Nonconstituent Coordination in Japanese: A Case of Phonological Reordering

1 Introduction

The Japanese particle to ‘and’ is generally considered a coordinator of nominals. Thus, it can coordinate two NPs, but not, for instance, two verbs, as demonstrated in (1) and (2). The particle is phonologically weak and cliticizes onto an immediately preceding element.

(1) John-ga [[_np Mary]-to [_np Bill]]-o mita.

  John-nom Mary-and Bill-acc saw

  ‘John saw Mary and Bill.’

(2)

Interestingly, to can also coordinate two conjuncts that do not appear to be syntactic constituents (Koizumi 1995, 2000). In (3), each conjunct consists of an indirect object, a direct object, and a quantifier associated with the latter. I will refer to such conjuncts as nonconstituents. Other kinds of nonconstituents, such as [subject, direct object] and [adverbial, subject, quantifier], can also be coordinated by to.

(3)

Considering that only constituents can usually be coordinated, data such as (3) are puzzling and have generated much discussion about how the coordination should be achieved. There are two main schools of thought. One argues that the conjuncts are remnant VPs, derived by across-the-board movement of the verb (Koizumi 1995, 2000); the other claims that they are complex nominals, created by particular operations (Fukui and Sakai 2003, Takano 2002) or base-generated as such (Fukushima 2003).

As one piece of evidence for their nominal analysis, Fukui and Sakai (2003:345) note a further peculiarity of to: namely, that it may be duplicated on the second conjunct and optionally followed by a case marker, as (4) illustrates.¹ They argue that since only nominals can bear case in Japanese, the whole coordinate structure must be a nominal, providing the bracketing indicated (see also Fukushima 2003). The duplication is also possible in NP-coordination, as (5) shows, but here a case marker following to is obligatory. The additional to seems to place the conjuncts in focus.

(4) Mary-ga [[John-ni ringo 2-tu]-to [Bob-ni

  Mary-nom John-dat apple 2-cl-and Bob-dat

  banana 3-bon]-to]-o ageta.

  banana 3-cl-and-acc gave

(5) John-ga [Mary-to Bill-to]-o mita.

  John-nom Mary-and Bill-and-acc saw

In this squib, I argue that data such as (4) do not in fact support treating nonconstituents as nominals. I will show that the case marker that seemingly appears outside the coordination belongs syntactically to the direct object in the second conjunct. It surfaces in a position after to because of a phonological reordering process that takes place between the two particles (section 3). The remnant-VP approach is compatible with the proposed account, while the nominal approach makes incorrect predictions (sections 4 and 5). First, I will spell out the two approaches in more detail in the following section.

2 Two Approaches to Deriving Nonconstituent Coordination

Koizumi (1995, 2000) argues that the nonconstituents are remnant VPs, derived by across-the-board movement of the verb. The example in (3) therefore has the structure shown in (6). The nominal coordinator to can coordinate the remnant VPs in this instance because it can cliticize onto a nominal-like element, namely, the quantifier.²

(6) S [_VP[_VP IO DO Q t_v] to [_VP IO DO Q t_v]] V-T

Koizumi (2000:231) claims that the verb is not part of the second conjunct, as it is possible to scramble the whole coordinate structure to a position higher than the subject.

(7)

An alternative approach is offered by Takano (2002), Fukui and Sakai (2003), and Fukushima (2003), who claim that the conjuncts are nominals. Fukui and Sakai argue that the conjuncts are VPs in the narrow syntax, but that the verb in the first conjunct is deleted under identity with the verb in the second conjunct and the latter undergoes morphological merger with the tense morpheme in T in the sense of Marantz 1988. The remaining elements are reanalyzed as NPs at PF and subsequently assigned case. Thus, the coordination in (3) has the following representations (slightly modified from Fukui and Sakai 2003:350).

(8)

a. Syntax

b. Phonology

Takano (2002) and Fukushima (2003) argue...