Movement as a Last Resort
Movement is pervasive in natural languages: it is quite typical that most or all of the pronounced elements in a given structure have moved from the positions in which they were first merged. Consider for instance a French interrogative clause like the following, illustrating OVS order:
Under current assumptions, this surface configuration is derived via three kinds of movement operations from the thematic nucleus expressed by the VP (or a more complex structure involving a vP layer, as in Chomsky 1995 and related works).
The lexical verb first moves to the inflectional field to pick up its in flectional affixes of tense, agreement, and so on, and then to the C field, possibly, in this case to pick up-t-, which may be analyzed as an interrogative affix (alternatively, to check the relevant affixes; on the ‘‘building’’ versus ‘‘checking’’ approach to the syntax-morphology interface, see Baker 2002). The subject moves to the specifier position (Spec) of some inflectional head to pick up Case and satisfy the EPP. The wh-object moves (possibly through intermediate steps) to its operator position in the C system. We thus end up with a representation like the following, with the thematic nucleus entirely vacated:
A wealth of empirical evidence has accumulated over the years showing that movement is ubiquitous in natural language syntax. But why does movement exist? The Minimalist Program hypothesizes that movement is a device to satisfy certain interface requirements. Consider the movement of the verb to I and then to C in (3): following a line of analysis that goes back to Syntactic Structures (Chomsky 1957), through Emonds 1978, Pollock 1989, Belletti 1990, and much related work, this kind of movement appears to be motivated by the need to form morpho logically well-formed words, by associating such affixes as-ra and-t-, which cannot survive as independent words, to an appropriate verbal root (but see Koster 2003 for an alternative approach).
As for the movement of the object to the C system, a different requirement seems to be operative, involving the interface with semantic interpretation. Chomsky (2001, 2004) suggests that (certain) natural language expressions may need to be associated to two kinds of semantic properties: argumental properties (a y-role in the case of a direct object) and scope-discourse properties (clausal scope in the case of a wh-element). Linguistic expressions can receive both kinds of interpretive properties by ‘‘picking them up’’ from positions, as it were: they move from positions dedicated to properties of the first kind to positions dedicated to properties of the second kind. So, qui receives the y-role ‘‘patient’’ from the verb in the position in which it is first merged, and then the scope discourse property ‘‘interrogative with main clause scope’’ in its final position. Movement is a device to achieve this duality of interpretations.
This kind of teleological motivation seems to be required of movement: movement is a last resort operation; there is no free or optional movement. I will assume the following informal version of the relevant economy principle (the concept goes back at least to Chomsky 1986; important elaborations, essentially along the lines assumed here, were introduced in Fox 1995 and Reinhart 1997):
The interface level involved may be the syntax-morphology interface, in ternal to the narrow computational system (as in the head movement case), or the external interface with semantics, as in the case of left peripheral A ¯-movement. As for the third kind of chain in (1), the one involving the movement of the subject from its y-position to the Spec of an inflectional head, things are less straightforward. One standard assumption is that this movement is motivated by Case requirements on the subject, which would suggest again the relevance of the interface with morphology. On the other hand, if the subject position is endowed with special discourse properties (quasi-topicality, and the like; Chomsky 2002), and/or if external systems require events to be expressed in subject predicate format (Rothstein 1983), the interface with semantics may be relevant as well to motivate movement of the subject from the VP internal position.
Curiously, if the case of A-chains is less straightforward than that of other chain types in terms of the teleological motivation of movement, it is also the case that has basically inspired the formal implementation of movement in general in minimalism. Without entering into technical details here, I will assume that movement is formally triggered by matching of features: a head X, endowed with feature F, acts as the probe searching for a goal Y in its c-domain, endowed with the same feature. Y(or, more precisely, a phrase containing this element, selected in accordance with the principles responsible for ‘‘pied-piping’’) is then remerged in the immediate structural environment of X: basically, Y adjoins to X if Y is a head, and is merged as X’s specifier if it is a phrase (see Chomsky 2000 and below for more detail and variants of these basic assumptions).
