Clarifications for Assignment 2
15 September 1997

Thanks to all of you who participated in the mailing list discussion
over the last day or so.  I was not able to get to my mail until this
afternoon, but it seems like you have done a nice job together of
reasoning out the right answers.  (Yes, assignments, like programs,
sometimes need to be debugged.)  To be clear, I will give full credit
for assignments that meet the letter of what I assigned, but the
following clarifications should answer the questions about what was
intended.

Assignment 2b

  The following should be accepted/*rejected:

    the second two contestants admired the first two contestants 
    the first and second contestants admired the second two contestants
    the second two contestants admired the second two contestants
    the first contestant admired the first contestant
    the first and second contestants admired the first and second contestants
    the second and second contestants
    *the second and contestants

  So yes, the general principle of expanding coverage while retaining
  the previous behavior is correct.  Your machine *will* overgenerate,
  however, especially in the case of order effects (e.g. '*the second
  and first contestant').  This is frequently the case, and rather more
  general than what we're doing here; e.g. MANY grammars will accept
  both 'the big blue house' and 'the blue big house' even though a
  native speaker of English would reject the latter.

Assignment 2c

  I am only concerned with sequences that can appear after
  the subject NP in a simple active declarative sentence.

  Yes, sequences with no auxiliaries must be accepted,
  e.g accept 'write' (because "They write" is a valid sentence 
  of English) but reject 'written'.

  No, you don't need to deal with 'to' or 'need', or any other
  lexical items other than the ones I included; nor the modal use 
  of 'have'; nor gerunds; nor whatever other fun stuff Kleanthes
  decides to throw in there.  Maybe I'll have him write a few of
  the exam questions, though.  :-)