under the hood as straight quotes, that would be fine!

The problem is that it’s pretty unpossible to do this automatic conversion correctly for display purposes. For my Text-TV scraper, I often see how my own best attempts to implement this gets borked up.

Douglas Adams once said it could be done in only twelve lines, but I’m not so sure.

It gets borked when there are contractions at the start of a word, like: ’Tis the season. It also gets borked when there’s a mistake or elided quote, as is often the case on Text-TV.

For 7off, I went a little heavier. 30 lines. And the lines are slow and look-ahead-y, but it needs to figure out when we’re in preformatted text or when we’re in plain text, and it has a list of stuff like “rock ‘n roll”, “‘twas the night before christmas” etc. But it’s not possible to detect all left-contractions since contractable words are an open word class. For example, some people might refer to pizza as ’za with an elision mark. Which needs to be a right quote and would get borked by 7off unless you type it manually, like I just did there. It could be any word! Maybe some dork wants to call a giraffe a ’raffe? The display code can’t figure this out on its own since the character has a larger scope of intended semantics than a single placeholder can work with.

That’s why I sometimes feel that it’s better to do it in doing it in the underlying source text, like using Emacs’ smart quotes mode. If you do the smart quotes on-the-fly, as Adams prefered, you can see right away when things go wrong and fix it manually.

So why am I not doing that anymore? Because I found that that went wrong when I sometimes forgot, or when I was quoting text. My source texts at this point is a borked up mish-mash across hundreds of files.

I can’t have all straight quotes, because I’ve pasted stuff in from Fedi or from other smart-quotified places. I can’t have all fancy quotes, because I’ve pasted stuff in from classic text files.

I kinda need both.

See also:

“The Mac is not a typewriter”, book by Robin Williams


Jesse writes in:

I know that apostrophe-like characters constitute letters in the alphabets of many other languages too, especially Austronesian ones, and now I’m curious whether there are widely-followed typographical conventions among users of those languages about how that letter should be rendered (on the page and on the screen).

This sort of thing runs both ways—I believe the unusual-looking orthography of SENĆOŦEN (an Indigenous language of the Pacific Northwest) is due to the limited capabilities of the typewriter owned by its creator.

I’ve often wondered the same thing! My unsubstantiated guess would be to use single right close. For example, Hawai’i.


November wrote a follow-up post: