Embedded wa-phrases, predication, and judgment theory

Abstract

Topic marking with the particle wa is a prevalent phenomenon in Japanese, and it is generally considered a root phenomenon, as its appearance is much more restricted under embedding. While it has been pointed out that there are some types of embedded clauses that allow wa-phrases, relative clauses present the most restricted linguistic environment for the particle. This exceptional restrictiveness cannot be attributed to a general constraint on wa in embedded contexts. Adopting Kuroda’s (Japanese syntax and semantics, 1992) theory of wa-marking, I argue that the restriction on wa in relative clauses is a result of the clash between two distinct notions of ‘predication’; the general understanding of the term relevant to the predicate abstraction process in a relative clause and Kuroda’s notion of predication, which is structurally manifested in a sentence with wa. Specifically, the lack of wa-marking in relative clauses is due to the illicit abstraction process that crosses over wa-phrases. It will be shown that the analysis has many consequences beyond relative clauses. In particular, it explains the wide scope tendency of topic phrases and the distribution of wa in as diverse sentence types as wh-interrogatives, scrambled sentences, cleft sentences, and clausal comparatives.

Appendix

I would like to use this Appendix to discuss known cases of ‘exceptions’ to the generalization about wa-embedding. At the onset of this paper, I specifically mentioned the exclusion of the two non-thematic uses of wa: the contrastive topic and the negative topic. The distribution of contrastive wa is not easily graspable, and it also shows rather confusing patterns in relative clauses. The only systematic study on the distributional patterns of contrastive wa that I am aware of is Hara (2006), who claims that contrastive wa is not allowed in relative clauses without point-of-view providers. Hara’s generalization makes the situation for the contrastive wa more or less comparable to the non-contrastive use of the particle.

• (71)

  

It is not clear, however, that the ill-formedness of (71) is due to the lack of a point-of-view provider, as Hara claims. It may well be due to the K-Pred Constraint. Since (71) is an instance of object relativization and wa is on the subject, the abstraction process would have violated the K-Pred Constraint if contrastive wa-marking is also subject to the constraint. The following examples indeed suggest that the K-Pred Constraint is a better candidate as the source of the unacceptability, as they are judged significantly better than the example (71).

• (72)

  

No K-Pred violations arise in these examples since the process of subject relativization does not cross over the wa-marked object. The difference between the two types is schematized in (73).

• (73)

  

So far the situation surrounding contrastive topics in relative clauses sends rather confusing messages. On the one hand, it indicates that the K-Pred Constraint is operative on contrastive wa-phrases as well. On the other hand, contrastive wa-phrases do not seem to require the presence of point-of-view providers. Therefore, there is no indication that a contrastive topic involves the Subject—K-Pred partition. To make the matter even more complicated, contrastive topics in ‘parallel listing’ environments are spared from the K-Pred Constraint.

• (74)

  

The second type of exceptional wa-phrases is found in negative sentences, as in (75a) or sentences with negative/adversative predicates (75b) (cf. McGloin 1987).

• (75)

  

The second wa-phrases in these examples are unusual in three respects. Typically, if two arguments bear wa, the second argument is interpreted as contrastive. That is not the case in (75). It could be understood to be contrastive (with appropriate prosody), but such an interpretation is not obligatory. In this sense, it behaves like a usual topic (or a thematic topic). However, the second wa-marked argument can be part of new information, which is atypical of non-contrastive wa-phrases. For instance, (75b) can be an answer to the question, ‘What kind of person is Yuka?’.¹⁹ Lastly but perhaps most importantly for our discussion, a negative wa can appear in a relative clause without any overt evidential marker or other point-of-view provider.

• (76)

  

In these examples, however, the relevance of the K-Pred Constraint is not obvious because they are subject-relatives and the abstraction processes do not cross over the negative wa-phrases. The following examples are some test cases.

• (77)

  

The first example illustrates the base structure, and the wa-marked dative argument is not contrastive. Crucially, (77b) is grammatical despite the fact that the gap is in the object position and the movement of the operator would have violated the K-Pred Constraint. Examples (77b) and (77c) constitute a minimal pair that demonstrates the contrast between a negative sentence and a positive sentence. Therefore, the negative wa constitutes a true exception to the K-Pred Constraint.

To sum up, the two non-thematic uses of wa do not seem to fit into the overall picture of the K-Pred Constraint. The contrastive wa sends a mixed message; sub-classes of it seem to obey the constraint, but there is a fairly clear case (i.e., the ‘parallel contrast’ type) that is spared from it. The negative wa is oblivious to the constraint altogether. If these non-thematic uses of the particle do not lead to the structural realization of Kuroda’s Predication, we predict that the constraint is not operative with non-thematic wa. The prediction is largely correct, but some cases of contrastive wa do show the K-Pred effects, I cannot offer a comprehensive account for this fact and must leave it as an unresolved issue.