[tei-council] pb, lb

[Lou Burnard]

Here is a small essay on this topic, with a suggestion at the end.

Most milestone units indicate a change of state of some kind. The 
default assumption is that the old state ceases to apply and the new 
state starts to apply at the point the milestone tag is inserted in the 
text. Consequently if for some state (say "being on page 42") there is 
no milestone, you are not yet in that state. This is the rationale for 
saying that you must put a <pb> before the material on page 42, not 
after it.

There is another interpretation of milestones, which is that they are 
solely there to mark points (like <anchor>s). In this sense, the <lb/> 
for example doesn't really mark a change of state (from "being on this 
line" to "being on a different line") but simply the notional point 
between the two lines. But this looks like sophistry to me.

I think there are two real usage issues that need to be addressed 
somehow. The first is that *some* milestone changes imply changes in 
others Changing from one page to the next implies of necessity changing 
from one line to the next and (if I am doing columns) one column to the 
next. So there is an implicit hierarchy in which inserting a <pb/> 
notionally inserts also a <lb/> and a <cb/> at the same point, just as 
inserting a <cb/> implies inserting an <lb/> : you just cannot have a 
column or a line which spans a page break, without doing violence to the 
definition of what "column" or "line" means.

Note that this is by no means true of all milestones -- indeed the main 
reason for introducing milestones is precisely to allow for units which 
do not behave in this way. But, I contend, it *is* true of the three 
mentioned here, which are also, probably, the three most frequently used!

If you grant the possibility of "implicit" milestones (and you can of 
course say "this is XML: we don't do minimization") then you also have 
to confront the fact that some milestones are implicit in the deployment 
of other elements. For example, part of the definition of <p> or <l> 
(but not, I suggest, <item> or <div> ) is that they have an implied 
<lb/> at their beginning. I am not, of course, talking about how we want 
to render the things, I am talking about how we recognize them. If we 
found in our text something which had other paragraph-like properties, 
but didn't start a new line, we'd probably have difficulty agreeing that 
it was a paragraph (in fact, I'm having difficulties thinking how else 
we'd identify it at all). If we are marking up a poem, there is probably 
an implicit <lb/> following each <l> -- so that when a metrical line 
gets fractured by the typography, we can happily insert just one <lb/> 
solely at that point.

The trouble is that we have no way, currently, of formally stating these 
rules of interdependence or implication between elements (assuming, 
indeed, that you think they are useful). We cannot easily record that a 
an occurrence of milestone type x implies an occurrence of milestone y, 
nor that the presence of element a implies an occurrence of milestone x.
And, as I suggested above, you may plausibly say "tough: we don't do tag 
minimization" and leave it at that.

Since however <pb> <lb> and <cb> are so widely (and inconsistently) 
used, I can't help wondering whether some tighter usage rules for these 
special cased milestones only -- along the lines indicated above -- 
might not help the community. What do you think?




Dan O'Donnell wrote:
> 
> On Wed, 2007-26-09 at 12:30 +0200, Matthew James Driscoll wrote:
>> I've always maintained that the "b" stood for "boundary", rather than
>> "break" (an idea I believe I got from Lou originally), and insisted that
>> they be put at the front of the thing in question -- don't make no sense
>> otherwise.
> 
> I think the impetus to put them at the end comes from html:lb usage.
> 
>> Matthew
>>
>> -----Original Message-----
>> From: Lou Burnard [mailto:lou.burnard at oucs.ox.ac.uk] 
>> Sent: Tuesday, September 25, 2007 8:34 PM
>> To: tei-council at lists.village.Virginia.EDU
>> Subject: Re: [tei-council] pb, lb
>>
>> Dan O'Donnell wrote:
>>> Hi all,
>>>
>>> This is probably silly, but I've been teaching a student research
>>> assistant P5 and the issue came up.
>>>
>>> Was there not discussion some time ago of changing the name of pb and lb
>>> from page break and line break to page begin and line begin?
>> No discussion about changing the name of the element sfaik. But 
>> agreement to clarify that th "b" should be understood as implying you 
>> put the thing at the "beginning"
>>>  The
>>> preferred convention is to put milestones at the beginnings of spans
>>> they involve (e.g. handShift), right?
>>>
>>>   
>> krekt. " By convention, pb/ 
>> <http://tei.oucs.ox.ac.uk/P5/Guidelines-web/en/html/ref-pb.html> 
>> elements should appear at the start of the page to which they refer." is 
>> what it says in the Good Book.
>>
>> _______________________________________________
>> tei-council mailing list
>> tei-council at lists.village.Virginia.EDU
>> http://lists.village.Virginia.EDU/mailman/listinfo/tei-council
>> _______________________________________________
>> tei-council mailing list
>> tei-council at lists.village.Virginia.EDU
>> http://lists.village.Virginia.EDU/mailman/listinfo/tei-council