Question about collapse_ChicagoAfterCollapse.txt

[jgm]

I'm confused about this test; probably I don't understand how after-collapse-delimiter is supposed to work, and I'm hoping someone can illuminate me.

The result is

(Whittaker 1967, 1975; Wiens 1989a, 1989b)
(Wong 1999, 328; 2000, 475; Garcia 1998, 67)
(Whittaker 1967; Wiens 1989b)

I fully understand the first entry: we use the after-collapse-delimiter, which is ;, after the collapsed "Whittaker 1967, 197". But what is going on in the third? Here there is no collapsing. Yet the after-collapse-delimiter is still used.

The spec gives this example:

]E.g. with collapse set to “year”, delimiter on cs:layout in cs:citation set to “, ”, and after-collapse-delimiter set to “; ”, citations look like “(Doe 1999, 2001; Jones 2000, Brown 2001)”.

Here the after-collapse-delimiter is only used after the collapsed citation; it is not used after "Jones 2000". On the basis of that, I'd expect the regular delimiter , to be used for the second example:

(Whittaker 1967, Wiens 1989b)

[jgm]

Can anyone help explain this?

[denismaier]

I have no idea, but how this is implemented, and if there is some further reasoning behind that. But after reading the specs, this looks like a bug to me. As no collapsing takes place in the third citation, I don't see why the after-collapse-delimiter should be used instead of just the regular delimiter. So yes, I agree the expected output should be:

(Whittaker 1967, Wiens 1989b)

@fbennett @adam3smith @bwiernik @bdarcus ?

[denismaier]

Hmm, maybe: Chicago uses a semicolon as delimiter between multiple sources in the same citation. So, actually I wonder why we have delimiter=", " instead of delimiter="; ...? Is this some hack to accommodate these rules?

By the way, the first two test outputs come directly from CMoS 15.30:

(Whittaker 1967, 1975; Wiens 1989a, 1989b)
(Wong 1999, 328; 2000, 475; Garcia 1998, 67)

The third output (i.e. (Whittaker 1967; Wiens 1989b)) isn't there.

[fbennett]

My memory on this was hazy, but with some detective work I've confirmed that @denismaier's intuition is exactly right. The report below is a bit long, but there is a prize at the end.

The after-collapse-delimiter attribute has been in there since at least 2009, with a test that conforms to the spec:

(Condit et al. 2000, 1995, 1998; Connell 1971, Cortes & Hunziker 1971)

Oddly, both this test and the collapse_ChicagoAfterCollapse.txt test raised here pass, although they reflect different behavior with after-collapse-delimiter.

There was a proposal in June of 2011 to add an attribute collapse-delimiter that was aimed at an APA (? and perhaps CMoS?) requirement. That led to an extensive discussion of author grouping, which led to the current author-grouping specification:

• Provide explicit delimiter for author groups or collapsed cites within cs:citation schema#52
• APA requires special delimiter between years of collapsed citation schema#40

The delimiter proposal seems to have dropped out of the mix in the discussion. The current off-spec behavior in the current collapse_ChicagoAfterCollapse.txt test dates from March 8, 2018:

• Juris-M/citeproc-js@84a3916

The change limits the previously tested (and conformant) behavior of after-collapse-delimiter to styles that set collapse="year-suffix". CMoS sets collapse="year", so the special delimiter is used everywhere but between collapsed citations. That's the role-reversal that @denismaier notes, and it seems to have been done by intention to work around the lack of a collapse-delimiter attribute.

Following that lead, we find that after-collapse-delimiter="; " is actually used in the current official release of chicago-author-date.csl``:

• https://github.com/citation-style-language/styles/blob/master/chicago-author-date.csl#L610

Tracing the origin of that change leads to this commit ...

• citation-style-language/styles@d9852a2

... which leads to this discussion on the Zotero forums:

• https://forums.zotero.org/discussion/70718/author-date-citations-with-page-references-not-separated-by-a-semicolon

And there we have it: a workaround introduced to get correct output from styles in wide use, while avoiding the need for a CSL schema update. The workaround is on me, but it was an escape manoeuver that everyone signed off on at the time, so I'll forgo the customary apology on this occasion. :-)

Edit: It turns out that git log has an option for sniffing out changes affecting particular areas of a file, using regular expressions to define the span. Here is the command that turned up the commit:

• git log -L /after-collapse-delimiter/,/layout/:chicago-author-date.csl

[bdarcus]

Wow; quite the detective work! As we wrap up 1.1, should we fix this in the schema? If yes, how?

[fbennett]

As @adam3smith notes in the Zotero forum thread, the delimiter attributes are a bit of a mess. It might be time for an audit of what styles require, and root-and-branch reform of the attributes and their respective names.

[bdarcus]

It might be time for an audit of what styles require, and root-and-branch reform of the attributes and their respective names

Hmm ... any thoughts on this @bwiernik?

[bwiernik]

Yeah, clarifying the names would be really good here.

[bdarcus]

This was partially resolved, it seems, by adding @cite-group-delimiter ...

citation-style-language/schema#52

... but this test doesn't use it; I guess because the comma is default.

In any case, the above linked issue is the primary record, I think.

Perhaps at minimum we could add to the docstrings for these two attributes, with an example; maybe even the one John posted.

Do we need any actual schema changes?

Also, could we added a description field to this test to clarify for future developers? If yes, what should the content be?

Something like this?

In these examples, the semi-colon serves to delimit the citation (author) groups, and the comma the individual citations within those groups. [insert brief attribute discussion]

[bwiernik]

I think we will need at least one new attribute. Working on a post.

[fbennett]

This was partially resolved, it seems, by adding @cite-group-delimiter

chicago-author-date.csl doesn't sort citations, so (in current citeproc-js) citation-group-delimiter would not be available.

(Does CMoS mandate grouping of citations by the same author, or is the order left to author discretion? If the latter, that would undermine the case for imposing author grouping in all styles, regardless of presence/absence of explicit citation sorting.)

[bwiernik]

It's left to discretion. Chicago recommends alphabetical, chronological, or based on rhetorical purpose.

From the manual:

Two or more references in a single parenthetical citation are separated by semicolons. The order in which they are given may depend on what is being cited, and in what order, or it may reflect the relative importance of the items cited. If neither criterion applies, alphabetical or chronological order may be appropriate. Unless the order is prescribed by a particular journal style, the decision is the author’s.

[bwiernik]

I didn't pull up specific manual references for these, but I can recall styles specifying the following delimiters for citation groups:

Delimiter	Example	Current name	New name
Delimiter between cites	Jones 1992, p. 123; Smith 1994	delimiter on cs:citation	delimiter on cs:citation
Delimiter between cite and locator	Jones 1992, p. 123	delimiter on cs:group	delimiter on cs:group
Delimiter between collapsed cites	Jones 1992, p. 123, 1994; Smith 1994	Not explicit; will pick up group delimiter if citations are sorted	collapse-delimiter
Delimiter between collapsed cites, if there is a locator	Jones 1992, p. 123; 1994, 1996; Smith 1994	hack of after-collapse-delimiter	collapse-with-locator-delimiter
Delimiter between grouped cites, even if not collapsed (e.g., citations by the same author)	Jones 1992, p. 123, Jones 1994; Smith 1994	cite-group-delimiter	cite-group-delimiter
Delimiter used after a collapsed cite before the next cite (typically used if , is the regular delimiter and the collapse delimiter, but ; should be used after collapse)	Jones 1992, p. 123, 1994; Smith 1994, Wilson	after-collapse-delimiter	after-collapse-delimiter
Delimiter between year suffixes	Jones 1999a, b	year-suffix-delimiter	year-suffix-delimiter

I think adding collapse-delimiter and collapse-with-locator-delimiter explicitly makes the most sense and avoids trying to be too clever.

Inherited default values for the delimiters

• delimiter on cs:citation -> cite-group-delimiter -> collapse-delimiter -> collapse-with-locator-delimiter
• delimiter on cs:citation -> after-collapse-delimiter
• , -> year-suffix-delimiter

[bdarcus]

That's super helpful; thanks!

We should move this to the schema repo.

I wonder, at least to clarify documentation:

Doesn't the "collapse" language, per my consistent argument over the years (as you can see in the linked threads), confuse the issue?

"Collapsing" is the outcome of the formatting, not logically what's happening in the styles; nor how a style guide would describe it. And collapsing is merely one way to represent the grouping, but the need to distinguish delimiters remains regardless.

Leaving aside the locator question, I see "author group" delimiters in the table above, and "within-author group" delimiters (if you will, author-year delimiters).

Also, @bwiernik, I can't tell from your examples how locators impact this, since the examples don't all include locators. Could you fix that please?

[bwiernik]

In style writing, cite grouping and cite collapsing are two different steps to accommodate two different types of style recommendations.

Some styles, like APA and Chicago, collapse adjacent citations by the same author. Chicago for example says:

Additional works by the same author (s) are cited by date only, separated by commas except where page numbers are required.

APA says:

Give the authors' surnames only once, for each subsequent work, give on the date.

They don't use the word "collapse", but that reasonable word to summarize the explicated behavior.

Other styles (e.g., some British Harvard styles) don't collapse, but do group (e.g., the Jones 1992, Jones 1994; Smith 1994 example).

[bwiernik]

Also, @bwiernik, I can't tell from your examples how locators impact this, since the examples don't all include locators. Could you fix that please?

I added page numbers to the Jones 1992 item in each of the examples. Sorry that I mistakenly left off the locator from the with-locator example before.

[jgm]

CMoS sets collapse="year", so the special delimiter is used everywhere but between collapsed citations.

I still don't quite get the expected results.

(Wong 1999, 328; 2000, 475; Garcia 1998, 67)

Here we have a collapsed citation: Wong 1999, 328 and Wong 2000, 475 collapse to Wong 1999, 328; 200, 475 -- but the after-collapse delimiter ; is NOT used between them. It seems that this is due to the presence of a locator, perhaps? But I'm still missing a clear statement of the rule to follow.
EDIT: Ah, I see "if there is a locator" in @bwiernik 's table above.