#treebank

Question 1: Single choice

A PCFG is estimated by maximum likelihood from a treebank of three trees:

(S (NP John) (VP (V1 said) (SBAR (COMP that) (S (NP Sally) (VP (VP (V2 snored)) (ADVP loudly))))))
(S (NP Sally) (VP (V1 declared) (SBAR (COMP that) (S (NP Bill) (VP (VP (V2 ran)) (ADVP quickly))))))
(S (NP Fred) (VP (V1 pronounced) (SBAR (COMP that) (S (NP Jeff) (VP (VP (V2 swam)) (ADVP elegantly))))))

The sentence "Jeff pronounced that Fred snored loudly" has two parses: loudly modifies the lower verb (snored) or the higher verb (pronounced). How do their probabilities compare?

They are equal, because both parses use exactly the same rules1
The low attachment is more likely, because every ADVP in the treebank is attached low2
The high attachment has probability 0, because it never occurs in the treebank3
The high attachment is more likely, because it uses fewer rules4

Timed test