Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo

Close

#445 Core chapter requires examples of linking quotes to refs

closed-accepted
Martin Holmes
5
2013-02-01
2012-09-27
Martin Holmes
No

This thread on TEI-L:

http://listserv.brown.edu/archives/cgi-bin/wa?A2=ind1204&L=TEI-L&F=&S=&P=57936

demonstrates that guidance on linking quotations with references should be improved, probably in the Core chapter, 3.3.3. Assigned by the Council FTF to MDH, September 2012. This arises out of the now-closed ticket http://purl.org/TEI/BUGS/3520414.

Discussion

1 2 > >> (Page 1 of 2)
  • Martin Holmes
    Martin Holmes
    2013-01-07

    • status: open --> open-accepted
     
  • Lou Burnard
    Lou Burnard
    2013-01-08

    Firstly you might want to revise that example to include a sharp sign! More seriously, did we really agree that @corresp is the right way to do this? I really don't think that's what @corresp is for. I know we use it in the Guidelines to say where an example is taken from but that's also a bit of a stretch. I believe that @corresp is intended for things that are reciprocally aligned in some sense -- like translation pairs, which is the only canonical example we have so far used. I would recommend reverting this addition for the moment.

     
  • Martin Holmes
    Martin Holmes
    2013-01-08

    @Lou: The idea for @corresp came from a posting by one L. Burnard:

    <http://listserv.brown.edu/archives/cgi-bin/wa?A2=ind1204&L=TEI-L&D=0&P=63546>

    If you've changed your mind and would like to suggest something else, please do. In the meantime, there's a clear need for guidance about how to do this, and we do need to provide something. The addition lays out the use-cases, and provides suggestions; I'm happy to replace the suggestions with better ones, but we do need something in there.

    I've added the missing hash.

     
  • Lou Burnard
    Lou Burnard
    2013-01-09

    That L Burnard is so impestuous. More detailed reflection, following other recent discussions about @corresp, suggests that this attribute isn't really fit for this purpose. I could see how you might legitimately use it to link a quotation to the corresponding part of an online instance of the work being quoted, or to a translation of that quotation. The association between a quote and a bibliographic reference for the work quoted seems quite different though: for one thing its not "mutual" in the same way . I think I would now recommend instead something close to Syd's first suggestion in this thread -- add a new attribute to link a quote with a citation. This could be called @citRef or just plain @ref and would have precisely the semantics wanted : this points to a bibliographic description for the source of this quotation .

    I am less convinced that there's any need for the other case Syd describes (a quote with a ref when the two are not consecutive in the text) : the <cit> element is meant for that specific combination iff it appears as such in a source document. If the two are not consecutive or grouped explicitly in the source, then imho that's not a <cit>. I could probably be bullied into saying you might represent the association using a <join result="join"/> or some such madness, but I don't think it's helpful.

     
  • Martin Holmes
    Martin Holmes
    2013-01-09

    So impestuous he's coined a new adjective for himself. :-)

    I do take your point about @corresp, although it's widely used for a variety of different kinds of relationship, based perhaps on the vagueness of the definition "correspond to the current element in some way".
    This says nothing about mutuality.

    It's also true that the Guidelines XML uses @corresp to link <egXML> elements to their source in the bibliography. This is a clear parallel, and if it's wrong, we'll really have to change that too.

    So I think, on balance, there's a strong case that this is a justified use of @corresp.

    However, if it's not, I like your idea of @citRef or @ref. I'd prefer to use the existing @ref, I think, for simplicity, and because it sort of parallels the use of the <ref> element in a <cit>/<quote>/<ref> block. We could add <quote> (and <q>???) to att.canonical, and perhaps then remove @key from it.

    I'm absolutely convinced of the need for a way to link a quote to a <bibl> which is not contiguous to it; in digital journal encoding, it's something I've wanted for a long time.

     
  • Lou Burnard
    Lou Burnard
    2013-01-09

    I agree that the current definition is a bit vague. However "correspond in some way" doesn't really licence an interpretation "is the source for" in my book! Looking up "correspond" in a dictionary shows a variety of senses which all seem to have in common some sense of equivalence between the things that correspond (even in the rather different epistolary context) which is conspicuously missing from the usage being proposed here. "His actions correspond with his intentions" means that there is reciprocity between the two; "this passage in Homer corresponds with that passage in Vergil" means both passages are about the same thing; "the adjective corresponds with the noun" means the noun also corresponds with the adjective....
    As to our practicre in the examples in the Guidelines, as previously noted, I agree that this would need changing and that would be a Good Thing. It's not a usage I have ever been comfortable with. If we'd had @citRef or @cit or @ref on <quote>, we'd have used that without thinking twice about it.

     
  • Kevin Hawkins
    Kevin Hawkins
    2013-01-09

    It sounds like both Lou and Martin are happy with @citRef; the only question, then, is how harshly to scold people who have used @corresp for this purpose instead. I think we if we add @citRef, anyone consulting the Guidelines going forward will see that this is the way to do it, and they still start doing so. If our release notes include mention of @citRef, that might draw people's attention to it in case they were using @corresp and want to update their practice.

     
  • i think there's a fair danger of lots of attributes doing similar things. do we really not already have one we can generalize? this, for example, from <writing>, can't be a million miles away:

    <attDef ident="source" usage="opt">
    <desc>points to a bibliographic citation giving
    a full description of the source or script of the
    writing.</desc>

    If there a vote, I'm inclined to come down in the field against @corresp.
    It will be no problem undoing its use in the Guidelines in favour of @XXX

     
  • Martin Holmes
    Martin Holmes
    2013-01-09

    I liked @source at first. I went looking for a class we might put it in, and found att.sourced, which contains @ed (and presumably now @edRef).

    So if we're not careful we're going to end up with a class called att.sourced that doesn't contain @source, and a different class that does.

    Where do we go from here? Should we move @source into att.sourced? That would add @ed* to <writing> and to <quote> and friends.

     
  • I don't think lb/@source would make sense, or quote/@edRef, tempting as it is.
    The @source on <writing> seems far too useful to keep just for this rather obscure element,
    but we cant add it to att.sourced. att.cited or something?

     
  • Martin Holmes
    Martin Holmes
    2013-01-10

    att.cited or att.cite works for me. But I really feel like att.sourced should be renamed, don't you? Unless we can convince ourselves that it means "source edition".

     
  • Laurent Romary
    Laurent Romary
    2013-01-10

    For the record, while making an ODD spec of EAG and incorporating TEI components, I have become a fan of @source, which is exactly appropriate for the current discussion. So +1. +1 also on revamping att.sourced which is some kind of a naming infelicity.

     
  • James Cummings
    James Cummings
    2013-01-10

    Re: @corresp

    Where another attribute exists to do what I want I'll use that other attribute. But where I just want to say "this element here is somehow vaguely related in some way but I'm not telling you how it is just something I know" then I use @corresp all the time for this. I disagree that it must be a strictly mutual relationship. (It may have been invented as such, but usage over time by many people has certainly departed from this and I view it as a general purpose relationship pointer these days.) See, I'm not entirely a purist.;-)

    I'm also a fan of having @ref on things where there may be more additional information. Why @citRef when we could use @ref? But having @source would certainly be more specific.

    Not sure that helps much, sorry.
    -James

     
  • James may be right in saying that @corresp's mutuality may be weaker than it was; but if it asserts a vague undefined relationship only, we should be _promoting_ its use qua @source (and its promotion and recommend that we came in on).

    I suspect Laurent is using a different @source, and I am worried that the correlation is not there.

     
  • Martin Holmes
    Martin Holmes
    2013-01-10

    I suggested adding <quote> to att.canonical earlier, and given James's view, I think that's a better solution than creating a new attribute just for this purpose.

    If we do that, we should presumably add <q> as well -- any others? Presumably not <said>?

     
  • att.canonical doesnt seem right for a bibliographical link. "canonical information about the object being named or referenced"? your example takes a fragment of what Tolkien wrote, there is no formal object there. you want to name the source of the quotation, not canonical information? equating a quotation with the list of named real-life objects using att.canonical seems wrong.

    alternatively, edit the description in att.canonical...

     
  • My concrete suggestion is:

    * rename att.sourced to att.edition, and change its members accordingly
    * create new att.source with members egXML, q, quote, writing (and said?)
    and an attribute @source, which is data.pointer
    * remove local @source from writing

     
  • Laurent Romary
    Laurent Romary
    2013-01-10

    Perfect. And don't be afraid, this exactly the kind of @source I would want

     
  • Martin Holmes
    Martin Holmes
    2013-01-10

    Agreed, this is the best solution.

    Re <said>: I would not include it, on reflection, since it's explicitly about thoughts and speech, rather than quotation; and in any case, it can be contained in a <quote>, and can contain one, so if an attribution of source is necessary, wrapping will solve the problem.

    Any objections? If not, I'll go ahead and do this, and then update the Core chapter section I added to use @source instead of @corresp.

     
  • +1 from me to just go and do it

    We need to decide whether to switch away from @corresp for the Guidelines themselves. this will involve global changes to the XML sources, and stylesheet changes, of course, fairly wide-ranging ones. If we're gonna do it, better bite the bullet now.

     
  • Martin Holmes
    Martin Holmes
    2013-01-10

    Re @corresp: let's not bite the bullet five days before a commit freeze for a release. We should do it after the release. :-)

     
  • Martin Holmes
    Martin Holmes
    2013-01-10

    [I mean the Guidelines use of @corresp, of course, not the @source plan.]

     
  • ah. i was hoping to have the stylesheets match the features in the new release. chiz chiz

     
  • Martin Holmes
    Martin Holmes
    2013-01-11

    I've carried out the planned changes in rev 11385. Now waiting for Jinks builds to see if all is OK, and inspect the results. UVic Jinks is temporarily borked by a test file -- working on that.

     
1 2 > >> (Page 1 of 2)