Quantity Expressions in Japanese

Abstract

After presenting some basic genetic, historical and typological information about Japanese this chapter outlines the quantification patterns it expresses. It illustrates various semantic types of quantifiers, such as generalized existential, generalized universal, proportional, definited and partitive which are defined in the Quantifier Questionnaire in Chapter 1. It partitions the expression of the semantic types into morpho-syntactic classes: Adverbial type quantifiers and Nominal (or Determiner) type quantifiers. For the various semantic and morpho-syntactic types of quantifiers it also distinguishes syntactically simple and syntactically complex quantifiers, as well as issues of distributivity and scope interaction, classifiers and measure expressions, and existential constructions. The chapter describes structural properties of determiners and quantified noun phrases in Japanese, both in terms of internal structure (morphological or syntactic) and distribution.

10.1 Preliminaries

Here we study quantification in Japanese addressing the questions raised in Chapter 1, the Quantifier Questionnaire, of the present volume. Throughout we use the terminology introduced there without explanation. We begin with some basic facts about the Japanese language so that non-native speaker linguists can follow the discussion of quantity expressions effectively.

10.1.1 Word Order, Case-Markers, and Postpositions

Japanese is an SOV language, and nominal expressions that serve as arguments of verbs are normally accompanied by a case-marker. Some sentence patterns that correspond to the English intransitive, transitive, and ditransitive constructions are schematized in (1), (3), and (5), and exemplified by the sentences in (2), (4), and (6), respectively.^{1 , 2}

(1)	Corresponding to the English intransitive construction:
	NP ga Verb

(2)	a.	Kenta ga	neta (yooda).
		Kenta nom slept seem
		‘(It seems that) Kenta fell asleep.’
	b.	Kaori ga	naita (rasii).
		Kaori nom cried seem
		‘(It seems that) Kaori cried.’

(3)	Corresponding to the English transitive construction:
	a.	NP ga NP o Verb
	b.	NP ga NP ni Verb
	c.	NP ga NP to Verb

(4)	a.	Taroo ga	Ziroo o		nagutta (sooda).
		Taro	nom Jiro	acc hit	heard
		‘(I heard that) Taro hit Jiro.’
	b.	Noriko ga	Taroo ni	aisatusita (yooda).
		Noriko nom Taro		dat greeted	seem
		‘(It seems that) Noriko greeted Taro.’
	c.	John ga	Mary to	kekkonsita (rasii).
		John nom Mary with got:married seem
		‘(It seems that) John got married to Mary.’

(5)	Corresponding to the English ditranstive construction:
	NP ga NP ni NP o Verb

(6)	a.	Megumi ga	Kimura sensei	ni	ronbun o	okutta (rasii).
		Megumi nom		Kimura teacher dat paper	acc sent	seem
		‘(It seems that) Megumi sent a paper to Prof. Kimura.’
	b.	Bill ga	John ni	Mary o	syookaisita (mitaida).
		Bill nom John dat Mary acc introduced seem
		‘(It seems that) Bill introduced Mary to John.’

The orders of the arguments given in (3) and (5) are often said to be the base orders (cf. Hoji 1985, 2003b; Hayashishita 2000a). They may be altered to create a new construction or simply to satisfy preferences pertaining to discourse and stylistic factors. Sentences with a marked order are often referred to as derived by scrambling.³

The speaker often utters a sentence to assert that a certain object possesses a certain property (in Kuroda’s 1992 terminology, he /she performs a categorical judgment as opposed to a thetic judgment). In this situation, the object is marked with the so-call topic marker wa; thus, we may not observe the constructions schematized in (1), (3), and (5) above as they are. For example, the first sentence in (7) makes use of the construction in (3a), and the second sentence in (8) the construction in (5).

(7)	Taroo wa	Ziroo o		nagutta. Sikasi, Ziroo wa			Taroo o
	Taro	top Jiro	acc hit	but	Jiro	top Taro	acc
	naguri-kaes-anak-tta.
	hit-return-neg-past
	‘Taro hit Jiro. But Jiro did not hit him back.’

(8)	A: John wa	kinoo	no	paatii de dare ni	deatta no	ka naa.
	John top yesterday gen party at who dat met				comp q emph
	‘At yesterday’s party, who did John meet?’
	B: John ni	wa	Bill ga	Mary o	syookaisiteita yo.
	John dat top Bill nom Mary acc introduced				emph
	‘To John, Bill was introducing Mary.’

It is worth noting that case-markers in Japanese are not necessarily the overt realization of the structural cases (i.e., nominative, accusative, dative, etc.).⁴ Although the marker o is often glossed as an accusative marker in the literature, it is sometimes attached to an item that is considered to be an adjunct; see (9).

(9)	a.	Kareha	ga	kawa o	nagareru.
		dead:leave	nom river		acc flow
		‘Dead leaves flow on rivers.’
	b.	(Based on Takai 2009:50 [100])
		Sono san-nenkan		o	kono ie	de kurasita.
		that three-year:period acc this			house at lived
		‘[We] lived in this house for those three years.’

Similarly, the particle ga, which is usually glossed as a nominative marker, often marks an item that does not have a thematic relation with a predicate (cf. Mikami 1959, 1960; Kuno 1973; Kuroda 1988); see (10).

(10)	(= Kuno 1973:71 [27c], slightly adapted)
	Bunmeikoku	ga	dansei ga	heikinzyumyoo	ga	mizikai.
	civilized:country	nom male		nom average:life:span nom short
	‘It is civilized countries that men – their average life-span is short in.’

Japanese has other particles regarded as postpositions. Phrases consisting of a nominal expression and a postposition may appear at any position before the verb in the clause in which they originate, provided that pragmatic factors are controlled. The sentences in (11), for example, make use of some of the postpositions listed in (12).

(11)	a.	Tookyoo kara	Oosaka made sinkansen			de	san-zika	gurai
		Tokyo	from Osaka	unti	bullet:train with three-hour about
		kakaru.
		require
		‘To go from Tokyo to Osaka by a bullet train requires about three hours.’
	b.	Susan wa	Jennifer to	Taroo no	ie		de keeki o	yaita.
		Susan top Jennifer with Taro			gen house at cake		acc baked
		‘Susan baked a cake with Jennifer at Taro’s house.’

(12)	Partial list of postpositions:
	kara ‘from’, made ‘to’, de ‘at, with’, to ‘with’, e ‘to’

Incidentally, given that case-markers in Japanese do not correspond to English structural cases, and that they behave similarly to postpositions in terms of the word order, one might reasonably argue that in Japanese the distinction between case-markers and postpositions is no more than an artifact.⁵

10.1.2 Nominal Expressions

In Japanese, nominal expressions can be used as a predicate. This is illustrated in (13).

(13)	Taroo ga	kaitagatteiru no	wa	kuruma desu.
	Taro	nom want:to:buy	comp top car	copula
	‘What Taro wants to buy is a car.’

They can also denote an object, and when they do, they can give rise to a wide range of interpretations (cf. Kuroda 1965, 1992; Hoji 1998; Tomioka 2003). For example, the sentence in (14) may be translated as any of the English sentences in (15).

(14)	John ga	kuruma o	aratta	(rasii).
	John nom car		acc washed seem
	‘(It seems that) John washed a car.’

(15)	a.	John washed a car.
	b.	John washed cars.
	c.	John washed the car.
	d.	John washed the cars.
	e.	John washed his (= John’s) car.
	f.	John washed his (= John’s) cars.
	g.	John washed Bill’s car.
	h.	John washed Bill’s cars.

In other words, kuruma ‘car’ may correspond to an indefinite (i.e., (15a)), a bare plural (i.e., (15b)), a singular definite (i.e., (15c)), or a plural definite (i.e., (15d)). In the context where we are talking about one’s own cars, it can be understood to mean John’s car or John’s cars (i.e., (15e) and (15f)). When someone else’s car or cars is/are salient, say Bill’s car or cars, it can be taken to mean Bill’s car or cars (i.e., (15 g) and (15 h)). In addition, as illustrated in (16), nominal expressions can also express the generic meaning.

(16)	a.	Neko wa	nezumi o	oikakeru.
		cat	top mouse	acc chase
		‘Cats chase mice.’
	b.	Nihonzin wa	osusi ga	sukida.
		Japanese top sushi nom like
		‘Japanese people like sushi.’

In what follows, the terms nominal expressions and noun phrases (= NPs) are used interchangeably without any theoretical commitments.

Japanese does not morphologically distinguish between definite and indefinite or between singular and plural.⁶ Given that no lexical item corresponds to an indefinite or definite article, the fact that the mass and count distinction is not morphologically expressed is not surprising. This is illustrated in (17) and (18).

(17)	a.	Mary ga	mizu	o	nonda.
		Mary nom water acc drank
		‘Mary drank water.’
b.	Mary ga	hon	o	yonda.
		Mary nom book acc read
		‘Mary read a book.’

(18)	a.	Mary wa	kaminoke ga	kireida.
		Mary top hair	nom beautiful
		‘Mary’s hair is beautiful.’
	b.	Mary no	suupu ni kaminoke ga	haitteita.
		Mary gen soup	in hair	nom entered
		‘A hair was found in Mary’s soup.’

Adjective phrases (= AdjPs) or verbal phrases (= VPs) may modify an NP with their appropriate inflection to create complex nominal expressions, as schematized in (19). Some examples are given in (20).

(19)	[NP AdjP /VP [NP]]

(20)	a.	[utukusii [zyosei]]
		beautiful woman
		‘a beautiful woman’
	b.	[kinbenna [gakusei]]
		diligent	student
		‘a diligent student’
	c.	[togatta [naihu]]
		sharp	knife
		‘a sharp knife’
	d.	[kinoo	Taroo ga	nakusita [tetyoo]]
		yesterday	Taro	nom lost	notebook
		‘the notebook that Taro lost yesterday’

Other phrases can also modify an NP, but we need a linker no in between, which is often glossed as a genitive marker.⁷ Thus, the resulting phrase has the form in (21).

(21)	[NP X no [NP]]

X in (21) can be an NP (e.g., (22a)), an adverbial phrase (= AdvP) (e.g., (22b)), or even a prepositional phrase (= PP) (e.g., (22c)).

(22)	a.	[amerika no	[kuruma]]
		America gen car
		‘an American car’
	b.	[totuzen no	[yuudati]]
		sudden	gen afternoon:shower
		‘a sudden shower’
	c.	[rosanzerusu e	no	[densya]]
		Los Angeles to gen train
		‘a train to Los Angeles’

Since quantity expressions in Japanese can appear in the position of X, the form in (21) appears repeatedly when we discuss them below.

10.1.3 Classifiers and Measure, Container, and Time Phrases

In Japanese, when nominal expressions are modified with a number, the number is accompanied by a classifier. (Some classifiers may also be used as nouns. For simplicity’s sake, we continue to label them as classifiers.) While there are (at least) two general classifiers that can be used for counting a wide range of objects, the other classifiers are designated to count specific types of objects. Here we provide partial lists of classifiers and examples to illustrate them; see (23)–(26). (See Iida 2004 for a more complete list.)

(23)	Classifiers (for countable objects in general):
	tu, ko

(24)	a.	omame itu-tu
		bean	five-cl
		‘five beans’
	b.	omame san-ko
		bean	three-cl
		‘three beans’

(25)	Classifiers (for specific objects):
	nin (for persons),8 hiki/biki/piki (for animals in general),9 tou (for cows and horses i.e., large animals), wa (for birds), hon/bon/pon (for long objects), mai (for sheets), tubu (for small round objects), satu (for books), ken (for houses), dai (for machines), kumi (for pairs/sets), hako/pako (for boxes), hai/bai/pai (for filled vessels)

(26)	a.	gakusei go-nin
		student	five-cl
		‘five students’
	b.	buta	san-biki
		pig	three-cl
		‘three pigs’
	c.	usi	ni-too
		cow two-cl
		‘two cows’
	d.	niwatori go-wa
		chicken	five-cl
		‘five chickens’
	e.	enpitu yon-hon
		pencil	four-cl
		‘four pencils’
	f.	kami	iti-mai
		paper	one-cl
		‘one sheet of paper’
	g.	omame san-tubu
		bean	three-cl
		‘three beans’
	h.	hon	ni-satu
		book two-cl
		‘two books’
	i.	ie	go-ken
		house five-cl
		‘five houses’
	j.	kuruma san-dai
		car	three-cl
		‘three cars’
	k.	gakusei huta-kumi
		student	two-cl
		‘two pairs/sets of students’
	l.	danbooru	san-hako/pako
		carton:box three-cl
		‘three carton boxes’
	m.	baketu go-hai
		bucket five-cl
		‘five filled buckets’

There are also classifiers that count events; see (27) and (28).

(27)	Classifiers (for events):
	kai (for rounds), do (for rounds)

(28)	a.	san-kai
		three-cl
		‘three times’
	b.	go-do
		five-cl
		‘five times’

Like English, Japanese uses measure or container phrases to ‘count’ mass items. We provide a partial list of measure phrases in (29) and a few illustrations in (30).¹⁰

(29)	rittoru ‘liter’, syoo ‘1.8 liter’, meetoru ‘meter’, kiroguramu ‘kilogram’, en ‘Japanese yen’, doru ‘dollar’

(30)	a.	mizu	ni-rittoru
		water two-liter
		‘two liters of water’
	b.	kin	san-kiroguramu
		gold three-kilogram
		‘three kilograms of gold’

To measure mass items with a container phrase, the container itself needs to be accompanied by an appropriate classifier, e.g., (26 l) and (26m). If we measure beads and water in terms of carton boxes and buckets, respectively, we have (31a) and (31b).

(31)	a.	biizu danbooru	go-hako
		bead carton:box five-cl
		‘five carton boxes of beads’
	b.	mizu baketu ni-hai
		water bucket two-cl
		‘two buckets of water’

The list in (32) includes unit phrases for measuring time. A few illustrations are given in (33).

(32)	byoo ‘second’, hun /pun ‘minute’, ka /nichi ‘day’, syuukan ‘week’ tuki ‘month’ nen ‘year’

(33)	a.	zyuugo-byoo
		15-second
		‘fifteen seconds’
	b.	san-syuukan
		three-week
		‘three weeks’
	c.	mi-tuki
		three-month
		‘three months’

So far we have said nothing about the order among (i) a classifier phrase, (ii) the NP it modifies, and (iii) the rest of the sentence. Besides the word order exemplified by (26), (30), and (31), several other orders are possible, which we will discuss in Sections 10.2, 10.3, 10.4, and 10.8 below.

10.1.4 Demonstratives

Japanese demonstratives consist of ko-, so-, and a-words, as illustrated in (34) and (35).

(34)	Simplex demonstratives:
	Ko-words So-words A-words
	kore ‘this thing’	sore ‘that thing’	are ‘that thing’
	koko ‘this place’	soko ‘that place’	asoko ‘that place’
	kotira ‘this direction’	sotira ‘that direction’	atira ‘that direction’
	koo ‘in this way’	soo ‘in that way’	aa ‘in that way’

(35)	Complex demonstratives:
	Ko-words So-words A-words
	kono NP ‘this NP’	sono NP ‘that NP’	ano NP ‘that NP’
	konna NP ‘this kind of NP’	sonna NP ‘that kind of NP’	anna NP ‘that kind of NP’
	kooyuu NP ‘NP as this’	sooyuu NP ‘NP as that’	aayuu NP ‘NP as that’
	kono yooni ‘in this way’	sono yooni ‘in that way’	ano yooni ‘in that way’
	konna huuni ‘in this way’	sonna huuni ‘in that way’	anna huuni ‘in that way’

In their deictic uses, ko-words are used to refer to an object that is within the speaker’s domain while so-words are used for an object within the hearer’s domain. To refer to objects that are neither in the speaker’s domain nor in the hearer’s domain, a-words must be used.¹¹ This is illustrated in (36)–(38). (We use ♯ to indicate those sentences or phrases which are infelicitous in the specified contexts—they may be used felicitously in some other contexts.)

(36)	[Context: the speaker talks to the hearer about the book that the speaker is holding in his hand.]
	{Kore / ♯Sore / ♯Are} wa			totemo omosiroi	no	desu.
	this	that	that	top very	interesting comp copula
	‘{This / That/ That} is very interesting.’

(37)	[Context: the speaker talks to the hearer about the book that the hearer is holding in his hand.]
	{♯Kono / Sono/ ♯Ano}			hon	wa	omosirokatta	desyoo.
	this	that	that	book top was:interesting isn’t:it
	‘{This/ That/ That} book was interesting, wasn’t it?’

(38)	[Context: the speaker talks to the hearer about the book that someone else is holding in his hand.]
	{♯Konna / ♯Sonna / Anna}		hon	ga	kaketa	ra ii	desu
	this:kind	that:kind that:kind book nom could:write if good copula
	ne.
	emph
	‘It would be good if [we] could write {this kind/ that kind/ that kind} of book, wouldn’t it?’

Turning to their non-deictic uses, a- and ko-words are used to refer to an object which the speaker came to know through his or her personal experience; cf., Kuroda (1979), Takubo and Kinsui (1996, 1997), Ueyama (1998), Hoji et al. (2003). So-words, on the other hand, refer to a linguistic antecedent. Thus, the contrasts in (39) and (40) are expected.

(39)	(= Hoji et al. 2003 [12], based on Ueyama 1998:ch.4 [10] and [20], slightly adapted)
	[Context: The detective is looking for a man. He believes that the man should be hiding in a certain room. He breaks into the room and asks the people there.]
	{♯Soitu / Aitu}	wa	doko	da?
	that:guy / that:guy top where copula
	‘Where is that guy?’

(40)	(= Hoji et al. 2003 [13], based on Ueyama 1998:ch.4 [16] and [23], slightly adapted)
	[Context: A wife told her husband on the phone that someone had called him. He has no idea who the person is. He asks her.]
	{Soitu /	♯Aitu}	wa	nante itteta?
	that:guy / that:guy		top what	said
	‘What did that guy say?’

The difference between ko- and a-words is related to the speaker’s perceived psychological distance from the referent. We, for example, observe the following contrasts.¹²

(41)	(Based on Masuoka and Takubo 1989:148 [14])
	A: Dakara,	uti	no	purozyekuto wa	kanari zitugen	no
	therefore our gen project				top very	realization gen
	kanousei	ga	detekiteiru no		da.
	possibility nom manifest			comp copula
	‘Therefore, the possibility of our project being carried out is becoming high.’
	B: {Kono / ♯Ano}		purozyekuto ni	wa	wareware mo	sankadekiru
	this	that	project	dat top we	also can:participate
	no	desu	ka.
	comp copula q
	‘Can we also participate in {this/ that} project?’

(42)	[Context: You happen to think about your old friend, John, whom you have not seen for a long time. You talk to yourself, saying ... ]
	{♯Kono / Ano}		otoko wa	genki	ni	siteiru	ka na.
	this	that	man	top healthy	dat is:doing		q emph
	‘Has {this/ that} man been doing fine?’

(43)	(= Hoji et al. 2003 [31], slightly adapted)
	a.	[Context: The leader of the anti-government movement has called an underground meeting in order to designate the members who will put into action the plan of bombing the embassy, which they have been working on for a couple of weeks. Every member is waiting for him to speak. The leader begins the meeting by making the following statement.]
		[{Kono / ♯Ano}		keikaku o	saisyoni kangaedasita mono] o
		this /	that	plan	acc first	proposed	person acc
		kondono	taisikan	bakuha	keikaku no	zikkoo	sekininsya
		upcoming embassy bombing plan			gen execution leader
		ni	siyoo.
		dat let:us:do
		‘[I] nominate the person who first proposed {this/ that} plan to be the execution leader of the upcoming embassy bombing plan.’
	b.	[Context: After the failure of the bombing at the embassy ten years ago, the group of anti-government guerillas became too weak, and they have decided to dissolve their organization. No one dares to speak a word at the meeting, except the leader.]
		[{♯Kono / Ano}		keikaku o	saisyoni kangaedasita mono] ga
		this /	that	plan	acc first	proposed	person nom
		zyuu-nen mae	no	taisikan	bakuha	keikaku no	zikkoo
		10-year	before gen	embassy	bombing plan		gen execution
		sekininsya ni		naru	bekidatta.
		leader	dat become should:have
		‘The person who first proposed {this/ that} plan should have become the execution leader of the embassy bombing plan ten years ago.’

10.1.5 Interrogatives

The inventory of the Japanese wh-words includes those that pattern with the demonstrative paradigms (i.e., (44) and (45)) and those that do not (i.e., (46)).¹³

(44)	Those that pattern with simplex demonstratives:
	a.	dore ‘which one’
	b.	doko ‘where’
	c.	dotira ‘which of the two’
	d.	doo ‘in what way’

(45)	Those that pattern with complex demonstratives:
	a.	dono NP ‘which NP’
	b.	donna NP ‘what kind of NP’
	c.	dooyuu NP ‘what kind of NP’
	d.	dono yooni ‘in what way’
	e.	donna huuni ‘in what way’

(46)	a.	dare ‘who’
	b.	itu ‘when’
	c.	nani/nan ‘what’
	d.	ikutu ‘how many’
	e.	ikura ‘how much’
	f.	nan + classifier ‘how many’
	g.	naze/doosite ‘why’

Unlike in English, in Japanese wh-words need not be fronted. This is illustrated in (47).

(47)	a.	Kaori san	wa	Taroo kun ni	nani o	agemasita ka.
		Kaori Miss top Taro		Mr. dat what acc gave		q
		‘What did Kaori give to Taro?’
	b.	John wa	kono ronbun o		doko	de happyoosimasita ka.
		John top this	paper	acc where at presented		q
		‘Where did John present this paper?’
	c.	Eriko wa	ano paatii ni	naze ik-anak-atta	to	omoimasu
		Eriko top that party dat why go-neg-past comp think
		ka.
		q
		‘Why do you think that Eriko did not go to that party?’

Generative grammar has been concerned with whether or not Japanese wh-words undergo covert movement (cf. Lasnik and Saito 1984; Saito 1985; Nishigauchi 1986; Pesetsky 1987). Supporting the covert movement analysis is the generalization that the structural distance between a given wh-word and its associated question particle ka cannot be too far.¹⁴ (In some cases ka may not be phonologically realized.) Recently, however, Kubo (1989) and Deguchi and Kitagawa (2002) argue against this generalization, demonstrating that if a certain intonation contour is assigned to the unit starting from the wh-word through its associated question particle, the long-distance relation can be established.

A terminological remark: In referring to the items in (44)–(46), we used the term wh-word. This may not be appropriate for the following reasons. First, in questions, the items under discussion must co-occur with the question particle ka; so they do not by themselves function as interrogative words. Second, as we observe below, adding certain particles to them produces existential and universal quantifier analogues.¹⁵ For these reasons, Kuroda (1965) terms them indeterminate pronouns rather than wh-words. However, in an effort to reduce the amount of terminology, we will continue to call them wh-words below.

10.2 Existential (Intersective) Quantifiers and Related Issues

10.2.1 Type (1,1) Quantifier Analogues—D-Quantifiers

To express what English intersective D-quantifiers mean in Japanese, we may use one of the three patterns in (48), where QE stands for quantity expression and CM signifies a case-marker or a postposition.

(48)	a.	QE + no + NP + CM
	b.	NP + QE + CM
	b.	NP + CM + QE

As we observe below, their distribution is different. Thus, for any analysis of quantity expressions to be considered viable, the distributional difference ought to be accounted for.^{16 , 17}

The phrases that serve as a QE are partially listed in (49), and we provide a few illustrations in (50)–(51).¹⁸

(49)	a.	Phrases consisting of (a modifier) + ♯ + a classifier:
		san-nin	‘three people’,	suu-sya ‘a few companies’,		san-nin	izyoo
		‘three or more people’	sukunakutomo	suu-sya	‘at least a few companies’
	b.	Phrases describing amount:
	syoosuu ‘a small number’, tasuu ‘a large number’, takusan ‘many’

(50)	a.	San-nin	no	dansigakusei ga	kita.	(Cf. (48a).)
		three-cl gen male:student nom came
		‘Three male students came.’
	b.	Dansigakusei san-nin ga kita. (Cf. (48b).)
	c.	Dansigakusei ga san-nin kita. (Cf. (48c).)

	(51)	a.	John wa	sukunakutomo suu-satu	no	hon	o	yonda.
			John top at:least	several-cl gen book acc read
			‘John read at least a few books.’ (Cf. (48a).)
		b.	John wa hon sukunakutomo suu-satu o yonda. (Cf. (48b).)
		c.	John wa hon o sukunakutomo suu-satu yonda. (Cf. (48c).)

As illustrated in (52) and (53), some items in (49b) are not compatible with the NP + QE + CM pattern (i.e., (48b)), though they seem to be with the other two patterns.

(52)	a.	John wa	takusan no	hon	o	yonda. (Cf. (48a).)
		John top many		gen book acc read
		‘John read many books.’
	b.	?*John wa hon takusan o yonda. (Cf. (48b).)
	c.	John wa hon o takusan yonda. (Cf. (48c).)

(53)	a.	Syoosuu	no	zyosigakusei	ga	kita. (Cf. (48a).)
		small:number gen female:student nom came
		‘A small number of female students came.’
	b.	?*Zyosigakusei syoosuu ga kita. (Cf. (48b).)
	c.	?Zyosigakusei ga syoosuu kita. (Cf. (48c).)

The items that are not compatible with the NP + QE + CM pattern seem to be those that cannot occur as an argument of a verb independently. This is illustrated in (54).

(54)	a.	{San-nin / Sukunakutomo san-nin / San-nin izyoo}		o
		three-cl	at:least	three-cl three-cl equal:more acc
		suisensita.
		recommended
		‘[He] recommended {three/ at least three/ three or more}.’
	b.	?*{Syoosuu /	Takusan} o	suisensita.
		Small:number many	acc recommended
		‘(Lit.) [He] recommended {small number/ many}’

There is another way to express what the English existential D-quantifier means; we can add the prefix boo ‘some’ or the nominal modifier aru ‘some’ to an NP as in (55). If, for example, the sentences in (56) are translated into English, the italicized parts can be rendered as some actress and some male student.

(55)	boo-NP ‘some NP’19, aru NP ‘some NP’

(56)	a.	Yamada sensei	wa	ima	boo-zyoyuu	to	ren’aisiteiru.
		Yamada teacher top now some-actress with be:in:love
		‘Prof. Yamada is in love with some actress now.’
	b.	Aru	dansigakusei ga		John to	kenkasita.
		some male:student nom John with fought
		‘Some male student fought with John.’

In addition, we may construct existential quantifier analogues, using wh-words. Adding the particle ka to the wh-words in (57), we obtain the items in (58).

(57)	a.	Among the items listed in (44):
		dore ‘which one’, doko ‘where’, dotira ‘which of the two’
	b.	Among the items listed in (46):
		dare ‘who’, nani /nan ‘what’, ikutu ‘how many’, ikura ‘how much’, nan + classifier ‘how many’

(58)	a.	dore ka ‘some of them’, doko ka ‘somewhere’, dotira ka ‘one of the
		two things/people’
	b.	dare ka ‘someone’, nani/nan ka ‘something’, ikutu ka ‘a few things’, ikura ka ‘a few things’, nan + classifier ka ‘a few NP (where the NP is compatible with the classifier)’

Using some of the items in (58), we may, for example, construct sentences like those in (59).²⁰

(59)	a.	Dare ka ga			kita	mitaida.
		who	p	nom came seem
		‘It seems that someone has come.’
	b.	John wa		nani ka o		katta.
		John top what p acc bought
		‘John bought something.’
	c.	Ano otoko wa		mada kono syuuhen no	doko	ka ni iru		hazuda.
		that	man	top still	this	fringe	gen where p	at exist must
		‘That man must still be somewhere near this area.’

We note, however, that the complex wh-words listed in (45) cannot be used to create existential quantifier analogues; see (60).

(60)	a.	*Dono hito	ka ga	kita.
		which person p	nom came
	b.	*Mariko wa	donna	hon	ka o	yonda.
		Mariko top what:kind book p			acc read

Existential quantifier analogues built from a wh-word may be used as the QE in the three patterns of (48), and the resulting unit as a whole also expresses what existential quantifiers mean. With the QE + no + NP + CM and NP + CM + QE patterns (i.e., (48a) and (48c)), all the items in (58) can be used,²¹ but with the NP + QE + CM pattern (i.e., (48b)), only some of them are usable. This is illustrated in (61)–(63).

(61)	a.	Nan-nin	ka no	gakusei ga	kita.	(Cf. (48a).)
		how:many-cl p	gen student		nom came
		‘A few students came.’
	b.	Gakusei nan-nin ka ga kita. (Cf. (48b).)
	c.	Gakusei ga nan-nin ka kita. (Cf. (48c).)

(62)	a.	John wa	matteiru aida	ni nan-satu	ka no	manga
		John top waiting	period in how:many-cl p		gen comic:book
		o	yonda.
		acc read
		‘John read a few comic books while he was waiting.’ (Cf. (48a).)
	b.	John wa matteiru aida ni manga nan-satu ka o yonda. (Cf. (48b).)
	c.	John wa matteiru aida ni manga o nan-satu ka yonda. (Cf. (48c).)

(63)	a.	John wa	matteiru aida	ni nani ka no	manga	o
		John top waiting	period in what p		gen comic:book acc
		yonda.
		read
		‘John read some comic book while he was waiting.’ (Cf. (48a).)
	b.	?*John wa matteiru aida ni manga nani ka o yonda. (Cf. (48b).)
	c.	John wa matteiru aida ni manga o nani ka yonda. (Cf. (48c).)

10.2.2 Type (1,1) Quantifier Analogues—A-Quantifiers

We now turn to intersective A-quantifier analogues. The adverbial expressions in (64) are comparable to intersective A-quantifiers in English.

(64)	a.	Phrases consisting of ♯ + a classifier:
		suu-kai ‘a few times’, san-do ‘three times’
	b.	Phrases describing frequency:
		tokidoki ‘sometimes’, sibasiba ‘often’

Using some of them, we can, for example, construct the sentences in (65).

(65)	a.	Taroo wa	suu-kai	Hanako to	deetosita.
		Taro	top several-cl Hanako with dated
		‘Taro went out with Hanako a few times.’
	b.	Mary wa	tokidoki	keeki o	yaku.
		Mary top sometimes cake		acc bake
		‘Mary sometimes bakes cakes.’

Existential A-quantifier analogues can also be built from a wh-word. For example, adding the particle ka to nan-kai ‘how many times’ and nan-do ‘how many times’, we obtain the items in (66); see (67).

(66)	nan-kai ka ‘a few times’, nan-do ka ‘a few times’

(67)	a.	Mary wa	nan-kai	ka keeki o	yaita.
		Mary top what-cl p		cake	acc baked
		‘Mary baked cakes a few times.’
	b.	Susan wa	nan-do	ka Bill ni	iiyotta.
		Susan top what-cl p Bill dat approached
		‘Susan approached Bill a few times’

10.2.3 Compound QEs with Boolean Connectives

In English, complex intersective quantifiers can be built with Boolean connectives productively, e.g., not more than ten students and at least two but not more than five students. Since the negation in Japanese does not combine with a QE or an NP, we do not observe many instances of compound QEs. For example, we have no way to express what neither A nor B means with an NP by itself.

One case we have thus far discovered is one in which QEs are combined with an item expressing disjunction; see the sentences in (68), for example.

(68)	a.	San-nin	{ka / mosikuwa / matawa}	yo-nin	no	gakusei ga
		three-cl or or	or	four-cl gen student nom
		kuru daroo.
		come likely
		‘Probably, three or four students come.’ (Cf. (48a).)
	b.	Gakusei san-nin {ka / mosikuwa / matawa} yo-nin ga kuru daroo. (Cf. (48b).)
	c.	Gakusei ga san-nin {ka / mosikuwa / matawa} yo-nin kuru daroo. (Cf. (48c).)

We can also build phrases that correspond to the English expression at least two but not more than five students, using an item expressing conjunction; see (69) and (70).

(69)	a.	Sukunakutomo	huta-ri		sikasi go-nin	miman no	gakusei o
		at:least	two-cl but		five-cl below	gen student	acc
		suisensuru	koto	ni	natta.
		recommend comp dat became
		‘It has been decided that [we] will recommend at least two but not more than five students.’ (Cf. (48a).)
	b.	Gakusei sukunakutomo huta-ri sikasi go-nin miman o suisensuru koto ni natta. (Cf. (48b).)
	c.	Gakusei o sukunakutomo huta-ri sikasi go-nin miman suisensuru koto ni natta. (Cf. (48c).)

(70)	a.	Huta-ri izyoo	katu go-nin	ika	no	gakusei o
		two-cl equal:more and		five-cl equal:less gen student		acc
		erab-anakerebanaranai.
		choose-must
		‘[We] must choose two or more but five or less students.’ (Cf. (48a).)
	b.	Gakusei huta-ri izyoo katu go-nin ika o erab-anakerebanaranai. (Cf. (48b).)
	c.	Gakusei o huta-ri izyoo katu go-nin ika erab-anakerebanaranai. (Cf. (48c).)

Incidentally, even if the item expressing conjunction katu ‘and’ is eliminated from the sentences in (70), the resulting sentences express the meanings of the original ones; see the sentences in (71).

(71)	a.	Huta-ri izyoo	go-nin	ika	no	gakusei o
		two-cl equal:more five-cl equal:less gen student acc
		erab-anakerebanaranai.
		choose-must
		‘[We] must choose two or more but five or less students.’ (Cf. (48a).)
	b.	Gakusei huta-ri izyoo go-nin ika o erab-anakerebanaranai. (Cf. (48b).)
	c.	Gakusei o huta-ri izyoo go-nin ika erab-anakerebanaranai. (Cf. (48c).)

10.2.4 Existential Sentences

In Japanese, there seems to be no special construction that is appositely used for existential sentences. When we express what the English existential sentence means, we may use the construction in (72), where iru is used for animate objects, and aru for inanimate objects. This is illustrated in (73).

(72)	B {ni / ni wa} A ga iru / aru, where B is a location, and A is the existing object.

(73)	a.	Kyoositu		{ni / ni wa}	gakusei ga	iru.
		classroom in	in top student nom exist
		‘There are students in the classroom.’
	b.	Kono heya		{ni / ni wa}	kuroi tukue ga		aru.
		this	room in	in top black desk		nom exist
		‘There is a black desk in this room.’

Some researchers assume (74a) instead of (74b) to be the base order for (72), and claim that the existential construction is ‘special’ with respect to the word order (cf. Kuno 1971; Takezawa 1987).

(74)	a.	B ni A ga iru /aru
	b.	A ga B ni iru /aru

However, given that the word order is relatively free in Japanese, it is difficult to examine this claim.

In Japanese, we do not observe the definite/indefinite distinction to which the English existential construction is sensitive (cf. Milsark 1974); all of the sentences in (75)–(77) are perfectly acceptable.

(75)	a.	Kyoositu		{ni / ni wa}	sukunakutomo suu-dai	no	konpyuutaa
		classroom in		in top at:least		several-cl gen computer ga aru.
		nom	exist
		‘In the classroom, there are at least several computers.’ (Cf. (48a).)
	b.	Kyoositu {ni / ni wa} konpyuutaa sukunakutomo suu-dai ga aru. (Cf. (48b).)
	c.	Kyoositu {ni / ni wa} konpyuutaa ga sukunakutomo suu-dai aru. (Cf. (48c).)

(76)	a.	Kono ie		{ni / ni wa}	san-biki	no	inu ga	iru.
		this	house in	in top three-cl gen dog nom exist
		‘In this house, there are three dogs.’ (Cf. (48a).)
	b.	Kono ie {ni / ni wa} inu san-biki ga iru. (Cf. (48b).)
	c.	Kono ie {ni / ni wa} inu ga san-biki iru. (Cf. (48c).)

(77)	a.	Huransu		{ni / ni wa}	ano yuumeina Efferutoo		ga	aru.
		France		in	in top that famous	Eiffel:Tower nom exist
		‘In France is that famous Eiffel Tower.’
	b.	Kono ie		{ni / ni wa}	senzitu	wadai ni	natta	Saburoo ga
		this	house in	in top recently topic		dat became Saburo nom
		iru.
		exist
		‘In this house is Saburo, who became a topic of conversation recently.’
	c.	Kyoositu		{ni / ni wa}		Chomsky no	hotondo no	ronbun ga
		classroom in		in top Chomsky gen almost			gen paper	nom
		aru.
		exist
		‘In this classroom are most of Chomsky’s papers.’
	d.	Paatii kaizyoo	{ni / ni wa}		sannensei	no	subete no
		party hall	in	in top third:year:student gen all				gen
		gakusei ga	iru.
		student nom exist
		‘In the party hall is every third-year student.’

Those who wish to maintain that the existential construction in Japanese is ‘special’ might claim that with the sentences in (75) and in (76) the base order is (74a); however, those in (77) have (74b) as the base order and involve scrambling. But it is difficult to evaluate this claim at this point.

10.3 Universal (Co-intersective) Quantifiers and Related Issues

10.3.1 Type (1,1) Quantifier Analogues—D-Quantifiers

Similarly to the cases of intersective D-quantifier analogues above, we may construct universal D-quantifier analogues, using one of the three patterns in (48), repeated here.

(48)	a.	QE + no + NP + CM
	b.	NP + QE + CM
	c.	NP + CM + QE

A partial list of the items that serve as a QE is provided in (78), and illustrations of the three patterns are given in (79) and (80).

(78)	a.	Phrases consisting of a prefix that means ‘all’ + a classifier22:
		zen-in ‘all members’, zen-sya ‘all companies’, zen-bu ‘all things’
	b.	Phrases describing amount:
		subete ‘all’, minna ‘all’

(79)	a.	Kimura bengosi wa	zen-bu no	seiyakugaisya	o
		Kimura attorney top all-cl gen pharmaceutical:company acc
		uttaeta.
		sued
		‘Attorney Kimura sued all the pharmaceutical companies.’ (Cf. (48a).)
	b.	Kimura bengosi wa seiyakugaisya zen-bu o uttaeta. (Cf. (48b).)
	c.	Kimura bengosi wa seiyakugaisya o zen-bu uttaeta. (Cf. (48c).)

(80)	a.	Subete no	dansigakusei ga	eiga	ni	ikitagatteita.
		all	gen male:student nom movie dat wanted:to:go
		‘All the male students wanted to go to a movie.’ (Cf. (48a).)
	b.	Dansigakusei subete ga eiga ni ikitagatteita. (Cf. (48b).)
	c.	Dansigakusei ga subete eiga ni ikitagatteita. (Cf. (48c).)

Another way to construct universal D-quantifier analogues is to add the prefix zen- ‘all’ to an NP and make a compound NP as in (81); see the illustrations in (82).

(81)	zen-NP ‘all NP’

(82)	a.	Seihu	wa	zen-hokengaisya	ni	keikoku o
		government top all-insurance:company dat warning acc
		hassita.
		gave
		‘The government gave a warning to all the insurance companies.’
	b.	Zen-tiimu ga		ano torofii o	kisoiau.
		all-team	nom that trophy acc compete
		‘All teams compete for that trophy.’

Universal D-quantifier analogues can be built from a wh-word. Interestingly, however, which wh-words can serve as the base is different from the existential D-quantifier analogue cases. Adding the particle mo to the wh-words in (83), we can create universal quantifier analogues as in (84).

(83)	a.	From the list in (44):
		dotira ‘which of the two’
	b.	From the list in (45):
		dono NP ‘which NP’, donna NP ‘what kind of NP’, dooyuu NP ‘what kind of NP’, dono yooni ‘in what way’, donna huuni ‘in what way’

(84)	a.	dotira mo ‘both of them’
	b.	dono NP mo ‘every NP’, donna NP mo ‘every kind of NP’, dooyuu NP mo ‘every kind of NP’, dono yooni mo ‘in any way’, donna huuni mo ‘in any way’

The sentences in (85) provide illustrations.

(85)	a.	Ziroo wa		dotira mo	tabeta.
		Jiro	top which also ate
		‘Jiro ate both.’
	b.	Dono	hito	mo	kita.
		which person also came
		‘Every person came.’
	c.	Kono kurabu wa		donna	hito	ni	mo	annaizyoo	o
		this	club	top what:kind person dat also invitation:letter acc
		okuru.
		send
		‘This club sends an invitation letter to every person.’
	d.	Watasi no	tugoo	wa	dono	yooni	mo	dekiru node,
		I	gen schedule top which condition also able						because
		sukina	zikan o		erande kudasai.
		convenient time		acc choose please
		‘As my schedule can be modified in any way, please choose a good time.’

Mo used in (84) is one of the so-called focus-sensitive particles, discussed in Section 10.7, and is often translated as ‘also’. Just as many other particles in this group, when mo is attached to an NP-α unit where α is a case-marker ga or o, α may not surface at least phonologically (e.g., (85a) and (85b)). If it is attached to an NP-α unit where α is a postposition or a case-marker other than ga and o, on the other hand, α must surface (e.g., (85c)).

There are a number of wh-words that may not become universal quantifier analogues with the addition of mo, e.g., those listed in (86).

(86)	a.	Among the items listed in (44):
		dore ‘which one’, doko ‘where’, doo ‘in what way’
	b.	Among the items listed in (46):
		dare ‘who’, nani /nan ‘what’, ikutu ‘how many’, ikura ‘how much’,
		nan + classifier ‘how many’

As illustrated in (87), adding the particle mo to (i) ikutu ‘how many’ or (ii) nan + classifier ‘how many’ yields a phrase that means many items.

(87)	a.	Hanako wa	keeki o	ikutu	mo	tabeta.
		Hanako top cake		acc how:many also ate
		‘Hanako ate a lot of cakes.’
	b.	Taroo wamangaonan-satu moyonda.
		Taro	top comic:book acc what-cl also read
		‘Taro read a lot of comic books.’

The rest of the wh-words in (86), when combined with mo, generally cannot be used in affirmative sentences; see (88).²³

(88)	a.	*Dare mo	kita.
		who	also came
	b.	*John wa	nani mo	yonda.
		John top what also read
	c.	*John wa	doko	ni	mo	itta.
		John top where dat also went

There are some fixed expressions that make use of some of the wh-words in (86), expressing the universal meaning; see (89).

(89)	a.	Tetuya	wa	dare mo	kare mo	ni	syootaizyoo o		okutta.
		Tetsuya top who also he				also dat invitation			acc sent
		‘Tetsuya sent an invitation to everyone.’
	b.	Watasi no	haha	wa	nani	mo	ka	mo	kaootosuru node
		I	gen mother top what also that also try:to:buy						because
		itumo	okane	ga	nai.
		always money nom not:exist
		‘Because my mother tries to buy everything, she always lacks money.’
	c.	Kurisumasu	siizun	wa	doko	mo	kasiko mo	nigiwatteiru.
		Christmas	season top where also there				also lively
		‘In the Christmas season, everywhere is lively.’

It should also be noted here that we can create universal D-quantifier analogues by embedding a wh-word in an NP and attaching mo to the NP rather than to the wh-word itself. With this strategy, not only the wh-words in (83) but also those in (86) can be used. This is illustrated in (90).²⁴

(90)	a.	Sakunen wa	[dono	sensei	ni	suisenzyoo		o
		last:year top which teacher dat recommendation:letter acc
		tanonda gakusei] mo		daigakuin		ni	haire-nak-atta.
		asked student also graduate:school dat able:enter-neg-past
		‘Last year, every student who solicited a letter of recommendation from any professor was not able to get into a graduate school.’
	b.	Kono gakkoo de wa		[donna	iiwake o		yuu hito]	mo
		this	school	at	top what:kind excuse acc say person also
		syobatusareru.
		is:disciplined
		‘At this school, anyone who provides any excuse gets disciplined.’
	c.	[Nani o	katta	hito]	mo	syoosai o	hookokusita.
		what	acc bought person also details				acc reported
		‘Everyone who bought anything provided an explanation in detail.’

Universal D-quantifier analogues built from a wh-word can be used as the QE in the NP + CM + QE pattern (i.e., (48c)), and the whole unit expresses what universal quantifiers mean. This is illustrated in (91).

(91)	a.	?Kono kurasu no		gakusei ga	dono	gakusei mo	eiga	ni
		this	class	gen student nom which student also movie dat
		ikitagatteiru. (Cf. (48c).)
		want:to:go
		‘Every student in this class wants to go to a movie.’
	b.	?Kimura bengosi wa		Kyooto no	seiyakugaisya			o
		Kimura attorney top Kyoto			gen pharmaceutical:company acc
		dono	kaisya	mo	uttaeta. (Cf. (48c).)
		which company also sued
		‘Attorney Kimura sued every pharmaceutical company in Kyoto.’

However, unlike existential D-quantifier analogues built from a wh-word, they cannot be used as the QE in the QE + no + NP + CM pattern, i.e., (48a) (see (92)) or in the NP + QE + CM pattern, i.e., (48b) (see (93)).²⁵

(92)	a.	*Dono gakusei mo no kono kurasu no gakusei ga eiga ni ikitagatteiru.
		(Cf. (48a).)
	b.	*Kimura bengosi wa dono kaisya mo no Kyooto no seiyaku gaisya o uttaeta.
		(Cf. (48a).)

(93)	a.	?*Kono kurasu no gakusei dono gakusei mo eiga ni ikitagatteiru. (Cf. (48b).)
	b.	?*Kimura bengosi wa Kyooto no seiyaku gaisya dono kaisya mo uttaeta.
		(Cf. (48b).)

10.3.2 Type (1,1) Quantifier Analogues—A-Quantifiers

Let us turn to universal A-quantifier analogues. We may express what universal A-quantifiers mean with the expressions in (94); see the illustrations in (95).

(94)	a.	Phrases consisting of a prefix that means ‘all’ + a classifier:
		mai-kai ‘every round’, mai-do ‘every time’ mai-syuu ‘every week’,
		mai-tosi ‘every year’
	b.	Phrases describing frequency:
		kanarazu ‘necessarily, always’, tuneni ‘always’

(95)	a.	Sakunen wa		takusan siai	o	sita ga	mai-kai
		last:year top many		game acc did but every-round
		maketesimatta.
		lost
		‘Last year, although [we] did many games, we lost every single time.’
	b.	Mary wa	ano resutoran		ni	iku to kanarazu keeki o		taberu.
		Mary top that restaurant dat go if always						cake acc eat
		‘When she goes to that restaurant, Mary always eats a cake.’
	c.	Watasi ga	au toki	John wa	tuneni	aoi	seetaa	o
		I	nom meet when John top always blue sweater acc
		kiteiru.
		is:wearing
		‘When I see John, he always wears a blue sweater.’

Attaching goto ni or tabi ni to a VP, an NP, or a QE as in (96) also produces an expression that serves as a universal A-quantifier. This is illustrated in (97).

(96)	a.	VP goto ni ‘every time someone VP’, NP goto ni ‘every NP’,
		QE goto ni ‘every QE’
	b.	VP tabi ni ‘every time someone VP’, NP no tabi ni ‘every NP’

(97)	a.	Yuuko wa	nihon ni	kikokusuru {goto		/ tabi}	ni otya
		Yuko	top Japan dat return		every:time every:time at tea
		o gohyaku guramu mottekaettekuru.
		acc 500 gram bring:back
		‘Whenever Yuko returns to Japan, she brings back 500 g of tea.’
	b.	Suzuki sensei	wa	zibun no	kenkyuu happyoo	no
		Suzuki teacher top self			gen research presentation gen
		tabi	ni dare ka o		kizutukeru.
		evert:time at who p acc harm
		‘Prof. Suzuki hurts someone at each occasion of his research presentation.’
	c.	Takasi	wa	yo-nin	goto	ni syootaizyoo o	tewatasita.
		Takashi top four-cl every:time at invitation				acc handed:out
		‘Takashi handed out an invitation to every four people.’

We may also construct universal A-quantifier analogues from wh-words. Similarly to D-quantifier cases, however, the wh-words that can be used here are different from those for intersective A-quantifier analogues. For example, adding the particle mo to itu ‘when’, we obtain (98), and with it we can construct sentences like (99).

(98)	itu mo ‘always’

(99)	Mary wa	itu	mo	keitaidenwa o	motteiku.
		Mary top when also cellphone		acc carry
	‘Mary always carries a cellphone.’

But adding the particle mo to nan-kai ‘how many times’ or nan-do ‘how many times’ produces an expression that means many times. This is illustrated in (100).

(100)	a.	Lynn wa	paatii de nan-do		mo	piano o	hiita.
		Lynn top party at what-time also piano acc played
		‘Lynn played piano repeatedly at parties.’
	b.	Masako wa	niku	tabehoodai	no	resutoran	ni	iku to
		Masako top meat all:you:can:eat gen restaurant dat go if
		nan-kai	mo	takusan no	syurui no	niku o		taberu.
		what-round also many		gen type		gen meat acc eat
		‘When Masako goes to an all-you-can-eat BBQ restaurant, she eats many kinds of meat repeatedly.’

In addition, we can construct universal A-quantifier analogues, by using a wh-word and a noun like toki ‘time’ and baai ‘occasion’, as schematized in (101). This is illustrated in (102) and (103).

(101)	a.	wh-word + {toki (+ CM) / baai (+ CM)} + mo
	b.	[[ ... wh-word ... ] {toki (+ CM) / baai (+ CM)}] + mo

(102)	a.	Suguru wa	dooyuu	toki (de) mo	Kanako o	tasukeyoo
		Suguru top what:kind time at		also Kanako acc try:to:help
		to	suru.
		comp do
		‘Suguru tries to help Kanako at any time.’
	b.	Mary wa	dono baai	(ni) mo	reiseini taisyosita.
		Mary top what occasion at			also calmly handled
		‘Mary did not lose her composure on any occasion.’

(103)	a.	Mary wa	[[donna	otoko to	deetosuru] toki		(ni)] mo
		Mary top what:kind man			with date		when at	also
		kireini	kikazaru.
		beautifully dress:up
		‘Mary dresses up beautifully when she goes out with a man, no matter what kind of man he is.’
	b.	Takesi	wa	[[doko ni	iku] baai	(de)] mo	okane	o
		Takeshi top where dat go				situation at	also money acc
		takusan motteiku.
		many bring
		‘Takeshi brings a lot of money when he goes out, no matter where the destination is.’

10.3.3 Distributivity

Regarding the issue of how distributivity is expressed in Japanese, we first point out that distributivity obtains without any overt distributor. For example, (104) can be understood to mean that each student baked a cake.

(104)	Gakusei ga	keeki o	yaita.
	student	nom cake	acc baked
	‘Students baked cakes.’

Similarly, the sentences in (105) can all mean that three students each baked a cake.

(105)	a.	San-nin	no	gakusei ga	keeki o	yaita. (Cf. (48a).)
		three-cl gen student		nom cake	acc baked
		‘Three students baked cakes.’
	b.	Gakusei san-nin ga keeki o yaita. (Cf. (48b).)
	c.	Gakusei ga san-nin keeki o yaita. (Cf. (48c).)

There are also expressions that serve as distributors, e.g., those in (106). If such an expression is used, the NP that it modifies is individuated.

(106)	Individual distributors:
	onoono ‘each’, sorezore ‘each’, meimei ‘each’

These seem to be able to occur rather freely, as illustrated in (107).

(107)	a.	{Onoono / Sorezore / Meimei} no			gakusei ga	ronbun o
		each	each	each	gen student nom paper	acc
		teisyutusita.
		submitted
		‘Each student submitted a paper.’
	b.	Gakusei {onoono/ sorezore/ meimei} ga ronbun o teisyutusita.
	c.	Gakusei ga {onoono/ sorezore/ meimei} ronbun o teisyutusita.
	d.	Gakusei ga ronbun o {onoono/ sorezore/ meimei} teisyutusita.
	e.	{Onoono/ Sorezore/ Meimei} gakusei ga ronbun o teisyutusita.

The distributors in (106) may occur with a QE; the paradigms in (108), in (109), and in (110) illustrate the QE + no + NP + CM pattern (i.e., (48a)), the NP + QE + CM pattern (i.e., (48b)), and the NP + CM + QE pattern (i.e., (48c)), respectively.

(108)	a.	*{Onoono / Sorezore / Meimei} no			san-nin	no	gakusei ga
		each	each	each	gen three-cl gen student		nom
		ronbun o	teisyutusita.
		paper acc submitted
		‘Three students each submitted a paper.’
	b.	*San-nin no {onoono / sorezore / meimei} no gakusei ga ronbun o teisyutusita.
	c.	San-nin no gakusei ga {onoono / sorezore / meimei} ronbun o teisyutusita.
	d.	San-nin no gakusei ga ronbun o {onoono / sorezore / meimei} teisyutusita.
	e.	{Onoono / Sorezore / Meimei} san-nin no gakusei ga ronbun o teisyutusita.

(109)	a.	??Gakusei {onoono / sorezore / meimei} san-nin ga ronbun o teisyutusita.
	b.	??Gakusei san-nin {onoono / sorezore / meimei} ga ronbun o teisyutusita.
	c.	Gakusei san-nin ga {onoono / sorezore / meimei} ronbun o teisyutusita.
	d.	Gakusei san-nin ga ronbun o {onoono / sorezore / meimei} teisyutusita.
	e.	{Onoono / Sorezore / Meimei} gakusei san-nin ga ronbun o teisyutusita.

(110)	a.	*?Gakusei {onoono / sorezore / meimei} ga san-nin ronbun o teisyutusita.
	b.	Gakusei ga {onoono / sorezore / meimei} san-nin ronbun o teisyutusita.
	c.	Gakusei ga san-nin {onoono / sorezore / meimei} ronbun o teisyutusita.
	d.	Gakusei ga san-nin ronbun o {onoono / sorezore / meimei} teisyutusita.
	e.	{Onoono / Sorezore / Meimei} gakusei ga san-nin ronbun o teisyutusita

By using the expressions in (111), we can individuate events.

(111)	Event distributors:
	a.	1 + classifier + zutu or 1 + classifier + 1 + classifier
		‘one thing at a time’
	b.	n + classifier + zutu, where n is any positive integer
		‘n things at a time’

Their distribution is illustrated in (112)–(113).

(112)	a.	?*Hito-ri	hito-ri	no	gakusei ga	bungakubu	o
		one-cl one-cl gen student nom school:of:humanities acc
		yameteitta.
		quitted
		‘Students walked away from the School of Humanities one by one.’
	b.	?*Gakusei hito-ri hito-ri ga bungakubu o yameteitta.
	c.	Gakusei ga hito-ri hito-ri bungakubu o yameteitta.
	d.	Gakusei ga bungakubu o hito-ri hito-ri yameteitta.
	e.	Hito-ri hito-ri gakusei ga bungakubu o yameteitta.

(113)	a.	?*John wa	has-satu	zutu	no	hon	o	heya	ni
		John top eight-cl at:the:time gen book acc room dat
		hakobikonda.
		brought:in
		‘John brought in books to the room eight at a time.’
	b.	?*John wa hon has-satu zutu o heya ni hakobikonda.
	c.	John wa hon o has-satu zutu heya ni hakobikonda.
	d.	John wa hon o heya ni has-satu zutu hakobikonda.
	e.	John wa has-satu zutu hon o heya ni hakobikonda.

10.3.4 Exception Phrases

It is questionable whether Japanese has an NP directly corresponding to an English exception phrase, e.g., everyone but John. One may argue that NPs whose form is (114) can be regarded as such instances. For example, we can construct sentences like (115).

(114)	A igai no ({subete / zen-bu} no) B ‘all B other than A’,
	A o nozoku ({subete / zen-bu} no) B ‘all B other than A’,
	A no hoka no ({subete / zen-bu} no) B ‘all B other than A’

(115)	Kimura sensei	wa	John	igai	no	({subete / zen-bu} no)
		Kimura teacher top John other:than gen all			all-cl	gen
		gakusei o	suisensita.
		student acc recommended
		‘Prof. Kimura recommended all the students other than John.’

However, the contrast between (116) and (117) illustrates that, unlike English exception phrases, phrases with the form of (114) may not give rise to the conventional implicature that the excepted individuals do not possess the relevant property.

(116)	Kimura sensei	wa	John igai	no	({subete / zen-bu} no)
	Kimura teacher top John other:than gen all				all-cl	gen
	gakusei o	suisensita	no	wa	motiron,	John mo
	student acc recommended comp top of:course John also
	suisensita.
	recommended
	‘Expectedly, Prof. Kimura recommended all the students other than John, and he also recommended John.’

(117)

♯Expectedly, Prof. Kimura recommended everyone except John, and he also recommended John.

Incidentally, we have the adverbial counterparts of the items in (114), i.e., those in (118), and these contribute to sentence meanings in a similar way; see (119).

(118)	A igai, {subete / zen-bu} no B ‘all B, excluding A’;
	A o {nozoite / nozoki}, {subete/ zen-bu} no B ‘all B, excluding A’;
	A no hoka, {subete / zen-bu} no B ‘all B, excluding A’

(119)	Kimura sensei	wa	John o	{nozoite / nozoki)	{subete / zen-bu}
	Kimura teacher top John acc excluding excluding all				all-cl
	no	gakusei o	suisensita.
	gen student acc recommended
	‘Prof. Kimura recommended all the students, excluding John.’

It should be noted that with the adverbial clauses in (118), the conventional implicature under discussion is likely to obtain; see (120).

(120)	♯Kimura sensei		wa	John o	{nozoite / nozoki}	{subete / zen-bu}
	Kimura	teacher top John acc excluding excluding all					all-cl
	no	gakusei o	suisensita	no	wa	motiron,	John mo
	gen student acc recommended comp top of:course John also
	suisensita.
	recommended
	‘Expectedly, Prof. Kimura recommended all the students excluding John, and he also recommended John.’

The closest to the English exception phrases we find would probably be (121), but it must occur with negation. As pointed out by Kataoka (2006:Section 5.6), (122), for example, has the conventional implicature that the excepted individuals do not possess the relevant property; see (123).²⁶

(121)

NP + sika

(122)	Kimura sensei	wa	John sika	suisensi-nak-atta.
	Kimura teacher top John only recommend-neg-past
	‘Prof. Kimura did not recommend anybody but John.’

(123)	♯Kimura sensei	wa	John sika	suisensi-nak-atta	no	wa
	Kimura	teacher top John only recommend-neg-past comp top
	motiron,	John mo	suisensi-nak-atta.
	of:course John also recommend-neg-past
	‘As expected, Prof. Kimura did not recommend anybody but John, and he also did not recommend John.’

We note that (121) can be used as an adverb, e.g., (124), and its adverbial use also gives rise to the conventional implicature under discussion.

(124)	Kimura sensei	wa	dansi gakusei o	John sika
	Kimura teacher top male student acc John only
	suisensi-nak-atta.
	recommend-neg-past
	‘Prof. Kimura did not recommend any male student other than John.’

10.4 Proportional Quantifiers

10.4.1 Type (1,1) Quantifier Analogues—D-Quantifiers

Like the intersective and universal D-quantifier analogues above, to express what English proportional D-quantifiers mean in Japanese, the three patterns in (48) may be used. (48) is repeated here for convenience. The items listed in (125) are among those that can be used as a QE.

(48)	a.	QE + no + NP + CM
	b.	NP + QE + CM
	b.	NP + CM + QE

(125)	Phrases describing amount:
	hotondo ‘almost all’27, hanbun ‘half’, iti-bu ‘one part’, ♯ paasento ‘♯ %’,
	♯ wari ‘♯ tenth(s)’, ♯ 1 bun no ♯ 2 ‘♯ 2 / ♯ 1’

Using some of the items in (125), we can construct examples like those in (126)–(128).²⁸

(126)	a.	Hotondo no	dansigakusei ga	tesuto ni	otitesimatta.
		most	gen male:student nom test	dat failed
		‘Most male students failed the test.’ (Cf. (48a).)
	b.	Dansigakusei hotondo ga tesuto ni otitesimatta. (Cf. (48b).)
	c.	Dansigakusei ga hotondo tesuto ni otitesimatta. (Cf. (48c).)

(127)	a.	Suzuki sensei	wa	iti-bu	no	gakusei o	suisensita
		Suzuki teacher top one-cl gen student acc recommended
		‘Prof. Suzuki recommended a portion of the students.’ (Cf. (48a).)
	b.	??Suzuki sensei wa gakusei iti-bu o suisensita. (Cf. (48b).)
	c.	Suzuki sensei wa gakusei o iti-bu suisensita. (Cf. (48c).)

(128)	a.	San wari no ginkoo ga enzyo o moosideta.
		three tenths gen bank nom support acc offered
		‘Three tenths of the banks offered support.’ (Cf. (48a).)
	b.	Ginkoo san wari ga enzyo o moosideta. (Cf. (48b).)
	c.	Ginkoo ga san wari enzyo o moosideta. (Cf. (48c).)

10.4.2 Type (1,1) Quantifier Analogues—A-Quantifiers

Turning to proportional A-quantifiers, there are adverbs in Japanese that correspond to mostly or usually, e.g., those in (129); see the illustrations in (130).

(129)	Phrases describing frequency:
	hutuu ‘usually’, hudan ‘usually’, daitai ‘mostly’, yoku ‘frequently’

(130)	a.	Taroo wa	hutuu	Hanako to	dansusuru.
		Taro	top usually Hanako with dance
		‘Taro usually dances with Hanako.’
	b.	Emily wa	sotoni	iku toki	daitai	kamera o	motteiku.
		Emily top outside go when mostly camera acc bring
		‘When Emily goes out, she mostly brings a camera.’

Japanese does not have an adverb that in itself corresponds to rarely or seldom. To express the meaning of rarely or seldom, we need to use adverbs like mettani or hotondo with negation, as in (131). A few illustrations are provided in (132).

(131)	[... {mettani / hotondo}... Verb + Neg], where {mettani / hotondo} and
	Neg are clause-mates

(132)	a.	Susumu wa	paatii ni	it-temo	mettani sake	o
		Susumu top party dat go-even:if hardly			alcohol acc
		nom-anai.
		drink-neg
		‘Even when Susumu goes to a party, he seldom drinks alcohol.’
	b.	Aya wa	hotondo zyugyoo ni	ik-anai.
		Aya top almost	class	dat go-neg
		‘Aya seldom goes to classes.’

10.5 Partitives

One way to express the partitive meaning in Japanese is to use the form in (133). This is illustrated in (134)–(135).²⁹

(133)

NP no QE CM

(134)	a.	Uti no	gakusei no	san-nin	ga	kita.
		our gen student gen three-cl nom came
		‘Three of our students came.’
	b.	Uti no	gakusei no	tasuu ga	kita.
		our gen student gen many nom came
		‘Many of our students came.’
	c.	Uti no	gakusei no	dare ka ga	kita.
		our gen student gen who p nom came
		‘One of our students came.’
	d.	Uti no	gakusei no	zen-in	ga	kita.
		our gen student gen all-member nom came
		‘All of our students came.’
	e.	Uti no	gakusei no	subete ga	kita.
		our gen student gen all nom came
		‘All of our students came.’
	f.	Uti no	gakusei no	hanbun ga	kita.
		our gen student gen half			nom came
		‘Half of our students came.’

(135)	a.	Taroo wa		ano kenkyuusyo no	menbaa no	suu-nin	ni
		Taro	top that laboratory		gen member gen several-cl dat
		zibun no	ronbun o	okutta.
		self	gen paper	acc sent
		‘Taro sent his paper to several of that laboratory’s members.’
	b.	Taroo wa		ano kenkyuusyo no	menbaa no	tasuu ni	zibun
		Taro	top that laboratory		gen member gen many dat self
		no	ronbun o	okutta.
		gen paper	acc sent
		‘Taro sent his paper to many of that laboratory’s members.’
	c.	Taroo wa		ano kenkyuusyo no	menbaa no	dare ka ni
		Taro	top that laboratory		gen member gen who p		dat
		zibun no	ronbun o	okutta.
		self	gen paper	acc sent
		‘Taro sent his paper to one of that laboratory’s members.’
	d.	Taroo wa	ano kenkyuusyo no		menbaa no	zen-in	ni
		Taro	top that laboratory		gen member gen all-member dat
		zibun no	ronbun o	okutta.
		self	gen paper	acc sent
		‘Taro sent his paper to all of that laboratory’s members.’
	e.	Taroo wa		ano kenkyuusyo no	menbaa no	subete ni	zibun
		Taro	top that laboratory		gen member gen all	dat self
		no	ronbun o	okutta.
		gen paper	acc sent
		‘Taro sent his paper to all of that laboratory’s members.’
	f.	Taroo wa		ano kenkyuusyo no	menbaa no	5% ni	zibun
		Taro	top that laboratory		gen member gen 5% dat self
		no	ronbun o	okutta.
		gen paper	acc sent
		‘Taro sent his paper to 5% of that laboratory’s members.’

Alternatively, the NP + CM + QE pattern (i.e., (48c)) allows us to express the partitive meaning; see (136).

(136)	a.	Kansyu ga		yosomi	o	siteiru uti	ni	syuuzin ga
		guard	nom look:away acc do				while dat prisoner nom
		hito-ri	nigedasita to	siyoo.
		one-cl escape	comp suppose
		‘Suppose that one of the prisoners escapes while the guard is looking away.’
		Soosuruto hoka no		syuuzin mo		zibun mo	itu	ka wa
		if:so	other gen prisoner also self				also when p		top
		nigerareru	to	kangaedasu	daroo.
		able:to:escape comp begin:to:think probably
		‘Then the other prisoners would start thinking that they can also escape someday.’
	b.	Kono kurasu kara	gakusei o			san-nin	erande kudasai.
		this	class	from student	acc three-cl select		please
		‘Please select three of the students from this class.’

It is also reported in Inoue (1978) that the combination of the (48a) and (48c) patterns, i.e., [QE + no + NP] + CM+ QE may express the partitive meaning; see (137).

(137)	(= Inoue 1978:175 [36])
	[Narande hasitteita suu-dai		no	torakku] ga	gaadoreeru ni
	lined:up	running	several-cl gen truck	nom guardrail	dat
	san	yon-dai	butukatta.
	three four-cl struck
	‘Three or four of the several trucks running abreast struck the guard rail.’

10.6 Expressions Involving Negation

10.6.1 Decreasing D-Quantifiers

First of all, there are no D-quantifiers in Japanese that correspond to no NP or few NP in English. To express what no NP means in Japanese, we may use a wh-word plus the particle mo co-occurring with negation, as schematized in (138).

(138)

[ ... wh-word + mo ... Verb + Neg ], where (i) wh-word + mo, and (ii) Neg are clause-mates

For example, in expressing what the sentences in (139) mean, we may use the sentences in (140).

(139)	a.	No student did homework.
	b.	John visited no place.
	c.	Mary applied to no company.

(140)	a.	Dono		gakusei mo	syukudai	o	si-nak-atta.
		which student also homework acc do-neg-past
		‘No student did homework.’
	b.	John wa	doko	ni mo	ik-anak-atta.
		John top where at also go-neg-past
		‘John did not go to any place.’
	c.	Mary wa	dono	kaisya	ni	mo	oobosi-nak-atta.
		Mary top which company dat also apply-neg-past
		‘Mary did not apply to any company.’

We can also use the form in (141) to express the meaning of no NP; see the illustrations in (142).

(141)

[ ... NP-CM ... 1-classifier + mo .... Verb + Neg], where (i) the NP, (ii) 1-classifier + mo, and (iii) Neg are clause-mates

(142)	a.	Yukiko wa	hon	o	is-satu	mo	yom-anak-atta.
		Yukiko top book acc one-cl also read-neg-past
		‘Yukiko did not read even one book.’
	b.	Gakusei ga	hito-ri	mo	ko-nak-atta.
		student	nom one-cl also come-neg-past
		‘No student came.’

To express what few NP means, we use a similar strategy; we use the form in (143), where hotondo roughly means almost all.

(143)

[ ... hotondo wh-word + mo ... Verb + Neg ], where (i) wh-word + mo and (ii) Neg are clause-mates

The sentences in (144) are, for example, translated into the sentences in (145).

(144)	a.	Few students did homework.
	b.	John visited few places.
	c.	Mary applied to few companies.

(145)	a.	Hotondo dono	gakusei mo	syukudai	o	si-nak-atta.
		almost which student	also homework acc do-neg-past
		‘Almost no student did homework.’
	b.	John wa	hotondo doko	ni	mo	ik-anak-atta.
		John top almost		where dat also go-neg-past
		‘John went to few places.’
	c.	Mary wa	hotondo dono kaisya	ni	mo	oobosi-nak-atta.
		Mary top almost	which company dat also apply-neg-past
		‘Mary applied to few companies.’

Having no D-quantifiers that correspond to no NP or few NP, one might wonder if Japanese has any decreasing D-quantifiers. We maintain that there are none. One might argue that an expression that corresponds to less than ♯ NP would be one such candidate. For example, the sentences in (146) arguably correspond to (147).³⁰

(146)	a.	Yukiko wa	san-satu	ika	no	hon	o	yonda.
		Yukiko top three-cl equal:less gen book acc read
		‘Yukiko read three or less books.’ (Cf. (48a).)
	b.	Yukiko wa	hon	o	san-satu ika		sika
		Yukiko top book acc three-cl equal:less only
		yom-anak-atta.
		read-neg-past
		‘Yukiko only read three or less books.’ (Cf. (48c).)

(147)

Yukiko read less than three books.

However, unlike (147), the sentences in (146) entail that Yukiko read some books. We thus observe the contrast between (148) and (149).

(148)	a.	Yukiko wa	san-satu ika	no	hon	o	yonda.
		Yukiko top three-cl equal:less gen book acc read
		♯Zitu wa	is-satu	mo	yom-anak-atta	no	da.
		truth top one-cl also read-neg-past comp copula
		‘Yukiko read three books or less. In fact, she did not read even one book.’
	b.	Yukiko wa	hon	o	san-satu	ika	sika
		Yukiko top book acc three-cl equal:less only
		yom-anak-atta.
		read-neg-past
		♯Zitu wa	is-satu	mo	yom-anak-atta	no	da.
		truth top one-cl also read-neg-past comp copula
		‘Yukiko only read three books or less. In fact, she did not read even one book.’

(149)

Yukiko read less than three books. Actually, she read no book.

10.6.2 NPIs and Negation-Sensitive Items

We now turn to consider what items can be considered NPIs in Japanese. As we have seen above (see (138), (141) and (143)), certain sets of items must co-occur with negation to give rise to a ‘special meaning’. As illustrated below, we can also identify items whose presence requires negation. (In Japanese, when negation is required, it must be a clause-mate of the relevant item(s); cf. McGloin 1976; Kato 1994.) However, we are not completely sure that they are indeed NPIs for the following reason. Since NPIs are by definition items that require a downward entailing environment, the presence of decreasing quantifiers is required to determine if a given item is an NPI (rather than an item merely requiring the presence of negation). Unfortunately, however, Japanese lacks decreasing quantifiers; see Section 10.6.1.

During the remainder of this section, we introduce several items whose presence requires negation. First, the expressions of the form in (150) require negation; see the illustrations in (151).

(150)	1-classfifier + mo
	E.g., hito-tu mo, ik-ko mo, hito-ri mo, iti-dai mo, etc.

(151)	a.	Paatii ni	gakusei ga	hito-ri	mo	{ko-nak-atta /	*kita}.
		party dat student nom one-cl also come-neg-past came
		‘To the party, no student came.’
	b.	John wa	hon	o	is-satu	mo	{yom-anak-atta / *yonda}.
		John top book acc one-cl also read-neg-past read
		‘John did not read even one book.’

Second, when the expressions of the form in (152) are used without a case-marker, negation is usually required; see (153).

(152)	NP + 1-classifier
	E.g., hanataba ito-tu ‘one bouquet’, hon is-satu ‘one book’, kuruma iti-dai ‘one car’

(153)	a.	Watasi no		kare	wa	hanataba hito-tu	{kure-nai /
		I	gen boy:friend top bouquet one-cl give-neg
		kure-nak-atta /		*kureru / *kureta}.
		give-neg-past give		gave
		‘My boyfriend {has not given / did not give / gives / gave} me even one bouquet.’
	b.	Taroo wa		hon	is-satu	yomooto {si-nai / si-nak-atta /
		Taro	top book one-cl read			do-neg do-neg-past
		*suru / *sita}.
		do	did
		‘Taro {does not read / did not read / reads / read} even one book.’

There are also a number of adverbs that need negation. Some of them are listed in (154); the sentences in (155) provide illustrations.

(154)	{zenzen / mattaku} + negation ‘not at all’
	toutei + negation ‘no matter what one does’
	{mettani / hotondo} + negation ‘hardly’
	{amari / sahodo/ sonnani} + negation ‘not much/many’
	dateni + negation ‘with one’s efforts not wasted’
	nidoto + negation ‘never’

(155)	a.	Tookyoo daigaku	ni	wa	toutei	{haire-nai /
		Tokyo	university dat top no:matter:what able:enter-neg
		*haireru} to	omoimasu.
		able:enter comp think
		‘I think I cannot get into the University of Tokyo no matter what I do.’
	b.	Anna	otoko to	wa	nidoto {asobi-masen / *asobimasu}.
		that:kind man	with top never play-neg			play
		‘I will never play with that kind of man.’

10.7 So-Called Focus-Sensitive Particles

In Japanese, there is a class of expressions that arguably correspond to the English so-called focus-sensitive particles such as only, even, and also. The meanings of so-called focus-sensitive particles are said to be focus-sensitive, i.e., to make reference to a set of alternative choices under consideration (cf. Kuroda 1965; Jackendoff 1972; Rooth 1985, 1992). Hereafter, we refer to such expressions as FPs. (156) presents a partial list of FPs.³¹

(156)	a.	X dake ‘only X’
	b.	X nomi ‘only X’
	c.	X bakari ‘only X’
	d.	X sae ‘even X’
	e.	X sura ‘even X’
	f.	X made ‘up to X’
	g.	X mo ‘also X’
	h.	X nado ‘X and so on’ or ‘X which/who is low in a list’
	i.	X nanka ‘X, which/who is low in a list’
	j.	X koso ‘X, which/who is the most appropriate or important for the relevant context’
		(Note: X can be an NP, or a phrase other than an NP for some FPs)

The FPs in (156) all generally indicate that X is among the set of alternative choices under consideration. Their distribution is also similar. As mentioned in Section 10.3.1, when they attach to an NP-α unit where α is a case-marker ga or o, α may not phonologically surface; see (157).

(157)	a.	Kimura sensei	wa	Kentaroo {dake / nomi / bakari / sae /sura /
		Kimura teacher top Kentaro	only	only	only	even even
		made / mo}	suisensita.
		up:to also	recommended
		‘Prof. Kimura recommended {only / only / only / even / even / up to
		/ also} Kentaro.’
	b.	Satoko {nanka / nado} ki-temo,			kono mondai	wa
		Satoko p	p	come-even:if this	problem top
		kaiketusi-nai.
		solve-neg
		‘(Lit.) The coming of Satoko, who is low in the list, will not solve this problem.’
	c.	John koso seitokaityoo		ni	naru	bekida.
		John p	student:representative dat become should
		‘It is John who should be the student representative.’

In contrast, when FPs are attached to an NP-α unit where α is a postposition or a case-marker other than ga and o, α must surface. This is illustrated in (158)–(160).

(158)	a.	Kimura sensei		wa	Kentaroo to	{dake / ?nomi / bakari / sae /
		Kimura teacher top Kentaro		with only	only	only	even
		sura / ?made / mo} ronbun o		kaita.
		even	up:to	also paper	acc wrote
		‘Prof. Kimura wrote papers {only / only / only / even / even / up to / also} with Kentaro.’
	b.	*Kimura sensei wa Kentaroo {dake / nomi / bakari / sae /sura / made / mo} ronbun o kaita.

(159)	a.	Watasi wa	Siroo to	{nanka / nado} dansusi-nai.
		I	top Shiro with p	p	dance-neg
		‘I do not dance with Shiro, who is low in the list.’
	b.	*Watasi wa Siroo {nanka/ nado} dansusi-nai.

(160)	a.	Bill wa		Suzuki sensei	ni	koso suisenzyoo	o
		Bill top Suzuki teacher dat p				recommendation:letter acc
		tanomu bekidatta.
		ask	should:have
		‘It is Prof. Suzuki from whom Bill should have solicited a letter of recommendation.’
	b.	*Bill wa Suzuki sensei koso suisenzyoo o tanomu bekidatta.

We have observed in (158a), (159a), and (160a) that when the FPs listed in (156) attach to an NP-CM unit, they can all follow the whole unit. Some of the listed expressions can also be inserted between the NP and the CM of the NP-CM unit. As illustrated in (161), dake, nomi, bakari, nado, and nanka can occur in such a position.

(161)	a.	Kimura sensei		wa	Kentaroo {dake / nomi / bakari} to	ronbun
		Kimura teacher top Kentaro	only	only	only	with paper
		o	kaita.
		acc wrote
		‘(Lit.) Prof. Kimura wrote papers with only Kentaro.’
	b.	Watasi wa		Siroo {nanka / nado} to	dansusi-nai.
		I	top Shiro p	p	with dance-neg
		‘I do not dance with Shiro {who is low in the list / and so on}.’

Sae, sura, and koso seem to depend on the speaker we consult with; some say that the sentences in (162) are as acceptable as those in (161), but others find them marginal or unacceptable.³²

(162)	a.	OK /?? /*Kimura sensei		wa	Kentaroo {sae / sura} to		ronbun
		Kimura	teacher top Kentaro		even	even	with paper
		o	kaita.
		acc wrote
		‘(Lit.) Prof. Kimura wrote papers with even Kentaro.’
	b.	OK /?? /*Bill wa		Suzuki sensei		koso ni
		Bill	top Suzuki teacher p			dat
		suisenzyoo		o	tanomu bekidatta
		recommendation:letter acc ask				should:have
		‘It is Prof. Suzuki from whom [we thought] Bill should have solicited a letter of recommendation.’

It is clear that made and mo cannot occur between the NP and α; for example, no speaker finds the sentence in (163) to be acceptable.³³

(163)	*Kimura sensei	wa	Kentaroo {made/ mo} to	ronbun o	kaita.
	Kimura teacher top Kentaro		up:to	also with paper	acc wrote
	‘(Lit.) Prof. Kimura wrote papers with {up to/ also} Kentaro.’

As we indicated in (156) above, some of the FPs may also appear next to a phrase other than an NP; see (164).

(164)	a.	John wa	[VP onnanoko o			oikakete]-bakari-de		amari
		John top	girl	acc chase-only-copula much
		benkyoosi-nai.
		study-neg
		‘John always pays attention to girls, and does not study much.’
	b.	(= Masuoka and Takubo 1989:137 [55])
		Suzuki san no		kotoba wa	[AdvP sugasugasiku]-sae-atta.
		Suzuki Mr. gen word		top		refreshing-even-existed
		‘Mr. Suzuki’s words were even refreshing.’
	c.	(= Masuoka and Takubo 1989:137 [56])
		Hanako wa	sono ue	[AdjP kinbende]-mo-aru.
		Hanako top that	top		diligent-also-exist
		‘In addition, Hanako is also diligent.’
	d.	(Based on Masuoka and Takubo 1989:137 [58])
		Konkai	wa	[VP A sya	no	syatyoo	to	menkaisita]
		this:time top		A company gen president with met
		dake desu.
		only copula
		‘For this [trip], I only met the CEO of Company A.’
	e.	[VP Syazaizyoo o		kaitari]-{nanka / nado} -si-temo,
		apology	acc write-		p	p	-do-even:if
		yurusitemorae-nai daroo.
		is:forgiven-neg	likely
		‘Even if [we] write a letter of apology {which is low in the list/ and so on}, [we] will not be able to be forgiven.’

We have noted above that any of the FPs in (156) generally indicates that the denotation of its sister is among the set of alternative choices under consideration. In accordance with this characterization, when the denotation of an NP is among the alternative choices, an FP cannot be attached to a VP that contains it, and conversely, when the denotation of a VP is among the alternative choices, an FP generally cannot be attached to an NP within it. For example, in direct response to (165), (167a) is felicitous while (167b) is not. By contrast, in directly replying to (166), (167b) is appropriate while (167a) is not.

(165)	Tookyoo ni		itte iroirona hito	ni	au	tte	itteta
	Tokyo	dat go	various	people dat meet comp were:saying
	kedo,	kekkyoku dare ni		attekita		no	desu	ka.
	although after:all		who dat met:come:back comp copula q
	‘You were saying that you would meet various people in Tokyo. But who did you end up meeting?’

(166)	Tookyoo ni		itte iroirona koto	o	suru tte	itteta	kedo,
	Tokyo	dat go	various	things acc do	comp were:saying although
	kekkyoku nani	o	sitekita	no	desu	ka.
	after:all	what acc do:come:back comp copula q
	‘You were saying that you would do various things in Tokyo. But what did you end up doing?’

(167)	a.	[Noriko] sae	ni	attekimasita.
		Noriko	even dat meet:came:back
		‘I met even Noriko.’
	b.	[Noriko ni	ai]-sae-sitekimasita.
		Noriko	dat meet-even-do:came:back
		‘I even met Noriko.’

There are, however, examples that appear to be contrary to the above characterization of FPs—a given FP generally indicates that the denotation of its sister is among the set of alternative choices under consideration; see the sentences in (168).

(168)	a.	(= Kuroda 1965:81 [20], slightly adapted)
		(Zyuu-nen tatte) [musuko mo			daigaku	ni	hairi], [musume mo
		10-year	past	son	also university dat enter	daughter	also
		yome ni itta].
		bride dat went
		‘(In the last ten years) my son also got into a university, and my daughter also got married.’
	b.	(= Aoyagi 2006:122 [6b], slightly adapted)
		Kinoo	no	konpa de mada miseinen no		Taroo wa	[sake
		Yesterday gen party	at	still	non:adult gen Taro		top alcohol
		o	nonda] bakarika [tabako sae sutta].
		acc drank	not:only	cigarette even smoked
		‘At the party yesterday, Taro, who is not yet legally an adult, not only drank alcohol but also even smoked cigarettes.’

Finally, we note that some of the FPs listed in (156) can attach to the QE in the NP + CM + QE pattern (i.e., (48c)), as illustrated in (169).

(169)	a.	Satoo sensei		wa	gakusei o	san-nin	dake suisensita.
		Sato	teacher top student			acc three-cl only recommended
		‘Prof. Sato recommended three students only.’
	b.	Kinoo	no	paatii ni	onnanoko ga	20-nin mo	kita.
		yesterday gen party dat girl				nom 20-cl also came
		‘To yesterday’s party, as many as 20 girls came.’

10.8 Floating Quantifiers

By definition, floating quantifiers are those quantifiers which are phonologically separated from the NPs they modify. According to this definition, Japanese has floating quantifiers. For example, we have seen in Sections 10.2, 10.3, and 10.4 that we can construct intersective, universal, and proportional D-quantifier analogues with the NP + CM + QE pattern (i.e., (48c)). With this pattern, since the CM intervenes between the NP and the QE, the QE is characterized as a floating quantifier. We can also find clearer cases: the QE can modify the NP in the configurations in (170). This is illustrated in (171)–(173).

(170)	a.	... NP + CM ... α ... QE ..., where α is any phrase
	b.	... QE ... α ... NP + CM ..., where α is any phrase

(171)	Intersective D-quantifier analogues:
	a.	Dansigakusei ga			kinoo	san-nin	kita.
		male:student	nom yesterday three-cl came
		‘Three male students came yesterday.’ (Cf. (50c).)
	b.	San-nin kinoo dansigakusei ga kita.
	c.	John wa	hon	o	sensyuu	sukunakutomo suu-satu	yonda.
		John top book acc last:week at:least					several-cl read
		‘John read at least several books last week.’ (Cf. (51c).)

(172)	Universal D-quantifier analogues:
	a.	Dansigakusei ga	kinoo	subete eiga	ni	ikitagatteita.
		male:student	nom yesterday all		movie dat wanted:to:go
		‘All the male students wanted to go to a movie yesterday.’ (Cf. (80c).)
	b.	Kimura bengosi wa	seiyakugaisya		o	sakunen zen-bu
		Kimura attorney top pharmaceutical:company acc last:year all-cl
		uttaeta.
		sued
		‘Attorney Kimura sued all pharmaceutical companies last year.’ (Cf. (79c).)
	c.	Kimura bengosi wa zen-bu sakunen seiyakugaisya o uttaeta.

(173)	Proportional D-quantifier analogues:
	a.	Dansigakusei ga		sengetu	hotondo tesuto ni	otita.
		male:student	nom last:month most		test	dat failed
		‘Most male students failed the test last month.’ (Cf. (126c).)
	b.	??Hotondo sengetu dansigakusei ga tesuto ni otita.
	c.	Suzuki sensei	wa	gakusei o	sengetu	iti-bu	suisensita.
		Suzuki teacher top student			acc last:month one-cl recommended
		‘Prof. Suzuki recommended one portion of the students last month.’
		(Cf. (127c).)

It should be noted, however, that a given QE can float only if the NP it modifies is in a particular relation to its clause-mate verb. The NPs that are marked with the nominative or accusative marker always allow their QEs to float; see (171)–(173). But those that are marked with the dative marker do so only occasionally; see (174).

(174)	a.	Kimura sensei		wa	kondo	no	paatii no	tameni gakusei
		Kimura teacher top this:coming gen party gen for						student
		ni	san-nin	dezaato o		mottekosaseta.
		dat three-cl dessert			acc made:bring
		‘Prof. Kimura made three students bring something for dessert.’
	b.	(= Shibatani 1978:352 [41b], slightly adapted)
		Boku wa	kankoku de gengogakusya ni			go	roku-nin
		I	top Korea		at linguist	dat five six-cl
		syookaisareta.
		was:introduced
		‘In Korea I was introduced to five, six linguists.’
	b.	*Suguru wa	sensei	ni	san-nin	aisatusita.
		Suguru top teacher dat three-cl greeted
		‘Suguru greeted three teachers.’

The NPs with other case-markers or postpositions rarely permit their QEs to float; see (175).³⁴

(175)	a.	*Kenta wa	onnanoko to	san-nin	dansusita.
		Kenta top girl	with three-cl danced
		‘Kenta danced with three girls.’
	b.	*Mariko wa	konpyuutaa de	san-dai	tegami o	kaita.
		Mariko top computer		with three-cl letter	acc wrote
		‘Mariko wrote letters with three computers.’

10.9 Scope Interaction

Having described various phenomena involving a QE in Japanese, we now proceed to examine the scope interaction among scope-bearing elements. In what follows, we use the term QNPs to cover D-quantifier analogues in general. We discuss three types of scope interaction: (i) among QNPs (Section 10.9.1), (ii) between QNPs and wh-words (Section 10.9.2), and (iii) between QNPs and negation (Section 10.9.3). We confine our discussion to the base order, leaving aside scope interaction in the scrambling construction. We choose to limit our discussion thus, as we believe it is more beneficial to emphasize those factors which need to be considered when discussing scope interaction generally than to make a rough and hasty sketch of scope interaction in various constructions. For an in-depth assessment of scope interaction in the scrambled order, please see Hayashishita (2000a, 2004). Nor do we discuss in this section the scope interaction among FPs, as it requires much background information. (We do, however, briefly touch on this matter in Section 10.10.1 when we introduce Type (2) quantifiers.) Readers who are interested in the scope interaction among FPs in Japanese may wish to consult Hayashishita (2011), which includes a detailed study on this topic.

10.9.1 Among QNPs

As we have seen in Sections 10.2, 10.3, and 10.4 above, intersective, universal, and propositional D-quantifier analogues may have the forms in (48), repeated here. We thus first describe the scope interaction among QNPs, referring to these forms.

(48)	a.	QE + no + NP + CM
	b.	NP + QE + CM
	c.	NP + CM + QE

In the configuration in (176), α is able to take wide scope with respect to β, whether α and β take the form of (48a), (48b), or (48c). Hereafter, the reading where α takes wide scope with respect to β is referred to as the surface scope reading.

(176)

[... α-ga ... β-ni /o ...], where α and β are QNPs and clause-mates

Let us illustrate this with respect to intersective, universal, and proportional D-quantifier analogues. For example, (177) illustrates cases where α is an intersective D-quantifier, and any combinations of the subject and object items can be taken to mean (178).

(177)	{Suu-nin	no	kyoozyu ga /		Kyoozyu suu-nin		ga /	Kyoozyu
	several-cl gen professor nom professor several-cl nom professor
	ga	suu-nin}	{san-nin	izyoo	no	gakusei o /		gakusei
	nom several-cl three-cl equal:more gen student						acc student
	san-nin	izyoo	o /	gakusei	o	san-nin	izyoo}
	three-cl equal:more acc student acc three-cl equal:more
	suisensita.
	recommended
	‘Several professors recommended three or more students.’

(178)

There are several professors such that each of them recommended three or more students.

(179) further illustrates that in the configuration of (176), α can take wide scope with respect to β when α is an intersective D-quantifier analogue; any combinations in (179) give rise to (180).

(179)	{Go-nin izyoo			no	kyoozyu	ga /	Kyoozyu go-nin
	five-cl equal:more gen professor nom professor five-cl
	izyoo		ga /	Kyoozyu ga	go-nin	izyoo}	{san-nin
	equal:more nom professor nom five-cl equal:more three-cl
	izyoo		no	gakusei o /	gakusei san-nin	izyoo	o /
	equal:more gen student		acc student	three-cl equal:more acc
	gakusei o		san-nin		izyoo}	suisensita.
	student	acc three-cl equal:more recommended
	‘Five or more professors recommended three or more students.’

(180)

There are five or more professors such that each of them recommended three or more students.

(181) presents cases where α is a universal D-quantifier analogue; any combinations of the subject and object items in (181) can give rise to (182).

(181)	{Subete no		kyoozyu	ga /	Kyoozyu subete ga /	Kyoozyu ga
	all	gen professor nom professor all				nom professor nom
	subete}	{san-nin izyoo	no	gakusei o /	gakusei san-nin
	all	three-cl equal:more gen student			acc student three-cl
	izyoo	o /	gakusei o	san-nin	izyoo}	suisensita.
	equal:more acc student acc three-cl equal:more recommended
	‘All professors recommended three or more students.’

(182)

For each professor, there are three or more students that he/she recommended.

Cases where α is a proportional D-quantifier analogue are illustrated in (183); any combinations in (183) can all be taken to mean (184).

(183)	{Sanbun no		iti	no	kyoozyu	ga /	Kyoozyu sanbun no	iti
	third	gen one gen professor nom professor third					gen one
	ga /	Kyoozyu ga		sanbun no	iti}	{san-nin	izyoo	no
	nom professor nom third				gen one three-cl equal:more gen
	gakusei o /	gakusei san-nin				izyoo	o /	gakusei o
	student acc student three-cl equal:more acc student acc san-nin izyoo} suisensita.
	three-cl equal:more recommended
	‘One third of the professors recommended three or more students.’

(184)

For one third of the professors, each of them recommended three or more students.

We now consider the availability of the readings where β takes wide scope with respect to α in (176), repeated here. This reading is referred to as the inverse scope reading below.

(176)

[... α-ga ... β-ni /o ...], where α and β are QNPs and clause-mates

Inverse scope readings seem more difficult to detect than surface scope readings. In fact, Kuroda (1969/1970) and Hoji (1985) maintain that inverse scope readings are impossible. Recently, however, a number of researchers reported that they are detectable (cf. Kitagawa 1990; Kuroda 1994; Kuno et al. 1999; Hayashishita 1999, 2000b, 2004; Hoji 2003b). In fact, we may detect an inverse scope reading if β has the QE + no + NP + CM form (i.e., (48a)) or the NP + QE + CM form (i.e., (48b)).

For illustrations, imagine the situation in (185).

(185)

You are a department administrative staff member. The head of the department asks you to count the number of students who have received recommendation from three or more professors. You check the students one by one. Did John get recommendation from three or more professors? How about Mary? And so on. You then reply to the head, saying ...

If any combinations of the subject and object items in (186) are uttered in this situation, they are understood to mean (187).

(186)	{San-nin izyoo		no	kyoozyu	ga /	Kyoozyu san-nin
	three-cl equal:more gen professor nom professor three-cl
	izyoo	ga /	Kyoozyu ga		san-nin	izyoo}	{suu-nin
	equal:more nom professor nom three-cl equal:more several-cl
	no	gakusei o /	gakusei suu-nin		o}	suisensimasita.
	gen student	acc student several-cl acc recommended
	‘Three or more professors recommended several students.’

(187)

There are several students such that each of them is recommended by three or more professors.

Similarly, in the same context, any combinations in (188), those in (190), and those in (192) give rise to (189), (191), and (193), respectively.

(188)	{San-nin izyoo		no	kyoozyu ga /		Kyoozyu san-nin
	three-cl equal:more gen professor nom professor three-cl
	izyoo	ga /	Kyoozyu	ga	san-nin	izyoo}	{go-nin
	equal:more nom professor nom three-cl equal:more five-cl
	izyoo	no	gakusei o /	gakusei go-nin		izyoo	o}
	equal:more gen student acc student five-cl equal:more acc
	suisensimasita.
	recommended
	‘Three or more professors recommended five or more students.’

(189)

There are five or more students such that each of them is recommended by three or more professors.

(190)	{San-nin izyoo		no	kyoozyu	ga /	Kyoozyu san-nin
	three-cl equal:more gen professor nom professor three-cl
	izyoo	ga /	Kyoozyu ga		san-nin	izyoo}	{subete no
	equal:more nom professor nom three-cl equal:more all						gen
	gakusei o /	gakusei subete o}			suisensimasita.
	student	acc student all		acc recommended
	‘Three or more professors recommended all the students.’

(191)

For each student, there are three or more professors who recommended him/her.

(192)	{San-nin izyoo			no	kyoozyu	ga /	Kyoozyu san-nin
	three-cl equal:more gen professor nom professor three-cl
	izyoo		ga /	Kyoozyu ga		san-nin	izyoo}	{sanbun no
	equal:more nom professor nom three-cl equal:more third							gen
	iti	no	gakusei o /	gakusei sanbun no		iti	o}	suisensimasita.
	one gen student acc student third					gen one acc recommended
	‘Three or more professors recommended one third of the students.’

(193)

For one third of the students, each of them was recommended by three or more professors.

Liu (1990) notes that, in English, when β in (176) is a certain type of QNP, the inverse scope reading is difficult to obtain. Among the items that do not support inverse scope readings, she lists decreasing QNPs and one type of intersective D-quantifiers called modified numerals. Modified numerals are expressions such as more than three boys and at least five girls, where an NP accompanied with a number phrase is modified by some expression. We cannot comment on her remark about decreasing QNPs because they are absent in Japanese (see Section 10.6.1 above), but her remark about modified numerals initially appears to extend to Japanese. When speakers are asked to provide judgments without any accompanying context, they generally have more difficulty in detecting an inverse scope reading with modified numerals than with other QNPs. As we alluded to above, however, if an appropriate context is given, the difficulty which speakers experience with modified numerals disappears—with the context in (185), we detect an inverse scope reading in (188) as easily as in the other sentences.

Hayashishita (2004, 2010) attempts to describe this state of affairs. One way to state Hayashishita’s claim is (194).

(194)

In [... α-ga ... β-ni /o ...], where α and β are QNPs and clause-mates, β takes wide scope with respect to α only if in the relevant context, there is one and only one set of individuals that can possibly be the extension of β.

We hereafter call the condition embedded in (194) the unique set condition, which we claim to be a necessary condition for the inverse scope reading. For example, no combinations of the subject and object items in (195) give rise to an inverse scope reading (cf. (188)).

(195)	Maitosi		{san-nin	izyoo		no	kyoozyu	ga /	kyoozyu
	every:year three-cl equal:more gen professor nom professor
	san-nin	izyoo		ga /		kyoozyu	ga	san-nin	izyoo}
	three-cl equal:more nom professor nom three-cl equal:more
	{go-nin	izyoo		no	itinensei	o /	itinensei	go-nin	izyoo
	five-cl equal:more gen freshman acc freshman five-cl equal:more
	o}	suisensimasu.
	acc recommend
	‘Every year, three or more professors recommend five or more freshmen.’

According to Hayashishita (2010), this is because the value of the first-year students changes each year—because the unique set condition cannot be met.

Regarding the issue of why people generally have more difficulty in initially detecting inverse scope readings with modified numerals than with the other types of QNPs, Hayashishita (2010) states the following. For some types of QNPs including modified numerals, the unique set condition cannot be satisfied by their lexical meanings alone—in the discourse domain that includes a lot of individuals, there is more than one set of individuals which can serve as their extensions. In other words, in those cases, to meet the unique set condition, the context must play a role. With modified numeral cases, to imagine a context which singles out one and only one set of individuals is especially difficult. If the judgments of a speaker are solicited out of blue, he/she is likely not to imagine a necessary context. It is hence expected that if no appropriate context is provided, people generally fail to detect an inverse scope reading with modified numerals.

Here we wish to reiterate that the detection of surface scope readings is much easier than that of inverse scope readings. Plus, as pointed out by Hayashishita (2004, 2010), the unique set condition does not apply to surface scope readings—α in the configuration of (176) can take wide scope with respect to β even if there is more than one set of individuals that can possibly be the extension of α. For example, any combinations of the subject and object items in (196) can give rise to a surface scope reading.

(196)	Maitosi		{san-nin	izyoo	no	sinnin		kyooin ga /
	every:year three-cl equal:more gen newly:hired teacher nom
	sinnin		kyooin san-nin		izyoo	ga /	sinnin	kyooin
	newly:hired teacher three-cl equal:more nom newly:hired teacher
	ga	san-nin		izyoo}	{go-nin izyoo		no	gakusei o /
	nom three-cl equal:more five-cl equal:more gen student acc
	gakusei go-nin			izyoo	o /	gakusei o	go-nin	izyoo}
	student five-cl equal:more acc student acc five-cl equal:more
	suisensimasu.
	recommend
	‘Every year three or more newly hired teachers recommend five or more students.’

One might thus wish to analyze surface scope readings differently from inverse scope readings (cf. Ben Shalom 1993; Hayashishita 2004, 2010).

As we alluded earlier, the inverse scope reading is not possible if β in (176) occurs in the NP + CM + QE form (i.e., (48c)). Any combinations of the subject and object items in (197), those in (198), those in (199), and those in (200), for example, fail to give rise to (187), (189), (191), and (193), respectively.

(197)	{San-nin izyoo		no	kyoozyu	ga /	Kyoozyu san-nin
	three-cl equal:more gen professor nom professor three-cl
	izyoo	ga /	Kyoozyu ga	san-nin	izyoo}	gakusei o
	equal:more nom professor nom three-cl equal:more student acc
	suu-nin	suisensita.
	several-cl recommended
	‘Three or more professors recommended several students.’

(198)	{San-nin izyoo			no	kyoozyu	ga /	Kyoozyu san-nin
	three-cl equal:more gen professor nom professor three-cl
	izyoo		ga /	Kyoozyu ga	san-nin	izyoo}	gakusei o
	equal:more nom professor nom three-cl equal:more student acc
	go-nin	izyoo		suisensita.
	five-cl equal:more recommended
	‘Three or more professors recommended five or more students.’

(199)	{San-nin izyoo			no	kyoozyu ga /	Kyoozyu san-nin
	three-cl equal:more gen professor nom professor three-cl
	izyoo		ga /	Kyoozyu ga	san-nin	izyoo}	gakusei
	equal:more nom professor nom three-cl equal:more student
	o	subete		suisensita.
	acc	all		recommended
	‘Three or more professors recommended all students.’

(200)	{San-nin izyoo			no	kyoozyu	ga /	Kyoozyu san-nin
	three-cl equal:more gen professor nom professor three-cl
	izyoo		ga /	Kyoozyu ga	san-nin	izyoo}	gakusei o
	equal:more nom professor nom three-cl equal:more student acc
	sanbun no		iti	suisensita.
	third	gen one recommended
	‘Three or more professors recommended one third of the students.’

In this paper, we leave open the question as to why the NP + CM + QE form (i.e., (48c)) does not support inverse scope readings. We note, however, that this form is usually unsuitable when there is one and only one possible referent for it (cf. (194)); for example, in the context of (201), (202) is usable while (203) is not.

(201)

The speaker is a father who has three and only three children.

(202)	Kodomo ga	hito-ri	sigotoba	ni	tazunetekita	no
	child	nom one-cl workplace dat come:to:visit comp
	desu.
	copula
	‘One child (of mine) came to my workplace.’

(203)	Kodomo ga	san-nin	sigotoba	ni	tazunetekita	no
	child	nom three-cl workplace dat come:to:visit comp
	desu.
	copula
	‘Three children (of mine) came to my workplace.’

(Incidentally, if the subjects in (202) and (203) are replaced with the QE + no + NP + CM form (i.e., (48a)) or the NP + QE + CM form (i.e., (48b)), the resulting sentences are both usable in the context of (201).)

We now turn to the scope interaction involving existential quantifier analogues built from a wh-word. In this section, we only discuss the reading with an existential quantifier analogue built from a wh-word taking narrow scope, as it is difficult to determine if the reading with an existing quantifier taking wide scope exists. To ensure that the reading with an existential quantifier taking wide scope exists, we need a situation that makes this reading true but makes the reading with the other scope order false. However, any situations which make the reading with an existential quantifier taking wide scope with respect to another quantifier true necessarily make the reading with the other scope order true (cf. Reinhart 1976).

When β in (176), repeated below, is an existential quantifier analogue with a wh-word, α can take wide scope with respect to β, no matter what type of QNP α is and whether or not the unique set condition is met.

(176)

[... α-ga ... β-ni /o ...], where α and β are QNPs and clause-mates

For example, any combinations of the subject and object items in (204), those in (206), those in (208), and those in (210) give rise to (205), (207), (209), and (211), respectively.

(204)	{Suu-nin	no	kyoozyu	ga /	Kyoozyu suu-nin	ga /	Kyoozyu
	several-cl gen professor nom professor several-cl nom professor
	ga	suu-nin}	dare ka o		suisensita.
	nom several-cl who p			acc recommended
	‘Several professors recommended someone.’

(205)

There were several professors such that each of them recommended someone.

(206)	{Go-nin izyoo		no	kyoozyu	ga /	Kyoozyu go-nin
	five-cl equal:more gen professor nom professor five-cl
	izyoo	ga /	Kyoozyu ga	go-nin	izyoo}	dare ka o
	equal:more nom professor nom five-cl equal:more who p acc
	suisensita.
	recommended
	‘Five or more professors recommended someone.’

(207)

There were five or more professors such that each of them recommended someone.

(208)	{Subete no		kyoozyu	ga /	Kyoozyu subete ga /		Kyoozyu ga
	all	gen professor nom professor all				nom professor nom
	subete} dare ka o		suisensita.
	all	who p	acc recommended
	‘All professors recommended someone.’

(209)

For each professor, there was someone whom he /she recommended.

(210)	{Sanbun no		iti	no	kyoozyu	ga /	Kyoozyu sanbun no	iti
	third	gen one gen professor nom professor third						gen one
	ga /	Kyoozyu ga		sanbun no		iti} dare ka o	suisensita.
	nom professor nom third				gen one who p		acc recommended
	‘One third of the professors recommended someone.’

(211)

For one third of the professors, each of them recommended someone.

When α in (176) is an existential quantifier analogue built from a wh-word, β may take wide scope with respect to α. But β has to have the QE + no + NP + CM form (i.e., (48a)) or the NP + QE + CM form (i.e., (48b)), and the unique set condition needs to be met. For example, in the context of (212), all the combinations of the subject and object items in (213), those in (215), those in (217), and those in (219) can be understood to mean (214), (216), (218), and (220), respectively.

(212)

You are a department administrative staff member. The head of the department asks you to count the number of students who have received recommendation from someone. You check the students one by one. Did someone recommend John? How about Mary? And so on. You then reply to the head, saying ...

(213)	Dare ka ga			{suu-nin	no	gakusei o /	gakusei suu-nin
	who	p	nom several-cl gen student acc student several-cl
	o}	suisensiteimasita.
	acc	recommended
	‘Someone recommended several students.’

(214)

There were several students such that each of them was recommended by someone.

(215)	Dare ka ga			{go-nin izyoo	no	gakusei o /	gakusei go-nin
	who	p	nom five-cl equal:more gen student acc student five-cl
	izyoo	o}	suisensiteimasita.
	equal:more acc recommended
	‘Someone recommended five or more students.’

(216)

There were five or more students such that each of them was recommended by someone.

(217)	Dare ka ga		{subete no	gakusei o /	gakusei subete o}
	who	p	nom all	gen student acc student all		acc
	suisensiteimasita.
	recommended
	‘Someone recommended all the students.’

(218)

For each student, there was someone who recommended him/her.

(219)	Dare ka ga			{sanbun no	iti	no	gakusei o /	gakusei sanbun
	who	p	nom third	gen one gen student			acc student third
	no	iti	o}	suisensiteimasita.
	gen one acc recommended
	‘Someone recommended one third of the students.’

(220)

For one third of the students, each of them was recommended by someone.

However, if the objects in the above examples are replaced with the NP + CM + QE form (i.e., (48c)), the inverse scope readings disappear. This is illustrated in (221)–(224).

(221)

Dare ka ga gakusei o suu-nin suisensiteimasita.

(222)

Dare ka ga gakusei o go-nin izyoo suisensiteimasita.

(223)

Dare ka ga gakusei o subete suisensiteimasita.

(224)

Dare ka ga gakusei o sanbun no iti suisensiteimasita.

Let us now come to the scope interaction involving universal quantifier analogues built from a wh-word. Here, we only investigate if they can take wide scope with respect to another quantifier, as it is difficult to examine the availability of the other scope order with simple sentences. When α in the configuration of (176), repeated here, is a universal quantifier analogue built from a wh-word, α can take wide scope with respect to β, no matter what form β has. For example, any combinations of the subject and object items in (225) can be construed as (226).

(176)

[... α-ga ... β-ni /o ...], where α and β are QNPs and clause-mates

(225)	Dono kyoozyu mo		{san-nin	izyoo	no	gakusei o /	gakusei
	which professor also three-cl equal:more gen student acc student
	san-nin	izyoo	o /	gakusei o	san-nin	izyoo}
	three-cl equal:more acc student acc three-cl equal:more
	suisensita.
	recommended
	‘Every professor recommended three or more students.’

(226)

For each professor, there are three or more students whom he /she recommended.

But when β is a universal quantifier analogue built from a wh-word, β cannot take wide scope with respect to α; for example, no combinations of the subject and object items in (227) can be taken to mean (228) even if they are uttered in the context of (185), repeated here.

(185)

You are a department administrative staff member. The head of the department asks you to count the number of students who have received recommendation from three or more professors. You check the students one by one. Did John get recommendation from three or more professors? How about Mary? And so on. You then reply to the head, saying ...

(227)	{San-nin izyoo				no	kyoozyu	ga /	Kyoozyu san-nin
	three-cl equal:more gen professor nom professor three-cl
	izyoo		ga /	Kyoozyu ga	san-nin	izyoo}	dono	gakusei
	equal:more nom professor nom three-cl equal:more which student
	ni	mo	suisenzyoo	o	kakimasita.
	dat also reference:letter acc wrote
	‘Three or more professors wrote a recommendation letter to every student.’

(228)

For each student, there are three or more professors who wrote a recommendation letter to him /her.

10.9.2 Between QNPs and Wh-Words

The scope interaction between QNPs and wh-words is difficult to describe, for we have to investigate it indirectly through possible answers to questions. Here in this paper, following Karttunen (1977), Groenendijk and Stokhof (1984, 1989), and Krifka (2001), among others, we assume that when a given question with the configuration of (229) is answered with a pair-list answer, the QNP takes wide scope with respect to the wh-word (contra Engdahl 1985 and Chierchia 1993, who assume that the wh-word scopes over the QNP, but its trace is a function variable bound by the QNP). The wide scope reading under discussion is referred to as the pair-list reading below.

(229)

[... α-ga ... β-ni /o ...] before wh-movement, where one of α and β is a QNP, and the other is a wh-word

It has been reported that in English, when α is a QNP and β is a wh-word, the question may be answered with a pair-list answer, but it is not so when α is a wh-word and β is a QNP (cf. Chierchia 1993). The same seems true with Japanese; for example, the A-B sequence in (230) is felicitous while that in (231) is not.

(230)	A: {Subete no			gakusei ga /	Gakusei subete ga}	dono	hon
	all	gen student nom student all				nom which book
	o	yonda ka osiete kudasai.
	acc read	q teach please
	‘Please tell me which book every student read.’
	B: Yosio	ga	LGB o,	Suzan ga	Barriers o, ..., sosite Takasi
	Yoshio nom LGB acc Suzan nom Barriers acc and						Takashi
	ga	MP o	yomimasita.
	nom MP acc read
	‘Yoshio read LGB, Suzan Barriers, ..., and Takashi MP.’

(231)	A: Dono		gakusei ga	{subete no	hon	o /	hon	subete o}
	which student	nom all		gen book acc book all				acc
	yonda ka osiete kudasai.
	read	q	teach please
	‘Please tell me which student read every book.’
	B: Yosio ga LGB o, Suzan ga Barriers o, ..., sosite Takasi ga MP o yomimasita.

Regarding what types of QNPs can support pair-list readings, researchers’ positions diverge. Some claim that a wide range of QNPs excepting decreasing QNPs give rise to pair-list readings (cf. Lahiri 2002), while others maintain that only universal quantifiers support them (cf. Groenendijk and Stokhof 1984, 1989; Krifka 2001). (Incidentally, many of the researchers in the latter group acknowledge that other types of QNPs appear to give rise to pair-list readings, but analyze them differently from the ‘genuine’ cases involving a universal quantifier; see for example the discussion in Krifka 2001.) We claim that as far as Japanese is concerned, all types of QNPs may support pair-list readings, provided that they have the QE + no + NP + CM form (i.e., (48a)) or the NP + QE + CM form (i.e., (48b)). (Recall that Japanese does not have decreasing QNPs; see Section 10.6.1.) For example, with the specified contexts, the A-B sequences in (232)–(233) are felicitous.³⁵

(232)	[Context: Person A is interested in knowing which book each student read, and A tries to discover this from Person B. A suggests that B pick any group consisting of several students and start with them.]
	A: Zya, mazu	(dare demo	ii	kara)	{suu-nin	no	gakusei ga /
	then at:first who		even:if good since several-cl gen student nom
	gakusei suu-nin	ga}	dono	hon	o	yonda ka ittemite	kudasai.
	student several-cl nom which book acc read					q	try:to:tell please
	‘Then, as a starter please (pick any group of several students and) tell me which book they read.’
	B: Yosio ga LGB o, Suuzan ga Barriers o, ..., sosite Takasi ga MP o yomimasita.

(233)	[Context: Person A is interested in knowing which book each student read, and A tries to discover this from Person B. A suggests that B pick any group consisting of one third of the students, and start with them.]
	A: Zya, mazu		(dare demo	ii	kara) {sanbun no	iti	no	gakusei
	then at:first who even:if good since third					gen one gen student
	ga /	gakusei sanbun no	iti	ga}	dono	hon	o	yonda ka
	nom student third		gen one nom which book acc read q
	ittemite	kudasai.
	try:to:tell please
	‘Then, as a starter please (pick any group of one third of the students and) tell me which book they read.’
	B: Yosio ga LGB o, Suuzan ga Barriers o, ..., sosite Takasi ga MP o yomimasita.

But we point out that, just like inverse scope readings, pair-list readings require that the unique set condition be met. In fact, in the contexts of (230), (232), and (233), in answering A’s question, B associates one and only one set of students with the subject QNP. In the context of (234), on the other hand, A’s question is not about one particular group of students. Thus, in answering A’s question, B would have more than one set of students that can possibly be the extension of the relevant QNP.

(234)	B: Watasi wa		donna	hon	ga	ninki	ga	atta	ka nado
	I	top what:kind book nom popularity nom existed q so:on
	zyookyoo o	yoku haakusiteiru tumori desu.
	situation	acc well know			assume copula
	‘I am aware of the situation well such as what kinds of books are popular.’
	A: Zya, ninki		no	aru	hon	o	siritai		kara,
	then popularity gen exist book acc want:to:know because
	{suu-nin	no	gakusei ga /		gakusei suu-nin		ga}	dono	hon
	several-cl gen student nom student several-cl nom which book
	o	yonda ka osiete kudasai.
	acc read	q	tell	please
	‘Then, since I would like to know what is popular, please tell me which book several students read!’
	B: Yosio ga LGB o, Suuzan ga Barriers o, ..., sosite Takasi ga MP o yomimasita.

In this situation, pair-list answers are not possible—the B-A-B sequence in (234) is not felicitous. Here B must reply to A’s request with a single constituent answer such as (235).

(235)	(Suu-nin	no	gakusei ga	yonda no	wa)	LGB desu.
	several-cl gen student nom read				comp top LGB copula
	‘(What several students read was) LGB.’

One may thus suggest that pair-list readings must be analyzed on a par with inverse scope readings (cf. Hayashishita 2004). In fact, the resemblance between them can be demonstrated extensively. Recall that inverse scope readings are not possible if the wide-scope taking expression (i.e., β in (176), repeated here) has the NP + CM + QE form (i.e., (48c)).

(176)

[... α-ga ... β-ni /o ...], where α and β are QNPs and clause-mates

Similarly, pair-list readings are not possible if the relevant QNP has the NP + CM + QE form. If the A’s utterances in (230), (232), and (233) are replaced with (236), (237), and (238), respectively, the following B’s utterances become infelicitous. After (236), (237), and (238), B must reply with a single constituent answer.

(236)

Gakusei ga subete dono hon o yonda ka osiete kudasai.

(237)

Zya, mazu (dare demo ii kara) gakusei ga suu-nin dono hon o yonda ka ittemite kudasai.

(238)

Zya, mazu (dare demo ii kara) gakusei ga sanbun no iti dono hon o yonda ka ittemite kudasai.

We have observed above that if β in (176), repeated above, is a universal quantifier analogue based on a wh-word, β cannot take wide scope with respect to α; see (227). Similarly, if the QNP in (229), repeated below, is a universal quantifier analogue based on a wh-word, the question cannot be replied to with a pair-list answer. For example, (unlike the A-B sequence in (230)) the A-B sequence in (239) is not possible, and in this situation B must reply with a single constituent answer.

(229)

[... α-ga ... β-ni /o ...] before wh-movement, where one of α and β is a QNP, and the other is a wh-word

(239)	A: Dono	gakusei mo	dono	hon	o	yonda ka osiete kudasai.
	which student also which book acc read				q teach please
	‘Please tell me which book every student read.’
	B: Yosio ga LGB o, Suuzan ga Barriers o, ..., sosite Takasi ga MP o yomimasita.

10.9.3 Between QNPs and Negation

We now turn to the scope interaction between QNPs and negation. Since we believe that negation is sensitive to focus—its meaning makes reference to a set of alternative choices under consideration, we describe the scope interaction under discussion, paying close attention to this factor.³⁶ In what follows, we call a phrase in a sentence whose denotation is among alternative choices a focused phrase. Following the standard practice, we mark focused phrases with the subscript F below.

We first describe the scope interaction between QNPs and negation, limiting our attention to cases where both the relevant QNP and negation are within a focused phrase. Consider the situation in (240).

(240)

People are wondering for what reason John has been mad. The speaker attempts to explain the reason.

In this situation, the speaker’s utterance would contain a reason, and the phrase expressing the reason would become a focused phrase. As we see shortly, in this situation, no matter what form the relevant QNP has—the QE + no + NP + CM form (i.e., (48a)), the NP + QE + CM form (i.e., (48b)), or the NP + CM + QE form (i.e., (48b))—it may take wide or narrow scope with respect to its clause-mate negation. (241), for example, can be understood to mean (242a) or (242b).

(241)	John wa		[Bill ga	{san-nin no	zyosei	o /	zyosei	san-nin
	John top Bill			nom three-cl gen woman		acc woman three-cl
	o /	zyosei	o	san-nin}	syootaisi-nak-atta	node]F		okotteiru.
	acc woman acc three-cl invite-neg-past because is:mad
	‘John is mad because Bill did not invite three women.’

(242)	a.	John is mad because there are three women whom Bill did not invite.
	b.	John is mad because it is not the case that Bill invited three women.

Similarly, (243), (245), and (247) give rise to both scope orders: (243) can be taken to mean (244a) or (244b); we can understand (245) to mean (246a) or (246b); (247) can give rise to both (248a) and (248b).

(243)	John wa	[Bill ga		{san-nin	izyoo	no	zyosei	o /	zyosei
	John top Bill		nom three-cl equal:more gen woman acc woman
	san-nin	izyoo	o /	zyosei	o	san-nin	izyoo}
	three-cl equal:more acc woman acc three-cl equal:more
	syootaisi-nak-atta		node]F	okotteiru.
	invite-neg-past because is:mad
	‘John is mad because Bill did not invite three or more women.’

(244)	a.	John is mad because there are three or more women whom Bill did not invite.
	b.	John is mad because it is not the case that Bill invited three or more women.

(245)	John wa		[Bill ga	{subete no	zyosei	o /	zyosei	subete o /
	John top Bill	nom all		gen woman acc woman all				acc
	zyosei	o	subete}	syootaisi-nak-atta	node]F	okotteiru.
	woman acc all			invite-neg-past because is:mad
	‘John is mad because Bill did not invite all women.’

(246)	a.	John is mad because Bill did not invite any women.
	b.	John is mad because it is not the case that Bill invited all women.

(247)	John wa		[Bill ga		{sanbun no	iti	no	zyosei	o /	zyosei
	John top Bill nom third					gen one gen woman acc woman
	sanbun no		iti	o /	zyosei	o	sanbun no	iti}
	third	gen one acc woman acc third						gen one
	syootaisi-nak-atta		node]F okotteiru.
	invite-neg-past because is:mad
	‘John is mad because Bill did not invite one third of the women.’

(248)	a.	John is mad because for one third of the women, Bill did not invite them.
	b.	John is mad because it is not the case that Bill invited one third of the women.

We now discuss cases where the verb phrase next to negation is a focused phrase, and the relevant QNP is in the verb phrase. As we demonstrate directly, in these cases, the negation necessarily takes wide scope with respect to the QNP. For example, imagine the situation in (249).

(249)

There is a project to be carried out. Before starting the project, several things need to be completed. The project leader asks the speaker to report what has been completed and what has not been.

In this situation, the set of alternative choices is those things needing to be completed before starting the project. Now consider in this situation the utterance in (250) together with its specified context.

(250)	[Context: Among the things needing to be completed are to secure 10 sawmill machines and to convince five workers to work for this project.]
	10-dai no	seizaikikai wa		karimasita ga,	mada [{go-nin no
	10-cl gen sawmill		top rented	but still	five-cl	gen
	sagyooin o /	sagyooin go-nin o /		sagyooin o	go-nin}
	worker	acc worker	five-cl acc worker		acc five-cl
	settokusitei]F-masen.
	convince-neg
	‘Although we rented 10 sawmill machines, we have not convinced five workers yet.’

Regarding the scope interaction in the second sentence of (250), the negation necessarily takes wide scope with respect to the relevant QNP—the second sentence can be taken to mean (251b) but not (251a).

(251)	a.	There are five workers that we have not convinced (to work for the project).
	b.	It is not the case that we have convinced five workers (to work for the project).

Similarly, the second sentence of (252) is understood to mean (253b), but not (253a); that of (254) gives rise to (255b) but not to (255a); (256) is taken to mean (257b) but not (257a).

(252)	[Context: Among the things needing to be completed are to secure 10 sawmill machines and to convince five or more workers to work for this project.]
	10-dai	no	seizaikikai wa		karimasita ga, mada [{go-nin izyoo
	10-cl gen	sawmill	top rented		but still	five-cl	equal:more
	no	sagyooin o /	sagyooin go-nin		izyoo	o /	sagyooin o
	gen worker	acc worker		five-cl equal:more acc worker			acc
	go-nin	izyoo}	settokusitei]F-masen.
	five-cl equal:more convince-neg
	‘Although we rented 10 sawmill machines, we have not convinced five or more workers yet.’

(253)	a.	There are five or more workers that we have not convinced (to work for the project).
	b.	It is not the case that we have convinced five or more workers (to work for the project).

(254)	[Context: Among the things needing to be completed are to secure 10 sawmill machines and to convince all of the workers to work for this project.]
	10-dai no		seizaikikai wa		karimasita ga, mada [{subete no
	10-cl gen sawmill		top rented	but still	all	gen
	sagyooin o /		sagyooin subete o /	sagyooin o	subete}
	worker	acc worker		all	acc worker	acc all
	settokusitei]F-masen.
	convince-neg
	‘Although we rented 10 sawmill machines, we have not convinced all the workers yet.’

(255)	a.	For each worker, we have not convinced him/her (to work for the project).
	b.	It is not the case that we have convinced each worker (to work for the project).

(256)	[Context: Among the things needing to be completed are to secure 10 sawmill machines and to convince one third of the workers to work for this project.]
	10-dai no		seizaikikai wa	karimasita ga, mada [{sanbun no			iti
	10-cl gen sawmill		top rented	but still	third	gen one
	no	sagyooin o /	sagyooin sanbun no	iti	o /	sagyooin o
	gen worker	acc worker	third	gen one	acc worker		acc
	sanbun no	iti}	settokusitei]F-masen.
	third	gen one convince-neg
	‘Although we rented 10 sawmill machines, we have not convinced one third of the workers yet.’

(257)	a.	There are one third of the workers that we have not convinced (to work for the project).
	b.	It is not the case that we have convinced one third of the workers (to work for the project).

Let us now turn to cases where QNPs are focused phrases. In these cases, as we illustrate directly, the relevant QNP takes wide scope with respect to its clause-mate negation. For example, consider the situation in (258), to which any utterance made in direct response would make the relevant QNP a focused phrase.

(258)

The department administrator asks the speaker to find out who Prof. Kimura recommended, who he did not recommend, how many students he recommended, how many students he did not recommend, and so on.

If (259) is uttered in the situation of (258), the second sentence may be taken to mean (260a) but not (260b).

(259)	Kimura sensei		wa		go-nin	no	dansigakusei o	suisensi,
	Kimura teacher top five-cl gen male:student acc recommend
	sosite {[san-nin no				zyosigakusei	o]F/ [zyosigakusei		san-nin
	and	three-cl gen female:student acc female:student three-cl
	o]F/ [zyosigakusei				o	san-nin]F} suisensi-masen-desita.
	acc female:student acc three-cl						recommend-neg-past
	‘Prof. Kimura recommended five male students, and he did not recommend three female students.’

(260)	a.	There are three female students that Prof. Kimura did not recommend.
	b.	It is not the case that Prof. Kimura recommended three female students.

The same point can be illustrated with other types of QNPs. In the situation of (258), the second sentence of (261), that of (263), and that of (265) give rise to (262a), (264a), and (266a) but not to (262b), (264b), and (266b), respectively.

(261)	Kimura sensei		wa	go-nin		no	dansigakusei o		suisensi,
	Kimura teacher top five-cl gen male:student acc recommend
	sosite {[san-nin izyoo				no	zyosigakusei		o]F / [zyosigakusei
	and	three-cl equal:more gen female:student acc female:student
	san-nin	izyoo		o]F / [zyosigakusei			o	san-nin	izyoo]F
	three-cl equal:more acc female:student acc three-cl equal:more
	suisensi-masen-desita.
	recommend-neg-past
	‘Prof. Kimura recommended five male students, and he did not recommend three or more female students.’

(262)	a.	There are three or more female students that Prof. Kimura did not recommend.
	b.	It is not the case that Prof. Kimura recommended three or more female students.

(263)	Kimura sensei		wa	go-nin	no	dansigakusei o	suisensi,
	Kimura teacher top five-cl gen male:student acc recommend
	sosite {[subete no			zyosigakusei		o]F / [zyosigakusei	subete o]F /
	and	all	gen female:student acc female:student all					acc
	[zyosigakusei		o	subete]F suisensi-masen-desita.
	female:student acc all				recommend-neg-past
	‘Prof. Kimura recommended five male students, and he did not recommend all the female students.’

(264)	a.	For each female student, Prof. Kimura did not recommend her.
	b.	It is not the case that Prof. Kimura recommended each female student.

(265)	Kimura sensei		wa		go-nin		no	dansigakusei o		suisensi,
	Kimura teacher top five-cl gen male:student acc recommend
	sosite {[sanbun no				iti	no	zyosigakusei		o]F / [zyosigakusei
	and	third	gen one gen female:student acc female:student
	sanbun no	iti	o]F / [zyosigakusei					o	sanbun no	iti]F}
	third	gen one acc female:student acc third								gen one
	suisensi-masen-desita.
	recommend-neg-past
	‘Prof. Kimura recommended five male students, and he did not recommend one third of the female students.’

(266)	a.	For one third of the female students, Prof. Kimura did not recommend them.
	b.	It is not the case that Prof. Kimura recommended one third of the female students.

10.10 Complex Quantifiers

We consider to what extent the complex quantifiers found in English are available in Japanese.

10.10.1 Type (2) Quantifier Analogues

In this section, we list some of what appears to be Type (2) quantifiers, which are functions expressing a property of binary relations—functions which are probably not reducible to the iterated applications of two functions of Type (1,1). First, to interpret the words that express the meaning of different or same requires the computation of two separate domains. Thus, arguably, they are type (2) quantifiers. Here we illustrate several cases.

(267)	a.	{Subete no		gakusei ga /	Gakusei subete ga /		Gakusei ga
		all	gen student nom student all			nom student	nom
		subete} tigau		kaisya	ni	syuusyokusita.
		all	different company dat got:employed
		‘All the students got a job offer from a different company.’
	b.	{Zen-bu no		zidoosyagaisya	ga /	Zidoosyagaisya
		all-cl	gen automobile:company nom automobile:company
		zen-bu ga /		Zidoosyagaisya	ga	zen-bu} onazi ginkoo
		all-cl nom automobile:company nom all-cl				same	bank
		to	torihikisiteiru.
		with is:dealing
		‘All the automobile companies are dealing with the same bank.’
	c.	Tigau	gakusei ga	tigau	situmon ni	kotaeta.
		different student nom different question dat answered
		‘Different students answered different questions.’

Second, the sentences in (268a) and in (269a) can be taken to mean (268b) and (269b), respectively, suggesting that the two wh-words in each sentence form Type (2) quantifiers.

(268)	a.	Kondo	no	ongakkai de wa	dare ga	nani o	hiku
		this:time gen concert		at top who nom what acc play
		koto	ni	narimasita ka.
		comp dat became		q
		‘At this coming concert, who plays what?’
	b.	What are the set of pairs (x, y) such that x is a person, y is a musical
		instrument, and x plays y at this coming concert?

(269)	a.	Kondo	no	dansu paatii de wa	dono	dansigakusei ga
		this:time gen dance party at top which male:student nom
		dono	zyosigakusei	to	dansusuru koto	ni	narimasita ka.
		which female:student with dance comp dat became q
		‘At this coming dance party, which male student dances with which female student?’
	b.	What are the set of pairs (x, y) such that x is a male student, y is a female student, and x dances with y at this coming dance party?

Our third example comes from so-called focus-sensitive particles (= FPs). We have observed in Section 10.7 that when they modify an NP, FPs may appear either (i) between the NP and the CM of the NP-CM unit or (ii) after the NP-CM unit. For convenience, we refer to (i) as the FP internal order and (ii) as the FP external order. As pointed out by Hayashishita (2011), if in a sentence, two or more instances of FPs appear both in the FP internal order, then one prominent reading associated with the sentence is that within which they are scopally independent from each other. For example, (270a) and (271a) are associated with (270b) and (271b), respectively. Thus, we suggest that two instances of FPs in the FP internal order may form Type (2) quantifiers.

(270)	a.	John wa	Kimura sensei	dake ni	Kyooto daigaku	dake de
		John top Kimura teacher only dat Kyoto university only at
		aisatusita.
		greeted
		‘(Lit.) John greeted only Prof. Kimura only at Kyoto University.’
	b.	There is no x other than Prof. Kimura and no y other than Kyoto
		University such that John greeted x at y.

(271)	a.	John dake ga	NELS dake de ronbun o	happyoosimasita.
		John only nom NELS only at paper acc presented
		‘Only John presented a paper only at NELS.’
	b.	There is no x other than John and no y other than NELS such that x presented a paper at y.

Incidentally, as Hayashishita (2011) points out, if one of the two instances of FPs above appears in the FP external order, the scope-independent reading under discussion cannot be obtained. For example, unlike (270a), the sentences in (272) are necessarily taken to mean (273).

(272)	a.	John wa Kimura sensei ni dake Kyooto daigaku de dake aisatusita.
	b.	John wa Kimura sensei dake ni Kyooto daigaku de dake aisatusita.

(273)

There is no person other than Prof. Kimura such that John greeted him at no place other than Kyoto University.

Our fourth example is Japanese comparatives. To express what the English more ... than comparative means, we may use one of the three constructions schematized in (274). The three constructions are exemplified in (275).

(274)	a.	[ ... [[NP yori] X] ... ], where X is a gradable expression
	b.	[ ... [[NP CM Verb yori] X] ... ], where X is a gradable expression
	c.	[ ... [[NP CM yori] X] ... ], where X is a gradable expression

(275)	a.	Taroo wa [[Hanako yori] sakini] Satiko ni hanasikaketa.
		‘Taro talked to Sachiko earli[er] than Hanako.’
	b.	Taroo wa	[[Hanako ni	hanasikakeru yori] sakini] Satiko
		Taro	top Hanako	dat talk	than early	Sachiko
		ni	hanasikaketa.
		dat talked
		‘Taro talked to Sachiko earli[er] than [he] talked to Hanako.’
	c.	Taroo wa [[Hanako ni yori] sakini] Satiko ni hanasikaketa.
		‘Taro talked to Sachiko earli[er] than to Hanako.’

As we illustrate directly, we can illustrate Type (2) quantifiers, using the constructions in (274a) and in (274b), but not the construction in (274c). When two instances of comparisons are expressed in a sentence, using the construction in (274a), the scope of one comparison may be independent from the scope of the other comparison. For example, (276a) and (277a) can be understood to mean (276b) and (277b), respectively.

(276)	a.	[[John yori] sakini] Bill ga			[[LGB yori] sakini] Aspects o
		John	than	early	Bill nom LGB	than early	Aspects acc
		yomioemasita.
		finished:reading
		‘Bill finished reading Aspects earli[er] than LGB earli[er] than John did.’
	b.	Bill finished reading a book earlier than John, and he read Aspects earlier than LGB.

(277)	a.	Kimura sensei		wa	[[John yori] sakini] Bill ni		[[sintakkusu
		Kimura teacher top John			than early	Bill dat syntax
		no	zyugyoo yori] sakini] semantikkusu no			zyugyoo de ronbun
		gen class		than early	semantics	gen class	at paper
		o	happyoo-sase-masita.
		acc present-cause-past
		‘Prof. Kimura made Bill present in the semantics class earli[er] than in the syntax class earli[er] than he made John do.’
	b.	Prof. Kimura made Bill present a paper earlier than John, and he made Bill present at the semantics course earlier than at the syntax course.

Similarly, (278) and (279), which make use of the construction in (274b), can give rise to (276b) and (277b), respectively.

(278)	[[John ga yomu yori] sakini] Bill ga [[LGB o yomu yori] sakini] Aspects o
	yomioemasita.

(279)

Kimura sensei wa [[John ni happyoo-sase-ru yori] sakini] Bill ni [[sintakkusu no zyugyoo-de happyoo-sase-ru yori] sakini] semantikkusu no zyugyoo de ronbun o happyoo-sase-masita.

On the other hand, using the construction in (274c), the scope of one comparison must be within the scope of the other comparison—with the construction in (274c), we cannot illustrate Type (2) quantifiers. For example, (280a) contrasts with (277a) and (279) in that it cannot give rise to the reading in (277b); it must be taken to mean (280b).³⁷

(280)	a.	Kimura sensei wa [[John ni yori] sakini] Bill ni [[sintakkusu no zyugyoo de yori] sakini] semantikkusu no zyugyoo de ronbun o happyoo-sase-masita.
	b.	Prof. Kimura made Bill present a paper at the semantics course earlier than at the syntax course, earlier than he made John do.

10.10.2 Type ((1,1),1) Quantifier Analogues

10.10.2.1 Comparative D-Quantifiers

In English, comparative D-quantifiers can be constructed as in (281). We claim that Japanese does not have their analogues.

(281)	a.	More students than teachers came to the party.
	b.	John invited more male students than female students.
	c.	At least as many students as teachers came to the party.
	d.	John invited at least as many male students as female students.

One might argue that the sentences in (281) correspond to those in (282).³⁸

(282)	a.	{Sensei yori takusan no			gakusei ga /		Gakusei ga		sensei
		teacher than many		gen student nom student nom teacher
		yori takusan} paatii ni			kita.
		than many	party dat came
		‘More students than teachers came to the party.’
	b.	John wa {zyosigakusei yori takusan no dansigakusei o /
		John top female:student than many				gen male:student acc
		dansigakusei o		zyosigakusei		yori takusan}		syootaisita.
		male:student acc female:student than many						invited
		‘John invited more male students than female students.’
	c.	{Sukunakutomo sensei			to	onazi gurai no		kazu	no
		at:least		teacher with same about gen number gen
		gakusei ga /	Gakusei ga	sukunakutomo sensei				to	onazi
		student nom student			nom at:least			teacher with same
		gurai no	kazu}	paatii ni	kita.
		about gen number party dat came
		‘At least as many students as teachers came to the party’
	d.	John wa	{sukunakutomo zyosigakusei				to	onazi gurai no
		John top at:least			female:student with same about gen
		kazu	no	dansigakusei o /			dansigakusei o
		number gen male:student acc male:student acc
		sukunakutomo zyosigakusei			to	onazi	gurai no		kazu}
		at:least		female:student with same about gen number
		syootaisita.
		invited
		‘John invited at least as many male students as female students.’

But the sentences in (282) are different from those in (281). For example, (281a) and (281c) compare the number of the students who came to the party and that of the teachers who came to the party. By contrast, with (282a) and (282c), the number of the students who came to the party is simply described in terms of the number of the teachers in the relevant context (possibly the number of the teachers in the students’ school). It may thus turn out that the teachers, whose number is compared with the number of the students, did not come to the party.

10.10.2.2 Combinations with Conjunctions

In English, it is possible that one quantifier takes two or more NPs. For example, (283a) and (284a) can be understood to mean (283b) and (284b), respectively.

(283)	a.	Every man, woman, and child jumped overboard.
	b.	Every man, every woman, and every child jumped overboard.

(284)	a.	Some man, woman or child works on Sunday.
	b.	Some man or some woman or some child works on Sunday.

In Japanese, we can do the same, using the NP + QE + CM or NP + CM + QE pattern (i.e., (48b) or (48c)). For example, (285a) and (286a) can be taken to mean (285b) and (286b), respectively.

(285)	a.	{A gumi no		gakusei to	B gumi no	gakusei subete ga /	A
		A	class gen student and B class gen student all				nom A
		gumi no	gakusei to	B gumi no	gakusei ga	subete} kita.
		class gen student and B class gen student nom all					came
		‘All the students from Class A and Class B came.’
	b.	Every student from Class A and every student from Class B came.

(286)	a.	Seihu		wa	{Mituikei		no	ginkoo to	Risonakei
		government top Mitsui:related gen bank						and Resona:related
		no	ginkoo zen-bu o /			Mituikei		no	ginkoo to
		gen bank	all-cl acc Mitsui:related gen bank						and
		Risonakei		no	ginkoo o		zen-bu} enzyosita.
		Resona:related gen bank				acc all-cl		supported
		‘The government supported every Mitsui-related bank and Resona-related bank.’
	b.	The government supported every Mitsui-related bank and every Resona-related bank.

Similarly, we may understand (287a) and (288a) to mean (287b) and (288b), respectively.

(287)	a.	{A gumi no	gakusei ka B gumi no	gakusei suu-nin	ga /
		A class gen student or B class gen student several-cl nom
		A gumi no	gakusei ka B gumi no	gakusei ga	suu-nin}
		A class gen student or B class gen student nom several-cl
		kita.
		came
		‘A few students from Class A or from Class B came.’
	b.	A few students from Class A or a few students from Class B came.

(288)	a.	Seihu		wa	{Mituikei	no	ginkoo ka Risonakei
		government top Mitsui:related gen bank or Resona:related
		no	ginkoo san-sya		o /	Mituikei	no	ginkoo ka
		gen bank		three-cl acc Mitsui:related gen bank				or
		Risonakei		no	ginkoo o	san-sya} enzyosita.
		Resona:related gen bank				acc three-cl supported
		‘The government supported three Mitsui related banks or Resona related banks.’
	b.	The government supported three Mitsui related banks or three Resona related banks.

With the QE + no + NP + CM pattern, the situation is different. (289a) does not necessarily give rise to (285b); it may mean (289b). Similarly, (290a) can be understood to mean (290b).

(289)	a.	Subete no A gumi no gakusei to B gumi no gakusei ga kita.
	b.	Every student from Class A and some students from Class B came.

(290)	a.	Seihu wa san-sya no Mituikei no ginkoo ka Risonakei no ginkoo o enzyosita.
	b.	The government supported three Mitsui related banks or some Resona related banks.

10.10.3 Type (1, (1,1)) Quantifier Analogues

In English, we observe Type (1, (1,1)) quantifiers—cases where there is just one conservativity domain but two predicate properties; e.g., the sentences in (291).

(291)	a.	More students came to the party than studied for their exam.
	b.	The same students came early as left late.

We are not sure that Japanese has such cases. For example, to express what (291a) means in Japanese, we use a sentence like (292), in which two ‘conservativity domains’ are mentioned.

(292)	Paatii ni	kita	gakusei no		kazu	wa	siken no	tameni
	party dat came student gen number top test						gen for
	benkyoosita gakusei no			kazu	yori ooi.
	studied	student gen number than many
	‘The number of students who came to the party is larger than that of students who studied for their tests.’

To express what (291b) means, we may use the sentences in (293).

(293)	a.	Hayaku kita		gakusei wa	osoku made nokotta		gakusei
		early	came student top late			until remained student
		da.
		copula
		‘The students who came early are those who remained until late.’
	b.	Osoku made nokotteita no			to	onazi gakusei ga	hayaku kara
		late	until remained	one with same student nom early			from
		kiteita.
		came
		‘(It turned out) the same students who remained until late came early.’

Like (292), (293a) explicitly mentions two conservativity domains. (293b), on the other hand, appears to have only one conservativity domain. We note, however, that since the word no can be a replacement of an NP, it is reasonable to assume it to mean gakusei ‘students’. It may thus turn out that (293b) also mentions two conservativity domains.