Auto-Apostrophe correction with CALT-Feature

Thomas Schostok

Dear typography enthusiasts

I just had some research on this forum, but I couldn't find anything related. I just want your opinion about an “Auto-Apostrophe” correction as an OpenType CALT-Feature.

We are all familiar with the incorrect use of quotesingle ' instead of the correct quoteright ’ for use as an apostrophe. Incorrect micro-typography that most people don’t care about or still consider correct, for whatever reason.

On the Apple Mac, the system-wide autocorrect usually kicks in, or the apps (e.g. TextEdit, Affinity, Adobe, etc.) use their own autocorrection to correct such errors. Does anyone know how Windows or Linux manages this? 

Recently, a customer sent us a graphic that he created himself using GIMP for Windows. It contained a German text with an incorrect apostrophe. Instead of geht’s he wrote geht's.

In such cases wouldn’t it be an advantage if an OpenType CALT feature could solve the problem?

The CALT feature would automatically replace the incorrect use of the quotesingle and also the grave and acute with an apostrophe (quoteright).

Put simply, the combination “lowercase + quotesingle + lowercase” would automatically result in “lowercase + quoteright (the correct apostrophe) + lowercase”.

Capital letters or single letters would of course also be taken into account. 

The advantage of the CALT-Feature would be that it is activated in most cases, but it can also be switched off if the designer wishes.

We have tested this in various combinations and it works perfectly.

The questions:
What would speak against it? Are there any languages or situations, apart from the display of code or passwords, where automatic replacement would be problematic?

When displaying code, a text font would probably not be used anyway. We have not yet tested password entries in the browser but while I am writing this text in the browser, the texts are not automatically corrected with regard to the typographical quotation marks.

The feature would be more or less intended for the end user who is not a designer and familiar with correct punctuation and typographical quotation marks. It would be a kind of paternalism, yes, but the same could be said for the regular autocorrection.

What do you think?

Screens:
1) Example of incorrect use of pseudo-apostrophes
2) CALT-Feature corrects the misspelled variations

Thomas Phinney

Microsoft Word does this automatically.

Having lived with this for many years, I can attest that there are unfortunately situations where the auto-apostrophe gets the wrong results. Two very common ones are (1) code references or the like where one actually might want “dumb” apostrophes ('calt' feature), and (2) cases where one should have a leading apostrophe rather than a leading open quote, such as when elision is being indicated (e.g. ’80s, ’til).

John Hudson

Despute algorithms occasionally getting this wrong, it is defintely something that should be handled at the character level, and not something to be attempted in glyph processing. Remember, calt and other glyph substitution features are only affecting the appearance of text, not the underlying character encoding, so even if you manage to program a calt feature that gets this right, or as right as other algorithms—for all languages and national conventions—, it only works in your font and fails when the font is changed.

Essentially, what you propose is hiding problems in text encoding behind visually correct glyph display.

The CALT feature would automatically replace . . . the grave and acute with an apostrophe (quoteright).

If someone has used e.g. ` as an opening quote, they have mistyped the text, and hiding this error behind a glyph substitution is a terrible idea.

Christian Thalmann

Much as I hate seeing graves used at apostrophes in the German-speaking world (I guess it's the most accessible quote-like thing on the keyboard), there are cases where substituting a grave would be a grave mistake. LaTeX code comes to mind, where ‘this’ text is encoded like `this'.

Craig Eliason

Your system would fail the "rock 'n' roll" test, I suspect.

Andreas Stötzner

We need new keyboards.

Nick Shinn

I addressed the issue in Bellefair by angling <quotesingle> and <quotedbl>.
(Perhaps not the ideal genre for this method!)
But this is over-ruled by “smart quote” apps.
The only really effective remedy would involve a dictionary and grammatical knowledge. And that is something that could occur with AI attached to the layout app—in keeping with John’s comment re. “character level”?

John Butler

I could swear that standalone grave is used as its own special delimiter in certain shell scripting languages. Standalone grave, ^ centered circumflex, ~ centered tilde, \ backslash and | pipe all bollixing up our keyboard layouts instead of proper quotes and proper en and em dashes.

In one custom layout I’ve set up, one keystroke does both “” quotes, likewise for ‘’ and ›‹ and »«. I wish I could get it to back the cursor up as well, so I can keep typing inside the delimiters then just → out of it when I’m done. Other fence-type delimiters could be set up similarly, () [] {} &c.

In another keyboard layout, I keep everything normal except replace QWERTY with ABCDEF. Dvorak notwithstanding, I’m amazed no one has bothered doing it this way yet. It’s no worse than QWERTY, and it’s far more intuitive for anyone learning to type. Ironically it was my least favorite interface—arrowing through the alphabet to enter search terms on a Roku TV—that gave me this mundane idea.

Andres Torresi

I believe fonts should not alter the encoded text, much less assume the user intention (although I really like the idea of solving real world problems). If a codepoint is entered, the font shouldn't change it, (unless there is a very specific requirement, controlled usage). You can't foresee all present and potential context scenarios, and you can't stop the user to write a specific string for whatever reason, the font should deliver anyway. It's a text shaping issue. For example: a new brand or online slang that uses a quotesingle intentionally. 
Similar principle should be applied to other features like localization, where encoded characters are swapped assuming the user intention. These hacks have some roots in the past, but as soon as there is a code point, is keyboard/layout/text shaping issue, not fonts. Again, as a general principle, there might be exceptions or situations you need or want it differently. 

Andres Torresi

And I'm doing exactly what you want to solve, because my phone does it. 

Thomas Schostok

The “Rock ’n’ Roll” argument got me! 
Yes, I agree, fonts shouldn’t alter the text. It was very enlightening and interesting to hear your examples and arguments, things I hadn't thought about before.

Maybe one day there will be a database-supported autocorrection within the operating systems that takes into account exactly such eventualities, if we actually need such a thing. Until then, we'll be stuck with the wrong spelling on signs, advertisements, etc. Let's enjoy it ;-)
Thank you very much for your feedback!
Thomas

Kent Lew

In terms of other language conventions, there are some where the replacement of ' by ’ would be considered incorrect. Many languages use some form of single-quote-like mark to represent the glottal stop. But not all of them accept ’ as a reasonable substitute.

For instance, in some South American languages, the correct character is ꞌ (U+A78C, saltillo). In the absence of font support for saltillo, a straight ' is used, and in this case the substitution of a curly quote would be downright inappropriate (though no doubt encountered, thanks to Word et. al.)

Similarly, there are Polynesian languages that use an ʻokina to represent the glottal stop. (Hawaiian among them, as in the word Hawaiʻi itself.) Unicode has encoded ʻ (U+02BB, modifier turned comma) for this. However, in absence of font support, the use of a left curly quote ‘ is preferred. A straight quote ' is also a frequent occurrence in less typographically aware environments; but substituting the straight quote with a curly right one ’ would be inappropriate.

Peter Constable

One nice thing about Word’s auto-correct replacement of straight quotation marks with curly quotation marks is that the replacement is treated as an edit that can be undone. So, for example, if I'm going to type 'calt' and on entering the first apostrophe it automatically gets changed to ‘, pressing cntl-z will undo the automatic replacement, changing ‘ back to '.

Nick Shinn

In USA English, single quotation marks only appear rarely (in quotations nested within quotations).
However, the apostrophe is much more frequent.

Therefore, a much less fail-prone “smart quotes” would occur in USA English with:
• <USA English> language tag.
• <quoteleft.USA> character, with apostrophe glyph, for use with that language tag.