Abstract

The term scrambling was coined in the 1960s by Ross (1967), who originally defined it as a truly stylistic rule applying freely in the grammar. This phenomenon is closely correlated with the fundamental issue of “basic word order” in scrambling languages. Hale (1980) was the earliest attempt to explore the availability of this operation within a larger context of the configurational parameter. He claimed that scrambling is available only in non-configurational languages with flat phrase structures, such as Japanese, because there is no structural difference among grammatical functions (such as subjects and objects). Ross’s intuition has also been variously reflected in subsequent work in the literature on scrambling in languages such as Japanese (Saito 1985; 1989; Kuroda 1988; Fukui 1986): these claim that this phenomenon is a semantically vacuous movement operation in the syntax which is undone at the level of semantic interpretation. This time-honored conception of scrambling fit well with the Move-a format of the so-called Government-Binding Theory within the Generative Framework (Chomsky 1981; 1986) where any syntactic constituent can move anywhere as long as the resulting outputs meet independently motivated constraints. However, this traditional description of scrambling as a truly optional movement operation has become a central issue in the Minimalist Program (MP) (Chomsky 1995). One of the most important precepts in minimalism is that movement occurs only as Last Resort, driven by morphosyntactic factors such as Case or wh-features. This framework, therefore, has effectively eliminated the very concept of optional movement, of which scrambling has hitherto been considered an exemplary case.