- A minimalist approach to scrambling: Evidence from Persian
Simin Karimi is the best-known generative syntactician working on Persian in North America. This book is the culmination of her long-held interest in scrambling, which has also motivated the organization of a conference, the publication of an edited volume, and several single-authored papers on the subject. K's research interests are not limited to scrambling, however; she has also published papers on, among other things, the object marker -ra and on complex predicates—both very popular subjects within the field of Persian linguistics.
As the title indicates, K uses Persian to investigate the nature of scrambling within the minimalist program. As such, the book is less about Persian than it is about generative syntactic theory, and K frequently includes other languages that exhibit scrambling in her discussions, notably German, Japanese, and Hindi. Taking scrambling to be free word order resulting from movement, she poses the following questions on the first page of her book: (i) What motivates the movement of various constituents within a clause? (ii) Is this movement optional? (iii) Does it have an effect on semantic interpretation?
In seeking to answer these questions, K starts with some fairly uncontroversial assumptions about Persian syntax. She proposes that Persian, which has default SOV word order, is underlyingly SOV as well; that sentential complements to verbs (as opposed to DPs and PPs) are base-generated after the verb, where they surface; and that T (as opposed to V and other lexical projections) is head-initial. Although none of these claims are radical departures from what can be found in the literature on Persian, they each present theoretical problems that K addresses in the first chapter.
K goes on to develop her theory of Persian clause structure in Ch. 3, after presenting a survey of past approaches to scrambling in the second chapter of the book. Her proposal here is that all constituents of a sentence remain within vP in the 'unmarked' case. (An unmarked sentence is one that is neutral with respect to information structure.) This view is possible given two recent developments in syntactic theory. The first is the move away from doing morphology in the syntax. Whether one assumes that verbs are inserted into syntax fully inflected (as K does on p. 9) or whether one assumes distributed morphology whereby morphemes are spelled out postsyntactically (as she does on p. 26), verbs are not required to pick up their inflectional morphology via movement.
The second development that plays an important role in K's theory of Persian clause structure is the nature of the vP projection itself. It is in the specifier of vP that subjects are checked for nominative case, whether they have been merged there (agents) or have moved there (themes). It is in the lower specifier of vP that specific direct objects can be checked for accusative case. (Nonspecific direct objects do not move to the lower specifier of vP.) Agreement on the verb can be checked within vP via the operation Agree. The only remaining reason why any constituent would move out of this domain is the EPP, and here K proposes that the EPP should be understood in a new way. [End Page 646]
K distinguishes two kinds of EPP. The first kind is the original extended projection principle of Chomsky 1981, which states that every clause must have a subject. She calls this the EPPg (where g stands for grammatical). Noting that the EPPg usually involves movement of an XP into [Spec, TP] or V-movement to T, K proposes as a third alternative that the EPPg can be satisfied morphologically, that is, that the rich agreement on the verb can satisfy the EPP feature. Given that this same argument has been used to explain the licensing of null subjects in languages that allow them, this opens up a potentially interesting avenue for exploration, namely whether null-subject languages tend to lack...