according to the APA style guidelines, which have many adoptions in
various CSL-styles, it is mandatory to have ‘and’ connecting authors in
narrative, and ‘&’ connecting authors in parenthetical citations[1].
None of the styles I tried were accommodating for this rule.
So then I’ve tried to build a if-else-control structure to identify
whether it is a narrative or a parenthetical citation, but it seems
that the processor would do that automatically by using “prefix” and
"suffix"?
Is there any way to identify in the csl-file what type of citation is
calling the evaluation?
Or even better, is there already an existing style that accommodates
for my issue (I could not find one, but that may be because of very
vague and ubiquitous search terms).
CSL has no concept of a “narrative citation,” which would explain why
there’s no way to distinguish formatting based on it.
There’s been discussions about it, but am not remembering where we ended up.
If we don’t already have it, we should have a general issue ticket on
GitHub for the broader feature, and include a note about this particular
detail there.
Even if we label it “won’t fix,” at least we have the documentation in one
place.
I have created a github issue, but I could unfortunately not tag it as
“wontfix” myself.
In my case, I will probably use latex (biblatex-apa) for this, as the
journal is very picky about their citations.
In case anyone is interested, another option to solve this is to use
and=“symbol” and use a regex for post-processing.
This one works:
“([&])(?=\s+(?:[\w-]+ ){1,3}[(])” -> “and”
(unless there are punctuation characters in between the ampersand and
the year, e.g. for first names, or there is random ampersands followed
by brackets in the text)
To be clear, I’m not sure we won’t fix. It’s just a possibility.
This issue is tied into a somewhat complex issue: whether we support the
feature (“narrative citations”), it needs to maintain a current original
design goal, which is that an author can switch between note and other
style types without editing their text.
I couldn’t figure out how to reconcile those in the past, but would be
happy if someone did.
what reference manager are you using? Pandoc? E.g. in Zotero that whole
idea doesn’t exist. You’d have to write out the narrative reference in the
text. Which has its own downsides, e.g. won’t adapt to et al. rules if the
citation moves.
I do think we’ll need to handle this. It’s one of the most frequent
complaints on Zotero and as Malte notes, it cause reference managers that
try to implement “narrative” citations to render incorrect citations.
Since it’s a relative big change, I’d not expect this to happen fast,
though.
I use zotero and export to bibtex; then pandoc-citeproc runs over the
references.
I usually thought everything works well, until editors and supervisors
complained about the incorrect citation style.
To make it compliant with APA, for now I run pandoc with --biblatex
command so that it turns the [@cite1234] into latex citations, and then
compile it with latexmk. Luckily, there is shell scripts to take care
of that, so my workflow does not change too much. I just cannot use the
word-output ‘correctly’ anymore then.
The general idea that seems to have emerged from various discussion here and on the Zotero forums that styles would need to specify two different citation formats. For author-date styles, these would generally take the forms: (Author, Date) and Author (Date). For numeric and note styles, these would be: [1] and Author [1].
Supporting two formats like this in CSL would be excellent, as it would dramatically improve the ability of users to change styles quickly (at the moment, changing rules for “and” vs “&”, formatting rules for “et al.”, etc. must be manually adjusted for “narrative citations”). As Sebastian mentions, this is probably the single most requested feature/complaint from Zotero users.