[Ltru] Matching Text

We've been talking about two different models for dealing with
macrolanguages: extlang and no-extlang. In our telephone conversations, a
third has come up: partial-extlang.

The partial-extlang model is where we only make X a secondary language for Y
when it can reliably be assumed that someone who understands X will also
understand Y. That would be the case if all of the microlanguages of Y are
mutually intelligible, or if Y is generally identified with a "prestige"
variant. With this model we wouldn't make all the microlanguages be
secondary, just certain ones like those having ar and zh as macrolanguages.
An example might be zh-yue.

To assess these options, I suggest that the right way to do it is to look at
the textual changes that we would need to make to RFC 4747 in order to have
each of these work -- AND describes the implications for implementers. That
way we can assess actual text that reflects the impact of the models on the
operations that use their structure.

Mark