Exercise 1 Formulate in your own words the condition that regulates the relation between who and the pronoun he in (1): (1) a. *the man who i he i praised left town b.??the man who i his i mother praised left town c. the man who i praised his i mother left town (2) a. *El hombre a quien i pro i alabó se fue de la ciudad. b.??el hombre a quien i su i madre alabó se fue de la ciudad. c. El hombre que i alabó a su i madre se fue de la ciudad. In the a and b examples, the wh-element crosses over the pronoun (hence the name of crossover assigned to this phenomenon; [?]), so that we obtain configuration like the following: (3) wh i... pronoun i... t i In the a. cases, the pronoun is c-commanding the trace of the wh-element, and if they share the index, it is binding it (strong crossover), which is forbidden, for traces are like names, and cannot be bound by a pronoun. Crucial concepts: ˆ binding = c-command plus the same reference, coded by means of an index ˆ Binding principles: Principle A: anaphors (= reflexives) must be bound in their own sentence. Principle B: pronouns cannot be bound in their own sentence. Principle C: referential expressions (= nouns, and traces) cannot be bound. Examples: (4) Principle A a. John i likes himself i. b. *Himself i likes John i. the reflexive is not bound. c. John i thinks that himself i is a genius. the reflexive is bound, but not in the same sentence. (5) Principle B a. *John i likes him i. the pronoun is bound in its own sentence. b. He i likes Mary k. the pronoun is not bound, so it gets reference by discourse principles (coreference). c. John i thinks that he i is a genius. the pronoun is bound, but not in the same sentence. (6) Principle C a. John i likes Mary k. the names are not bound. b. *He i likes John i. the name is bound. c. He i thinks that John i is a genius. the name is bound. Hence, the examples are ruled out. (7) strong crossover a. wh i binds pronoun i b. pronoun i binds t i In the b. cases, the situation is different, for the pronoun is not c-commanding the trace of the wh-element, and even though they share the index, it is not binding it (weak crossover). 1
(8) weak crossover a. wh i binds pronoun i b. wh i binds t i There is no consensus on the proper analysis of these cases, which show a weaker ungrammaticality. One proposal is arguing that in (8) the wh-element is binding two elements at the same time, which creates an unacceptable semantic ambiguity [Koopman & Sportiche(1982)]. Finally, note the c. examples, where there is no crossover effect: the wh-element binds its trace, and the trace doesn t bind the pronoun, so it can corefer with it: (9) no crossover a. wh i binds t i b. pro i does not bind t i c. wh i does not bind pro i Exercise 2 With the rules in (10), derive The girl has bought a bike and A boy believes the girl is renting the car. (10) a. S DP AUX VP b. DP D NP c. NP N d. VP V; VP V DP; VP V S e. D the; D a f. N girl; N boy; N car N bike g. V sleeps; V runs; V thinks; V believes; V bought V renting; V wants V sold h. AUX has; AUX is 1. S DP AUX VP 2. D NP AUX VP 3. D N AUX VP 4. D N AUX V DP 5. D N AUX V D NP 6. D N AUX V D N Now we apply lexical insertion rules. 1. the N AUX V D N 2. the girl AUX V D N 3. the girl has V D N 4. the girl has buy 1 D N 5. the girl has buy a N 6. the girl has buy a bike 1. S DP AUX VP 2. D NP AUX VP 1 Later application of morphophonemic rules will adjust the form of the verbal head to ensure it is the past participle bought. 2
3. D N AUX VP 4. D N AUX V S 5. D N AUX V DP AUX VP 6. D N AUX V D NP AUX VP 7. D N AUX V D N AUX VP 8. D N AUX V D N AUX V DP 9. D N AUX V D N AUX V D NP 10. D N AUX V D N AUX V D N Now we apply lexical insertion rules. 1. a N AUX V D N AUX V D N 2. a boy AUX V D N AUX V D N 3. a boy V D N AUX V D N 4. a boy believes D N AUX V D N 5. a boy believes the N AUX V D N 6. a boy believes the girl AUX V D N 7. a boy believes the girl is V D N 8. a boy believes the girl is renting D N 9. a boy believes the girl is renting a N 10. a boy believes the girl is renting a car Is the grammar in (10) overgenerating? Offer two examples. The answer is affirmative: we cannot make sure that the grammar doesn t generate *The boy has sleeps a bike or *A girl is wants. To avoid this problem we must include some adjustments in the system. This was the role of subcategorization rules, which adjusted the requirements of verbs:transitive verbs were required to combine with the adequate direct object, for example a DP or a sentence, depending on the case. A different solution is letting latter components do the job. In the Principles and Parameters framework, the Theta Criterion or the Projection Principle would rule out the wrong cases. Exercise 3 Try to formulate in your own words the rule that underlies the phenomenon in (11): (11) a. Mary has bought a car. b. Has Mary bought a car? This is the subject-auxiliary inversion rule: (12) DP AUX VP AUX DP VP if the sentence is interrogative. Could a phrase-structure rule do this job? This could be done by means of a phrase structure rule like (13) S AUX DP VP 3
Then we should add some subcategorization rule to be sure that this pattern only applies to interrogatives. Exercise 4 Can we link the rule in the previous exercise with the phenomena in (14)? (14) a. What has Mary bought? b. I don t know whether Mary bought a car. c. I don t know what Mary bought. Here we must add a transformation moving wh-elements like what or whether to the initial position. And then we would apply subject-auxiliary inversion. 1. Mary has bought what? 2. What Mary has bought? 3. What has Mary bought? In the cases in b. and c., we still have the wh-movement rule at work, but subject-auxiliary inversion is not applying, for in English it is restricted to main clauses (compare with similar asymmetries between main and subordinate clauses in German or Dutch). In most Romance languages, we have both rules in main and subordinate clauses. Could a phrase-structure rule do this job? Phrase structure grammars cannot formulate such long distance relations in a meaningful way. Exercise 5 Formulate in your own words the condition that restricts the application of certain movement transformations in (15) and (16): (15) a. Where did he insisted on going to? b. Where did she forgot about going to? c. *Where did he knew Latin before going to? d. *Where did he knew Latin without going to? (16) a. Qué fuiste a comprar a la farmacia? b. * Qué fuiste a la farmacia para comprar? c. La calle de la que vivo cerca... d. *La calle de la que almuerzo cerca... The crucial difference concerns the nature of the extraction constituent: in the grammatical cases, the wh-element moves from a complement, whereas in the ungrammatical ones, it is moving from an adjunct. Hence, adjuncts are islands for extraction. This was formulated as (17) Condition on extraction domains (CED) [Huang(1982)] A phrase A may be extracted out of a domain B only if B is properly governed. Typically, complements are properly governed by a lexical head, whereas subjects and adjuncts are not. Exercise 6 Formulate in your own words the condition that restricts the application of certain movement transformations in (18) and (19): (18) a. Which problem don t you know how to solve? b. Qué problema no sabes como solucionar? c. Quin problema no saps com resoldre? 4
(19) a. *How don t you know which problem to solve? b. * Cómo no sabes qué problema solucionar? c. * Com no saps quin problema resoldre? The phenomenon exemplified in (19) and (18) is called superiority: when two wh-elements move, the higher one must move higher than the lower one. (20) Superiority Condition [Chomsky(1973)] No rule can involve X, Y in the structure... X... [ α... Z... W Y V... ]... where the rule applies ambiguously to Z and Y and Z is superior to Y. Z is superior to Y if Z asymmetrically c-commands Y. Essentially, this prevents the movement of a wh-word over another one. Exercise 7 Formulate in your own words the condition that restricts the application of certain movement transformations in (21): (21) a. Quién saben todos que está estudiando? Who does everybody know is studying? b. Qué saben todos que está estudiando Pedro? What does everybody know that Peter is studying? c. * Quién saben todos lo que está estudiando? Who does everybody know the things that is studying? d. * Qué saben todos lo que está estudiando Pedro? What does everybody know the things that Peter is studying? Whereas in (21)-a-b the wh-word moves from a declarative clause, it seems that in c-d the free relative is working as an island for extraction. Namely, we have a typical Complex NP Constraint effect: even though there is no antecedent, the free relatives in c-d yield the same effect regarding extraction than relatives. Exercise 8 Consider the alternations in Klamath [Barker(1964)], and the schema of the rules proposed for dealing with them. Try to simplify the system ordering the rules. /honli:na/ /honl i/ /honl a:l a/ /pa:ll a/ /jaljall i/ [holli:na] [holhi] [holpa:l a] [pa:lha] [jaljalpi] 1. nl ll 2. nl lh 3. nl lp 4. ll lh 5. ll lp It is clear that one rule converts any n followed by l into l (1, 2, 3). As for the laterals, aspirated [l ] and glottalized forms [l ] lose their lateral articulation and become [h] and [P], respectively debuccalization. As for the ordering of the rules, the change nl lh clearly shows that lateralization is taking place before [l ] converts to [h]. Hence, 5
/honli:na/ /honl i/ /honl a:l a/ /pa:ll a/ /jaljall i/ lateralization [holli:na] [holl i] [holl a:l a] [pa:ll a] [jaljall i] debuccalization [holli:na] [holhi] [holpa:l a] [pa:lha] [jaljalpi] If we had ordered the rules in the opposite order, we would have obtained a bad result for /honl i/: /honli:na/ /honl i/ /honl a:l a/ /pa:ll a/ /jaljall i/ debuccalization [honli:na] [honhi] [holpa:l a] [pa:lha] [jaljalpi] lateralization [holli:na] *[honhi] [holpa:l a] [pa:lha] [jaljalpi] Exercise 9 When we studied the role of modifying adjectives with respect to nouns (e.g. tall student), we typically resorted to the intersection of the sets that each item denoted: [[tall student]] = [[tall]] [[student]]. Do you think that the same mechanism is valid for adjectives like false, former or fake? Try to explain, in your own words, the semantic import of these adjectives. The problem with these adjectives is that they do not partition the set denoted by the nominal they attach to. Whereas white house is a subset of house, an adjective like false does a completely different thing: a false dollar is not a dollar, hence we are not creating a subset of dollars, but rather placing the element in the complementary set. Take also former: a former president is not an actual president, so (s)he does not belong to the set of presidents; rather we are creating a complementary set of people that were presidents. The case of fake is just as the case of false. In summary, these adjectives are not operating on the extension of sets, but rather on a higher level of complexity, what we called the intension. References [Barker(1964)] Barker, M A R. 1964. Klamath Grammar (Klamath Grammar v. 32-33). University of California Press. https://books.google.es/books?id=c6pvjweacaaj. [Chomsky(1973)] Chomsky, Noam. 1973. Conditions on Transformations. A Festschrift for Morris Halle, edited by Stephen Anderson & Paul Kiparsky, 232-286. [Huang(1982)] Huang, C T James. 1982. Logical Relations in Chinese and the Theory of Grammar: dissertation. [Koopman & Sportiche(1982)] Koopman, Hilda & Dominique Sportiche. 1982. Variables and the Bijection Principle. The Linguistic Review 2(2). 139 160. 6