[osis-core] quotes

[Chris Little]

Patrick & Steve,

Steven J. DeRose wrote:

> Seems to me there are a couple questions in here....
> 
> 1) Quotes interrupted by something, like "she said with a smile; then 
> ocntinuing with a less friendly expression,", etc -- in this case, there 
> is a good case that the stuff on both sides form a discontiguous but 
> single quotation.
> 
> In such cases, I think it would be valuable to be able to say that. I 
> know two ways to do that:
> 
> a: co-index the quotes before and after (like with "continued-in" and 
> "continues" attributes, or some such -- a lot like segmentation, but I 
> agree with Troy that it's not the same thing conceptually..
> 
> b: Mark the interrupting text as such (semantically, flagging it so that 
> it doesn't inherit the property "is a quote" (or "is a quote of the same 
> level", in certain grosser cases), from the quote; then mark the whole 
> quote as a unit. This seems trickier, especially to process, so I favor 
> (a).

I agree completely.  (a) sounds good, (b) sounds icky.  Patrick's 
solution of reviving splitID seems good.  I'm not sure that it should be 
global again however.

I'm unclear on the initial example you give since it doesn't seem to 
have any interruption, but I interpret you to mean something akin to:
"Soon," she said, "I will go to town."

With splitID added, I would encode that as:
<p><q splitID="ref">Soon,</q> she said, <q splitID="ref">I will go to 
town.</q></p>

> The other question involves the relationship of these structures to 
> typography. English, I think, is one of the easier languages. It does 
> have the silliness (IMHO) about leaving the close quote off of a 
> paragraph when the quote continues into the next paragraph, but *not* 
> leaving off the balancing re-open quote at the start of the next para), 
> but that's not too hard to process as a special case in your XSL (I 
> think). But I believe Spanish typography has much more complicated 
> conventions, that let you tell from the punctuation whether you're in 
> the first part, a medial part, or the last part of a discontiguous 
> quotation -- I don't know the rules, but I'm pretty sure they'd rquire 
> at least knowing which quoted bits are part of the same conceptual 
> quotation.

I'll repeat my proposed markup of Troy's example from the NASB (updated 
to include splitIDs as necessary):

<p><verse>Then Peter, filled with the Holy Spirit, said to them, <q 
type="spoken" level="1" eID="someRef" who="Peter"/>Rulers and elders of 
the people, if we are on trial today for a benefit done to a sick man, 
as to how this man has been made well, let it be known to all of you and 
to all the people of Israel, that by the name of Jesus Christ the 
Nazarene, whom you crucified, whom God raised from the dead--by this 
name this man stands here before you in good health.</verse>
<verse>He is the <q type="otPassage" osisRef="Ps.118.22" 
splitID="split">stone which was rejected</q> by you, <q type="otPassage" 
osisRef="Ps.118.22" splitID="split">the builders</q>, but <q 
type="osPassage" osisRef="Ps.118.22" splitID="split">which became the 
chief corener</q> stone.</verse>
<verse>And there is salvation in no one else; for there is no other name 
under heaven that has been given among men by which we must be saved."<q 
sID="someRef"/></verse></p>

As I mentioned, the most common presentation used for the NASB is to 
print every verse on a new line.  With that presentation, verse 
boundaries that occur in the middle of a quotation cause an extra open 
quotation mark to be generated at the start of the new verse.

There is however, another presentation format common to printed NASBs 
that uses traditional paragraph formatting.  In this case, extra open 
quotation marks would only be generated if a quotation crosses a 
paragraph boundary.

The two styles of presentation can be controlled by separate stylesheets 
without much difficulty. Adding <q/> milestones for continuation quotes 
would result typographically-motivated markup.  Encoders will place the 
milestones where they believe quotation marks should appear according to 
the presentation they envision, without regard to alternative 
presentation forms.  For English texts that add extra open quotation 
marks, encoders will tend to add only start milestones without matching 
end milestones.  And if they DID actually encode texts correctly, it 
would still void Troy's position that using <q/> must result in the 
generation of a quotation mark character of some type.

It seems to me that putting different sorts of quotation marks at the 
start & end of initial, medial, and final blocks of quotations spanning 
paragraphs (or verses, as in NASB) is no more difficult than putting 
open quotes on all blocks & close quotes only on the final, as in English.

> This knowledge is also needed to do sensible searching: "Where did Jesus 
> say X?" can't just be treated as Select element e where e.type = "Q" and 
> e.text contains X.... because His utterance may have been split across 
> Qs -- even though the SQL to do the query right will be gross, I think 
> we should at least have enough markup to make it *possible".
> 
> You could typeset a red-letter edition fine without this, but search 
> imposes more requirements.

I assume this addresses the problem for discontinuous quotations 
(problem 1), not continuous quotation that span paragraph boundaries 
(problem 2).  For problem 1, I think the need does exist and Patrick's 
solution of using splitID should solve it.  For problem 2, <q/> 
milestones already solve the issue.

--Chris