A few points about the questions.
Almost everybody got full marks for the minimal pair question, but I
think there is a bit of confusion on how the method works. Some people
gave cases like (1) or (2) as examples of minimal pairs.
(1) a. John likes to talk. b. John likes to talk with Mary. (2) a. John has talked to Mary b. John has to Mary
In (1) "with Mary" changes the meaning. In (2), removing "talked"
makes the sentence ungrammatical. This is true, but it is utterly
trivial. Normally, then we change something we are interested in
changes that are not so immediately and obviously connected with the
element inserted. One example one gave me is:
John is sick John is sick of Mary
Her the suprising fact is that the appearence of "of Mary" changes the
meaning of "sick". Other example is
John allowed Bill to leave. John promised Bill to leave.
Where changing the verb changes the understood subject of "leave".
Few people got the question on Type 3 grammar right, and indeed, it required some thinking on what counted as a termina element. P, N, Comp, Adj, Det, V where terminals: as you can see they where not specified by any rule in the left column. If you included the Lexicon, then no rule was Type 3 (but then no grammar with abstract terminal symbols ever is). Some people wrote that every rule had to have the form A --> cB where c is terminal and B is non-terminal. Note however that a grammar of this form can only produce infinitely long sequences. The rule S --> DP VP was definitely non-Type 3, and so was DP --> NP, though as some observed one could get this effect in Type 3 by choosing an empty terminal.
There was some confusion in the overgeneration task as well, and probably I wasn't clear enough in stating the question. Many gave as an example of overgeneration sentences that were perfectly grammatical in English but nonsensical (the "Colorless green ideas sleep furiously" type). This is a bit cheating, because if you do this any grammar overgenerates: grammar does not exhausts semantic constraints, or world knowledge (I cannot claim that my grammar overgenerates if it produces "Barak Obama is a panda", which is unacceptable given what I know of BO). Better answers had sentences which violated some grammatical constraints: bare singulars (*"integer is sum") or verb selectional restrictions (*"Goldbach was that the integers is prime"). Nobody thought of modifiers ordering (*"Prime two numbers"), which is clearly a grammatical fact in many cases.
Most people did not break down "entangling" down to en+tangle (cf. enliven, embrace, enlight, encase, enclose ... and Italian "a+nod+are"). "Tangle" is primarily a noun, but can also be a verb. Likewise few reduced "grammatical" to "grammar" (possibly via "grammatic", cf. "program/programmatic"). Again, pairs like "adjective/adjectival", "verb/verbal", can be helpful.
Some people applies "dis" at the end of the derivation: [dis
[entangle+ing]] but added that "-ing" is an inflectional suffix
(go/going). But then this order must be wrong, since as you have
studied derivational affixes never apply outside inflectional
ones. One could argue that "-ing" is derivational, since some "-ing"
forms are nouns ("He took a beating"). Yes, "dis-" applies to other
verbal forms: "to dis-entangle" "he dis-entangl-ed/-s", so a route via
a nous to get back a verb is highly unlikely.
Moral: morphology IS difficult.
There was also a lot of confusion in the question on content
vs. grammatical morphology. Many confused it with the distinction
between inflectional and derivational morphology. But
"inflectional/derivational" is a distinction which only applies to
affixes (there are no "inflectional/derivational" roots). All
inflectional affixes are grammatical, but so are most (probably all)
derivational ones (e.g. "X-able" means "which can be Xed", but "can"
is a modal, hence part of the grammatical lexicon; "anti-" means
"against", which is a preposition, so again grammatical).
Agents are volitional entities. When an agent does an action you can always append the adverb "deliberately/involuntarily". You cannot do that with "the ice" beaking the bottle. It is a natural clause. Likewise "The ice" cannot be an 'experiencer' when it melts, unless you are talking of Olaf in the film Frozen. I will bring the exams in my office, in case you want to take a look.