#705 <lg> cannot contain <q>, <seg>, etc.

[Andrew Dunning]

It is possible in the real world to have quotations that span multiple verse lines, but in TEI <lg> cannot contain <q>, <said>, <seg>, and the like, making this invalid:

<lg>
<said>
<l>speech</l>
<l>speech</l>
</said>
<l>verse</l>
<l>verse</l>
</lg>

If you decide to change this, please also allow <seg> to contain <l> (for example, to allow one to mark a source for several lines without specifying that they are treated as a quotation in the text).

[Lou Burnard]

This is not strictly speaking a bug, but rather a consequence of the fact that XML only allows one hierarchy. In the TEI scheme of things, phrase-level elements such as <q> or <said> cannot both contain chunk-level elements such as <p> or <l>and be contained by them: you need to represent this kind of structure using either standoff or milestones. I am not certain from your example, but if what you want is to show that the first two lines of a stanza (or parts of other lines) are to be read as direct speech, an easy way would be to use a milestone element to mark where the direct speech begins and ends. <said> is for direct speech, by the way, not quotation from another source, which is <quote;

[Andrew Dunning]

Sorry not to have been clearer; I am having problems with marking up both quotation and speech in verses. What I find confusing in the example above is that the lines become valid if one removes the <lg> element, but I suppose that it's taking <q> and its allies as equivalent to <lg>? The idea of using <milestone/> is interesting, though using it as an alternative to <quote> or <seg> is complicated by its lack of @source.

Perhaps the 'bug', then, is simply that the guidelines do not say anything specific about dealing with anything that requires quotation marks within <l>: encoding a simple dialogue that runs across several lines can be quite time-consuming, and from the several alternatives available it's not altogether clear which method is most likely to be interpreted correctly by other software.

[Syd Bauman]

I'm not sure Lou has the details right, there, but I think he has the right big picture. It's probably a bad idea to allow <said> and friends inside <lg> if for no other reason than an <lg> is a group of metrical lines, not of other stuff.</lg></lg></said>

Details:

1) <said> and <q> may already both contain <l> and be contained by it.</l></said>

2) I think (or at least hope) Lou means to suggest using an "empty element used as a boundary marker" rather than an actual "<milestone> element", because it is not a true milestone. (<said> and <q> do not tessalate the way, say, pages do). The Guidelines recommend <anchor> for this purpose. (In chapter 20.9.) </anchor></said></milestone>

I think the solution here is to provide an example or two in the Guidelines of how to do this sort of thing. The following examples are adapted from some found in the WWP's corpus by my colleague Sarah Connell.

  <lg type="tercet" rend="braced()">
    <l><said rend="pre(“)post(”)">Sewell is dead.</said> Swift-pinion'd Fame thus cry'd.</l>
    <l><said rend="pre(“)post(”)">Is Sewell dead,</said> my trembling tongue reply'd,</l>
    <l>O what a bleſſing in his flight deny'd!</l>
  </lg>
      <!-- from Phillis Wheatley's _Poems on Various Subjects, Religious and Moral_ -->

    <lg type="couplet">
      <l>
        <said xml:id="said1297" next="#said1298">Our lives</said>, ſaid he,
        <said xml:id="said1298" next="#said1299" prev="#said1297">wee'll give before we yield</said>,
      </l>
      <l>
        <said xml:id="said1299" prev="#said1298">Wee'll win your battles, or dye in the field</said>.
      </l>
    </lg>
  <!-- from Margaret Cavendish's _Nature's Pictures_ -->

  <lg type="stanza">
    <l><said>What do ye fear?</said> the baron cried—</l>
    <l rend="indent(1)">For Zorietto trembled.—</l>
    <l>
      <said xml:id="wolf1" next="#wolf2-1" rend="post()">A wolf,</said> she sigh'd with whisper low,
    </l>
    <l>
      <said xml:id="wolf2-1" rend="post()" next="#wolf2-2" prev="#wolf1">Hark how the angry whirlwinds blow,</said>
    </l>
    <l rend="indent(1)">
      <said xml:id="wolf2-2" rend="post()" next="#wolf3" prev="#wolf2-1">Like demons dark assembled!</said>
    </l>
  </lg>
  <!-- from Mary Robinson's _Poetical Works_ -->

[rviglianti]

I like the idea of using @prev and @next: it allows to use the full functionality of <said> (which is what the encoder needs in this case) while allowing to break across the principal hierarchy, the verse structure.</said>

As Andrew said, one question remains: should we recommend one encoding on the guidelines? If yes, should it be @prev/@next?

[Martin Holmes]

I think it would be a good idea to show two approaches: the one nicely exemplified by Syd above, using @prev and @next, but also perhaps a "pure" standoff version using <join>, with @result="said". The same example could be encoded using both mechanisms to show the difference.

[rviglianti]

+1

[rviglianti]

Fixed in revision 13190.