Document colon, comma, semicolon

[bwiernik]

@abdealikhurrum @p-heckler Can you take a look at these changes? Do they sound correct in terms of where the punctuation should be changed and what should happen to other characters beside them (e.g., spaces)?

For example, if the title of a journal is "Journal of Research: Theory and Practice", because the title actually uses the colon character, I assume that this should not be changed?

[p-heckler]

Absolutely, this is exactly the expected behaviour that I had in mind!

[denismaier]

Well, but what about titles? With the new title-split feature this would be no longer true.

[bwiernik]

I think including title-delimiter in this list makes the behavior clear enough? What do you think @bdarcus?

[p-heckler]

Well, but what about titles? With the new title-split feature this would be no longer true.

I am not sure I fully understand this new feature, but if a style makes use of title-split, localised punctuation should be the expected behaviour, shouldn't it?

[denismaier]

but if a style makes use of title-split, localised punctuation should be the expected behaviour, shouldn't it?

It will depend on normalize-title-delimiters and title-delimiter. Regarding the "Journal of Research: Theory and Practice" example: With title-delimiter=". " you'll get "Journal of Research. Theory and Practice"

[p-heckler]

So if the title-delimiter is a colon, assuming that localised punctuation exists, the result would be "Journal of Research: Theory and Practice" if the style is set to English, but "Journal of Research : Theory and Practice" in French. Would that be problematic?

[denismaier]

So if the title-delimiter is a colon, assuming that localised punctuation exists, the result would be "Journal of Research: Theory and Practice" if the style is set to English, but "Journal of Research : Theory and Practice" in French. Would that be problematic?

I don't think that this is problematic. Isn't that what you would want in most cases?

[p-heckler]

Isn't that what you would want in most cases?

I agree: if a style wants titles to be rendered as a block, it should not use the title-split feature. If it does in order to alter the rendering of parts of the title, the punctuation should be adjusted as well. Should title-delimiter and normalize-title-delimiters be added to the documentation on localised punctuation then?

[denismaier]

Of course, if you normalize title punctuation with title-delimiter, you could also just add the correctly localized punctuation there.

[p-heckler]

The main advantage of using punctuation terms as delimiters is linguistic portability, especially for styles designed as multiligual. For stricly monolingual styles, it changes nothing, unless the style uses non-localised punctuation, which I would find very odd and which could be overriden on a per-style basis anyway.

[bdarcus]

Is this now RTM?

[bwiernik]

Of course, if you normalize title punctuation with title-delimiter, you could also just add the correctly localized punctuation there.

No, for example, someone using MHRA in Arabic might want a different symbol than using it in English.

[bwiernik]

@bdarcus Can you weigh in on 4d73edf#r450999690? Then it's read to merge.

[bwiernik]

As a comparison, here is where citeproc-js handles a missing page-range-delimiter attribute: https://github.com/Juris-M/citeproc-js/blob/46fc91837d2e849a7b13c39df72fd282991ac2ba/src/util_locale.js#L57

The analogous proposition would be to include a test like this in the style parsing:

if ("undefined" !== typeof this.locale[langspec.best].terms["semicolon"]) {
  ...semicolon replacement code...
}

If "semicolon" is not defined, then the entire replacement process can be skipped. Otherwise, the processor needs to run it every single time it generates references for any style in any locale.

[bwiernik]

I can't see any way that replacing ";" in a style's attributes is something other than a discrete step. Given that, providing a mechanism to skip the replacement for the vast majority of locales where it isn't needed is a good idea.

If semicolon is defined in en-US, the processor could similarly skip the step by testing whether semicolon is equal to ";", but the logic seems much clearer to me as "if semicolon is undefined, skip semicolon normalization" versus "if semicolon is ';', skip semicolon normalization".

[bwiernik]

I looked into how French publications handle spacing in title delimiters (: ; ? !). The most common convention is to space items according to their language conventions (e.g., to put a space before the punctuation for French titles, but not for English titles). For example http://mapageweb.umontreal.ca/mageaug/Articles/Trottier,%20Mageau,%20Trudel,%20&%20Halliwell,%202008.pdf.
I couldn't find an example, where a French-language publication inserted a space before a delimiter in an English title.

Given that, could we expect an author writing in French to include the space in their data for French items? I'm not sure that we need to add question-mark and exclamation-mark to the punctuation list, along with specification to replace : ; ? ! at the end of main or sub title parts with the corresponding term. (cf. citeproc-js currently does insert a space in all titles when the locale is French when uppercase subtitling is applied).

Thoughts @bdarcus @denismaier @fbennett ?

[p-heckler]

The most common convention is to space items according to their language conventions (e.g., to put a space before the punctuation for French titles, but not for English titles

This is in fact the default rule in French styling (see the ISO690 norm : https://bibliotheques.icp.fr/medias/fichier/norme-z-44-005_1446046708358-pdf?ID_FICHE=1567&INLINE=FALSE at point 2.2.2. or the guidelines of the Sorbonne, so I think it is fair to assume that authors writing in French take care of the spacing directly in their data. If anything, parsing titles and subtitles delimiters automatically would be quite inconvenient for French styles…

[denismaier]

The most common convention is to space items according to their language conventions (e.g., to put a space before the punctuation for French titles, but not for English titles

This is in fact the default rule in French styling (see the ISO690 norm : https://bibliotheques.icp.fr/medias/fichier/norme-z-44-005_1446046708358-pdf?ID_FICHE=1567&INLINE=FALSE at point 2.2.2. or the guidelines of the Sorbonne, so I think it is fair to assume that authors writing in French take care of the spacing directly in their data. If anything, parsing titles and subtitles delimiters automatically would be quite inconvenient for French styles…

Hmm, the Sorbonne regulations say: "Le sous-titre d’un ouvrage français est séparé du titre principal par un
point."

And ISO690 says: "Indiquer le sous-titre s’il y en a un, en l’introduisant par le signe deux-points (« : »)."
Maybe I've missed it, but I could not find any reference to English language titles in this document.

[p-heckler]

The "." délimiter in the Sorbonne style only applies to French titles, foreign titles must retain their original formatting, and the example given for foreign titles does contain a delimiter : E. Narmour, The Analysis and Cognition of Basic Melodic Structures: The Implication- Realization Model, Chicago, Chicago University Press, 1990.

It has always been my understanding that the colon delimiter in the ISO format also applied to French titles only.

[abdealikhurrum]

I think this would work for the one-to-one substitution required for Arabic and Persian.