複合形容詞と名詞編入 - Geert Booij's Page
[manuscript version, published in Word Structure 7 (2014), 55-87]
Complex negative adjectives in Japanese: The relation between syntactic and morphological constructions[i]
Hideki Kishimoto and Geert Booij
Abstract
Japanese has a fairly large set of complex adjectives formed by combining a noun with the adjective nai ‘null, empty’. The complex negative adjectives have the remarkable property that they allow nominative case marking to appear inside them optionally. We argue that these complex negative adjectives can be classified into three classes, and that the differences in their syntactic behaviour can be accounted for by positing three distinct morpho-syntactic configurations (ranging from a fully phrasal structure to a complex word involving morphological compounding). Some, though not all, complex negative adjectives with case marking show a behaviour which suggests that they constitute single lexical units, while at the same time their components are susceptible to syntactic operations that normally do not apply to the internal structure of words. The data on the Japanese complex adjectives illustrate that adjectives formed via quasi-noun incorporation do not constitute words in a strict morphological sense, in that the entire complexes behave as lexical units, while their components remain visible syntactically.
1 Introduction
In many languages, nouns are combined with predicates to give complex predicates—the grammatical process often referred to as noun incorporation. More often than not, the resulting predicates are verbs, but they can also be adjectives. In Japanese, both types are possible. Complex verbs can be created by combining nouns (more specifically, verbal nouns) with the verb suru ‘do’ (Kageyama 1982, 1999). In a similar vein, complex adjectives can be formed by combining nouns with the adjective nai ‘null’. The complex adjectives are of particular interest, because they show varying degrees of cohesion. We will argue that the ambiguous properties of some complex negative adjectives that seem to blur the boundary between syntax and morphology can be optimally characterized by positing a quasi-incorporation structure in which two head elements are compounded in such a way that they remain visible syntactically.
Japanese has several types of complex adjectival expressions where a noun is combined with an adjective (Namiki 1988, Yumoto 1990, 2008). Among them, there is a fairly large inventory of complex adjectives consisting of a noun plus the adjective nai ‘null, empty’ (Kudo 2000, Fukushima 1998). The following examples illustrate the morphological form of the complex adjectives.[ii]
(1) (a) otonage(-ga)-na-i (b) darasi(-ga)-na-i
adult(-nom)-null-pres tidiness(-nom)-null-pres
‘childish’ ‘untidy’
The complex adjectives in (1) show a peculiar morphological property, in the sense that nominative case marking can be placed optionally inside the complex form (without affecting the core meaning of the whole).[iii] This means that in combining a noun with the negative adjective nai, two structural options are available—a morphological compound structure without case marking, and a structure with case marking. They do not differ in their semantics, but they do differ in their degree of cohesion and syntactic flexibility. The forms without case marking represent single-word NA adjectives. On the other hand, when nominative case marking appears inside the complex form, the noun part looks like an argument syntactically separate from the adjectival head nai, but this is not necessarily the case, as we will discuss below.
Complex nai-adjectives with case-marked nouns can be classified into three classes (Classes I, II, and III), depending on the degree of cohesion in their constituency. We propose that complex negative adjectives with case-marked nouns occur in three distinct configurations. Specifically, Class I adjectives are shown to have phrasal status in the syntax, their components being transparent to the syntax. Class II adjectives represent a case of quasi-noun incorporation, where the sequence involves syntactic compounding (Booij 2010: Chapter 4). Class III adjectives are a case of morphological compounding with a frozen case marker (comparable to the compound-internal linking elements found in various Indo-European languages). It is seen that Class II adjectives show partial sensitivity to syntactic operations due to their being created via quasi-noun incorporation. The facts of the complex adjectives provide insight into the nature of the syntax-morphology demarcation, in particular, into the issue of lexical integrity (or lexical atomicity) (Lapointe 1980, Di Sciullo and Williams 1987, Bresnan and Mchombo 1995, Booij 2009).
The subsequent discussion proceeds as follows. In Section 2, after introducing a tripartite classification of complex adjectives consisting of a noun plus the adjective nai (the Classes I, II and III), some notable features of the complex adjectives are discussed. In Section 3, we illustrate how the complex negative adjectives show distinct behaviour depending on the class they belong to, and we show how this differential behaviour can be derived from the different structures we assign to the complex negative adjectives. Section 4 provides a summary of our findings.
2 Three classes of complex negative adjectives
In Construction Morphology (Booij 2010), and, more generally, in Construction Grammar, complex predicates are analyzed as constructions, that is, pairings of form and meaning. In this framework, complex phrasal units are seen as syntactic constructs, instantiations of phrasal constructions. On the other hand, single-word entities are regarded as morphological constructs. These constructs are stored in the lexicon, and the correspondence between form and meaning is defined by means of constructional schemas. Such schemas define the general properties of both syntactic and morphological constructs. Thus, the notion ‘constructional schema’ as implemented in Construction Grammar is instrumental in analyzing Japanese complex adjectives, because it can do justice to their various degrees of flexibility, which seem to blur the boundary between syntax and morphology. At the same time, this notion allows us to account for the fact that the use of constructions may depend on the presence of specific words, since some constructions are constructional idioms, with lexically specified slots. This is the case for the complex negative adjectives discussed here, as the use of their constructional frames depends on the presence of the negative adjective nai. By appealing to the notion ‘constructional idiom’, we can maintain the boundary between phrasal and morphological entities, but can still handle certain puzzling properties that crosscut syntax and morphology. In the following discussion, we will discuss certain properties of complex negative adjectives, and illustrate how they can be analyzed in the framework of Construction Morphology, with a view to providing a better understanding of the morphology-syntax interface.
The complex nai-adjectives under investigation have surface morphological forms in which the noun part appears with or without case marking. The complex adjectives have varying degrees of cohesion when the nouns appear with case marking. According to the degree of cohesion of the noun+nai sequence, the complex negative adjectives are divided into the following three classes, which can be ordered on a scale of morphological (and corresponding semantic) transparency, with Class I adjectives being the most transparent ones, and Class III adjectives being the least transparent ones.[iv]
(2) (a) Class I:
yurugi(-ga)-na-i [shake-nom-null-pres] ‘unshakable’
atogusare(-ga)-na-i [later.concern-nom-null-pres] ‘without later trouble’
nukari(-ga)-na-i [fault-nom-null-pres] ‘shrewd’
abunage(-ga)-na-i [danger-nom-null-pres] ‘safe, without danger’
sonsyoku(-ga)-na-i [inferiority-nom-null-pres] ‘comparable’
(b) Class II:
sikata(-ga)-na-i [doing.way-nom-null-pres] ‘unavoidable’
moosiwake(-ga)-na-i [excuse-nom-null-pres] ‘inexcusable’
darasi(-ga)-na-i [tidiness-nom-null-pres] ‘untidy’
{tawai/ta’ai}(-ga)-na-i [{solidity/solidity}-nom-null-pres] ‘silly, trivial’
otonage(-ga)-na-i [adult-nom-null-pres] ‘childish’
hugai(-ga)-na-i [viscera.worth-nom-null-pres] ‘shiftless’
toritome(-ga)-na-i [stop-nom-null-pres] ‘rambling’
moosibun(-ga)-na-i [remark-nom-null-pres] ‘perfect’
itaike(-ga)-na-i [gulity-nom-null-pres] ‘young, little’
kitan(-ga)-na-i [restraint-nom-null-pres] ‘candid, frank’
itasikata(-ga)-na-i [doing.way-nom-null-pres] ‘cannot help’
yondokoro(-ga)-na-i [source-nom-null-pres] ‘unavoidable’
(c) Class III:
syoo-ga-na-i [doing.way-nom-null-pres] ‘cannot help’
syoo-mo-na-i [doing.way-also-null-pres] ‘trivial’
doo-siyoo-mo-na-i [how-doing.way-also-null-pres] ‘no way of doing’
tohoo-mo-na-i [means-also-null-pres]‘extraordinary’
totetu-mo-na-i [reason-also-null-pres] ‘incredible’
nanige-na-i [intention-null-pres] ‘casual’
sarige-na-i [so.looking-null-pres] ‘casual’
The complex adjectives are idiomatic, in the sense that the entire sequences are taken to carry idiosyncratic meanings, which are conventionalized. In fact, the complex adjectives where the nouns are case-marked are sometimes referred to as “idiomatic adjectives” (Nishio 1985). The nouns appearing in the complex adjectives have generic interpretations. The component nouns have a tendency to lose transparency in meaning progressively from Class I to Class III, which can be seen by looking at the meanings of negative predicates in each class, listed in (3a-c).
It is important to note that the three classes of complex negative adjectives share the property that the combination of N and nai denotes a name-worthy state of affairs. The complex adjectives at issue carry conventionalized meanings. This suggests that the N + nai sequences are stored in the lexicon, and the following general schema can be posited for these complex adjectives.
(3) [ [N x ] i (-nom) [A na ] ] j ↔ [ WITHOUT SEM i ] j
The meanings of the entire expressions differ depending on the nature of the nouns, but the schema characterizes the core meaning obtained for the complex adjectives. The schema shows that the entire adjective is taken to specify the reverse or negative state of affairs denoted by its component noun. For instance, if the noun abunage ‘danger’ is combined with nai, the entire complex abunage(-ga)-nai means ‘without danger’. Owing to their idiomaticity, the N+nai sequences of complex negative adjectives should be regarded as idioms stored in the lexicon.[v]
As we will discuss at length below, the complex adjectives show a number of behavioural differences depending on the class they belong to. We will show how the distinct behaviour can be accounted for by differences in their structure. With regard to the structures of these complex adjectives, we suggest that the combination of N(-nom) + A can receive three different structural interpretations (see Booij 2010: Chapter 7), as in (4).
(4) (a) [AP [NP N0-nom ] A0 ] [A0 N A ] (Class I)
(b) [A0 N0-nom A0 ] [A0 N A ] (Class II)
(c) [A0 N-nom A ] ([A0 N A ]) (Class III)
Class I adjectives which have their nouns case-marked are APs consisting of an NP plus the head A0, where the noun part has phrasal status. Class II adjectives with case-marked nouns have the structure of syntactic compounds, as a consequence of quasi-noun incorporation, i.e. the N-A sequence is analyzed as having the structure [A0 N0-nom A0], where the case-marked N0 is adjoined to A0. Class III adjectives with case-marked nouns involve morphological compounding, with the structure [A0 N-nom A] , where the compound N-A as a whole is inserted into A0 position. As indicated in (4), the alternative forms without case marking are always available for Class I and II adjectives, which behave as single words, (and there are a few of such adjectives in the Class III, too). When the nominative case marker nom is absent, the nai-adjectives are all NA morphological compounds, where the entire sequence is inserted into the A0 position, as in [A0 N A].
Note that case-marked complex negative adjectives can be assumed to originate from periphrastic structures where the component noun and nai are separate syntactically, even when they may now serve as single lexical entities with no syntactic transparency, perhaps by virtue of lexicalization. This is precisely because (nominative) case marking, which appears with the component noun, is licensed at the level of syntax (Takezawa 1987). Complex words are often analyzed as being formed by lexical rules in the lexicon (i.e. word formation rules), but the case-marked complex adjectives should have phrasal origins.
Importantly, the Class II adjectives formed via quasi-noun incorporation do not constitute words in a strict morphological sense; the entire complexes behave as lexical units, while their components are visible syntactically, as opposed to full-blown noun incorporation, which gives rise to morphological compounds (see Booij 2010: Chapter 4). In Class II adjectives, the case-marked component noun undergoes quasi-incorporation to combine with the adjective nai, but the grammatical process does not give rise to morphological compounds, i.e. the quasi-noun incorporation structure does not count as form a single morphological word. This is because, in quasi-noun incorporation, N0 is adjoined to A0, and this operation creates a syntactic compound structure where both the component noun and nai appear in two distinct head positions.
As mentioned above, the nouns appearing in the complex adjectives have a tendency to lose transparency in meaning progressively from Class I to Class III. The differences in their syntactic behaviour are correlated with their degree of idiomaticity, and thus illustrate the transition process from a phrasal expression to a single lexical item (Brinton and Traugott 2005, Los et al. 2012). We claim that the distinct syntactic configurations proposed above bring about the differences in the negative adjectives’ susceptibility to syntactic operations. In effect, the three classes of complex negative adjectives display a cluster of properties that can be best described in terms of the structural differences (ranging from a structure having a phrasal syntax to a single complex word involving morphological compounding), as depicted in (4).
Note, in this connection, that all variants of Class III adjectives are morphological compounds, i.e. single-word adjectives, which are not susceptible to rules invoking case-marking change (such as nominative-genitive conversion and mo-replacement; see section 2.2), as we will discuss below. Nevertheless, we can readily see from (2c) that the inventory of Class III adjectives consists of three distinct forms that are found in Class I and II adjectives (i.e. N-nom nai, N-mo nai, and N-nai). Since nai-adjectives with nominative case marking, as well as nai-adjectives with the particle mo (derived via mo-replacement), should come from periphrastic forms, the fact suggests that the phrasal constructs of nai-adjectives, regardless of their forms, may develop into morphological constructs (i.e. single-word nai-adjectives), as a consequence of lexicalization (leading to the creation of a single word from a complex expression).[vi]
At this point, it should be mentioned that the negative expression nai can be used as a grammatical word—the functional negator—or an adjective (Kishimoto 2007, 2008). The expression nai can also be used as the negative form of the verb aru ‘be’, which expresses a possessive/existential meaning. Superficially, clauses headed by complex negative adjectives bear resemblance to clauses headed by the negative possessive/existential nai. Nevertheless, the complex negative adjective clauses differ from negated possessive/existential clauses, because the positive counterparts of the adjectival clauses cannot be derived by replacing nai with aru, as shown in (5).
(5) (a) Karera-ga yuuki-ga {na-i/ar-u}.
they-nom courage-nom {neg-pres/have-pres}
‘They {do not have/have} courage.’ (Possessive/Existential)
(b) John-no sinnen-ga yurugi-ga {na-i/*ar-u}.
John-gen confidence shake-nom {null-pres/have-pres}
‘John’s confidence is {unshakable/shakable (?)}.’ (Adjective: Class I)
(c) Mary-no taido-ga otonage-ga {na-i/*ar-u}.
Mary-gen attitude-nom adult-nom {null-pres/have-pres}
‘Mary’s attitude is {childish/adult-like (?)}.’ (Adjective: Class II)
(d) Ano-ziko-ga syoo-ga-{na-i/*ar-u}.
that-accident-nom doing.way-nom-{null-pres/have-pres}
‘That accident is {unavoidable/avoidable (?)}.’ (Adjective: Class III)
Needless to say, the affirmative forms are not available for the clauses headed by complex adjectives without overt case marking on the nouns, either. This is shown in (6).
(6) (a) John-no sinnen-ga {yurugi-na-i/*yurugi-ar-u}.
John-gen confidence {shake-null-pres/shake-have-pres}
‘John’s confidence is {unshakable/shakable (?)}.’ (Adjective: Class I)
(b) Mary-no taido-ga {otonage-na-i/*otonage-ar-u}.
Mary-gen attitude-nom {adult-null-pres/adult-have-pres}
‘Mary’s attitude is {childish/adult-like (?)}.’ (Adjective: Class II)
(c) Sonna taido-ga {sarige-na-i/*sarige-ar-u}.
such attitude-nom {so.looking-null-pres/so.looking-have-pres}
‘Such an attitude is {casual/formal (?)}.’ (Adjective: Class III)
Even though the complex adjective clauses often carry meanings similar to those expressed by the possessive/existential clauses, the two types of clauses behave differently, and hence they need to be distinguished.[vii]
The three classes of complex adjectives with the noun+nai sequence display a number of behavioural differences, if their component nouns are case-marked. The complex adjectives show different degrees of susceptibility to grammatical processes at the syntax level, which include rearrangements of constituents like substitution/replacement and modification via adjunction (with or without movement), as well as generation of syntactic scope. There are both data distinguishing Class I from Class II/III adjectives, and data distinguishing Class I/II adjectives from Class III adjectives. It is argued that the syntactic aspects of behavioural differences observed for the case-marked complex adjectives can be optimally characterized by means of the structures depicted in (4). In the subsections that follow, we will take a look at a number of indications for the structural differences among the three classes of complex adjectives.
2.1 Properties distinguishing between Class I and Class II/III adjectives
Class I adjectives differ from Class II and III adjectives with regard to adverb insertion, the addition of a prenominal modifier, the nominative-dative alternation, and negative polarity item (NPI) licensing. In this section, we will illustrate how the three classes of complex adjectives behave differently.
To begin, one difference that distinguishes Class I adjectives from the two other classes of complex adjectives concerns adverb insertion. Class I adjectives allow an adverb to appear after the component noun marked with nominative case, but this possibility is excluded when no case marking appears on the noun.
(7) (a) Kare-no unten-wa abunage-ga (mattaku) na-i.
he-gen driving-top danger-nom (entirely) null-pres
‘His driving is (entirely) without danger.’
(b) Kare-no unten-wa abunage (*mattaku) na-i.
he-gen driving-top danger (entirely) null-pres
‘His driving is (entirely) without danger.’ (Class I)
In contrast, Class II adjectives do not allow an adverb to intervene between the two elements even if the component noun appears with case marking, as illustrated in (8).
(8) (a) Ken-no taido-wa otonage-ga (*mattaku) na-i.
Ken-gen attitude-top adult-nom (entirely) null-pres
‘Ken’s attitude is (entirely) childish.’
(b) Ken-no taido-wa otonage (*mattaku) na-i.
Ken-gen attitude-top adult (entirely) null-pres
‘Ken’s attitude is (entirely) childish.’ (Class II)
Similarly, for the Class III adjectives syoo-ga-nai ‘unavoidable’ and sarige-nai ‘casual’, an adverb cannot intervene between the component noun and the adjectival base, as illustrated in the examples in (9).
(9) (a) Sono-ziko-wa syoo-ga (*mattaku) na-i.
that-accident-top doing.way-nom (entirely) null-pres
‘That accident is unavoidable (entirely).’
(b) Kanozyo-no taido-wa sarige (*mattaku) na-i.
she-gen attitude-top so.looking (entirely) null-pres
‘Her attitude is casual (entirely).’ (Class III)
The data illustrate that Class I adjectives behave differently from Class II and III adjectives with regard to adverb insertion.[viii]
A second difference between Class I versus Class II/III adjectives concerns prenominal modification, i.e. the possibility of adjoining modifiers to the component nouns.[ix] The addition of a premodifier used for the purpose of emphasis is possible with the case-marked Class I adjective abunage-ga nai ‘without danger’, as in (10).
(10) John-no unten-wa [{isasaka-no/koreppotti-no/nan-no} abunage]-mo nakat-ta.
John-gen driving-top {a.little-gen/this.little-gen/any-gen} danger-also null-past
‘John’s driving was the least dangerous.’ (Class I)
The addition of a premodifier is not possible with the case-marked Class II adjective otonage-ga nai ‘childish’, however, as illustrated in (11).
(11) *Ken-no taido-wa [{isasaka-no/koreppotti-no/nan-no} otonage]-mo nakat-ta.
Ken-gen attitude-top {a.little-gen/this.little-gen/any-gen} adult-also null-past
‘Ken’s attitude was highly childish.’ (Class II)
No prenominal modifier can be adjoined to the nouns of Class III adjectives like syoo-mo-nai ‘uninteresting’, either, as shown in (12).
(12) *Hanasi-ga [{isasaka-no/koreppotti-no/nan-no} syoo]-mo-nakat-ta.
story-nom {a.little-gen/this.little-gen/any-gen} doing.way-also-null-past
‘The story was the least interesting.’ (Class III)
The addition of a prenominal modifier is not allowed if complex adjectives appear without case marking, as shown in (13).
(13) (a) *isasaka-no abunage-na-i
a.little-gen danger-null-pres
‘without the least danger’ (Class I)
(b) *isasaka-no otonage-na-i
a.little-gen adult-null-pres
‘the least childish’ (Class II)
(c) *isasaka-no sarige-na-i
a.little-gen so.looking-null-pres
‘the least casual’ (Class III)
The negative adjectives in (13), which are all morphological compounds, do not tolerate the addition of a prenominal modifier regardless of the class to which they belong.
Incidentally, the most typical premodifier nan-no [any-gen] and koreppotti-no [this.little-gen] that can be added to the complex adjectives have an overlapping property with the NPI adverb sukosimo ‘in the least’, since these premodifiers are construed as negative polarity items. A premodifier like isasaka-no ‘a little’ stands in contrast with them, because it does not count as an NPI, as can be seen from (14).
(14) (a) Kare-ni-tuite-wa [{isasaka-no/koreppotti-no/nan-no} huan]-mo na-i.
he-about-top {a.little-gen/this.little-gen/any-gen} concern-also neg-pres
‘There is no concern about him at all.’
(b) Kare-wa i-i-ga, [{isasaka-no/*koreppotti-no/*nan-no} huan] {-ga/-mo}
he -top good-pres-but {a.little-gen/this.little-gen/any-gen} concern{-nom/-also}
ar-u.
be-pres
‘He is good, but there is (also) a little concern about him.’
The examples in (14) show that isasaka-no can occur in a positive, as well as in a negative context.[x] Even though prenominal modifiers are often counted as NPIs that need to be licensed by a grammatical negator, the fact shows that whether or not a prenominal modifier can be added to the component noun of complex adjectives is determined independently of whether or not it counts as an NPI (see section 3.2).
Third, the subjects of complex negative adjectives, which appear alongside the component nouns in the clauses, show variation with regard to the so-called ‘nominative-dative’ alternation (which is generally available for transitive adjectives; see Section 3.1 for the discussion of this point). The subjects of Class I adjectives may bear either dative or nominative marking, when the component noun of the nai-adjective bears nominative case marking, as shown in (15).
(15) [John-no unten] {-ga/-ni} abunage-ga na-i.
John-gen driving{-nom/-dat} danger-nom null-pres
‘John’s driving is not dangerous.’ (Class I)
Class II adjectives differ from Class I adjectives, in that the nominative-dative alternation is not possible, the nominative case being the only choice for the subject case marking, as shown in (16).
(16) [Ken-no taido]{-ga/*-ni} otonage-ga na-i.
Ken-gen attitude {-nom/-dat} adult-nom null-pres
‘Ken’s attitude is childish.’ (Class II)
In the case of the non-case-marked Class I adjective abunage-nai, a change of nominative case marking to dative case marking on the subject is not permitted, as shown in (17).
(17) [John-no unten] {-ga/*-ni} abunage-na-i.
John-gen driving{-nom/-dat} danger-null-pres
‘John’s driving is not dangerous.’ (Class I)
With the non-case-marked Class II adjective otonage-nai as well, the subject can only have nominative marking, as shown in (18).
(18) [Ken-no taido]{-ga/*-ni} otonage-na-i.
Ken-gen attitude{-nom/-dat} adult-null-pres
‘Ken’s attitude is childish.’ (Class II)
The Class I adjective abunage-nai with no nominative case marking on the component noun patterns with the non-case-marked Class II adjective otonage-nai with respect to the nominative-dative alternation. Moreover, with Class III adjectives, just like Class II adjectives, the nominative-dative alternation on the subject is not possible, regardless of whether the noun carries case marking, as illustrated in (19).
(19) (a) Hanasi{-ga/*-ni} syoo-mo-na-i.
story{-nom/-dat} doing.way-also-null-pres
‘The story is boring.’ (Class III)
(b) Sono-ziko{-ga/*-ni} syoo-ga-na-i.
that-accident{-nom/-dat} doing.way-nom-null-pres
‘That accident is unavoidable.’ (Class III)
(c) Sonna taido{-ga/*-ni} sarige-na-i.
such attitude{-nom/-dat} so.looking-null-pres
‘Such an attitude is casual.’ (Class III)
The noun part of the adjective syoo-ga-nai in (19b) bears nominative case marking, but the nominative marking appearing on its subject cannot be changed to dative marking. Thus the nominative-dative alternation provides a case where Class I adjectives are distinguished from Class II and III adjectives.
An asymmetry in projecting negative scope provides yet another indication of the distinction between Class I versus Class II/III adjectives. First, observe that an adverb like sukosimo ‘at all’ serves as an NPI, which is sanctioned only in a negative context, i.e. it is licensed by falling under the scope of negation.
(20) (a) John-ga sukosimo hon-o {yom-anakat-ta/*yon-da}.
John-nom at.all book-acc {read-neg-past/read-past}
‘John {did not read/read} books at all.’
(b) *Eiga-ga sukosimo tumar-ana-i.
movie-nom at.all uninteresting-pres
‘The movie is uninteresting at all.’
The examples in (20) show that sukosimo is allowed to occur when the predicate is negated by the grammatical negator nai, which serves to reverse the polarity of a proposition, but that its occurrence is not licensed by the suffixal negator nai, which is found in an adjective like tumar-anai ‘uninteresting’.[xi]
The complex adjectives under investigation show a difference with regard to the licensing of an NPI like sukosimo. (21) shows that in Class I adjectives like abunage-ga nai, the adjectival base nai licenses the NPI adverb sukosimo.
(21) Kare-no unten-wa sukosimo abunage-ga nakat-ta.
he-gen driving-top at.all danger-nom null-past
‘His driving was not dangerous at all.’ (Class I)
In (21), the adjective base nai can license an NPI, serving as an operator that projects negative scope, presumably due to the fact that it carries a logical meaning identical to the one expressed by the grammatical negator nai ‘not’. Nevertheless, the same adjective nai does not license the NPI sukosimo when the component noun occurs without case marking.
(22) *Kare-no unten-wa sukosimo abunage-nakat-ta.
he-gen driving-top at.all danger-null-past
‘His driving was not dangerous at all.’ (Class I)
This example illustrates that when the adjective nai forms a complex word with a noun, it loses its ability to project negative scope, and hence does not license an NPI, behaving on a par with the suffixal negator nai (in tumara-nai ‘uninteresting’). In contradistinction, Class II adjectives like otonage(-ga)-nai do not license the NPI adverb sukosimo, regardless of whether or not the noun part is case-marked.
(23) (a) *Ken-no taido-wa sukosimo otonage-ga nakat-ta.
Ken-gen attitude-top at.all adult-nom null-past
‘Ken’s attitude was childish at all.’ (Class II)
(b) *Ken-no taido-wa sukosimo otonage-nakat-ta.
Ken-gen attitude-top at.all adult-null-past
‘Ken’s attitude was childish at all.’ (Class II)
In Class III adjectives as well, the adjectival base nai does not license the NPI sukosimo, regardless of whether the component noun is case-marked or not.
(24) (a) *Sono-ziko-wa sukosimo syoo-ga-na-i.
that-accident-top at.all doing.way-nom-null-pres
‘That accident is unavoidable at all.’
(b) Sono taido-ga sukosimo sarige-na-i.
that attitude-nom at.all so.looking-null-pres
‘That attitude is casual at all.’ (Class III)
The examples in (23) and (24) show that the negative nai associated with Class II and III adjectives does not project negative scope and hence does not license an NPI like sukosimo even when their component nouns are case-marked.
2.2 Differences between Class I/II and Class III adjectives
Thus far, we have seen a number of linguistic phenomena that draw a line between Class I adjectives versus Class II and III adjectives. There are also phenomena where Class I and II complex adjectives pattern together, to the exclusion of Class III adjectives. We will discuss a number of such phenomena below.
A first difference between Class I/II and Class III adjectives has to do with the nominative-genitive conversion applying to the noun parts of these complex adjectives. In Japanese, nominative case marking on a noun can be changed to genitive case optionally when embedded under a larger nominal (i.e. in a relative clause or in a noun-complement clause) (Harada 1971, Watanabe 1996), as exemplified in (25).
(25) (a) [John{-ga/-no} hasit-ta] koto
John{-nom/-gen} run-past fact
‘the fact that John ran’
(b) [John{-ga/-no} yon-da] hon
John{-nom/-gen} read-past book
‘the book which John read’
This alternation is possible with Class I and II adjectives. The examples in (26) show that when the noun part of Class I adjectives appears with case marking, as in abunage-ga nai, the nominative-genitive conversion is allowed.
(26) (a) [John-no unten-ni abunage{-ga/-no} na-i] koto
John-gen driving-dat danger{-nom/-gen} null-pres fact
‘the fact that John’s driving is without danger’
(b) [abunage{-ga/-no} na-i] unten
danger{-nom/-gen} null-pres driving
‘driving without danger’ (Class I)
The Class II adjective otonage-ga nai also allows the genitive-nominative conversion, as illustrated in (27).
(27) (a) [Ken-ga otonage{-ga/-no} na-i] koto
Ken-nom adult{-nom/-gen} null-pres fact
‘the fact that Ken is childish’
(b) [otonage{-ga/-no} na-i] hito
solidity{-gen/-nom} null-pres man
‘a childish man’ (Class II)
In the case of Class III adjectives, the nominative-genitive conversion, which is available for the two other classes of adjectives, is not possible, however.
(28) (a) [sono-ziko-ga syoo{-ga/*-no}-na-i] koto
that-accident-nom doing.way{-nom/-gen}-null-pres fact
‘the fact that that accident is unavoidable’
(b) [syoo{-ga/*-no}-na-i] ziko
doing.way{-nom/-gen}-null-pres accident
‘an unavoidable accident’ (Class III)
As shown in (28), the genitive case marking is not available for the Class III adjective syoo-ga-nai, showing that no case alternation can be implemented on the noun part of this complex adjective, despite the presence of nominative case marking. In this respect, Class III adjectives differ from Class I and II adjectives.
Second, both Class I and II adjectives allow the insertion of an emphatic particle mo ‘also’ to the right of the noun part (and mo replaces the original nominative case marking). This is the same type of replacement that can be implemented on case-marked noun phrases in ordinary clauses, as illustrated in (29).
(29) Ken{-ga/-mo} ki-ta.
Ken{-nom/-also} come-past
‘Ken (also) came.’
(30) shows that the Class I adjective abunage-ga nai allows the nominative marking on the component noun to be replaced with mo.[xii]
(30) Mary-no unten-wa abunage{-ga/-mo} nakat-ta-si….
Mary-gen driving-top danger{-nom/-also} null-past-and
‘Mary’s driving was without danger, and …’ (Class I)
The Class II adjective otonage-ga nai behaves in the same way, since the particle mo can appear with the component noun, as seen in (31).
(31) Mary-no kotoba-wa otonage{-ga/-mo} nakat-ta.
Mary-gen speech-top adult{-nom/-also} null-past
‘Mary’s talk was childish.’ (Class II)
Class I and II adjectives both allow a free alternation between the nominative case marker ga and the particle mo (without changing their core meanings). In contrast, Class III adjectives do not allow an alternation between the nominative case and the adverbial particle mo, as shown in (32).
(32) (a) Sono-ziko-wa doo-siyoo{-ga /*-mo}-na-i.
that-accident-top how-doing.way{-nom/-also}-null-pres
‘There is no way of avoiding that accident.’ (Class III)
(b) Keikaku-wa totetu{-mo/*-ga}-naka-ta.
plan-top reason{-also/-nom}-null-past
‘The plan was incredible.’ (Class III)
The data illustrate that both Class I and II adjectives allow the replacement of nominative case marking with the particle mo (perhaps, for the purpose of emphasis), but this optional replacement is not allowed for Class III adjectives.
Note that the Class III adjective syoo-mo-nai (doing.way-also-null) ‘trivial’ does have the adverbial particle mo inside. However, this does not result from an optional grammatical operation substituting mo for the nominative case found in syoo-ga-nai (doing.way-nom-null) ‘cannot help’. Although, historically, both expressions originate from the same form, as often described in Japanese dictionaries (e.g. Nihon kokugo daiziten [The grand dictionary of the Japanese language]), these two expressions constitute distinct lexical items now, since they carry distinct lexical meanings. If these two expressions were related by the particle replacement operation that applies to the other classes of complex adjectives (i.e. Class I and II adjectives), they would carry the same basic meaning. Thus, it is reasonable to say here that syoo-mo-nai is not construed as an alternant of syoo-ga-nai, and that they are not related by the optional mo-replacement rule.
Another indication that Class II adjectives pattern together with Class I adjectives is found in the case-marking change that takes place under nominalization (Sugioka 1992, Kageyama 1993, Kishimoto 2006). In Japanese, when an adjectival clause is nominalized by attaching the suffix -sa to the adjectival predicate, an obligatory change of nominative/accusative case marking to genitive case marking takes place, and no adverbial particle is admitted, as seen in (33) (Martin 1975, Kishimoto 2005).
(33) (a) Mary{-ga/-mo} kawai-i.
Mary{-nom/-also} cute-pres
‘Mary is (also) cute.’
(b) Mary{-no/*-ga/*-mo} kawai-sa
Mary{-gen/-nom/-also} cute-nmlz
‘Mary’s cuteness’
Sa-nominalization is quite productive, and virtually any adjective can be turned into a derived nominal. As argued by Sugioka (1986) and Kishimoto (2006), this type of nominalization takes place at a phrasal level, rather than at a word-level. Thus, a syntactically complex predicate such as damas-are-yasu-i ‘easy to be deceived’, where passivization applies to the embedded verb, can be nominalized.
(34) (a) Kare-ga damas-are-yasu-i.
he-nom deceive-pass-easy-pres
‘He is easy to be deceived.’
(b) kare{-no/*-ga} damas-are-yasau-sa
he{-gen/-nom} deceive-pass-easy-nmlz
‘his easiness to be deceived’
In (34b), when sa-nominalization applies to the adjective yasu ‘easy’, taking the passive verb as its complement, the nominative case marking must be changed to genitive marking. The embedded verb is syntactically independent of the adjective with the affix -sa, as can be seen from the fact that soo-suru ‘do so’ replacement is possible with its complement; for instance, soo-si-yasu-sa [so-do-easy-nmlz] ‘easy to do so’ can be derived from tabe-yasu-sa [eat-easy-nmlz] ‘easy to eat’ via this replacement. In the light of the fact that no tense appears in the derived nominals, it is reasonable to assume that an adjectival clause is nominalized if the tense is replaced by the suffix –sa, (as in [NP [AP …] –sa ]).
Class I and II adjectives are susceptible to the case-marking change rule triggered by sa-nominalization. Example (35) shows that sa-nominalization brings about the effect that Class I adjectives allow only the genitive marking on their component nouns.
(35) abunage{-no/*-ga/*-mo} na-sa
danger{-gen/-nom/-also} null-nmlz
‘no danger’ (Class I)
Further, (36) shows that Class II adjectives as well invoke an obligatory case-marking change under sa-nominalization.
(36) otonage{-no/*-ga/*-mo} na-sa
adult{-gen/-nom/-also} null-nmlz
‘childishness’ (Class II)
In both Class I and II adjectives, nominative marking on the component noun must be changed to genitive marking under sa-nominalization, and the particle mo is excluded from the nominalized form. In Class III adjectives, by contrast, no genitive marking is allowed to occur, and also, the forms with the particle mo need to be preserved even under sa-nominalization, as shown in (37).
(37) (a) syoo{-ga/*-no}-na-sa
doing.way{-nom/-gen}-null-nmlz
‘unavoidability’
(b) tohoo*(-mo)-na-sa
means(-also)-null-nmlz
‘extraordinariness’ (Class III)
In the complex adjectives belonging to the Classes I and II, the morphological change of nominative to genitive case on the component noun is obligatory. By contrast, in the Class III adjectives, this case-marking change is not implemented even under sa-nominalization, and in fact, if the change is instantiated, ungrammaticality results. Moreover, with the Class III adjective tohoo-mo-nai ‘extraordinary’ comprising mo inside, the particle must be retained even under sa-nominalization, as shown in (37b), which is unacceptable for the other classes of complex adjectives.
The obligatory nature of case-marking change under nominalization can be further confirmed by the examples which contain the subjects of nominalized clauses.
(38) (a) unten{-no/*-ga} abunage{-no/*-ga} na-sa
driving{-gen/-nom} danger{-gen/-nom} null-nmlz
‘the safeness of driving’ (Class I)
(b) hurumai{-no/*-ga} otonage{-no/*-ga} na-sa
behaviour{-gen/-nom} adult{-gen/-nom} null-nmlz
‘the childishness of behaviour’ (Class II)
(c) ziko{-no/*-ga} syoo{-ga/*-no}-na-sa
accident{-gen/-nom} doing.way{-nom/-gen}-null-nmlz
‘the unavoidability of the accident’ (Class III)
For the Class I and II adjectives abunage-ga nai and otonage-ga nai, the concomitant case-marking change needs to be induced on the subjects and the noun parts, as indicated in (38a-b), when the clauses are nominalized with the addition of the suffix –sa to the predicates. This distribution is expected if the rule affecting case marking is implemented obligatorily. Nevertheless, as seen in (38c), while the case change is invoked on the subject of the Class III adjective syoo-ga-nai, the same case change does not take place on its component noun even under sa-nominalization. This shows that the component noun in this class of adjectives lies outside the domain where the obligatory case changing rule is implemented.
Note that syoo-ga nai is categorized as a Class III adjective when it is used in the sense of ‘inescapable, unavoidable’, and not when it describes a human attribute, meaning ‘bad, helpless, worthless’. In the latter case, the adjective falls into the Class II (rather than the Class III). The difference can be readily discerned by the fact that the adjective syoo-ga nai can participate in the nominative-genitive case conversion only when it takes a human subject, as in (39).
(39) (a) [syoo-no na-i] {yatu/*ziko}
doing.way-gen null-pres {fellow/accident}
‘a worthless fellow/unavoidable accident.’
(b) {ano-otoko-no/*ano-ziko-no} syoo-no na-sa
{that-man-gen/that-accident-gen} doing.way-gen null-pres
‘worthlessness of that man/unavoidability of that accident’
In (39), acceptability varies, depending on the choice of the subject, illustrating that the Class III syoo-ga-nai ‘inescapable, unavoidable’, which can be predicated of a non-human subject, constitutes a distinct lexical item from the Class II syoo-ga nai ‘bad, worthless, helpless’, which takes a human subject.[xiii]
2.3 Summary of differences between the three classes of nai-adjectives
Thus far, we have seen that the three classes of complex adjectives show different behaviour with regard to [1] adverb insertion (AdvI), [2] the addition of a prenominal modifier (PreMod), [3] NPI licensing (NPI), [4] the nominative-dative alternation on the subject (NomDatA), [5] the nominative-genitive conversion on the noun part of the adjective (NomGenC), [6] mo-replacement (Mo-R), and [7] obligatory case-marking change from nominative to genitive case (and exclusion of the particle mo) under phrasal sa-nominalization (OCCN). (40) sumarizes the observed patterns for the three classes of complex adjectives.
(40)
AdvI PreMod NPI NomDatA NomGenC Mo-R OCCN
Class I (abunage-ga nai) √ √ √ √ √ √ √
Class II (otonage-ga nai) * * * * √ √ √
Class III (syoo-ga-nai) * * * * * * *
The grammatical operations applicable to the complex adjectives are divided into two groups. The nominative-dative alternation, adverb insertion, NPI licensing, and the addition of a prenominal modifier are applicable to Class I adjectives, and not to Class II adjectives. On the other hand, the nominative-genitive conversion, mo-replacement, and obligatory case-marking change under nominalization apply to both Class I and II adjectives. Class III adjectives are not susceptible to any of these syntactic operations.
This section has illustrated some of the behavioural differences found among the three classes of complex adjectives. In Section 3, we will discuss how we can account for their behavioural patterns.
3 The structural explanation of the behavioural differences of nai-adjectives
As previously discussed, the complex negative adjectives under investigation display distinct behavioural patterns according to which class they belong to. This raises the question why the case-marked complex adjectives behave in the way they do. In this section, we propose that the different behavioural patterns of the complex adjectives with case marking can be accounted for by means of the distinct syntactic configurations proposed in Section 2.1, repeated here as (41).
(41) (a) [AP [NP N0-nom ] A0 ] [A0 N A ] (Class I)
(b) [A0 N0-nom A0 ] [A0 N A ] (Class II)
(c) [A0 N-nom A ] [A0 N A ] (Class III)
Under this proposal, the behavioural differences between Class I and Class II/III adjectives with case marking follow from whether or not the component noun is syntactically independent of the adjectival base. On the other hand, the syntactic behaviour distinguishing Class I/II adjectives from Class III adjectives comes from a difference in the syntactic visibility of the noun components of these complex adjectives. Note at this point that the presence of grammatical marking on nouns, such as case and inflection, is often taken to be an indication that incorporation has not taken place, since incorporated nouns usually lack such marking. Nevertheless, in Japanese, incorporated nouns can retain case marking, as clearly indicated by the fact that Class III adjectives which we interpret as morphological compounds can have overt nominative case marking, as in syoo-ga-nai ‘cannot help’. Given that this nominative marking cannot be affected by any syntactic operations, the nominative case marking appearing with the component nouns of Class III adjectives should therefore be interpreted as a case of frozen marking, which functions as a linking element.[xiv]
We will now show how the divergent properties of the three classes of complex adjectives follow from the proposed structural analysis. In Section 3.1, we will illustrate how Class I adjectives are distinguished from Class II/III adjectives, with regard to adverb insertion, premodifier insertion, the nominative-dative alternation (applying to the subject), and NPI licensing. In Section 3.2, we will discuss what gives rise to the differences between Class I/II and Class III adjectives with regard to the nominative-genitive conversion, mo-replacement, case-marking change under sa-nominalization, and prenominal modification.
3. 1 The differences caused by the surface position of component nouns
In this section, it is argued that the divergent behaviour of Class I adjectives vis-à-vis Class II/III adjectives with regard to adverb insertion, prenominal modification, NPI licensing, and the nominative-dative alternation (applying to the subject) emerges depending on whether the component noun constitutes part of a compound adjective or appears as a syntactically independent argument.
First, the facts of adverb insertion can be accounted for straightforwardly, because, under our analysis, the case-marked component nouns of Class I adjectives have a full phrasal status. In Japanese, an adverb can intervene between an independent argument and a predicate. Class I adjectives, but not Class II and III adjectives, allow an adverb to intervene between the adjectival base and the case-marked component noun (see the examples in (7)). In the Class I adjective abunage-ga nai ‘without danger’, the case-marked component noun abunage-ga is syntactically separate from the adjectival base nai.
(42) [AP [NP abunage0-nom ] AdvP A0 ] (Class I)
With an adjectival phrase having a regular syntax, there is a place where an adverb can be inserted between the adjectival nai and the component noun. By contrast, adverb insertion is not possible with the Class II adjective otonage-ga nai ‘childish’, nor with the Class III adjective syoo-ga-nai ‘cannot help’ (see the examples in (8) and (9)), since the component noun is part of an adjectival (syntactic or morphological) compound even if it is case-marked.
(43) (a) *[A0 otonage0-nom AdvP A0 ] (Class II)
(b) *[A0 syoo-nom AdvP A ] (Class III)
In the Class II adjective otonage-ga nai, the overtly realized component noun otonage-ga is adjoined to A0, and hence it is included in the A0 projection of the adjectival base nai. If an adverbial phrase is inserted between otonage-ga and nai, ungrammaticality results. This is because the A0 structure cannot be maintained if a phrasal element intervenes between the two heads. In a similar vein, no adverb can be inserted between syoo-ga and nai in the Class III syoo-ga-nai, because the complex word as a whole appears in the A0 position. Note further that Class I, II, and III adjectives behave as morphological compounds when they appear without case marking. In this case, no adverbs are allowed to be inserted into the complex adjectives, irrespective of their class (see the examples in (7b), (8b) and (9b)).
(44) (a) *[A0 abunage AdvP na ] (Class I)
(b) *[A0 otonage AdvP na ] (Class II)
(c) *[A0 sarige AdvP na ] (Class III)
The complex adjectives abunage-nai ‘without danger’, otonage-nai ‘childish’, and sarige-nai ‘casual’ all involve morphological compounding, where the entire sequence appears in A0 position, and therefore there is no possibility that an adverb can occur between the noun and nai.
The contrast in grammaticality observed between the case-marked negative adjectives abunage-ga nai (Class I) on the one hand, and otonage-ga nai (Class II) and syoo-ga-nai (Class III) on the other hand, with regard to the addition of a prenominal modifier isasaka-no ‘a little’ can be accounted for along the same line. First, since the component noun abunage projects to a phrase, an adnominal modifier can be added.
(45) [AP [NP NP-gen abunage0-nom ] A0 ] (Class I)
Second, the noun in the Class II adjective otonage-ga nai appears in N0 adjoined to A0, and this means that a full NP cannot be projected from the component noun. Accordingly, the component noun cannot accommodate a prenominal modifier.
(46) (a) *[A0 [NP NP-gen otonage0-nom ] A0] (Class II)
(b) *[A0 [NP NP-gen syoo-nom ] A] (Class III)
Third, in the Class III adjective syoo-ga-nai, which involves morphological compounding, the component noun is inserted into the A0 head, in which nai appears, and hence does not have the potential to project to NP. Thus, syoo-ga-nai does not accept a prenominal modifier. The same holds true for non-case-marked types of negative adjectives (see the examples in (13)), because the complex sequence appears in A0 position, where no phrasal element is admitted.
(47) (a) *[A0 [NP NP-gen abunage]-na ] (Class I)
(b) *[A0 [NP NP-gen otonage]-na ] (Class II)
(c) *[A0 [NP NP-gen sarige]-na ] (Class III)
Needless to say, prenominal modification is possible if the noun does not constitute part of a complex element, as in isasaka-no anzen ‘a little safety’. The facts regarding the addition of prenominal modifiers to the noun parts with overt case marking fall out naturally, given that in Class I adjectives, but not in the two other classes of adjectives, the case-marked component noun is syntactically separate from the adjectival base (with no compounding involved).
A fourth relevant fact concerns the nominative-dative alternation on the subjects of the three classes of complex negative adjectives. As we saw in Section 2.2, the subjects of Class I adjectives participate in the nominative-dative alternation when the component nouns are case-marked, but the subjects of Class II adjectives do not (see the examples in (15) and (16)). We suggest here that whether or not the nominative-dative alternation is allowed is determined by the general requirement in Japanese that at least one nominative argument is necessary in a tensed clause (Shibatani 1978). A key to understanding the issue of the case marking of complex negative adjective clauses lies in the fact that the clause headed by nai ‘null, empty’ can take a ‘nominative-dative’ as well as a ‘nominative-nominative’ case-marking pattern when the predicate is used transitively.
In order to illustrate how the nominative-dative alternation (on the subject) takes place, let us take a brief look at the (nominal) adjectives hituyoo-da ‘necessary’ and siawase-da ‘happy’. First, hituyoo-da ‘necessary’ is a transitive nominal adjective, taking two arguments. The experiencer subject of hituyoo-da in (48) is allowed to carry dative as well as nominative case, as there is another argument marked with nominative case.
(48) Ken{-ga/-ni} okane-ga hituyoo-da.
Ken{-nom/-dat} money-nom necessary-pres
‘Money is necessary for Ken.’
This type of alternation is allowed only when an adjective is transitive. Thus, the intransitive nominal adjective siawase-da ‘happy’ cannot have dative marking on its sole argument, i.e. the subject, as shown in (49).
(49) Ken{-ga/*-ni} siawase-da.
Ken{-nom/-dat} happy-pres
‘Ken is happy.’
Dative case marking is not allowed on the subject of the nominal adjective siawase-da, because the clause does not have any other argument that can be marked with nominative case.
With this case-marking fact in mind, let us proceed to consider the complex adjectives abunage-(ga)-nai ‘without danger’ (Class I), otonage(-ga)-nai ‘childish’ (Class II), and syoo-ga-nai ‘cannot help’ (Class III). As often noted in the literature (Baker 1988, 1996; Mithun 1984), when a noun is incorporated into a transitive verb, the entire complex verb is likely to behave as intransitive, reducing the syntactic valency. This is exactly what happens in the complex negative adjectives, and in fact, the facts follow straightforwardly, given that nouns appearing within A0 projections do not count as arguments.
To begin with, the component nouns of the complex adjectives without case marking do not trigger the nominative-dative alternation regardless of whether they fall into the Class I, the Class II or the Class III (see the examples in (17), (18) and (19)). This fact falls out naturally if the adjectives abunage-nai, otonage-nai and sarige-nai (with no case marking) constitute morphological compounds, i.e. single-word adjectives.
(50) (a) [TP SUBJ-nom/*-dat … [AP [A0 abunage-na ] ]-i ] (Class I)
(b) [TP SUBJ-nom/*-dat …. [AP [A0 otonage-na ]] –i ] (Class II)
(c) [TP SUBJ-nom/*-dat …. [AP [A0 sarige-na ]] –i ] (Class III)
When the noun part is not case-marked, it forms part of the complex nai-adjective, and hence does not constitute an independent argument. Consequently, the subjects of the non-case-marked complex adjectives do not participate in the nominative-dative alternation.
The nominative-dative alternation is possible when the component noun bears nominative case marking in Class I adjectives, but not in Class II adjectives (see the examples in (15) and (16)). This follows, if the nominative noun of a Class I adjective like abunage-ga nai is considered as a phrasal element, which behaves as an independent argument, whereas the nominative noun appearing in a Class II adjective like otonage-ga nai is not.
(51) (a) [TP SUBJ-nom/-dat [AP [NP abunage-nom ] [A0 na ]] -i ] (Class I)
(b) [TP SUBJ-nom/*-dat [AP [A0 [N0 otonage-nom ] [A0 na ]]] -i ] (Class II)
(c) [TP SUBJ-nom/*-dat [AP [A0 syoo-nom na ]] -i ] (Class III)
In the Class I adjective abunage-ga nai, dative case can be assigned to the subject when the component noun occurs with nominative case. This is precisely because this component noun counts as an independent phrasal argument, as illustrated in (51a). In the Class II adjective otonage-ga nai, the component noun does not count as an independent argument, despite its nominative case marking, because the adjective has the structure [A0 N0-nom A0 ], where the noun constitutes part of A0-head via adjunction, as (51b). The Class III adjective syoo-ga-nai, just like the Class II adjective otonage-ga nai, is not susceptible to the nominative-dative alternation (see the example in (19b)), because the expression has the structure [A0 N-nom A ], where the case-marked component noun syoo-ga is inserted into A0, as in (51c).
It is worthwhile to mention here that neither syntactic nor morphological compounding takes place when a case marker is simply dropped (in colloquial speech). This is shown by the fact that in (52), the dative marking on the subject can be retained even if the nominative marking on the other argument is dropped (and in the case of case marker drop, an intonation break is imposed between the noun and nai).
(52) John{-ga/-ni} sonna yuuki(-ga) na-i desyo!
John{-nom/-dat} such courage(-nom) neg-pres emph
‘John does not have such courage!’
The same case alternation pattern is found in the positive counterpart, as (52) illustrates.
(53) John{-ga/-ni} sonna yuuki(-ga) ari masu ka?
John{-nom/-dat} such courage(-nom) have polite q
‘Does John have such courage?’
The experiencer subjects in (52) and (53) can be marked with dative case. This fact shows that the theme NP yuuki ‘courage’ is counted as a full-fledged nominative argument syntactically even when its surface case marking is dropped. Therefore, the mere absence of nominative case marking on the theme NP does not imply the loss of the status of an independent nominative argument.
(54) [TP …. yuuki(-ga) na-i/ar-u]
Given that the nominative-dative alternation is allowed on the subjects in (52) and (53), it should be clear that the subject of a case-marked complex negative adjective can participate in the nominative-dative alternation when its nominative-marked component noun is a phrasal element separate from the A0 head nai syntactically, as in [AP [NP N0-nom ] A0 ] (and whether or not compounding takes place is not determined by the presence or absence of surface case marking per se).
With regard to NPI licensing as well, we observe a contrast in grammaticality between Class I and Class II/III adjectives. The phenomenon regarding NPI licensing is specific to the negative adjectives, where the adjectival base nai has the function of reversing the polarity of a proposition (P ( ¬ P), and is not found in other types of adjectives. Nevertheless, the fact can be accounted for along the same line. This is because the adjectival nai projects negative scope only when it stands as a syntactically independent element.
First, the compound adjectives abunage-nai, otonage-nai, and sarige-nai (with no accompanying case marking) do not project negative scope, and hence do not license NPIs (see the examples in (22), (23b) and (24b)). With these adjectives, nai is combined with their associated component nouns, forming complex heads appearing in the A0 position, as (55) illustrates.
(55) (a) [TP …. *NPI [AP [A0 abunage-na ] ]-i ] (Class I)
(b) [TP …. *NPI [AP [A0 otonage-na ] ]-i ] (Class II)
(c) [TP …. *NPI [AP [A0 sarige-na ] ]-i ] (Class II)
When nai forms a morphological compound with its component noun, negative scope is not projected; hence, NPIs are not licensed.
Second, when the component noun occurs with case marking, the Class I abunage-ga nai can license an NPI, but the Class II and Class III adjectives otonage-ga nai and syoo-ga-nai cannot (see the examples in (21), (23a) and (24a)). These facts follow, given that the case-marked nouns appearing in otonage-ga nai and syoo-ga-nai are compounded either syntactically or morphologically, whereas no compounding takes place with the case-marked noun of the Class I adjective abunage-ga nai, i.e. the component noun is not incorporated to nai, and is syntactically separate from A0 (without involving compounding).
(56) (a) [TP …. √NPI [AP [NP abunage0-nom ] [A0 na ]] -i ] (Class I)
(b) [TP …. *NPI [AP [A0 otonage0-nom na0 ]] -i ] (Class II)
(c) [TP …. *NPI [AP [A0 syoo-nom na ]] -i ] (Class III)
The negative adjective nai, which occurs in the Class I adjective abunage-ga nai, can license an NPI because it occurs in isolation, as illustrated in (56a). By contrast, the Class II otonage-ga nai and the Class III syoo-ga-nai do not license an NPI, because otonage-ga and syoo-ga constitute part of the adjectival nai within the A0 projection, despite their different syntactic status, as depicted in (56b) and (56c). Note at this point that a negative sentence like (57), which involves case marker drop, does not change the possibility of NPI licensing.
(57) Kare-ni-wa sukosimo sonna yuuki(-ga) na-i desyo!
he-dat-top at.all such courage-nom neg-pres emph
‘He does not have such courage at all!’
The fact of NPI licensing in (57) illustrates that the noun whose case marker is dropped via case marker drop (in colloquial speech) stands as an independent phrasal argument even if it appears contiguous with nai. This shows that the adjectival base nai can license an NPI when it occurs in isolation i.e. no head element other than the adjective nai appears inside the A0 structure.
To summarize, the behavioural differences between Class I adjectives, on the one hand, and Class II/III adjectives, on the other, with regard to the nominative-dative alternation, adverb insertion, and NPI licensing, can be accounted for by assuming that the component nouns of the Class II and III complex adjectives are included in the A0 structure of the adjective nai, as syntactic and morphological compounds, while the case-marked component nouns in the Class I negative adjectives count as syntactically independent phrasal arguments.
3.2 Visibility of component nouns
As seen above, the Class II adjective otonage-ga nai should involve compounding of nai with its component noun in a way to similar to the Class III adjective syoo-ga-nai. Nevertheless, these two adjectives show divergent behaviour with regard to the nominative-genitive conversion, mo-replacement, and case-marking change under phrasal sa-nominalization (in particular, see the examples (26)-(28), (30)-(32), and (35)-(37)). The facts regarding the status of case markers and particles follow, given that the Class II adjective otonage-ga nai has the constituent structure in (58a), whereas the Class III adjective syoo-ga-nai has the structure in (58b).
(58) (a) [A0 N0-nom A0 ] (Class II)
(b) [A0 N-nom A ] (Class III)
As noted earlier, the crucial difference that distinguishes between the two classes of adjectives lies in the fact that the case-marked noun in Class II adjectives is adjoined as N0 to A0, as in (58a), while the entire sequence of Class III adjectives including the case-marked noun forms a morphological compound and is inserted into A0 position, as in (58b). We suggest that this structural difference should be held responsible for the divergence in the syntactic behaviour between Class II and Class III adjectives.
The nominative-genitive conversion, mo-replacement, case-marking change under sa-nominalization all involve syntactic processes targeting the nominative case marker. In the Class II adjective otonage-ga nai, as well as in the Class I adjective abunage-ga nai, the nominative case marker on the component noun can be affected (see (26)-(27) and (35)-(36)). This suggests that the component nouns are syntactically visible in both classes of adjectives. Crucially, in the Class II adjective otonage-ga nai, the component noun appears in N0 (adjoined to A0,), as shown in (58a). Since the component noun of otonage-ga nai occupies a distinct head position from that of nai, the two heads are both syntactically visible, and hence accessible to certain syntactic operations. By contrast, in the Class III adjective syoo-ga-nai, the entire N+nai sequence is inserted as a whole into A0 position, as in (58b), and accordingly, the component noun of syoo-ga-nai is not visible to the syntax.
To be more specific, the genitive-nominative conversion is allowed on the case-marked noun appearing in otonage-ga nai, due to the presence of a larger nominal, i.e. when the adjectival clause is embedded in a relative or a noun-complement clause (see the examples in (26)). In phrasal sa-nominalization, the nominalizing suffix -sa, which turns an adjectival clause into a nominal, licenses genitive case marking, while excluding the nominative case and the particle mo (see the examples in (36)). In these cases, the noun in the Class II adjective behaves as an independent argument, on the grounds that it appears in N0 position syntactically distinct from A0. Given that the incorporated noun, which is adjoined to A0, is syntactically visible, the Class II adjective otonage-ga nai is naturally expected to behave on a par with the Class I adjective abunage-ga nai with regard to syntactic operations affecting case marking (see the examples in (27)).
In the Class III adjective syoo-ga-nai, in contrast, the component noun of syoo-ga-nai appears together with the adjective nai in a morphological compound, as in (58b). Consequently, the component noun cannot be affected by the syntactic operations targeting elements outside nai, including the genitive-nominative conversion and mo-replacement (see the examples in (28) and (32)). Notably, the obligatory change of case marking under sa-nominalization is not implemented in syoo-ga-nai, and further, with tohoo-mo-nai ‘extraordinary’, the particle mo must be retained even under nominalization (see the examples in (37)). This shows that the nominative case marker ga and the particle mo associated with Class III adjectives such as syoo-ga-nai and tohoo-mo-nai appear in a position where they are not accessible to syntactic operations.
3.3 Consequences for lexical integrity
The three classes of complex adjectives (Class I, II and III adjectives) possess distinct syntactic properties. Class I adjectives have a constituent structure where the nominative-marked component noun serves as a syntactically independent argument separate from the adjectival head, with no compounding involved. Accordingly, these negative adjectives are accessible to all the syntactic processes discussed—the nominative-dative alternation on the subject, adverb insertion, NPI licensing, and prenominal modification, the nominative-genitive conversion, mo-replacement, and obligatory case-marking change from nominative to genitive case under phrasal sa-nominalization.
Class II adjectives have a constituent structure in which their case-marked component nouns undergo quasi-noun incorporation into the adjective nai, as in [A0 N0-nom A0 ]. This class of adjectives does not allow the nominative-dative alternation on the subject, adverb insertion, NPI licensing or prenominal modification, because the component noun and the adjectival base nai form a syntactic compound. Nevertheless, owing to the configuration where N0 is adjoined to A0, syntactic processes affecting the case marker are applicable to the Class II adjectives, as seen in the nominative-genitive conversion, mo-replacement, and case change under sa-nominalization. In contrast, Class III adjectives include their component nouns inside the adjectival base appearing in A0 position, and none of the syntactic operations are applicable to the Class III adjectives, due to the absence of their qualified targets.
The facts of Japanese complex adjectives have a bearing on the issue of the lexical integrity (or lexical atomicity) principle, which prevents syntactic operations from applying to word-internal elements (Lapointe 1980, Selkirk 1982, Di Sciullo and Williams 1987, Anderson 1992, Bresnan and Mchombo 1995, Booij 2009). Let us consider the following definition of lexical integrity provided by Anderson (1992: 84).
(59) The syntax neither manipulates nor has access to the internal structure of words.
The lexical integrity principle in (59) prevents syntax from seeing parts of words, but this principle can be implemented in a number of different ways, depending on what is taken to constitute ‘the internal structure of words’.
Note that Class II adjectives, where the component noun undergoes quasi-noun incorporation into nai, are accessible to syntactic processes. As discussed above, the partial visibility for syntactic operations comes from the fact that N0 is adjoined to A0 rather than being incorporated morphologically. Thus, we can say that complex lexical items consisting of more than one distinct X0 head form a domain where syntactic operations are applicable and thus, are not constrained by the lexical integrity principle. In Class III adjectives, in contrast, no distinct syntactic heads appear, so that their constituents are located within the domain of the morphological compounds where no syntactic operation is applicable.
Haspelmath (2002: 161-162) proposes to restrict the applicability of lexical integrity to the blocking of syntactic operations rearranging word order and constituency, in the light of the fact that in some languages, incorporated nouns, which cannot undergo rearrangement or movement operations, do inflect for agreement. Instead, we propose that the fact that some (quasi-incorporated) nouns can be visible to syntactic operations should not be taken care of by weakening the principle of lexical integrity, but rather, by assuming a syntactic compound status for the relevant lexical items. If this is correct, we can maintain the strong form of the lexical integrity principle, stating that syntactic operations do not have access to morphological structure.
Essentially, Class II adjectives are not to be interpreted as words in a strictly morphological sense, in that their components are visible syntactically, with the structure where N0 is adjoined to A0. In this case, some, if not all, syntactic operations are applicable. The restricted applicability of syntactic operations to Class II adjectives comes from constraints independent of lexical integrity. On the other hand, no syntactic operations apply to parts of the complex morphological sequences appearing in A0 position in Class III negative adjectives, owing to lexical integrity. Thus, we can maintain the claim that syntactic operations do not apply to the internal structure of morphological words (Spencer 2005, Lieber and Scalise 2007, Booij 2009).
4 Concluding remarks
Japanese complex adjectives in which nouns are combined with the adjectival base nai can be classified into three classes (the Classes I, II and III), according to the degree of cohesion in the noun+nai sequence. Within the framework of Construction Morphology, we have proposed that the three classes of case-marked complex adjectives possess distinct constituent structures, which reflects the degree of idiomaticity that increases progressively from the Class I to the Class III.
(60) (a) [AP [NP N0-nom ] A0 ] [A0 N A ] (Class I)
(b) [A0 N0-nom A0 ] [A0 N A ] (Class II)
(c) [A0 N-nom A ] [A0 N A ] (Class III)
While Class I adjectives have a surface structure where their case-marked component noun appears as an NP outside the adjective nai, the component nouns appearing in Class II and III adjectives are incorporated in the adjectival predicate, even when they appear with case marking. In Class II adjectives, the component nouns, if case-marked, are adjoined to the adjectival base nai, by virtue of quasi-noun incorporation (i.e. N0 is adjoined to A0). In contrast, the component nouns of the Class III adjectives constitute part of an element appearing in a single A0 position by virtue of morphological compounding.
While Class I adjectives can undergo syntactic operations adding phrasal elements to part of adjectival elements, Class II adjectives are only amenable to syntactic operations affecting head elements. The facts of Class II adjectives indicate that if quasi-noun incorporation takes place, the component noun remains visible syntactically despite the fact that it forms a complex lexical item. The incorporated noun is adjoined to A0 rather than inserted into an A0 position as part of a morphological word, and accordingly, can be targeted by certain syntactic operations affecting head elements. The data regarding Class II adjectives illustrate that quasi-noun incorporation (which gives rise to a syntactically complex multi-headed word structure) brings out the effect of blurring the boundary of syntax and morphology. Nevertheless, given that the case-marked nouns in the Class II adjectives lie in the domain to which syntactic operations have access, but those in Class III adjectives do not, it can be concluded that syntactic operations do not ignore the difference between morphology (word structure) and syntax. The facts of Japanese complex negative adjectives show that syntactic operations do not affect word-internal elements, as expressed by the lexical integrity principle.
Japanese shows an interesting parallel with other languages in which a certain type of word sequence can receive various structural interpretations, correlating with various degrees of cohesion and idiomaticity. For instance, Dutch preverbal elements that form a lexical unit with a verb can have a full phrasal status, can be quasi-incorporated, or can have the status of part of a morphological compound (Booij 2010: Chapter 5, Los et al. 2012). Given the conceivable transition process of lexicalization from a full phrase to a single lexical item, it is plausible to state that full syntactic constructs can be reduced to ones having quasi-incorporated status via lexicalization, and further, to morphological constructs, which are the endpoint of the lexicalization of phrasal syntactic constructs.
The proposed analysis highlights and supports an important claim of Construction Grammar and Construction Morphology, the idea that syntactic and morphological constructions may exhibit a high degree of structural similarity, that both syntactic and morphological constructs may have to be listed in the lexicon, and that syntactic constructional schemas can develop into morphological ones. In addition, the notion of ‘constructional idiom’ appeared to be essential for a proper analysis of Japanese complex negative adjectives, as the structures involved are crucially licensed by a specific adjectival head, i.e. the negative adjective nai.
References
Anderson, Stephen. 1992. A-morphous morphology. Cambridge: Cambridge University Press.
Baker, Mark. 1988. Incorporation: A theory of grammatical function changing. Chicago: University of Chicago Press.
Baker, Mark. 1996. The polysynthesis parameter. Oxford: Oxford University Press.
Booij, Geert. 2009. Lexical integrity as a formal universal: a constructionist view. In Sergio Scalise, Elisabetta Magni , & Antonietta Bisetto (eds.) Universals of language today. Berlin: Springer Science+Business Media, 83-100.
Booij, Geert. 2010. Construction morphology. Oxford: Oxford University Press.
Bresnan, Joan, & Sam Mchombo. 1995. The lexical integrity principle: Evidence from Bantu. Natural Language & Linguistic Theory 13: 181-254.
Brinton, Laurie, & Elizabeth Traugott. 2005. Lexicalization and language change. Cambridge: Cambridge University Press.
Di Sciullo, Anna Maria, & Edwin Williams, 1987. On the definition of words. Cambridge, MA: MIT Press.
Fukushima, Kazuhiko. 1998. Negation in the Lexicon, not in NegP. CLS 34-1: 171-184.
Harada, Shin-Ichi. 1971. Ga-no conversion and idiolectal variations in Japanese, Gengo Kenkyuu 60: 25-38.
Harada, Shin-Ichi. 1973 Counter-equi NP deletion. Annual Bulletin, Research Institute of Logopedics and Phoniatrics 7: 113-147. University of Tokyo.
Hathpelmath, Martin. 2002. Understanding morphology. London: Arnold.
Kageyama, Taro. 1982. Word formation in Japanese. Lingua 57: 215-258.
Kageyama, Taro. 1993. Bunpoo-to gokeisei [Grammar and word formation]. Tokyo: Hituzi Syobo.
Kageyama, Taro. 1999. Word formation. In Natsuko Tsujimura (ed.) The handbook of Japanese linguistics. Oxford: Oxford University Press. 297-325.
Kato, Yasuhiko. 1985. Negative sentences in Japanese. Sophia Linguistica XIX.
Kishimoto, Hideki. 2005. Toogokoozoo-to bunpookankei [Syntactic structure and grammatical relations]. Tokyo: Kurosio Publishers.
Kishimoto, Hideki. 2006. Japanese syntactic nominalization and VP-internal syntax. Lingua 116: 771-810.
Kishimoto, Hideki. 2007. Negative scope and head raising in Japanese. Lingua 117: 247-288.
Kishimoto, Hideki. 2008. On the variability of negative scope in Japanese. Journal of Linguistics 44: 379-435.
Kudo, Mayumi. 2000. Hitei-no hyoogen [Negative expressions]. In Yoshio Nitta, & Takashi Masuoka (eds.) Toki, hitei-to toritate [Time, negation, and emphasis]. Tokyo: Iwanami. 94-150.
Kuno, Susumu. 1973. The structure of the Japanese language. Cambridge, Mass.: MIT Press.
Lapointe, Steven. 1980. The theory of grammatical agreement, Ph.D. dissertation. University of Massachusetts, Amherst.
Lieber, Rochelle, & Sergio Scalise (2007). The lexical integrity hypothesis in a new theoretical universe. In Geert Booij, Luca Ducceschi, Bernard Fradin, Emiliano Guevara, Angela Ralli, & Sergio Scalise (eds.), On-line Proceedings of the Fifth Mediterranean Morphology Meeting (MMM5). Università degli Studi di Bologna. 1-24.
Los, Bettelou, Corrien Blom, Geert Booij, Marion Elenbaas, & Ans van Kemenade. 2012. Morphosyntactic change: A comparative study of particles and prefixes. Cambridge: Cambridge University Press.
Martin, Samuel. 1975. A reference grammar of Japanese. New Haven: Yale University Press.
Mithun, Marianne. 1984. The evolution of noun incorporation. Language 60: 847-895.
Namiki, Takayasu. 1988. Fukugoogo-no nichi-ei-taisho: Fukugoo-meisi/fukugoo-keiyooshi [Complex words in contrast between English and Japanese: Compound nouns and compound adjectives]. Nihongogaku 7-5: 68-78.
Nihon Daiziten Kankokai, 2000-2002. Nihon kokugo daiziten [The grand dictionary of the Japanese language], 2nd ed. Tokyo: Shogakkan.
Nishio, Toraya. 1972. Keiyooshi-no imi yohoo-no kijutsuteki kenkyuu [Descriptive studies on the meaning and use of adjectives]. Tokyo: Shuuei Shuppan.
Nishio, Toraya. 1985. Keiyooshi-kanyooku [Adjective idioms]. Nihongogaku 4-1: 45-53.
Shibatani, Masayoshi. 1978. Nihongo-no bunseki [An analysis of Japanese]. Tokyo: Taishukan.
Selkirk, Elizabeth, 1982. The syntax of words. Cambridge, MA: MIT Press.
Spencer, Andrew. 1991. Morphological theory. Oxford: Blackwell.
Spencer, Andrew. 2005. Word-formation and syntax. In Pavol Štekauer & Rochelle Lieber (eds.), Handbook of word-formation. Amsterdam, Springer. 73-97.
Sugioka, Yoko. 1986. Interaction of derivational morphology and syntax in Japanese and English. New York: Garland.
Sugioka, Yoko. 1992. On the role of argument structure in nominalization. Language and Communication 10: 53-80. Keio University.
Takezawa, Koichi. 1987. A configurational approach to case marking in Japanese. Ph.D. dissertation, University of Washington.
Watanabe, Akira. 1996. Nominative-genitive conversion and agreement in Japanese: A cross-linguistic perspective. Journal of East Asian Linguistics 5: 373-410.
Yumoto, Yoko. 1990. Nitieitaishoo fukugoo-keiyooshi-no koozoo: Meishi+keiyooshi/keyoodooshi-no kata-nituite [A contrastive study of the structure of Japanese-English complex adjectives: On the patterns of noun plus adjective/nominal adjective]. Gengobunka Kenkyuu 16: 353-372.
Yumoto, Yoko. 2008. On the modularity of word formation seen from complex adjective formation. In Yoko Yumoto & Hideki Kishimoto (eds.) Goi-no imi-to bunpoo [Lexical meanings and grammar]. Tokyo: Kurosio Publishers. 209-229.
Authors’ addresses: (Hideki Kishimoto)
Department of Linguistics
Graduate School of Humanities
Kobe University
1. Rokkodai-cho, Nada-ku
Kobe, Japan
e-mail: kishimot@lit.kobe-u.ac.jp
(Geert Booij)
Leiden University Centre of Linguistics
P.O. Box 9515
2300 RA Leiden
The Netherlands
e-mail: g.e.booij@hum.leidenuniv.nl
-----------------------
Notes
[i] This is an extensively revised version of the paper presented at the 8th Mediterranean Morphology Meeting held at the University of Cagliari (September, 2011). Part of the material was also presented at the Workshop on “Formation of argument-included complex predicates: Its constraint and mechanisms” at the 142nd Annual Meeting of the Linguistic Society of Japan (June, 2011), and in a seminar at the School of Oriental and African Studies, University of London (September, 2011). We are grateful to Akira Watanabe, Kazuhiko Fukushima, Amir Zeldes, Mark Aronoff, Ryosuke Shibagaki, Peter Sells, Shin-Shook Kim, Yoko Yumoto, Yoko Sugioka, Takane Ito, Saeko Urushibara, Nobuko Hasegawa, Koji Fujita, Hiroshi Aoyagi, Taro Kageyama, Jenny Audring, and the audiences of the meetings, as well as the anonymous reviewers, for comments and suggestions.
[ii] In Japanese, adjectives, nominal adjectives, and verbs belong to inflectional categories, and hence have present and past forms. Morphologically, these forms are created by combining the stems with a tense morpheme, as in na-i [null-pres] and nakat-ta [null-past]. The present forms are also used as citation forms, as in nai [null]. The glosses of the component nouns of complex adjectives are assigned based on the meanings inferred from the meanings of the whole expressions, or, if possible, on the meanings of their source words, which are often no longer in use in contemporary Japanese.
[iii] The presence or absence of case marking in the complex forms more often than not gives rise to semantic differences in the other classes of complex adjectival expressions, i.e. when compound adjectives have non-compound counterparts, the two variants usually do not express the identical meanings. For instance, the compound adjective ha-gayu-i [tooth-itchy-pres] means ‘irritating’, but ha-ga kayu-i [tooth-nom itchy-pres] means ‘the tooth is itchy’, although, in some cases, the meanings of the paired expressions are approximately the same; e.g. yoku-buka-i [desire-deep-pres] ‘greedy’ and yoku-ga huka-i [desire-nom deep-pres] ‘mean’.
[iv] Native speakers tend to easily understand the meaning of the nouns in the Class I adjectives, but have difficulty in understanding the meaning of some nouns in Class II, if not all. In the Class III adjectives, some speakers do not even recognize that they have this morphological constituency until they are told.
[v] Owing to their idiomatic nature, a lot of nouns included in the complex adjectives do not serve as independent words now, and as such, they are not used meaningfully in other contexts; e.g. the noun tawai originally means ‘solidity in ideas or behaviour’, but it is no longer used as a meaningful word in isolation. Even though the meanings of the nouns are no longer transparent, the complex adjectives with N(-nom)+nai are conceived of as instantiating the schema in (3).
[vi] It is generally true that the Class III adjectives do not allow any case alternations that can be implemented in the other classes of nai-adjectives. Nevertheless, the suppression of nominative case marking on the Class III adjective syoo-ga-nai is possible in colloquial speech, as in syaa-nai. This shortened form is not derived from syoo-mo-nai. Note here that both syoo-ga-nai and syoo-mo-nai contain the same noun ‘syoo’, which has originated from si-yoo ‘way of doing’, but that they are not construed as alternants because of their distinct lexical meanings (see section 3). In general, when the Class III adjectives comprise the particle mo, it is not possible to remove mo from them. The Class III adjectives sarige-nai ‘casual’ and nanige-nai ‘casual’ have neither nominative-marked nor mo-marked counterparts owing to their fixedness. In the case of sarige-nai, the compound form can be thought of as originated from a periphrastic (or phrasal) expression consisting of a noun phrase plus nai, because sarige in sarige-nai was originally a phrasal expression, which can be seen by the fact that sarige can be further segmented into sa-ari-ge [so-be-nmlz] ‘being so’.
[vii] There are some other observable differences between the complex-adjective and the possessive/existential clauses, but we will not discuss them in this paper. The positive form of nai is aru ‘have’. We assume here that the negative form nai is derived by dropping the verb aru from *ara-nai [have-neg], since *ara- [have] is not a possible morphological form in Standard Japanese (see Kato 1985).
[viii] In colloquial speech, the case marker on the noun part of the complex adjectives may be dropped, and in such a case, a prosodic break is placed between the noun and the adjective. If (7b) is taken to involve case marker drop in a colloquial register (with an intonation break between the noun and nai), the example might be acceptable even when the adverb intervenes between the noun and the adjectival base. In contrast, (8b) and (9b) are simply rendered unacceptable with the intervening adverb, for no such possibility exists.
[ix] Even with the Class I adjectives, prenominal modification is fairly restricted, and only a limited set of prenominal modifiers can be used, if possible at all. These are minimizing expressions, and often need to co-occur with the particle mo. Note that mo is allowed to attach to the noun in the two types of complex adjectives, without affecting their acceptability.
[x] The particle mo needs to occur with isasaka-no in a negative context, but in a positive context, the presence of mo is not required for this modifier.
[xi] The expression tumar-ana-i ‘uninteresting’ has the morphological sequence of the verb ‘clog’ + nai, but functions as a single-word adjective.
[xii] For one reason or another, when mo is simply added, an adverbial use is often preferred (Nishio 1985).
[xiii] The complex adjective sokke-ga nai [interest-nom null] ‘curt, blunt’ seems to be ambivalent in its classification. When the adjective takes a non-human subject, it describes a value of a contingent action or an attitude, a[pic]#1BDLMUV«¬®¯ÏÐÑÒÓÛÜÝA D
5
B
k
l
q
×
Ù
E
nd falls into the Class I. On the other hand, when the adjective describes an inherent disposition or attribute of a human, it falls into the Class II. With sokke-ga nai, the Class II property seems to obtain in a restricted predicative context, which might be taken to be a sign that with this adjective, a shift from the Class I to the Class II is in progress. Speaker variation might arise here, due to a rather subtle difference in meaning between the two uses.
[xiv] Japanese has one nominal adjective (i.e. wagamma-da ‘selfish’) that comprises frozen case marking in it, which can be seen by the fact that the nominal part of this adjective wagamama is morphologically decomposable as wa-ga-mama [1.sg-nom-will] ‘do as one’s will’. (In Japanese, nominative case has originated as genitive case historically, so that the nominative case here can be seen a remnant of genitive marker). Needless to say, the nominative case marking inside this adjective is not visible to the syntax, so that no nominative-dative alternation is allowed for the subject of the adjective. The complex expression ki-o-tuke-o suru [mind-acc-attach-acc do] ‘pay attention’ provides another case. Here, the complex head ki-o-tuke ‘attention’ includes accusative case, but it can occur with another accusative marker, devoid of the so-called ‘double-accusative’ or ‘double-o’ constraint (Harada 1973, Shibatani 1978). Obviously, this is due to the fact that the accusative case marking inside the complex head is not visible to the syntax.
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related searches
- 1 page business plan pdf
- 2 page business plan sample
- product page examples
- school home page elementary
- 1 page business plan template
- 1 page business case template
- s vs s vs s s grammar
- when to use s or s s grammar
- s or s s grammar girl
- 50 s 60 s 70 s music playlist
- 50 s 60 s 70 s music youtube
- dictionary page by page view