PLINS101 INTRODUCTION TO GENERATIVE GRAMMAR - [7] X theory and (Head) Movement {Read Radford, 2004, pp.78-80, and ch. 5, (you can ignore 5.7 and 5.9); Adger, 2003:115-116} The Phrase-structures developed so far all conform to the constraints of X theory, as shown in (1): 1. XP Specifier X X Complement This has been claimed to be the basis of all structures in all languages. All phrases are endocentric, meaning that they are the projection of a head, X, where X is a variable over categories: i.e. it stands for any of N, V, Adj, P, C, D, etc., and X is an intermediate projection. So every NP has an N as its head; every VP has a V as its head, and so on. Every head has two local relations: to its complement and to the specifier of its projection. The head may be initial (as in English) or final (as in Japanese). One advantage of using X theory is that it simplifies the statement of complement structures. For instance, all of the lexical categories V, N, Adj and P may select a CP complement, as shown in (2): 2a. Terry believes [that mermaids are extinct] b. The fact [that mermaids are extinct] bothers me c. I am certain [that mermaids are extinct] d. Our aquatic hypothesis is unfortunate in [that mermaids are extinct] Structures like (1) are elaborated by selecting a numeration of items from the lexicon, using the operation merge to build up larger and larger constituents, and checking the resultant combinations for well-formedness. Generative grammar has long been (in)famous for allowing a further type of operation, movement, known in earlier versions of the theory as 'transformations'. Consider the following examples: 3a. Blunkett can deny the allegations b. Can Blunkett deny the allegations? 4a. He might convince the queen b. Might he convince the queen? 5a. He has lost his mind b. Has he lost his mind? 6a. The Samaritans are helping b. Are the Samaritans helping? In each of the (b) examples in (3-6) the auxiliary verb {can, might, has, are} has moved from its original position to the front of the sentence (precisely where will become apparent below). This movement takes place in main clauses ('root' sentences) but not in embedded sentences, so we have (7a), parallel to (3a), but we cannot have (7b) parallel to (3b). The example of so-called 'style indirect
2 libre' (free indirect style) in (7c) shows that defining the notion 'root sentence' is not straightforward. 7a. I wonder if Blunkett can deny the allegations b. *I wonder if can Blunkett deny the allegations c. I wonder: can Blunkett deny the allegations? Movement of this kind, previously referred to as 'Subject-Auxiliary inversion' is now described as 'T to C Movement', a kind of HEAD movement. It is not restricted to questions, but occurs in a variety of other constructions in a number of different languages. Examples in English are provided by 'negative inversion' of the kind illustrated in (8a, b) and by the pair of counterfactual conditionals in (8c, d). (A counterfactual conditional is one where the proposition expressed is presupposed to be false - that is, (8c, d) assume that she did not come). Note that negative inversion is also complicated, as is evident from examples (8e, f), which have quite different meanings. 8a. Under no circumstances would I cross a picket line b. *Under no circumstances I would cross a picket line c. If she had come, (I would have been happy) d. Had she come, (I would have been happy) e. With no job John would be happy f. With no job would John be happy g. *If had she come, (I would have been happy) (8g) shows that movement is impossible if the complementiser C (if) is present, suggesting that the auxiliary moves into the complementiser position. That is, the * is explained if there is only one position C, which can be filled either by a Complementiser or by another element (here T) moving into it. The tree in (9) represents the structure of (7a) with no movement. The tree in (10) shows the derivation of (3b), where the Tense element (T), i.e. the auxiliary verb can, has moved from T to C, as shown by the arrows. 9. TP PRN T' I T VP V CP wonder C TP if N T' Blunkett T VP can V DP deny D N the allegations 2
3 10. CP C TP N T' Blunkett T VP can V DP deny D N t the allegations (where t = trace, a kind of empty category) The question arises why the movement described in (10) takes place in English but not in some other languages. The standard answer (which is somewhat arbitrary) is given in terms of the difference between strong and weak features. For each language it is necessary to specify whether particular instances of T and C are strong, and can therefore (like a magnet) attract elements to them, or are weak, and so cannot attract elements to them. The situation for English is given in (11): 11. Interrogative C in English is strong, and therefore it triggers movement so that the COMP node is filled. In contrast, T is weak, so it does not need to be filled. In many languages (e.g. German), main Verbs and not just auxiliary verbs can move, giving rise to V to T to C movement. The COMP node has a number of other features, largely determined by the superordinate verb, (i.e. the verb in the higher clause) in addition to [± strong]. That is, like N and V, COMP is a bundle of features. As shown in (12): it can be [± finite], [± for], [± WH], etc. 12a. I believe that you ate a caterpillar [+finite] (i.e. allows a choice past/present) b. I deplore your eating caterpillars on a Sunday [- finite] c. I want desperately for Joan to elope with me [+ for] d. *I believe implicitly for John to have gone [- for] e. I know that you ate the penguin [- WH] f. I know whether you ate the penguin [+ WH] f. I know which penguin you ate [+ WH] The 'strength' of C is also the basis for an explanation of another unusual property of English: the use of the 'dummy' do. In English, unlike (standard) German for instance, only auxiliary verbs (and not main verbs) move. That is we have the examples in (13a - c) and (13d) is ungrammatical: 13a. They eat caterpillars b. They do eat caterpillars c. Do they eat caterpillars? ['dummy' do] d. *Eat they caterpillars? {Compare German: Essen sie Raupen?} 3
4 If we assume, as last week, that the T node is always present, even if it is left empty, we have the following analysis for (13): 14a. TP 14b. CP PRN T' C TP they T VP do PRN T' V N they T VP eat caterpillars t V N eat caterpillars 15. Dummy do is used as a last resort: i.e. to satisfy some grammatical requirement; specifically here, the need to fill a strong COMP. It is called a 'dummy' as it is supposed to be meaningless. In fact, it has pragmatic (e.g. emphatic) and syntactic effects, but plays no role in determining truth conditions. The trace theory of movement: 16. Any moved constituent leaves behind a 'trace' (t), an empty category with the same features as the item moved. The moved constituent functions as the antecedent of the trace and binds it (under c- command). The antecedent and the trace together form a chain. That is, antecedent-trace relations are similar to antecedent-anaphor relations of the kind seen in Johni admires himselfi. Although traces are, by definition, phonetically empty, they may nonetheless have audible effects. In Handout 5 (example 18c) we saw how an empty category blocked the phonological contraction of have to 've. Exactly comparable effects are seen with the movement we have been illustrating here. (17a - b) show that contraction is permissible, indeed usual, so that we ve in (17b) rhymes with weave. However, it is not possible to make he ve rhyme with heave in (17d) or (18b), because the structure of each of these sentences contains the trace of should or could, which intervenes between the he and the ve as indicated in (17e) and (18c): 17a. We have eaten the caterpillars b. We ve eaten the caterpillars c. He should have eaten the caterpillars d. *Should he ve eaten the caterpillars? {*[hi:v]} e. Should he t ve eaten the caterpillars? 18a. He could have denied the allegations b. *Could he ve denied the allegations? {*[hi:v]} c. Could he t ve denied the allegations? Note that some contraction is possible: the 'have' in "Should he have eaten the caterpillars" is pronounced with a schwa and no aspiration. Crucially, however, the sequence 'he have' must be 4
disyllabic. Things move in Nupe too 5 NVS - 22.11.2005 5