Ling 566: Intro Syntax for Computational Linguistics

Midterm Exam

Due 11/11 11:59pm

No collaboration. Send questions directly to me by email, and I will post Q&A on Canvas. You are strongly encouraged to subscribe to the Canvas discussion for the midterm so that you get updates by email. I won't be answering email over the weekend, including Monday 11/11 (which is a holiday).

The late policy for the midterm is as follows: Unless you make prior arrangments with me, midterms submitted late but by 11:59pm on 11/12 will receive 80% credit, and by 11:59pm on 11/13 will receive 70% credit. Midterms will not be accepted after 11:59pm on 11/13. Don't be late!

Please follow the same formatting policy as for homework (as specified on the main course page).

• Chapter 10, Problem 1

  Note: Your answer should include an ARG-ST for each passive verb in (i)-(viii) as they appear in those sentences, as well as an explanation of whether and how these ARG-STs together with the Binding Principles predict the acceptability judgments shown in (i)-(viii). You should account for each sentence explicitly -- no general statements about all of them together.

  The ARG-ST values should only include abbreviations (e.g. NP), INDEX values (so NP_i) and MODE values. For example, for the ARG-ST of seen as used in the sentence Kim_i was seen by Sandy_j, I'd like to see:

  [ ARG-ST < NP_i[ref], PP_j[ref] > ]

  There should only be one ARG-ST shown per sentence. It should correspond to the passive participle in that sentence (i.e. the ARG-ST of the word structure that is licensed by the OUTPUT of the Passive LR). Do not include any other ARG-STs.

• Chapter 10, Problem 3

  NOTE 10.3 was revised somewhat between printings. Please answer this version.

  For part C, please give ARG-STs with the same level of detail as requested in 10.1 above, with the exception that here FORM on the PPs is relevant and MODE isn't necessary.

• Question 3, with parts A-F, as follows:

  Consider the sentence:

  • Allocate or assign Lee's account to Sam!

  Using the grammar as developed through Ch 10:

  1. For each word in the sentence, identify the lexical type of the lexical entry that licenses it.
  2. For each word in the sentence, specify what lexical rule licenses it.
  3. Draw a tree showing the structure assigned to this sentence, using abbreviations defined on p.279 for the nodes. Use D as the abbreviation for nodes that are [HEAD det] and have a non-empty SPR value. Use CONJ as the abbreviation for any node that is [ HEAD conj ].
  4. Draw another tree showing the same structure. This time, use feature structures for the nodes, as follows:
     1. Show only the value of SPR, COMPS and ARG-ST (where appropriate) on each node, using abbreviations such as NP for the elements of these lists.
     2. Use tags to show identities between elements of these lists and elements of other lists or nodes in the tree.
     3. Show all identities required by all principles and rules, but don't need to show identity between empty lists.
  5. In the fully resolved tree, what is the PER value on the node (of type word) directly above Assign (at the path SYN | HEAD | AGR)? What rules and principles constrain this PER value? [Hint: I find 3.]
  6. Describe the chain of identities that links the ASSIGNEE role of the assign predication to the NAMED role of the name predication for Sam. Each link should say what two values are identified and what part of the grammar (lexical entry, rule, or principle) identifies them. [Hint: I find 9 links in this chain, including two required by the Argument Realization Principle. None of the links are due to the Semantic Compositionality Principle.]

     Please format your answer as a table, as in the following example:
     
     

     	Grammar entity (lex entry, phrase structure rule, principle)	First member of identity	Second member of identity
     1.	Lexical entry for assign	ASSIGNEE role of the assign predication	INDEX value of the second argument on assign's ARG-ST list

     Here is a latex template for the table, for those who would like to use it.

Please create a pdf file and upload it to Canvas.