Practice for Homework 7: Phrase structure rules

Question 0 (for future assignments). Lexical categories Which kind of complement does each word below take? Pick the option that better matches the subcategorization frame of the word, and justify your choice by creating a sentence using that complement option. Example: devour DP or nothing Choice: DP (in fact, it would be ungrammatical to use devour with no object, as in *John devours. Sentence: John devours pizza. Eat Observe Load Load Sickness Conversation Key points about doing this problem: 1. Always make up example sentences to represent both complement options. 2. Beware of confusing complements with modifiers: without the complement, the phrase is not complete; whereas modifiers are optional. While this problem does not appear on this homework, you should know how to do this, since we covered complements in class. Question 1. Structural ambiguity One of the following sentences has a structural ambiguity. a. A girl saw a boy with a telescope b. The old man the bridge Provide complete labelled tree diagrams for both interpretations of the ambiguous sentence use the phrase structure rules linked from the schedule! (also at end of last handout) identify by paraphrasing which interpretation corresponds to which structure. Make sure that your paraphrases are not ambiguous and that they match the trees! For the sentence that is not ambiguous, draw the one appropriate tree, and explain what rules out interpretation as a different structure, using trees and rules to illustrate. (a) is ambiguous, while (b) is not ambiguous.[YES, (b) is a grammatical sentence of English]

nothing or DP DP DP or nothing nothing or DP PPloc DP PPon or DP PPfor nothing or PPwith PPabout PPfrom or PPwith PPabout

1. A girl saw a boy who had a telescope

2. A girl used a telescope to see a boy

The old man the bridge initially seems ambiguous (and confusing!) why? Because we want to interpret the words the old man as the subject of the sentence (DP, where old is an adjective, and man is a noun), but then the next constituent should be the IP/VP, which should have an inflected verb or auxiliary as its head (no such verb is found), and should NOT begin with a determiner the!

So, were confused until we see the word the and then we reinterpret this sentence. Thus, the confusion is created by the fact that old could be an adjective or a noun, and man could be a noun or a verb (similar category ambiguity is in the homework). But the sentence is not ambiguous, because the can only be a determiner, and bridge can only be a noun (in our grammar). This also illustrates a VERY IMPORTANT POINT about the proper treatment of PLURAL AND MASS NOUNS. They are listed as plural in our lexicon. We need to keep track of which DPs and NPs contain plural/mass nouns, because different rules might apply to plurals vs. singulars, for example Big cats run. => this is ok without a determiner *Big cat runs. => this is ungrammatical without a determiner

One way to do it is to have a whole separate set of rules just for plurals. So, for instance, to get Big cats run, wed need the rule NPplural AdjP NPplural. This is bad, since it would almost double our little grammar with redundant rules saying the same thing twice for singular and plural nouns. So, we wont do this. Our solution: special principle for keeping track of plural feature: keep passing the plural feature up the tree, until you use it in a rule, or until you dont need it anymore. At that point you can just erase the feature. This is what happens with old above, when NP -> NPplural Question 2. Constituency tests In the lecture we discussed a series of tests for constituency. Are the indicated word strings in the following sentences constituents? (answer yes or no) Use three for each word string: substitution, movement, and question/stand-alone. Provide all three test sentences for each word string (whenever using the stand-alone test, give the questions as well). a) [the tragedy] upset the entire family - to test if the tragedy is a constituent of this S: Sub: It upset the entire family Mov: What upset the entire family was the tragedy Stand-Alone: What upset the entire family? The tragedy! the tragedy [upset the entire family] - to test if upset the entire family is a constituent: Sub: The tragedy did it. The tragedy happened. Mov: What upset the entire family was the tragedy What the tragedy did was upset the entire family [I knew it was going to upset the entire family, and] upset the entire family, the tragedy did. SA: What did the tragedy do? Upset the entire family! FOR STAND-ALONE always make sure to QUESTION the ORIGINAL SENTENCE

b) The [computer was very] expensive Sub: *The it expensive. *The one expensive. *The did so expensive. Mov: *Computer was very, the expensive. *What computer was very, is the expensive SA: *What the expensive? What did the expensive? *Computer was very! c) We ate our lunch [near the river bank] yesterday. Sub: We ate our lunch there yesterday. Mov: Near the river bank, we ate our lunch yesterday. SA: Where did you eat your lunch yesterday? Near the river bank! d) Steve looked [up the number] in the book. Sub: *Steve looked there in the book. *Steve looked it in the book. Mov: *Up the number, Steve looked in the book. SA: What did Steve look in the book? Up the number! NOTE: This is the kind of number up which I shall not look! Look up is a constituent that cannot be broken-up! Up the number is not a constituent! Churchill This is the kind of nonsense up with which I shall not put!

Questions 3, 4. Comparative syntax Question 3 in the homework asks you to compare Tshangla with the table of word-order correspondences discovered by Joseph Greenberg. The table is given in the Tom Payne reading (also below):

Look at the following Russian data, and

Describe the various head and complement orders in the various syntactic constituents illustrated. In what respects do these data conform to Greenbergs observations (see reading/table above), and in which they do not conform? Note any ambiguous or problematic data.

For Question 4 in the homework (Japanese), youll have to first figure out the break-down of the various morphemes, since there is no second line with morpheme-by-morpheme gloss, as below. To do that, just remember what you did with Turkish data. Reading what Tom Payne has to say about Japanese earlier in the chapter will also help.

Write phrase structure rules and a lexicon that will generate these sentences of Russian. Give one more string/sentence that your lexicon and rules would allow, and one that would be ungrammatical according to the rules.

Constituent orders to pay attention to (from Payne reading): Verbs and Objects, Pre- or Post-positions, Possessor (genitive) and Possessed, adjectives and the nouns they modify, relative clauses and the nouns they modify, inflected auxiliaries and the verbs. To answer the first question, just find these elements in the 10 sentences, and describe their order. Below each sentence I give order of relevant elements for that sentence. You dont have to do this, just give me the generalisation based on all the sentences. But you have to look at sentences to figure out what the generalisation is. Word order in Russian below is V O, Pre-positions, Possessor precedes head noun, Head noun follows adjective. To answer the second question, compare this with the table in Payne reading describing Greenbergs generalisations Russian is like the Greenberg VO languages in having VO, Prepositions, and Possessor preceding head noun. However, the nouns in Russian follow adjectives (like in English), so this goes against Greenbergs generalisation.

1. Moj djed posadjil etu boljshuju rjepku. 1stsgPOSS grandfather plant.PRF.PAST this big turnip My grandfather planted this big turnip VERB (plant) precedes OBJECT (this big turnip) Possessor (my) precedes head noun (grandfather) Head noun (turnip) follows adjective (big) 2. Jejo dom stojal na tom vysokom holmje. rd 3 sgFEM.POSS house stand.IMP.PAST on that high hill Her house stood on that tall hill Adposition (on) precedes its complement DP (that high hill) Possessor (her) precedes head noun (house) Head noun (hill) follows adjective (high) 3. V ljesu zhil odjin volk. in forest live.IMP.PAST one wolf In the forest lived a wolf / A wolf was living in the forest Adposition (in) precedes its complement DP (forest) 4. Mjikji spal. Mickey sleep.IMP.PAST. Mickey slept / Mickey was sleeping 5. V tom krasjivom parkje bjegal maljchik. in that beautiful park ran.IMP.PAST boy A/The boy used to run in that beautiful park Adposition (in) precedes its complement DP (that beautiful park) Head noun (park) follows adjective (beautiful) 6. Mjikji jel bulku. Mickey eat.IMP.PAST bun Mickey was eating a bun Verb (eat) precedes its complement DP (bun) 7. Odjin maljchik jel boljshuju bulku. one boy eat.IMP.PAST big bun A/One boy was eating a/the big bun Verb (eat) precedes its complement DP (big bun) Head noun (bun) follows adjective (big) 8. Mjikji kupjil bulku s izjumom. Mickey buy.PRF.PAST bun with raisins Mickey bought a/the bun with raisins = Mickey bought a/the raisin bun Verb (bought) precedes its complement DP (bun with raisins) Adposition (with) precedes its complement DP (raisins) 9. Mjikji bjegal s kazhdoj sobakoj. Mickey run.IMP.PAST with every dog Mickey used to run with every dog Adposition (with) precedes its complement DP (every dog) 10. Moj djed dal jabloko Mjinji . 1stsgPOSS grandfather give.PRF.PAST apple Minnie My grandfather gave a/the apple to Minnie Verb (gave) precedes both of its complement DPs (apple, Minnie) Possessor (my) precedes head noun (grandfather)

For Question 4, the first part, to write phrase structure rules and a lexicon, you need to draw trees for all the sentences. First, draw a plausible tree for the first sentence. o Assign grammatical categories that make sense in English (apple should be a noun, this should be a determiner, etc.). o If youre not sure about the category of a word, put down a couple of options, and hope that subsequent sentences will help you figure it out. Once you have the tree, read the rules from the tree. Next, do the next sentence, making sure that the second tree follows the same rules that you wrote for the first. o If things dont match up, change the rules so they fit both sentences at once. By the time youre halfway through the sentences, you should have a more or less complete grammar, and youll just need to add or tweak a few things to accommodate all 10 sentences. Follow the general phrase structure principles we discussed in the lecture (all your rules should basically be either specifier rules or complement rules or modifier rules). EXAMPLE: Here is how you would go about answering Question 4 for the Russian data above:

Etc. - after repeating this for (1-10), we get the following grammar (this is the answer for this part): Rules: S => VP DP S => DP VP DP => D NP DP => Nproper DP => NP NP => N NP => AdjP NP NP => NP PP VP => Vtrans DP VP => Vintrans VP => Vditrans DP DP VP => VP PP VP => PP VP PP => P DP AdjP => Adj Lexicon D => moj (my), etu (this), jejo (her), tom (that), odin (one), kazhdoj (every) N => djed (g'pa), rjepku (turnip), dom (house), holmje (hil), volk (wolf), ljesu (forest), parkje (park), maljchik (boy), bulku (bun), izjumom (raisins), sobakoj (dog), jabloko (apple) Nproper => Mjikji (Mickey), Mjinji (Minnie) Vintrans => stojal (stood), zhil (lived), spal (slept), bjegal (ran) Vtrans => posadjil (planted), jel (ate), kupjil (bought) Vditrans => dal (gave) Adj => boljshuju (big), vysokom (high), krasjivom (beautiful) P => na (on), v (in), s (with) Question 4, last part, to give some grammatical and ungrammatical sentences, you DONT need to KNOW Japanese (or Russian). Just answer according to what your little grammar rules tell you. (see below)1. E.g. Using S => VP DP rule, we can expand the VP using the VP => Vtrans DP, and eventually build j Kup il jabloko odin djed Bought apple one grandfather => 'One grandfather bought a/the apple' For ungrammatical string (ungrammatical according to our little grammar), we can violate, e.g. PP => P DP to get *Mjikji spal parkje v.
1

NOTE: Payne treats Japanese case markers ga o and ni as Determiners. Thats fine. In Russian, there is also case, which I ignored in the glosses above. So, a more precise way to gloss (1) is: (11) *Moj djed posadjil etu boljshuju rjepku st 1 sgPOSS grandfather.NOM plant.PRF.PAST this.ACC big.ACC turnip.ACC My grandfather planted this big turnip Since ACC case goes with objects, while NOM case goes with subjects, the following would be ungrammatical in Russian: Etu boljshuju rjepku rosla v parke this.ACC big.ACC turnip.ACC grow.IMP.PAST in park intended: This big turnip grew in a/the park To avoid doing this, we need our rules to be able to keep track of which DP is the subject, and which one is the object, etc. But this is something that English rules that I gave in class dont do, its making the rules more complex. SO, its ok to ignore case information, and just have the rule DP -> Det N, which will allow all 10 sentences, and also the ungrammatical sentence (11), because your task is just to have the rules for the 10 sentences, rather than for the entire language, so its ok if the rules are not precise enough.