Case and Coordination in Japanese
This talk provides new arguments that Japanese permits remnant VPs to be coordinated. There is a
type of coordination in Japanese, in which the conjuncts appear not to form syntactic constituents. In
addition, the coordinator to, which is affixal, can be duplicated on the second conjunct for a focus
effect and be followed by a case marker. This yields the appearance that the whole coordinate
structure is case-marked, as in (1). I argue specifically against an approach that treats the conjuncts as
nominals and offer an account of the case marker on the second conjunct in terms of a mismatch
between morpho-syntax and morpho-phonology of the sort attested in clitic clusters cross-
linguistically (Marantz 1986, Sproat 1985, Schütze 1994, Embick & Noyer 2001).
The literature offers two approaches to deriving the kind of coordination in (1). One approach
argues that the conjuncts are remnant VPs, created by across-the-board movement of the verb, as
shown in (2) (Koizumi 1995, 2000). The other approach claims that they are nominals, derived from
VPs by particular operations (Fukui & Sakai 2003, Takano 2002), or base-generated as such
(Fukushima 2003), having a resultant representation like (3). Citing the example in (1), Fukui & Sakai
argue that the whole coordinate structure must be a nominal, since only nominals are case-marked in
Japanese. However, the nominal approach makes two incorrect predictions. First, the Japanese
accusative case marker on an object may be omitted if no argument intervenes between the object and
the verb, a fact that underlies the distinction in (4). Under the nominal approach the coordinate
structure is case-marked like an accusative object. The case marker should therefore be obligatory
when the structure is scrambled. However, as illustrated in (5), this is not true. Second, considering
that the coordinate structure is treated like a direct object, it should be able to undergo passivisation.
In particular, the accusative phrase in the first conjunct should remain accusative under passivisation,
as this is not the phrase that is passivised. As shown in (6), the prediction is not borne out.
My proposal is that the case marker on the second conjunct belongs syntactically to the direct
object in the second conjunct, but is phonologically realized externally to the coordinator. The
proposal derives from observations concerning the distribution of the case marker in the second part
of the coordination. The generalization is that the accusative marker can be realized only once on one
of three elements, which are the direct object, to if present and the quantifier if to is absent, as
schematized in (7). The fact that the appearance of the case marker on to is in complementary
distribution with that on the direct object suggests that the latter is an instance of the realization of
accusative case associated with the direct object in the second conjunct. In Japanese, a quantifier and
its host noun may appear either as distinct constituents with the host noun bearing case, as in (8a), or
as one constituent with the case being realized on the quantifier as in (8b). A mismatch in the mapping
between morpho-syntax and morpho-phonology occurs when the direct object has the form in (8a)
and to is realized on the second conjunct, morphologically attaching to the case marker. Specifically,
the coordinator and the case marker surface in the reversed order, as illustrated in (9). The mismatch
occurs because to has a phonological requirement that it attach to a nominal-like element, in this
instance, the quantifier (Koizumi 1995, 2000). That the phonological characteristic of the coordinator
is indeed the relevant factor can be seen from the fact that other particles exhibiting comparable
behavior, such as ka, yara, and toka, all meaning `or', are also affixal. Moreover, the same
observation obtains with nominal coordination.
The proposed analysis is compatible with the remnant-VP approach and has advantages over the
nominal approach. In addition to capturing the distribution of the case marker in the second conjunct,
which is difficult on the nominal approach, the present analysis makes a further correct prediction. It
predicts that an adverbial may intervene between the direct object and the quantifier, only if the
accusative marker appears on the former, indicating that the quantifier is a distinct constituent. The
absence of a case marker on the direct object indicates the structure in (8b). As demonstrated by (10),
this is true. By contrast, under the nominal approach, an adverbial should not be able to appear
internally to the conjuncts at all, as the conjuncts are considered nominals.
In sum, the data such as (1) does not provide support for the treatment of the coordinate structure
as a nominal. A closer examination of the data demonstrates clearly that the apparent occurrence of a
case marker outside the coordination is not an indication of the syntactic category of the conjuncts.
Moreover, the observations regarding case-marker-drop, passivisation and adverbial placement all
show that the `non-constituent' conjuncts are best analyzed as remnant VPs.
(1) Mary-ga [John-ni ringo 2-tu]-to [Bob-ni banana 3-bon]-to-o ageta.
Mary-NOM John-DAT apple 2-CL-and Bob-DAT banana 3-cl-and-ACC gave
`Mary gave [two apples to John] and [three bananas to Bob].' (cf. Fukui & Sakai 2003: 345)
(2) S [VP [VP IO DO Q tv] to [VP [VP IO DO Q tv] V-T
(3) S [NP [NP IO DO Q]-to [NP IO DO Q]] V-T
(4) a. John-ga ringo(-o) tabeta. b. ringo-*(o) John-ga tabeta.
John-nom apple-ACC ate apple-ACC John-NOM ate
`John ate an apple.' `John ate an apple.'
(5) [[ John-ni ringo 2-tu tv]-to [Bob-ni banana 3-bon tv]-to(-o) Mary-ga ageta
John-DAT apple 2-CL-and Bob-DAT banana 3-CL -and-ACC Mary-NOM gave
(6) [ John-ni ringo-ga/*o 2-tu]-to [Bob-ni banana 3-bon]]-(to)-ga/*o
John-DAT apple-NOM/ACC 2-CL-and Bob-DAT banana 3-CL-and-NOM/ACC
Mary-niyotte wata-sare-ta
Mary-by hand-PASS-PAST
`[Two apples to John] and [three bananas to Bob] were handed by Mary.'
(7) a. banana(*-o) 3-bon(*-o) to-o
banana-ACC 3-CL-ACC and-ACC
b. banana-o 3-bon(*-o) to(*-o)
c. banana(*-o) 3-bon-o (*to(-o))
(8) a. NP-case ... Q b. [NP Q]-case
(9) a. Syntax: ... [NPIO [NP.DO NP Q]-o] to
b. Phonology: ... NPIO NP Q-to-o
(10) a. Mary-ga [[John-ni ringo-o kinoo 2-tu]-to
Mary-NOM John-DAT apple-ACC yesterday 2-CL-and
[Bob-ni banana-o kyoo 3-bon](-to)] ageta
Bob-DAT banana-ACC today 3-CL -and gave
b. *Mary-ga [[John-ni ringo-o kinoo 2-tu]-to
Mary-NOM John-DAT apple-ACC yesterday 2-CL-and
[Bob-ni banana kyoo 3-bon-to-o]] ageta
Bob-DAT banana-ACC today 3-CL-and-ACC gave
References:
Embick, D. & R. Noyer, 2001. Movement operations after syntax. Linguistic Inquiry 32:555-595.
Koizumi, M. 2000. String vacuous overt verb movement. Journal of East Asian Linguistics 9: 227-
285.
Fukui, N. & H. Sakai. 2003. The visibility guideline for functional categories: verb raising in Japanese
and related issues. Lingua 113: 321-375.
Fukushima, K. 2003. Verb-raising and numeral classifiers in Japanese: Incompatible bedfellows.
Journal of East Asian Linguistics 12: 313-347.
Marantz, A. 1988. Clitics, morphological merger, and the mapping to phonological structure. In M.
Hammond & M. Noonan (eds.) Theoretical morphology: Approaches in modern linguistics. San
Diego: Academic Press, 253-270.
Schütze, C. T. 1994. Serbo-Croatian second position clitic placement and the phonology-syntax
interface. In MIT Working Papers in Linguistics 21, 373-473, Cambridge MA.
Sproat, R. 1985. On deriving the lexicon. Ph.D. Thesis, MIT, Cambridge MA.
Takano, Yuji. 2002. Surprising constituents. Journal of East Asian Linguistics 11: 243-301.