DCS-LCSR / SignStream3

Sign language linguistics annotation software
2 stars 0 forks source link

pre-release 3.4.1: Sign Bank does not properly insert sign type for loan signs with prefix ns- #650

Closed cneidle closed 2 weeks ago

cneidle commented 1 year ago

unless the initial gloss entry included #

Screen Shot 2023-02-14 at 10 51 27 AM

This shows up as lexical rather than loan sign.

Screen Shot 2023-02-14 at 10 51 50 AM

Presumably because of the issue with counting NAME SIGN as a sign type, when it is not.

This does work properly if this is initially inserted with a # prefix.

Screen Shot 2023-02-14 at 10 53 54 AM Screen Shot 2023-02-14 at 10 54 06 AM

After entering all data from sign bank, the correct settings from the initial gloss remain in place:

Screen Shot 2023-02-14 at 10 54 15 AM Screen Shot 2023-02-14 at 10 54 21 AM

This is related to other tickets concerning the way "name sign" is handled, which needs to be fixed...

cneidle commented 1 year ago

Cf #607 . Name sign is not a sign type. As you say, we need "either an independent property for 'name sign true/false' in the sign bank file OR a new way to identify signs which are " PLUS " "

cneidle commented 1 year ago

This problem has MOSTLY been resolved.

The new Sign Bank file that Augustine is about to release will no longer replace the sign type by "name sign" for name signs, which is what was causing the basic problem above.

However, name signs are not explicitly marked in any way in the SignStream sign bank. On the SignStream side, you are interpreting anything that starts with "ns-" as a name sign. This is correct, and it works properly in such cases.

However, this misses the cases where the "ns-" is preceded by something in parentheses, e.g. (A)ns-TEXT or (2h)ns-TEXT. So what would remain to be fixed on the SignStream side would be to recognize as name signs not only cases where "ns-" is the first thing in the gloss, but also cases where "ns-" is preceded only by "(" wildcards ")". It is possible to have cases like (2h)(A)ns- but I'm not sure we actually have any of these...

THANKS.

@kwasiopoku FYI

cneidle commented 2 weeks ago

This appears to be working properly now.