[sword-devel] OSIS Variants - do we need a delimiter?

[Chris Little]

On 2/18/2014 7:06 AM, David Haslam wrote:
> When using a front-end to select All Readings, rather than either Primary
> Reading or Secondary Reading, do you think we should have a delimiter
> character between the two variant readings?

I think this is a good idea, particularly since it can be implemented in 
the filter code and won't require module updates.

> If we borrow a tip from Backus–Naur Form, then perhaps the vertical bar
> character would be appropriate?

We should probably approximate the practices of critical editions as 
much as possible, which would suggest that we use the broken bar ¦ to 
split readings within a single instance of variation.


We should probably indicate the starting & ending points of the whole 
variant instance, for which there are a few sensible options:

1) EBNF would suggest using (...|...) for grouping, but:
--DBG's LXX uses ( ) to indicate words/letters missing from some mss.
--Parenthesis are likely to appear in some base texts
--NA28 uses ( ) to indicate slightly varying witnesses

2) [...¦...] looks good, visually, but:
--DBG's NA28 uses [ ] to indicate words/letters of doubtful authenticity
--DBG's LXX uses [ ] to indicate words not in the original text
--[ ] is typical in epigraphy for reconstructions (e.g. to indicate 
abraded text)

3) We could pick another pair of uncommon brackets not appearing in the 
common critical apparatuses, such as any of the following, but this runs 
the risk of lacking font support:
--⸨...¦...⸩ - double parenthesis
--⧘...¦...⧙ - wiggly fence
--【...¦...】 - lenticular brackets

4) Perhaps most ideal, but slightly difficult to implement would be to 
generate real text critical replacement marks:
--⸂....⸃ for multi-word replacement
--⸀.... for single-word replacement
Then the variant reading would appear in an attached note.


In anticipation of wanting to identify the source of textual variants, I 
added a couple of new .conf attributes to the GNT releases a couple days 
ago. Siglum1 and Siglum2 identify the text source of variants 1 and 2 
respectively. So we could conceivably render something like:
(WH reading1 ¦NA reading2 ) with WH & NU subscripted.

--Chris