Computational Linguistics 6, Exercise 1: testing the grammar

Code (run with py -X utf8, using the chapter's parse):

for sent in ["the man saw i", "i saw", "the telescope saw the man", "the man saw the telescope"]:
    print(sent, len(parse(sent.split())))

Output from the run:

the man saw i 1
i saw 0
the telescope saw the man 1
the man saw the telescope 1

Does it match English?
- "the man saw i": accepted (1 parse), but ungrammatical in standard English
  ("me" is needed). The grammar has no case; "i" is just a pronoun NP.
- "i saw": rejected (0 parses), but it is a fine English sentence. Every VP rule
  needs an object, so intransitive uses are impossible.
- "the telescope saw the man": accepted, although odd; the grammar has no
  notion of what can see. This is a meaning limit, not a syntax limit.
- "the man saw the telescope": accepted, correctly.

Limits: the first is a missing-agreement/case limit (the article says
context-free rules do not capture agreement and reference); the second is
undergeneration by my own incomplete rules; the third is that the grammar knows
nothing about meaning. All three are properties of the grammar I wrote.