Reminder: There is no collaboration (consultation with other students) allowed on the midterm. Please send any questions to me via e-mail. If the answers are relevant for the whole class, I will post them to E-Post. Don't be shy about asking questions! In particular, you don't have to decide whether or not it's an okay question to ask. If I think answering it will give away too much information, I'll say so :-)
What follows are some specific notes on each of these problems.
Most importantly: You only need to do Part A of this problem.
This is a problem that asks you to consider a broken model (where the AGR value of possessive pronouns is as given in (i)) and explain why it is broken. In essence, this amounts to stating what it predicts about certain relevant sentences, stating why it makes those predictions, and stating what the actual facts are (i.e., actual grammaticality of the sentences).
Part A. An ideal answer to this part of the question will include relevant examples as well as discussion about how the grammar will license any grammatical examples you cite or fail to license the ungrammatical ones, given the assumption provided in the problem (possessive pronouns are underspecified for number).
Part B. All you need to provide here is the AGR value for each of the three words (ma, mon, mes). Since we're working with another language, you should make explicit your assumptions about which (if any) of the subtypes of agr-cat you posit for French.
This problem asks you to consider the interaction of two separate aspects of the theory: The Binding Theory and the analysis of passive. As noted in the problem description, an ideal answer to this question will specifically address the predictions that the grammar makes about each of the examples given ((i)-(viii)). Use your understanding of the Passive Lexical Rule and the Binding Theory to determine which lexical item's ARG-ST is relevant (just one in each case) and what that ARG-ST is in the example in question. For each example, determine whether both Binding Principles are satisfied. Explicitly state whether the predictions of the grammar match the judgments given.
Note 1: The examples involve argument-marking prepositions, so you'll need to be sure you understand how they interact with the extended definition of Outrank.
Note 2: Remember, our account of Passive does not involve relating sentences to each other, but rather relating lexical sequences to each other. In fact, for this problem, the input lexical sequences are not relevant at all. You only need to consider the output of the Passive Lexical Rule (i.e., the particular lexical sequences involved in licensing the passive sentences in question).
Part A. At least two examples means for one verb, give the two valence patterns. The best examples keep everything else in the sentence constant, except what you are focusing on (in this case, the Dative Alternation).
Part B. The lexical rule should contain sufficient information to constrain the class of inputs to the right type of verb (although you needn't worry about lexical exceptions to the Dative Alternation), as well as enough information to constrain the output lexical sequences. At the same time, you don't want to add extraneous information. Taking advantage of the lexeme type hierarchy can help make a very compact statement of the rule. One way to check your work is to pick a lexeme from the lexicon, calculate what the OUTPUT value would be, and then try to see if it "breaks" the grammar: e.g., would it allow a D (instead of an NP) as one of the complements? Also check whether your rule gets the semantics right. The examples in (i) have the same meaning as their counterparts in (ii).
Part C. Here the ideal answer shows the ARG-ST of the original lexical sequence (from the lexical entry that would need to be posited in the lexicon), as well as the linking of ARG-ST elements to roles in the verb's predication. It then shows the corresponding output when that lexical sequence is the input of the first lexical rule to apply (and says which rule this is!). In the case where there is a second lexical rule, the ideal answer continues by showing the output value when the output of the first lexical rule is used as the input of the second lexical rule (and says which rule this is!). Finally, this should be done for both examples (iii) and (iv).
Note: Lexical rules do not relate sentences to sentences, so I do not want to see anything about "input sentences" or "output sentences" or "when we apply Rule XYZ to sentence (i)".
Part D. It is useful in this case to work backwards from (v) and figure out what the ARG-ST for handed in this example must be (and what the linking to roles in the predication would look like on the sensible reading). Then describe what the rules would have to do to create that ARG-ST, and why your Dative Alternation LR and the Passive LR can't do this. To make your answer concrete, be sure to include that target ARG-ST.
Note: Lexical rules do not relate sentences to sentences, so I do not want to see anything about "input sentences" or "output sentences" or "when we apply Rule XYZ to sentence (i)".