In the LC NACO authority file, two titles have been established under "Arroyo Pizarro, Yolanda" which widen the range of AAP possibilities to three:
Arroyo Pizarro, Yolanda. $t Short stories. $k Selections -- with 400s for three separate titles of earlier collections.
Arroyo Pizarro, Yolanda. $t Short stories. $k Selections (Golpes de gracia) -- the title of a fourth story collection, not included among the 400s above.
Arroyo Pizarro, Yolanda. $t Negras -- Pete Wilson's case, another short story collection entered this time under its own title in existing bib records.
How can we make sense of this? One step is to understand that ".. $t Short stories. $k Selections" and "...$t Negras" are AAPs for different things. The first has been used to name several short story collections collectively, not one. The third option names a particular collection only as a work. The middle option tries to have it both ways, borrowing the collective AAP and adding a qualifier title to make it particular.
Since we don't as a general rule insert a conventional collective title between an author and a work title (i.e., we don't use "Shakespeare, ... $t Plays. $k Selections (King Lear)"), the middle option seems dubious. It also seems dubious to suppose that the first option represents only the three compilations which its 400s name and not Golpes de gracia. Either a resource has a particular work title or it doesn't. If it has one, it doesn't need the conventional collective title. If a compilation doesn't have a particular distinctive title, then any qualifying information added to make a conventional collective title specific should come from somewhere else.
So the question becomes, can a compilation have a particular work title? RDA 184.108.40.206.3's Alternative instruction says, "When identifying two or more works in a compilation, identify the parts
collectively by recording a conventional collective title (see 220.127.116.11.1 or 18.104.22.168.2, as applicable), followed by Selections. Apply this instruction instead of or in addition to recording the preferred title for each of the works in the compilation." The LC-PCC PS for this instruction adds, "Give an authorized access point for the first or predominant work."
This is generally taken as applying to the name of the compilation; but that's not really what the instruction says. The conventional collective title is meant as a replacement for separately identifying the parts of a compilation with AAPs, not as a name for the compilation itself. A strict application of the 22.214.171.124.3 alternative instruction and LC-PCC PS would look something like this:
100 1 $a Arroyo Pizarro, Yolanda.
240 10 $a Negras
245 14 $a Las negras / $c ...
505 0 $a Wanwe -- Matronas -- Saetas.
700 12 $a Arroyo Pizarro, Yolanda. $t Short stories. $k Selections.
700 12 $a Arroyo Pizarro, Yolanda. $t Wanwe.
As for the compilation, RDA 126.96.36.199, Recording the Preferred Title for a Work, says, "This instruction applies to individual works and to compilations of works.
Record the title chosen as the preferred title for a work by applying the basic instructions at 6.2.1." RDA 188.8.131.52 defines the title of the work as "A word, character, or group of words and/or characters by which a work is known." The instruction at 184.108.40.206, Works Created after 1500, says to "choose as the preferred title the title or form of title in the original
language by which the work is commonly identified either through use in
resources embodying the work or in reference sources." In this case, "Negras" is the title in the original language by which the compilation work is commonly identified through use in resources embodying the work; so "Negras" is the preferred title.
The important thing is that RDA does not impose a choice between "Negras" and a conventional collective title. Both are appropriate as part of the resource description, since they signify different things--the name of the compilation vs. an alternative to enumerating the contents of the compilation with separate AAPs.
And yes, I know this is a novel reading of RDA, and suspect that I've missed some important instruction that will easily unravel everything I've said; but before that happens, look again at the brave little record up there. Wouldn't this be a better way to handle compilations with distinctive titles?
I'd answer Pete's original question about whether the addition of a story to a later edition of "Negras" constitutes a new work in the negative, preferring "expanded as (expression)" to clarify the relationship between the original three-story expression and the later expanded four-story one. One source I saw for the latter called it a second edition, which aligns with treating it as an expression. This may also reflect differences in the way the term "work" is valued. I generally prefer to see a creator's production defined as fewer works and more expressions, based on a sense that this sort of analysis concentrates the work entities and thereby enhances their value. Treating a resource which adds relatively little to an existing work as a new work has the effect of diluting the significance of the work designation. But that's an argument more from sentiment than from reason.