Patch for particle sorting

Frank and I (think we) worked out a solution for customizing particle
sorting in CSL:

http://zotero-dev.googlegroups.com/web/Particle-sorting.patch

In short, two new global options on cs:style allow to:
a) toggle the way particles are handled in name sorting. Particles can
either take priority over the family name (“von Braun W.”) or take the
backseat to the family name (“Braun von W.”).
b) toggle the way particles are handled in the rendered name output.
Particles can either be prepended to the family name (“von Braun, W.”), or
take the backseat to both the family and given names (“Braun, W. von”).

The default options are set to replicate the current behavior in Zotero
(append particles in sorting, prepend in rendered output). In addition a
small cosmetic change has been made: the name-as-sort-order attribute of
cs:name has been renamed to invert-name, as with the new options, inverted
names (with the family name before the given name) in the final output may
differ in name order from the order used in sorting (e.g. “Braun, W. von”
(rendered output) and “Braun von W.” (sorting)).

Additional documentation can be found in the patch (essentially the same,
but in slightly different wording).

Rintze

Frank and I (think we) worked out a solution for customizing particle
sorting in CSL:

http://zotero-dev.googlegroups.com/web/Particle-sorting.patch

In short, two new global options on cs:style allow to:
a) toggle the way particles are handled in name sorting. Particles can
either take priority over the family name (“von Braun W.”) or take the
backseat to the family name (“Braun von W.”).

The basic idea seems fine.

But I think we should explicitly define the sort order of the
pieces—as an ordered list—rather than give examples of strings where
this is merely implicit.

Also …

b) toggle the way particles are handled in the rendered name output.
Particles can either be prepended to the family name (“von Braun, W.”), or
take the backseat to both the family and given names (“Braun, W. von”).

The default options are set to replicate the current behavior in Zotero
(append particles in sorting, prepend in rendered output).

I don’t know what this means :wink:

Also, my assumption is that default is that particles are not included
in sorting.

So in other words, my proposed spec language would be something like:

The default sorting priority of personal name parts is: family, given,
suffix. This may be modified with a global parameter [insert name]
that offers two options for dealing with particles (“von”, “de”, "bin"
and so forth):

  1. [insert value] [insert explanation]
  2. [insert value] [insert explanation]

Except it seems to me there’s only one option really: that which
changes the ordered list to: particle, family, given, suffix.

Can you clarify?

In addition a
small cosmetic change has been made: the name-as-sort-order attribute of
cs:name has been renamed to invert-name,

At face value, this doesn’t make much sense, since the inversion only
happens for Western names? What am I missing here?

as with the new options, inverted
names (with the family name before the given name) in the final output may
differ in name order from the order used in sorting (e.g. “Braun, W. von”
(rendered output) and “Braun von W.” (sorting)).

Additional documentation can be found in the patch (essentially the same,
but in slightly different wording).

I have a hard time reading the patches, and then trying to also figure
out how to translate this into spec language. It might be easier to
just include the text you’d propose for the spec (as I do above), and
then just add the changes to the schema directly if we agree on the
wording.

Bruce

Current behavior and appropriate default behavior

Zotero currently removes all-lowercase words from the beginning of the
last name field in a name when generating sort keys (L660-662,
csl.js), but the particle is included for display purposes. This will
produce results like the following:

Abner, Lisle
di Doe, Jack
van Doe, John
de Doe, Stephen
Goode, Charles

This rather chaotic result is produced by orderly sort keys that look
like as follows:

[Abner, Lisle]
[Doe, Jack]
[Doe, John]
[Doe, Stephen]
[Goode, Charles]

According to Rintze, the display form currently used is correct (for a
significant category of publications, at least), but the particle
should in this instance be appended to the last name when composing
the sort key. This would display the same names in the following
order:

Abner, Lisle
de Doe, Stephen
di Doe, Jack
van Doe, John
Goode, Charles

Note that the “de” entry has moved to the top of the list of persons
with last name “Doe”. According to Rintze, this is normal sorting
behaviour in Holland, where names containing particles are common.
The sort keys that produce this ordering are like as follows:

[Abner, Lisle]
[Doe de, Stephen]
[Doe di, Jack]
[Doe van, John]
[Goode, Charles]

We feel that the ordering given in the first example above (reflecting
current Zotero behavior) is difficult to read, and neither of us is
aware of a publication that requires it. We feel that either the
display order or the sort order ought to be changed. Regarding
display order, altering it (by removing the particle, or by changing
its position) would be a significant and unexpected change to the
appearance of bibliographies in the hands of end users. We would
expect this to produce a volume of support queries in the forums,
which would be undesirable.

As the second pair of examples given above is close to current Zotero
behavior in appearance, and is known to be actually used in
publishing, we believe that it should be the default.

It might be easier to
just include the text you’d propose for the spec (as I do above),

Your sample spec language reads as follows:

The default sorting priority of personal name parts is: family, given,
suffix. This may be modified with a global parameter [insert name]
that offers two options for dealing with particles (“von”, “de”, "bin"
and so forth):

  1. [insert value] [insert explanation]
  2. [insert value] [insert explanation]

I believe that by this you mean that there should be three options for
sorting. One would place the particle before the family name.
Another would place the particle after the family name. And a third
would omit the particle from the sort key altogether. Please confirm
that this is a correct understanding.

If that is a correct understanding, then each of these sort orderings
might be associated with one or more display orderings. In the
current proposal, sort ordering and display ordering would be
controlled by separate options. We believe that this will be easier
to manage than a scheme that attempts to control sorting behavior and
display behavior through a single option. Please also confirm that we
have agreement on this point.

I can post again with proposed text for the spec once we are sure
we’re on the same page re the points above.

Frank

Current behavior and appropriate default behavior

Zotero currently removes all-lowercase words from the beginning of the
last name field in a name when generating sort keys (L660-662,
csl.js), but the particle is included for display purposes. This will
produce results like the following:

Abner, Lisle
di Doe, Jack
van Doe, John
de Doe, Stephen
Goode, Charles

This rather chaotic result is produced by orderly sort keys that look
like as follows:

[Abner, Lisle]
[Doe, Jack]
[Doe, John]
[Doe, Stephen]
[Goode, Charles]

According to Rintze, the display form currently used is correct (for a
significant category of publications, at least), but the particle
should in this instance be appended to the last name when composing
the sort key.

This would display the same names in the following
order:

Abner, Lisle
de Doe, Stephen
di Doe, Jack
van Doe, John
Goode, Charles

Note that the “de” entry has moved to the top of the list of persons
with last name “Doe”. According to Rintze, this is normal sorting
behaviour in Holland,

I was a little lost, but this helps clarify.

Am just not sure it works in English on the display of the sorting
name. I’m pretty sure (though not 100% so) a name like “Alexander von
Humbolt” would end up as “Humbolt, Alexander von”.

where names containing particles are common.

The sort keys that produce this ordering are like as follows:

[Abner, Lisle]
[Doe de, Stephen]
[Doe di, Jack]
[Doe van, John]
[Goode, Charles]

We feel that the ordering given in the first example above (reflecting
current Zotero behavior) is difficult to read, and neither of us is
aware of a publication that requires it.

This is where this discussion is confusing. I’ve been meaning "sort"
form literally as the logic to sort the list of references. But here
you’re mean (to introduce an awkward new term) the “as sorted” display
form. Seems feasible they may differ a bit; not sure. But we probably
need to clarify this in the spec.

We feel that either the
display order or the sort order ought to be changed. Regarding
display order, altering it (by removing the particle, or by changing
its position) would be a significant and unexpected change to the
appearance of bibliographies in the hands of end users. We would
expect this to produce a volume of support queries in the forums,
which would be undesirable.

Here I’m lost. What do you mean by the “display order” here? Or, how
does it relate to my point just above?

As the second pair of examples given above is close to current Zotero
behavior in appearance, and is known to be actually used in
publishing, we believe that it should be the default.

It might be easier to
just include the text you’d propose for the spec (as I do above),

Your sample spec language reads as follows:

The default sorting priority of personal name parts is: family, given,
suffix. This may be modified with a global parameter [insert name]
that offers two options for dealing with particles (“von”, “de”, "bin"
and so forth):

  1. [insert value] [insert explanation]
  2. [insert value] [insert explanation]

I believe that by this you mean that there should be three options for
sorting.

This is only what I was understanding Rintze was arguing.

One would place the particle before the family name.
Another would place the particle after the family name. And a third
would omit the particle from the sort key altogether. Please confirm
that this is a correct understanding.

Well, this is just what I’m not clear on. I’m not sure.

If that is a correct understanding, then each of these sort orderings
might be associated with one or more display orderings. In the
current proposal, sort ordering and display ordering would be
controlled by separate options. We believe that this will be easier
to manage than a scheme that attempts to control sorting behavior and
display behavior through a single option. Please also confirm that we
have agreement on this point.

It seems to make sense.

I can post again with proposed text for the spec once we are sure
we’re on the same page re the points above.

Hopefully the above at least shows where there’s still some confusion.

Bruce

One would place the particle before the family name.
Another would place the particle after the family name. And a third
would omit the particle from the sort key altogether. Please confirm
that this is a correct understanding.

Well, this is just what I’m not clear on. I’m not sure.

We can provide for all three possibilities, and leave it up to style
authors to pick their preferred sorting arrangement. Rather than
discarding the particle completely in the latter case, it might make
sense to place it after the given name.

We can also provide three values for the inverted-name-display-order
(or whatever) option, so that particles in the “inverted” display form
can be placed either before the family name, after the family name, or
after the given name, according to the taste of the journal.

If that is a correct understanding, then each of these sort orderings
might be associated with one or more display orderings. In the
current proposal, sort ordering and display ordering would be
controlled by separate options. We believe that this will be easier
to manage than a scheme that attempts to control sorting behavior and
display behavior through a single option. Please also confirm that we
have agreement on this point.

It seems to make sense.

Excellent. I’ll leave the spec and schema language to Rintze, he is
more familiar with
the specific use cases, and can do a better job of condensing the
description. If there are further issues with the plan, I’ll let
Rintze cover those was well, tomorrow (Delft time).

Frank and I (think we) worked out a solution for customizing particle
sorting in CSL:

http://zotero-dev.googlegroups.com/web/Particle-sorting.patch

In short, two new global options on cs:style allow to:
a) toggle the way particles are handled in name sorting. Particles can
either take priority over the family name (“von Braun W.”) or take the
backseat to the family name (“Braun von W.”).

The basic idea seems fine.

But I think we should explicitly define the sort order of the
pieces—as an ordered list—rather than give examples of strings where
this is merely implicit.

This is just a question about how things should be documented, right? Would
be enough to add something like “For the name “W. von Braun” (name-parts:
“W.” = given, “von” = particle, “Braun” = family) …” to the annotation? I
think using a real name in the examples is clearer. We can also use the real
name examples in the spec, and limit our-selves to “given”, “family” and
"particle" in the schema annotation.

b) toggle the way particles are handled in the rendered name output.
Particles can either be prepended to the family name (“von Braun, W.”),
or
take the backseat to both the family and given names (“Braun, W. von”).

The default options are set to replicate the current behavior in Zotero
(append particles in sorting, prepend in rendered output).

I don’t know what this means :wink:

I probably should have tested this first, but apparently Zotero drops the
particle(s) altogether when sorting. Some output I generated:

von Braun, A. (2000).
Braun, A. (2000).
van Braun, B. (2000).
vam Braun, B. (2000).
Zaun, A. (2000).

The order of “van Braun, B.” and “vam Braun, B.”, and of “Braun, A.” and
"von Braun, A." is entirely depending on the order of the items in the
middle column of Zotero, so the particle isn’t used for sorting at all. So I
drop my comment that we want to replicate Zotero’s behavior (since it’s very
messy: it hurts my Dutch eyes).

Also, my assumption is that default is that particles are not included

in sorting.

Does that make any sense? I see Zotero’s current behavior as a bug, not a
feature. I must admit I haven’t seen a style guide rule discussing this (but
I’m willing to dig a bit more if you’d like to see some real life examples),
but the particle is still part of the family name. It seems very logical to
me to either sort on particle first, then on family name, and finally on
given name, or, for Dutch-style sorting, family name, particle, given name.

So in other words, my proposed spec language would be something like:

The default sorting priority of personal name parts is: family, given,
suffix. This may be modified with a global parameter [insert name]
that offers two options for dealing with particles (“von”, “de”, "bin"
and so forth):

  1. [insert value] [insert explanation]
  2. [insert value] [insert explanation]

Except it seems to me there’s only one option really: that which
changes the ordered list to: particle, family, given, suffix.

Can you clarify?

We have different opinions here on what options are needed. See above.

In addition a
small cosmetic change has been made: the name-as-sort-order attribute
of
cs:name has been renamed to invert-name,

At face value, this doesn’t make much sense, since the inversion only
happens for Western names? What am I missing here?

Well, then we should just document that only Western names are affected.
Keep in mind that the newly introduced cs:name-part element has the same
problem (defining custom family/given name formatting also wouldn’t affect
non-Western name rendering).

as with the new options, inverted
names (with the family name before the given name) in the final output
may
differ in name order from the order used in sorting (e.g. “Braun, W. von”
(rendered output) and “Braun von W.” (sorting)).

Additional documentation can be found in the patch (essentially the same,
but in slightly different wording).

I have a hard time reading the patches, and then trying to also figure
out how to translate this into spec language. It might be easier to
just include the text you’d propose for the spec (as I do above), and
then just add the changes to the schema directly if we agree on the
wording.

I thought it would be better to start with the schema, so you can more
easily see how things are affected.

Rintze

Am just not sure it works in English on the display of the sorting
name. I’m pretty sure (though not 100% so) a name like “Alexander von
Humbolt” would end up as “Humbolt, Alexander von”.

Well, some journals (I’m looking at you, Applied and Environmental
Microbiology) display as “von Humbolt, Alexander”. So we need the option
(for name display) to either put the particle(s) in front of the family name
(“von Humbolt, Alexander”), or after the family and given names “Humbolt,
Alexander von”. This would of course be limited to “inverted” (Western)
names.

where names containing particles are common.

The sort keys that produce this ordering are like as follows:

[Abner, Lisle]
[Doe de, Stephen]
[Doe di, Jack]
[Doe van, John]
[Goode, Charles]

We feel that the ordering given in the first example above (reflecting
current Zotero behavior) is difficult to read, and neither of us is
aware of a publication that requires it.

This is where this discussion is confusing. I’ve been meaning "sort"
form literally as the logic to sort the list of references. But here
you’re mean (to introduce an awkward new term) the “as sorted” display
form. Seems feasible they may differ a bit; not sure. But we probably
need to clarify this in the spec.

I don’t understand what exactly the issue is here.

One would place the particle before the family name.
Another would place the particle after the family name. And a third
would omit the particle from the sort key altogether. Please confirm
that this is a correct understanding.

Well, this is just what I’m not clear on. I’m not sure.

I greatly question the need for omitting the particle altogether. But if
this is the cause of most of the confusion, it might be worth looking into a
bit more.

Rintze

Am just not sure it works in English on the display of the sorting
name. I’m pretty sure (though not 100% so) a name like “Alexander von
Humbolt” would end up as “Humbolt, Alexander von”.

Well, some journals (I’m looking at you, Applied and Environmental
Microbiology) display as “von Humbolt, Alexander”. So we need the option
(for name display) to either put the particle(s) in front of the family name
(“von Humbolt, Alexander”), or after the family and given names “Humbolt,
Alexander von”. This would of course be limited to “inverted” (Western)
names.

But my point is that Frank suggested it would be “Humbolt von,
Alexander”. That seems wrong.

where names containing particles are common.

The sort keys that produce this ordering are like as follows:

[Abner, Lisle]
[Doe de, Stephen]
[Doe di, Jack]
[Doe van, John]
[Goode, Charles]

We feel that the ordering given in the first example above (reflecting
current Zotero behavior) is difficult to read, and neither of us is
aware of a publication that requires it.

This is where this discussion is confusing. I’ve been meaning "sort"
form literally as the logic to sort the list of references. But here
you’re mean (to introduce an awkward new term) the “as sorted” display
form. Seems feasible they may differ a bit; not sure. But we probably
need to clarify this in the spec.

I don’t understand what exactly the issue is here.

This entire discussion has been confusing, because we’re talking at
cross-purposes. Let’s keep apart two different things:

  1. how to sort personal names (e.g. what happens internally in some processor)

  2. how to display the sorted form (what is looks like when rendered)

I’m just saying it might be feasible that 1 and 2 differ. But I’m not
sure, since I don’t really understand all of this.

Another thing we haven’t addressed: what happens to a particle if
given names are initialized?

  1. Humbolt, A. von
  2. Humbolt, A.
  3. Humbolt, A. v.

…?

One would place the particle before the family name.
Another would place the particle after the family name. And a third
would omit the particle from the sort key altogether. Please confirm
that this is a correct understanding.

Well, this is just what I’m not clear on. I’m not sure.

I greatly question the need for omitting the particle altogether. But if
this is the cause of most of the confusion, it might be worth looking into a
bit more.

Bruce

I have a hard time reading the patches, and then trying to also figure
out how to translate this into spec language. It might be easier to
just include the text you’d propose for the spec (as I do above), and
then just add the changes to the schema directly if we agree on the
wording.

I thought it would be better to start with the schema, so you can more
easily see how things are affected.

For me, no. This is just syntax; the real question is how it’s
supposed to be implemented.

E.g. I should be able to read some proposed spec text and know
instantly exactly how I might implement it.

Bruce

But my point is that Frank suggested it would be “Humbolt von,
Alexander”. That seems wrong.

It is wrong for display, but can occur for sorting.

where names containing particles are common.

The sort keys that produce this ordering are like as follows:

[Abner, Lisle]
[Doe de, Stephen]
[Doe di, Jack]
[Doe van, John]
[Goode, Charles]

We feel that the ordering given in the first example above (reflecting
current Zotero behavior) is difficult to read, and neither of us is
aware of a publication that requires it.

This is where this discussion is confusing. I’ve been meaning "sort"
form literally as the logic to sort the list of references. But here
you’re mean (to introduce an awkward new term) the “as sorted” display
form. Seems feasible they may differ a bit; not sure. But we probably
need to clarify this in the spec.

I don’t understand what exactly the issue is here.

This entire discussion has been confusing, because we’re talking at
cross-purposes. Let’s keep apart two different things:

  1. how to sort personal names (e.g. what happens internally in some
    processor)

For sorting, the particle(s) can either take priority over the family name
(AEM style, particle-family-suffix-given, e.g. “von Humbolt III A.”), or the
family name can take priority over the particle(s) (Dutch/CMOS-style,
family-particle-suffix-given, e.g. “Humbolt von III A.”).

  1. how to display the sorted form (what is looks like when rendered)

For display, the particle(s) can either come before the family name (AEM
style, particle-family-suffix-given, e.g. “von Humbolt III, A.”), or the it
can come after the given name (Dutch/CMOS-style,
family-suffix-given-particle, e.g. “Humbolt III A., von”).

I’m just saying it might be feasible that 1 and 2 differ. But I’m not
sure, since I don’t really understand all of this.

They can differ.

Another thing we haven’t addressed: what happens to a particle if
given names are initialized?

  1. Humbolt, A. von
  2. Humbolt, A.
  3. Humbolt, A. v.

…?

Option (1).

Rintze

Also, I’ve come across cases of AEM-style display with either AEM-style or
CMOS-style sorting. I haven’t seen any cases of CMOS-style display with
AEM-style sorting (that also seems a bit weird).

Rintze

Am just not sure it works in English on the display of the sorting
name. I’m pretty sure (though not 100% so) a name like “Alexander von
Humbolt” would end up as “Humbolt, Alexander von”.

Well, some journals (I’m looking at you, Applied and Environmental
Microbiology) display as “von Humbolt, Alexander”. So we need the option
(for name display) to either put the particle(s) in front of the family name
(“von Humbolt, Alexander”), or after the family and given names “Humbolt,
Alexander von”. This would of course be limited to “inverted” (Western)
names.

But my point is that Frank suggested it would be “Humbolt von,
Alexander”. That seems wrong.

Not rejoining the discsussion, but for the sake of clarity, what I did
not suggest this. I wrote that sort keys would be composed in
this way. I later suggested that a family-particle-given-suffix
display order might be offered, because you seemed keen to have that
option available.

where names containing particles are common.

The sort keys that produce this ordering are like as follows:

[Abner, Lisle]
[Doe de, Stephen]
[Doe di, Jack]
[Doe van, John]
[Goode, Charles]

We feel that the ordering given in the first example above (reflecting
current Zotero behavior) is difficult to read, and neither of us is
aware of a publication that requires it.

This is where this discussion is confusing. I’ve been meaning "sort"
form literally as the logic to sort the list of references. But here
you’re mean (to introduce an awkward new term) the “as sorted” display
form. Seems feasible they may differ a bit; not sure. But we probably
need to clarify this in the spec.

I don’t understand what exactly the issue is here.

This entire discussion has been confusing, because we’re talking at
cross-purposes. Let’s keep apart two different things:

  1. how to sort personal names (e.g. what happens internally in some processor)

  2. how to display the sorted form (what is looks like when rendered)

This distinction was amply clear in the examples I posted earlier.
The examples in free text represent the display form. The
examples in braces represent (and were described as) sort keys.

I’m sorry if my attempt to clarify things has compounded the
confusion. When the schema has been settled and checked in, let me
know.

  1. how to display the sorted form (what is looks like when rendered)

For display, the particle(s) can either come before the family name (AEM
style, particle-family-suffix-given, e.g. “von Humbolt III, A.”), or the it
can come after the given name (Dutch/CMOS-style,
family-suffix-given-particle, e.g. “Humbolt III A., von”).

Given that, maybe we ought to go back to calling it an “article” :wink:

So if I understand you right, there are three different options:

  1. article-first: article, family, suffix, given (though I’m not sure
    whether the two should be swapped)

  2. article-last: family, suffix, given, article

  3. ???: family, article, suffix, given

3 only applies for sorting; not display.

Is that right?

I’ll take a look at CMS later today to see what it says about this.

Bruce

  1. how to display the sorted form (what is looks like when rendered)

For display, the particle(s) can either come before the family name (AEM
style, particle-family-suffix-given, e.g. “von Humbolt III, A.”), or the it
can come after the given name (Dutch/CMOS-style,
family-suffix-given-particle, e.g. “Humbolt III A., von”).

Given that, maybe we ought to go back to calling it an “article” :wink:

Not fussy about what it gets called, but we went over this terrain
before, I think. In normal usage, articles include “le” or “la”.
Prepositions include “von” and “van”. In the CMS, at least,
“particle” is used as a generic term for both.

So if I understand you right, there are three different options:

  1. article-first: article, family, suffix, given (though I’m not sure
    whether the two should be swapped)

  2. article-last: family, suffix, given, article

  3. ???: family, article, suffix, given

3 only applies for sorting; not display.

Is that right?

I’ll take a look at CMS later today to see what it says about this.

If I remember right, you’ll find the sorting schemes documented in the
indexing section, which is referenced in the rules on bibliography
composition.

  1. how to display the sorted form (what is looks like when rendered)

For display, the particle(s) can either come before the family name (AEM
style, particle-family-suffix-given, e.g. “von Humbolt III, A.”), or the it
can come after the given name (Dutch/CMOS-style,
family-suffix-given-particle, e.g. “Humbolt III A., von”).

Given that, maybe we ought to go back to calling it an “article” :wink:

Not fussy about what it gets called, but we went over this terrain
before, I think. In normal usage, articles include “le” or “la”.
Prepositions include “von” and “van”. In the CMS, at least,
“particle” is used as a generic term for both.

Yeah, that’s fine.

So if I understand you right, there are three different options:

  1. article-first: article, family, suffix, given (though I’m not sure
    whether the two should be swapped)

  2. article-last: family, suffix, given, article

  3. ???: family, article, suffix, given

3 only applies for sorting; not display.

Is that right?

I’ll take a look at CMS later today to see what it says about this.

If I remember right, you’ll find the sorting schemes documented in the
indexing section, which is referenced in the rules on bibliography
composition.

Yup; starting around 18.69.

They have an arabic name that seems to conflict with the earlier rule I quoted:

Jamal, Muhammad Hamid al-

Also, they don’t have an example with both particle and suffix.

Bruce

I tried to work out the specification entry for particles that would
accompany a new particle-option, demote-particles. I welcome any comments
before this gets added to the schema and specification. Bruce/Frank: this
version includes more details on how names with dropped particles should be
handled:

Name Particles

The display format and sort order of (Western) names containing particles
(e.g. “de” in the Dutch name “Willem de Koning” [“William the King”]) can
differ per locale, per citation style and per name. These differences,
discussed below, are observed when only the last name is shown (common in
in-text citations, e.g. “[de Koning 2002]”) or when the full name is
inverted (e.g. “Koning, W. de”):

  1. (display) When only the last name is shown, the particles may or may not
    be dropped (e.g. “Alexander von Humboldt” can appear in in-text citations
    either as “(von Humboldt 1815)” or as “(Humboldt 1815)”). Whether particles
    are dropped is purely name-dependent. For some names only some of the
    particles are dropped (e.g. the French name "Gérard de la Martinière"
    becomes “La Martinière”).
  2. (display) When the full name is inverted, non-dropped particles may be
    either prepended to the family name (e.g. “de Koning, W.”) or appended at
    the end of the name (after initials or given names, e.g. “Koning, W. de”).
    Dropped particles are always appended.
  3. (sorting) When the full name is inverted and the non-dropped particles
    are prepended to the family name, these particles may either remain part of
    the family name (as part of the primary sort key: sort order A), or may be
    separated from the family name and become (part of) a secondary sort key,
    joining any dropped particles, before the given names or initials: sort
    order B:

Sort order A
(1) “de Koning”
(2) "W."
and
(1) “la Martinière”
(2) “de”
(3) “Gérard”

Sort order B
(1) “Koning”
(2) “de”
(3) "W."
and
(1) “Martinière”
(2) “de la”
(3) “Gérard”

Particles can be included in either the family name or given name field of
the input, and are automatically recognized if written in lowercase (e.g.
given:“Alexander von”, family:“Humboldt” and given:“Willem”, family:“de
Koning”). Particles included in the given name field become “dropped
particles”. The option of how particles should be handled for inverted names
is set with the demote-particles option:

demote-particles
: Sets the display format and sort order of particles extracted from the
family name field in case of inverted names (e.g. “Koning, W. de”). The
possible values are:
- “never”: non-dropped particles are treated as part of the family
name, whereas dropped particles are appended (e.g. “de Koning, W.”, “la
Martinière, Gérard de”). Sorting occurs according to sort order A (“de
Koning, W.” appears under “D”).
- “sort-only”: as “never”, but names are now sorted according to sort
order B (“de Koning, W.” appears under “K”).
- “display-and-sort” (default): dropped and non-dropped particles are
appended to the rest of the name (e.g. “Koning, W. de” and “Martinière,
Gérard de la”). Names are sorted according to sort order B ("Koning, W. de"
appears under “K”).—

Rintze

I tried to work out the specification entry for particles that would
accompany a new particle-option, demote-particles. I welcome any comments
before this gets added to the schema and specification. Bruce/Frank: this
version includes more details on how names with dropped particles should be
handled:

Name Particles

The display format and sort order of (Western) names containing particles
(e.g. “de” in the Dutch name “Willem de Koning” [“William the King”]) can
differ per locale, per citation style and per name. These differences,
discussed below, are observed when only the last name is shown (common in
in-text citations, e.g. “[de Koning 2002]”) or when the full name is
inverted (e.g. “Koning, W. de”):

  1. (display) When only the last name is shown, the particles may or may not
    be dropped (e.g. “Alexander von Humboldt” can appear in in-text citations
    either as “(von Humboldt 1815)” or as “(Humboldt 1815)”). Whether particles
    are dropped is purely name-dependent. For some names only some of the
    particles are dropped (e.g. the French name "Gérard de la Martinière"
    becomes “La Martinière”).
  2. (display) When the full name is inverted, non-dropped particles may be
    either prepended to the family name (e.g. “de Koning, W.”) or appended at
    the end of the name (after initials or given names, e.g. “Koning, W. de”).
    Dropped particles are always appended.
  3. (sorting) When the full name is inverted and the non-dropped particles
    are prepended to the family name, these particles may either remain part of
    the family name (as part of the primary sort key: sort order A), or may be
    separated from the family name and become (part of) a secondary sort key,
    joining any dropped particles, before the given names or initials: sort
    order B:
    Sort order A
    (1) “de Koning”
    (2) "W."
    and
    (1) “la Martinière”
    (2) “de”
    (3) "Gérard"
    Sort order B
    (1) “Koning”
    (2) “de”
    (3) "W."
    and
    (1) “Martinière”
    (2) “de la”
    (3) "Gérard"
    Particles can be included in either the family name or given name field of
    the input, and are automatically recognized if written in lowercase (e.g.

Certainly getting there! I’m not sure this part about parsing out the
input needs to be in the formal CSL spec – but clarity on how it
should work with only a given name and last name field is certainly
important for Zotero.

Comparing Sort Order A with Sort Order B, it looks like this scheme
differs from our first-cut plan in two ways: (1) it has a new “shy
particle” that always drops behind the family name (the “de” in the
example); and (2) no particle is ever treated as a fixed part of the
family name for either Sort Order.

Just to confirm, is that correct and intended?

Frank

Sort order A
(1) “de Koning”
(2) "W."
and
(1) “la Martinière”
(2) “de”
(3) "Gérard"
Sort order B
(1) “Koning”
(2) “de”
(3) "W."
and
(1) “Martinière”
(2) “de la”
(3) “Gérard”

Comparing Sort Order A with Sort Order B, it looks like this scheme

differs from our first-cut plan in two ways: (1) it has a new “shy
particle” that always drops behind the family name (the “de” in the
example)

We recognized the need for “dropped” particles before (“Humboldt 1815” for
the name “Alexander von Humboldt”), but I never contemplated what dropped
particles meant for inverted names. The solution I came up with seems
sensible to me, but YMMV. If someone wants to output “de la Martinière,
Gérard” for whatever reason, they’d have to set “de” as a non-dropping
particle. That doesn’t seem like a big trade-off.

; and (2) no particle is ever treated as a fixed part of the
family name for either Sort Order.

Uhm. In sort order A, the non-dropped particles (“de” for “de Koning” and
"la" for “la Martinière”) are part of the primary sort key, right? Is that
what you meant with “fixed part of the family name”?

RintzeOn Mon, Sep 14, 2009 at 5:48 PM, Frank Bennett <@Frank_Bennett>wrote:

I haven’t had time to look at this, but I’m against any distinction
among particles. A particle is a particle. If it is what you call
"non-dropping" then it is by definition part of the family name.

Does that not work?

Bruce

No. “de” in the name “W. de Koning” is non-dropping, but it may be appended
in inverted names (“Koning, W. de”). So it’s not a fixed part of the family
name.

Rintze