Citation in a note containing narrative text: expected behavior for `if position="ibid"`

[denismaier]

A discussion with @jgm regarding his new citeproc lead to a couple of question regarding ibid-behavior:

What is the expected behavior for if position="ibid" if the citation is not the only content in a note. Say we have this:

[1] @doe

[2] Bla bla bla, bla bla bla... See also @doe

Or the other way round:

[1] Bla bla bla, bla bla bla... See also @doe

[2] @doe

Or:

[1] @doe. Bla bla bla, bla bla bla..

[2] @doe

Or,

[1] @doeB. Bla bla bla, bla bla bla.. See also @doeA.

[2] @doeA

[bdarcus]

I don't see how the narrative content per se is relevant to whether it's ibid, which simply means the same source as the last cited one.

Obviously people can abuse that such that it makes no sense, but I don't think that's our problem.

[denismaier]

I think I've seen requirements like: only use ibid. if the preceding citation is in a citation only note, and if it wasn't part of a group of multiple citations.

[bwiernik]

I've seen requirements like that. My vote would be to state that position="ibid" and position="ibid-with-locator" never apply when a citation contains more than one item or when the citation contains affixes. Alternatively, we could add a new value position="ibid-alone" to mean that case--"ibid and no other citations of affixes". @adam3smith

See also https://github.com/citation-style-language/documentation/issues/122#issuecomment-721786693

[denismaier]

I've seen requirements like that. My vote would be to state that position="ibid" and position="ibid-with-locator" never apply when a citation contains more than one item or when the citation contains affixes. Alternatively, we could add a new value position="ibid-alone" to mean that case--"ibid and no other citations of affixes". @adam3smith

See also https://github.com/citation-style-language/documentation/issues/122#issuecomment-721786693

Or we could add a new attribute on cs:style, ibid-strictness or so. ibid-strictness="strict" would prevent ibids in these cases while ibid-strictness="relaxed" would allow them. What do you think?

[bwiernik]

Styles that might use ibid in these cases would typically be okay with not using it to increase clarity, so I lean strongly toward not adding yet another ibid-related option unless there is a clear demand for it.

[denismaier]

Following up on #122, I'll restate one question here:

[1] @doeB. Bla bla bla, bla bla bla.. See also @doeA
[2] @doeA.

In #122 we've seemed to favor this option:

Restrict position="ibid" to single-item citations on both sides, including this case.

I think that's reasonable.

Just, here the citation processor would need to know that footnote 1 contains two citations and some text, to prevent an "ibid" in footnote 2. Should be possible with pandoc, but I'm not sure if that is possible with a Zotero/Word-Combination. @jgm @fbennett ? Sure, we can require users to enter the text in between citations as affixes, but that can quickly become quite awkward, especially in disciplines that make—for better or worse—extensive use of discursive footnotes.

[bwiernik]

I think that's outside of the scope of CSL. To properly handle ibid there, the user should enter both items into the same citation and use affixes for the commentary text, rather than directly inserting the citation into the footnote. There isn't any way that a processor could detect otherwise. There isn't a really a distinction between these in pandoc, but it matters with Word.

Based on my understanding of what pandoc is doing, it's not "recognizing that the two citations are in the same footnote"--it's treating the discursive text as affixes on items in a single citation.

[adam3smith]

Zotero/citeproc-js in Word currently handles this case correctly: If footnote 1 contains 2 separate citations, (i.e. footnote inserted with Word, 1st citation inserted, some text, 2nd citation inserted) then neither of the 2 cited items if cited again if FN 2 is treated as ibid. Or using the example above:

[1] @doeB. Bla bla bla, bla bla bla.. See also @doeA
[2] @doeA.

Results in [1] Berthold Doe, "I'm the longform of an item" Bla bla bla, bla bla bla.. See also Albert Doe, "I'm also a longform" [2] Doe, "I'm the subsequent/short form"

So this is clearly technically possible and given that citeproc-js (and pandoc-citeproc) already does it right, I think it'd be good to codify it in the specs.

[denismaier]

So this is clearly technically possible and given that citeproc-js (and pandoc-citeproc) already does it right, I think it'd be good to codify it in the specs.

Yes, we should codify it.

Btw: There have been some regressions introduced by @jgm's new citeproc. (Currently, testing for ibid. returns "true" in @adam3smith's example.

[bwiernik]

@fbennett How is that working technically?

[denismaier]

@fbennett How is that working technically?

I guess by comparing using the noteIndex associated with each citation, see https://citeproc-js.readthedocs.io/en/latest/csl-json/markup.html#citations

So maybe something like this: if citation[current].noteIndex - citation[current-2].noteIndex == 1 and citation[current-1].noteIndex - citation[current-2].noteIndex == 0

[bwiernik]

What exactly do we need to codify here?