[HPSG-L] Highly ambiguous sentences with lots of quantifier scoipings and psycholinguistic work on underspeification

[Alexandre Rademaker]

Hi Stefan,

Indeed, Utool has rewriting rules to remove equivalences. I discussed it with Alexander Koller in 2023, see https://github.com/coli-saar/utool/issues/8. This approach deals with not only the trivial ambiguities pointed out by Rui Chaves (only existentials and their all equivalences permutations) but also with the notion of weak and strong readings, so we can choose to keep a more ‘generic’ readings and ignore the more specific ones.

Best,
Alexandre

> On 11 Mar 2025, at 08:19, Stefan Müller <St.Mueller at hu-berlin.de> wrote:
> 
> Thanks Emily,
> 
> Wow! 10^12 readings! This is indeed impressive. But the sentence is not:
> 
> Myrdal is the mountain terminus of the Flåm rail line (or Flåmsbana) which makes its way down the lovely Flåm Valley (Flåmsdalen) to its sea-level terminus at Flåm.
> 
> I guess a lot of the ambiguity is caused by using generalized quantifiers for proper names and they should not scope with respect to each other. So it is an engineering problem, although an interesting one. Utool has some trick to cope with such unwanted ambiguities. There is some way to specify equivalences. I forgot how it works but want to find out.
> 
> So what I am looking for is something like:
> 
> Every daughter of an employee with at least four children knows somebody with a daughter who owns a pony.
> 
> But naturally occurring and with lots of scopings. It would be fine if it were only a million readings. =:-)
> 
> Something that convinces semanticsist or at least students that underspecification is a good thing to have. =:-)
> 
> Best
> 
>     Stefan