On Wh-Movement
OWM adopts and builds on the 1973 ‘‘Conditions on Transformations’’ framework, in which Chomsky initiated a research program that would lead to, among other things, a general theory of movement. The transformational rule system of this theoretical framework consists of general movement rules such as ‘‘Move wh-phrase’’ and ‘‘Move NP,’’ which essentially apply blindly. It should be noted that in OWM (p. 85, (43)), the formulation of the wh-movement rule specifies the target of the movement operation: ‘‘Move wh-phrase into COMP.’’
The generative force of the movement rules is constrained by a set of conditions. One such condition is the condition of the (strict) cycle, which governs the order in which transformational processes apply. Another, the Subjacency Condition, states that movement is local, where locality is defined in terms of the number of cyclic nodes (also called bounding nodes) that the moved constituent crosses (a single cyclic node being the maximum). More specifically, as stated in OWM (p. 73, (6)):
(1) A cyclic rule cannot move a phrase from position Y to position X (or conversely) in:
Phenomena of long-distance movement are analyzed as movement dependencies created by the successive-cyclic application of local movement steps (COMP-to-COMP movement in the case of wh-movement). A third type of condition governing (wh-)movement concerns the constituent structure derived by the application of movement. The movement derived structures, involving a relationship between a moved element and its coindexed trace, are taken to be subject to conditions such as the Propositional Island Condition (PIC) and the Specified Subject Condition (SSC) (see OWM, 73, (4)–(5)), which also govern anaphoric relations that are defined by so-called rules of construal. The dependency between the wh-phrase and its trace is interpreted as a quantifier-variable relationship at Logical Form (LF) by a rule of interpretation for wh-phrases. Importantly, as Chomsky points out (OWM, 83), the trace left behind by the moved phrase cannot always be identified as the variable within the scope of the wh-quantifier. This holds specifically in contexts where pied-piping is involved in the wh-movement process (e.g., [Whose book]i did Mary read ti?). In those cases, pied-piped material (here, book) must be placed back—reconstructed—into the extraction site.
With the ‘‘big picture’’ of movement theory presented in ‘‘Conditions’’ as a theoretical background, in OWM Chomsky analyzes a variety of clausal constructions from English, such as comparatives (2a), topicalization (2b), clefts (2c), (finite and infinite) indirect questions (2d), (finite and infinitival) relatives (2e), infinitival complements within the adjective phrase (2f), and easy-to-please constructions (2g).
Until the appearance of OWM, generative analyses of these constructions mostly took the route of linguistic inquiry that goes from studying structured expressions of a language to constructing a language-particular grammar, and then to inquiring into the language faculty. An automatic consequence of this route was that syntactic analyses often did not go be yond the construction-specific and language-particular description of hu man language. Each clausal construction in (2) was derived by its own transformational rule (Comparative Deletion, Topicalization, Clefting, Indirect Question Formation, Relativization, a rule for adjective-qualifier complements, Tough-Movement), each stated in a rather language specific way. The final result of this type of inquiry was not greater in sight into the nature of the language faculty, but a bigger grammatical apparatus (e.g., a large class of possible transformational rules) for describing a particular language.
OWM turned generative linguistic inquiry in a new direction. It showed that a particular conception of UG (specifically, the subtheories that concern the operation Move wh) influences the way one assigns structures to linguistic expressions of a specific language and how it in fluences the form of language-particular grammars. More specifically, Chomsky showed convincingly how the descriptively rich transformational rule system for English (e.g., for the constructions in (2)) could be eliminated, and how a varied set of English constructions like those in (2) could be explained in terms of a general (i.e., universal) movement rule, Move wh, in combination with general principles of language, each ac counting for some general (universal) property of the wh-movement process. This integration of general principles of grammar with compelling, detailed analyses of particular phenomena in English makes OWM an exemplary publication in generative syntax.
Importantly, this reduction operation not only led to dismantling the grammatical apparatus for the description of English, it also limited the class of possible movement rules, as defined by UG (see OWM, 111). Movement, being one of the computational operations defined by UG, is no longer defined in terms of an information-laden structural description and a structural change. The class of movement rules (e.g., Move NP, Move wh) just states that something is moved. Movement ingredients such as ‘‘What is moved?’’, ‘‘Where is it moved to?’’, ‘‘How local is the movement operation?’’, and ‘‘How is the movement operation ordered with respect to other movement rules?’’ are no longer part of the transformational rule itself; instead, they are ‘‘reallocated’’ as general, configurational properties to different components of the grammar (UG).
The importance of OWM also lies in the contention that these general configurational movement properties that have been isolated can serve as a kind of diagnostic tool for identifying the application of wh-movement in some syntactic construction. The key properties that Chomsky con siders to be diagnostic of wh-movement are these (see OWM, 86, (49)):
Property (3a) follows from the assumption that wh-movement moves a phrase, which in turn implies that a gap (i.e., a category voided of phono logical content) is left behind in the extraction site. Property (3b) refers to the fact that wh-movement can circumvent a violation of Subjacency, the PIC, and the SSC in structural contexts involving so-called bridge verbs. An important presupposition here is that wh-movement can proceed successive-cyclically (i.e., stepwise on successive cycles, making use of all the intermediate COMP positions that separate the extraction site from the ultimate landing site). The island properties (3c) and (3d) are implied by Subjacency.
Chomsky formulates the issue about the diagnostic function of the wh properties in (3) as the central research question of OWM (see p. 86, (50)).
The aim of OWM is thus to investigate whether or not the configuration in (3) can serve as a diagnostic for wh-movement on the basis of the con structions in (2). Much of OWM then consists of demonstrating that each of the clausal constructions in (2) has (fully or to a great extent) the properties in (3). To give an example: the following sentences, most drawn from OWM, illustrate that an infinitival indirect question construction like (2d) is based on wh-movement:
(5a) illustrates that movement of who leaves behind a gap in the direct object position after to see. (5b) and (5b0) show that who can be moved from the complement position of visit to the COMP position of the highest embedded clause thanks to the bridge properties of the verbs to promise, to persuade, and to order. (5c) and (5d) show, respectively, that infinitival indirect questions obey the CNPC and the Wh-Island Constraint.
In an infinitival indirect question like (2d), a wh-word (who) is overtly present at the surface, which, of course, is very helpful for identifying wh movement. Identifying wh-movement in constructions lacking an overt wh-phrase is arguably more exciting, since there is no direct (overt) element signaling the presence of a displaced wh-phrase. Take, for example, a comparative construction like (2a). Even though no overt wh-expression is present in the comparative clause, the wh-diagnostics lead us to ana lyze this construction in terms of wh-movement: that is, a wh-phrase is moved to COMP (Spec,CP in present-day terms). Consider in this regard the following examples, where t indicates the original extraction site:
(6a) illustrates the gap property. (6b) shows that an apparently un bounded dependency can be created if the necessary bridge requirements are fulfilled. (6c) exemplifies the CNPC effect, and (6d) shows that comparative formation is subject to the Wh-Island Constraint. The presence of the configuration of wh-properties suggests that a rule of wh-movement is also involved in the derivation of comparatives. According to OWM, this rule moves an (overt) wh-phrase to COMP, where it subsequently gets deleted by a (local) wh-phrase deletion rule (see Chomsky’s deletion rule (47), p. 86).
Thus, OWM made a crucial contribution to the abandonment of a construction-based view of syntactic rules. Furthermore, it introduced the idea of a diagnostic tool in generative syntax—an instrument that can be used to find out whether some computational operation is involved in the derivation of a linguistic expression. Ever since its introduction, this tool has been recognized as an important instrument in syntactic research and it has been employed by numerous linguists in studying a variety of constructions (see, e.g., Van Riemsdijk 1978; Groos and Van Riemsdijk 1981; Cinque 1982, 1990). Besides making these contributions to linguistic description and theorizing, the approach taken in OWM led to a drastic reduction in the number of available rules of grammar, and therefore in the class of available grammars, which Chomsky considered to be ‘‘the major goal of linguistic theory’’ (p. 125). OWM also led to further refinement of the universal conditions that restrict the expressive power of the general transformational rules (Move wh, Move NP).
Regarding the reduction of the available rules, OWM contributed in an important and compelling way to the debate over whether or not syntactic theory should allow unbounded transformations. OWM defends the position that there is no distinction between bounded and unbounded rules. All (cyclic) movement rules are subject to the Subjacency Condition and are therefore bounded. Apparently unbounded movement dependencies are the result of successive-cyclic movement.
