Morphological Boundaries of Japanese Adjectives: Reply to Namai

1 Introduction

In Nishiyama 1999, I proposed that the two types of Japanese adjectives illustrated in (1) share fundamentally identical phrase structure.

(1)

1. a. Canonical adjectives (CA)

   Yama-ga     taka-ku-mo  ar-u.

   mountain-NOM high-ku-even be-PRES

   'The mountain is even high.'

2. b. Nominal adjectives (NA)

   Yoru-ga  sizuka-de-mo ar-u.

   night-NOM quiet-de-even be-PRES

   'The night is even quiet.'

The major claim that I made in Nishiyama 1999 is that CA-k(u) in (1a) and NA-de in (1b) are parallel and that k(u) and de are allomorphs of Pred, a head essential for predication (Bowers 1993). In a later article, Namai (2002) illuminates issues not discussed thoroughly in Nishiyama 1999. The two major points made by Namai are that (a) k(u) cannot be Pred, and (b) k(u) is not an independent word but is part of a CA. (A third point has to do with modification, which will not be discussed here. See Yamakido 2000 and Bouchard 2002 for relevant discussion.) The main purpose of this squib is to examine (b) closely and argue that despite the new data that Namai presents, my original claim that k(u) is an independent syntactic head can still be maintained. Section 2 discusses claim (a), that k(u) cannot be Pred. The remaining sections concentrate on claim (b), that k(u) is not an independent morpheme but is part of a CA.¹ Section 3 shows that syntactic atoms do not always constitute one syntactic node. On the basis of this result, section 4 reexamines Namai's examples. Section 5 is concerned with adjectives and transitivity.

2 Adverbs and Predication

Namai (2002) argues that since k(u) appears with adverbs (see (2a)), it cannot be Pred. The argument is based solely on the assumption that adverbs do not involve predication (see Namai 2002:343). But it is not clear whether this assumption is valid. For example, Parsons (1990) argues that (manner) adverbs are predicates of events and gives a semantic formula like (2b).

(2)

1. a. Brutus-ga  haya-ku arui-ta.

   Brutus-NOM fast-ku  walk-PAST

   'Brutus walked fast.'

2. b. ∃e [Walking(e) & Agent(e, Brutus) & Fast(e)]

   (adapted from Parsons 1990:45)

(2b) says that there is an event that is characterized as walking, and its Agent is Brutus, and it is fast. Note that the adverb is predicated of the event argument (e). In one of the most comprehensive works on adverbs, Ernst (2002) identifies a class of adverbs that he dubs "predicational adverbs." These ways of treating adverbs are quite consistent with (or actually support) the identification of k(u) as Pred. Moreover, if Radford (1988), whose proposals Namai strongly endorses, and Baker (2003:sec. 4.5) are correct in claiming that adverbs and adjectives are in fact the same syntactic category, and if adjectives are predicates, then adverbs are indeed predicates. Note that even in attributive contexts, adjectives are semantically predicates.

(3)

1. a. [_N red book]'

2. b. λx [Red'(x) & Book'(x)]

(4)

1. a. The book is red.

2. b. Red' (the book)

What (4b) indicates is that Red' is the predicate of the book. In this respect, (4b) is the same as (3b), although the semantic type is different (<e, e> vs. <e, t>). See Nishiyama, to appear, for more discussion on the relations between Pred, adjectives/adverbs, and pre/postpositions. The following sections discuss the syntactic status of k(u).

3 Syntactic Atoms and Phrase Structure

To argue that CA-k(u) is one lexical word, Namai (2002) cites the following examples of coordination:

(5)

1. a. *[taka sosite utukusi]-k(u)        (CA)

   high and  beautiful-k(u)

2. b. [sizuka sosite haruka]-de         (NA)

   quiet  and  far-de

   (Namai 2002:345, slightly modified)

On the basis of this example (and others to be discussed below), Namai argues that k(u) does not occupy an independent syntactic node the way de does. In what follows, I show that (5) does not necessarily refute the alleged parallelism of k(u) and de in (1), once we consider the bound...