RE: [jboske] scope issues

["And Rosta" <a.rosta@lycos.co.uk>]

xorxes:
> 1- I assume that
> 
> (1a) lo cipni cu blabi gi'e vofli
> 
> expands to:
> 
> (1b) su'o da poi cipni zo'u ge da blabi gi da vofli
> 
> and not as:
> 
> (1c) ge su'o da poi cipni zo'u da blabi gi su'o de poi cipni zo'u de 
> vofli
> 
> 2- I assume that:
> 
> (2a) lo cipni cu blabi gi'e na vofli
> 
> therefore expands as:
> 
> (2b) su'o da poi cipni zo'u ge da blabi gi da na vofli
> 
> If this is correct, here we have a {na} (in 2a) that does not export to the 
> first place in the prenex. 

I assume (2) too, but it follows from this that the official na-scope
rule is indefensible (except on the grounds of its very officialness).

OTOH, if its very officialness does render the na-scope rule Right,
then assumption (2) must be wrong...

> In 2b it does, because there is a prenex after gi:
> 
> (2b) su'o da poi cipni zo'u ge da blabi gi naku zo'u da vofli
> 
> 3- I assume that:
> 
> (3a) lo cipni cu na vofli gi'e blabi
> 
> expands to:
> 
> (3b) su'o da poi cipni zo'u ge da na vofli gi da blabi
> 
> and not to:
> 
> (3c) naku su'o da poi cipni zo'u ge da vofli gi da blabi
> 
> i.e., I'm assuming {(na vofli) gi'e (blabi)} and not {na (vofli gi'e blabi}.
> I believe this is how the parser would group it, though that is no
> guarantee of anything.

My natural inclination would be to read it as (3b), albeit with a
fair dollop of doubt. I think the safest rule for afterthough
connectives would be to take the narrowest possible scope.

> If this is correct, then comparing {lo cipni cu na vofli} with
> {lo cipni cu na vofli gi'e blabi} we see that the final {gi'e blabi}
> completely turns around the first part. Very weird...

... which shows that it can't be correct. Something's gotta give.

--And.