[sc34wg3] Supporting variants in TMCL

[Steve Pepper]

* Lars Marius Garshol
|
| The paper is interesting, but unfortunately does not show the precise
| modelling of the variant names. Specifically, it's not clear exactly
| what scopes are used for the variants. This makes it difficult to use
| the examples in the paper as a use case for the TMCL constraint.

I will try to find the time to model Xuân's example in detail and distribute it
before the Leipzig meeting (which I won't be attending, by the way, even though
I plan to go to TMRA). I suspect my modeling will differ somewhat from Xuân's,
so I encourage him to do his own model.

Otherwise I agree that the paper is interesting, although I disagree with the
main conclusion. (No time here and now to explain why, unfortunately. Maybe at
TMRA.)

The major weakness, in my opinion, is that Xuân does not discuss the interaction
with scoped names.* Up to now scoping has been widely used for names in
different natural languages, and I personally see no reason to abandon this
convention. However, we still need variants for different orthographic
renditions of a single (phonological) name, such as the two variants of the
Ottoman Turkish equivalent of Constantinople in Xuân's example.

Steve

* Or typed names, for that matter.