KACL Lecture Series

#13 : Sigeru Miyagawa (宮川 繁) 氏
Professor, MIT
Title: Some Consequences of the EPP Analysis of Scrambling
Date : January 15, 2002, 17:00--18:30
Place : Meeting Room, the 3rd Floor of Graduate School of Humanities and Social Sciences Building, Kobe University

In Miyagawa (2001), I propose that A-scrambling is triggered by the EPP requirement of T. This analysis extends the notion in Miyagawa (1995, 1996, 1997) that scrambling is not due to an optional movement, and is compatible with the view that UG does not tolerate optional operations. The crucial minimal pair is the following.
  1. Zen'in-ga siken-o uke-nakat-ta (yo/to omotta).
    'All didn't take the test.' *NOT > ALL
  2. Siken-o zen'in-ga uke-nakat-ta (yo/to omotta).
    'The test, all didn't take.' NOT > ALL
In (1) "zen'in" can only be interpreted outside the scope of sentential negation (cf. Kato 1988), but in the object-scrambled example in (2), it may be interpreted inside the scope of negation. On the assumption that for something like 'all' to be interpreted inside the scope of negation, it must be c-commanded by negation (Klima 1964), I propose that in both (1) and (2), something MUST move to TP-Spec. In (1), it is the subject "zen'in," which puts it outside the c-command domain of negation, while in (2), it is the object "siken," which allows the subject "zen'in" to stay in situ in vP-Spec, where it is c-commanded by negation. What is crucial is the obligatory movement of subject/object to TP-Spec. I propose that this movement is triggered by the EPP requirement of T.

In this paper, I will take up some problems that arise with this analysis, some of which I note in Miyagawa (2001). One problem has to do with the "finite/subjunctive" distinction. As noted in Miyagawa (2001), in contrast to (1), it is possible for "zen'in" to be interpreted inside the scope of negation even in the "normal" SOV word order if the entire sentence is embedded under a nominal such as "koto" 'fact'.
  1. zen'in-ga siken-o uke-nakat-ta koto
    'the fact that all didn't take the test' NOT > ALL
As it turns out, this embedding phenomenon vis-a-vis negative scope occurs precisely where, in Classical Japanese, the predicate is in the so-called "rentai" (as opposed to "shuushi) inflection. Although the shuushi/rentai distinction has been lost in modern Japanese (cf. Miyagawa 1989 for references), recently, linguists have adopted the term "subjunctive" to refer to verbs in modern Japanese that occur in the Classical Japanese "rentai" position (Watanabe 1996, Hiraiwa 2001, Uchibori 2001). It is important to note that not all embedded clauses evidence the phenomenon noted above. Embedded clause with the "to" complementizer behaves like a root clause as in (1).
  1. Taroo-ga [zen'in-ga siken-o uke-nakat-ta to] omotta.
    'Taro thought that all didn't take the test' *NOT > ALL
In Classical Japanese, the embedded predicate under the "to" complementizer occurred in the "shuushi" inflection instead of the "rentai" inflection. Thus, the "shuushi/rentai" (finite/subjunctive) distinction is apparently maintained in modern Japanese even though the distinction in inflection has been lost. I will look at the consequences of this for the EPP analysis of scrambling. One possibility which I will entertain is a "V2" analysis of Japanese, building on proposals by Whitman (1991) and Koizumi (1995). I will explore the pros and cons of this alternative to Miyagawa (2001).



[back to KACL-home page]