"Comparatives" in English and Japanese
Emi Mukai (Presentation, 10/10/05)
1.Introduction
1.1.Comparatives in English and Japanese
(1)Comparative constructions in English
a.This book is longer than XP
b.Bill bought a longer book than XP
(2)'Comparative' constructions in Japanese
a.kono hon-wa [YORIXP-yori]nagai
this book-top XP-yorilong
b.Bill-wa[YORIXP-yori]nagaihon-okatta
Bill-topXP-yorilongbook-accbought
[yoriXP-yori] part:The yori constituent
XP part:--- The yori phrase if XP is a simple noun
--- The yori clause if XP is not a simple noun
1.2.Main Claims and arguments for them
(3)Main claims
a.Every adjective, regardless of whether it is gradable or not,isuniversally of type <e,t>. The difference between gradable adjectives and non-gradable adjectives is whether a standard (which varies depending on a context) is supplied or not; it is in the case of the former, but not in the case of the latter.
b.The functional category #, which is of type <et, det>, introduces a scale (i.e., a point in a totally ordered set).
c.A gradable adjective(<e,t>)in English has some feature, which should be checked off by # (<et, det>). The mother node of a gradable adjective and # is therefore of type <d,et>.
d.Japanese does not have #. As a consequence, Japanese lacks a constituent of type <d,et>.
e.The yori constituent is an adjunct modifying an AP.
2.The standard analysis (Bresnan 1973; von Stechow 1984; Heim 2000 among others)
2.1.Outline of the standard analysis
Let us first review the standard analysis briefly. First of all, it has been widely analyzed that while non-gradable adjectives are a one-place predicate, gradable adjectives are a two-place predicate; they are functions from degrees on a scale to the set of individuals (<d,et>).
(4)a.[[Adjective]] = d. x. x is Adjective to d<d,et>
(e.g., [[tall]] = d. x. x is tall to d)
b.[[Non-gradable Adjective]] = x. x is Adjective<e,t>
(e.g., [[Japanese]] = x. x is Japanese)
Adjectives then take DegPs as their first argument; i.e., the combination of them is done through Functional Application.
(5)Functional Application (FA)(Heim & Kratzer 1998: 44: (3))
If is a branching node, {, } is the set of 's daughters, and [[]] is a function whose domain contains [[]], then [[]] = [[]]([[]]).
Under the analysis, whether adjectives are realized as comparative (taller) or as absolute (tall) depend on what the first argument of the adjectives, DegP, are like. In the former, the DegP is headed by the degree morpheme (er/more/less etc.). In the latter case, the DegP is POS ('positive') operator (see(7)).
(6) 3
John e 3
is AP <e,t>
3
DegP tall <d,et>
POS <dt,t>
6 feet d
DegP =er-than Bill is <dt,t>
1 inch er-than Bill is <dt,t>
(7)[[POS]] = P<d,t>.d[P(d)=1 and d>ds](Cf. von Stechow 1984: 60, R6)
(where ds is the standard degree which is contextually identified.)
The semantics of degree morpheme er is either (8a) or (8b), which contains the maximality operator (see(9)).
(8)a.[[er]] = D'<d,t>. D<d,t>. max(D) > max(D')<dt,dtt>
b.[[er2]] = D'<d,t>. D"d. D<d,t>. max(D) - max(D') = D"<dt,<d,dtt>
(9)von Stechow 1984: (117)
Max(P) is true of d iff P(d) and ~d'[P(d') & d' > d]
The than clause involves the degree abstraction at LF followed by the deletion, and the whole than clause undergoes extraction at PF.
In sum, the sentences in (10) are analyzed as in (11), (12), (13) and (14), respectively, under the analysis.
(10)a.John is tall.
b.John is 6 feet tall.
c.John is taller than Bill is.
d.John is 1 inch taller than Bill is.
(11)John is tall(=(10a))
a.BeforeSpell-Out & PFb.LF
c.[[John is POS tall]]= [[POS]] ([[1]]([[John is t1 tall]]))
= P. d[P(d)=1 and d>ds] (d1. John is d1-tall)
= d[d. John is d-tall (d)=1 and d>ds]
=d[John is d-tall and d>ds]
(where ds is the standard degree which is contextually identified.)
d.John is tall is true iff for the unique d, John is d-tall and d exceeds the standard degree ds.
(12)John is 6 feet tall(=(10b))
a.Before Spell-Out & PF & LF
b.[[John is 6 feet tall]]= [[tall]] ([[6 feet]]) ([[John]])
= d. x. x is d-tall (6 feet) (John)
= John is 6 feet tall
c.John is 6 feet tall is true iff John is tall to 6 feet.
(13)John is taller than Bill is(=(10c))
a.Before Spell-Outb.PF
c.LF
d.[[John is taller than Bill is]]= [[er]] ([[than Op1 Bill is t1 tall]]) ([[2](John is t2 tall]]))
= [[er]] (d1. Bill is d1-tall) (d2. John is d2-tall)
= max(d2. John is d2-tall) > max(d1. Bill is d1-tall)
e.John is taller than Bill is is true iff the degree d2 such that John is d2-tall exceeds the degree d1 such that Bill is d1-tall.
(14)John is 1 inch taller than Bill is(=(10d))
a.Before Spell-Outb.PF
c.LF
d.[[John is 1 inch taller than Bill is]]
= [[er2]] ([[than Op1 Bill is t1 tall]]) ([[1 inch]]) ([[2](John is d2-tall]]) )
= [[er2]] (d1. Bill is d1-tall) (1 inch) (d2. John is d2-tall)
= max(d2. John is d2-tall) max(d1. Bill is d1-tall) = 1 inch
e.John is 1 inch taller than Bill is is true iff the degree d2 such that John is d2-tall minus the degree d1 such that Bill is d1-tall equal 1 inch.
--There has been sometimes assumed in the literature (Kikuchi 1987, Ishii 1991; 1999, Nakanishi 2004) that Japanese comparatives (and a sentence with an adjective as its predicate) are analyzed in line with the standard analysis. However, any of those should fail to capture the observations in 2.2 below.
2.2.Differences between English and Japanese
2.2.1.The interpretation of a measure phrase right before an adjective
(15)a.John is 1m tall.1m: direct MP
b.John is 1m taller.1m: differential
(16)John-wa1mse-ga-takai
John-top1mtall
= 'John is 1m taller (than someone else)'
=/= 'John is 1m tall'
---Incidentally, the acceptability between (17a) and (17b) may imply that the yori constituent is an adjunct.
(17)(Out of the blue:)
a.John-gakowasita-ndesu!
John-nombreak:past-copula(declarative)
'John broke (it)!'
b.?#John-wa3cmse-ga-takai-ndesu!
John-nom3cmtall-copula(declarative)
'John is 3cm taller (than someone else)!'
2.2.2.Presupposition requirement (PR) (initial observation)
---It seems the case for both English and Japanese that neither the subject of a matrix clause (e.g., John) nor the noun in the yori/than clause/phrase (e.g., Bill) need tosatisfy the property indicated by the predicate (e.g., tall), as shown in (18).
(18)a.John is taller than Bill (is).NeitherJohn norBill has to be tall.
b.John-wa Bill-yori segatakaiNeitherJohn nor Bill has to be tall.
John-top Bill-yori tall
---However, when the than/yori clause contains another adjective, English and Japanese cometo differ from each other.
(19)a.*John is taller than Bill is tall.
b.John-wa [Bill-ga segatakai] yori segatakai.John and Bill should be tall.
John-top Bill-nom tallyori tall
---I introduce (20).
(20)Presupposition Requirement (PR)
The argument of a predicate should satisfy the property expressed by the predicate.
--Incidentally, it seems that the underlying structure for the sentence in (18b) is not the same asthat for the sentence in (19b); otherwise the difference with respect to PR would be mysterious.
2.2.3.Subcomparatives (initial observation)
--Sentences in which the than clause contains a different adjective from the matrix clause are called subcomparatives.
(21)a.The shelf is taller than the door is wide.
b.#The shelf is taller than the bridge is old.
---Kennedy provides the generalization for the acceptability of subcomparatives as in (22).
(22)Kennedy 1997: chap. 1, (23)/(106)/(136)
A comparative construction is semantically well-formed only if the compared adjectives have the same dimensional parameter.
Cf.#The class was longer than this table is.(Kennedy 1997: chap 1, (15))
---To be theory-neutral, I modify (22) and assume as a description that subcomparatives are subject to the requirement in (23).
(23)Same Dimension Requirement(SDR)
A comparative construction is semantically well-formed only if the compared objects are compared along the same dimension.
--It has been observed that subcomparatives are not possible in Japanese (Snyder et al. 1995, Beck et al. 2004, Nakanishi 2004 among others).
(24)*Kono tana-wa [ano doa-ga hiroi (no) yori] takai.(Beck: (5a)/(74a))
This shelf-top that door-nom wide one. yori tall
In sum so far; the observations in 2.2 show that any approach along the line with the standard analysis cannot capture the properties of comparatives in Japanese.
3.Observations revisited
Beck et al. (2004) propose a novel analysis that yori is not Japanesecounterpart of than. Basically following the standard analysis, they claim that (i) the yori constituent is not the argument of soundless er, and (ii) yori takes an individual as its complement (i.e., the yori phrase/clause is always NP/DP).
Problem for Beck et al. 1:They do not explicitly argue in which position the yori constituent is located at LF and how they are combined with the rest of the sentence.
Problem for Beck et al. 2:They attribute the unacceptability of (24) (and the lack of subcomparatives in Japanese in general) to the unavailability of nominalization of the yori phrase with no, as in (25).
(24)*Kono tana-wa [ano doa-ga hiroi (no) yori] takai.(Beck: (5a)/(74a))
This shelf-top that door-nom wide one. yori tall
(25)*Watasi-wa[doa-gahiroi/ookii]-no-oaketa(Beck: (80))
I-top door-nomwide/large-one-accopened
'I opened the door that was wide/large.'
However, the nominalization of the yoriclauseitself is not bad to me (and to the speakers who I consulted with). See(26a/b).
(26)a.Kono heya-wa [ano heya-ga hiroi yori] hiroi
This room-top that room-nom wide yori wide
b.Kono heya-wa [ano heya-ga hiroi no yori] hiroi
This room-top that room-nom wide no yori wide
Moreover, some people find the sentence in (27)significantly better than (24).
(27)Kono ido-wa [ano koosoo biru-ga takai (no) yori] fukai
This well-top that high-rise building-nomtall no yori deep
Importantly, in (27), PR is satisfied quite easily thanks to the word koosoo biru 'high-rising building'. When we change it into somatuna inugoya 'humble kennel', the sentence becomes bad.
(28)*Kono ido-wa [ano somatuna inugoya-ga takai (no) yori] fukai
This well-top that humble kennel-nom tall no yori deep
Puzzle to be solved in the rest of the paper:
What makes the subcomparative in (27) better than thosein (24) and (28)?
The purpose of the rest of section 3:
To show that PR and DSR between Japanese and English.
3.1.Gradable vs. non-gradable
(29)Two defining characteristics for gradable adjectives (Kennedy 1997: 1, citing Klein 1980: 6)
a."[G]radable adjectives can be modified by degree adverbials such as quite, very, and fairly. [...] Although non-gradable adjectives like dear do sometimes occur with degree modifiers, as in e.g., Giordano Bruno is quite dead, such uses are marked, and tend to convey a sense of irony or humor."
b."[Gradable adjectives] can appear in a class of complex syntactic environments," which Kennedy calls degree constructions. (E.g., (30))
(30)Degree Constructions (identified by Kennedy 1997: 2)
(Examples here are cited from Kennedy 1997: chap. 1, (8)-(14). Highlights are mine.)
a.Venus is brighter than Mars.
b.Neptune is not as distant as Pluto.
c.The equipment is too old to be of much use to us.
d.Current spacecraft are not fast enough to approach the speed of light.
e.The black hole at the center of the galaxy is so dense that nothing can escape the pull of its gravity, not even light.
f.How bright is Alpha Centauri?
3.2.Non-gradable adjectives: pregnant, Bulgarian, wooden, golden, fake, extinct, dead, octagonal, former etc.
---As shown in (31) to (33), it is not quite clear whether the two properties in (29) can be a diagnostic test to differentiate between gradable and non-gradable adjectives.
(31)Examples cited from Roumi Pancheva's lecture note 1 (Spring 2005), with the judgments reported by Roumi.
a.Mary is more Bulgarian than Sue
b.???The ring is more golden than the necklace.
c.???The coin is more fake than the dollar bill.
d.???Dinosaurs are too extinct.
e.Mary is very/so Bulgarian
f.This coin is very/so fake.
(32)Examples cited from Kennedy 1997: chap 1, (4)-(6) and (15)-(17), with Kennedy's judgments.
a.??Giordano Bruno is very dead.
b.??I want the new spacecraft to be quite octagonal.
c.??Carter is a fairly former president, and Lincoln is an extremely former president.
d.??Giordano Bruno is too dead to fly on the space shuttle.
e.??How former a president is Carter.
(33)Examples provided by Hoji-sensei at GGES (October, 2005)
a.How pregnant is Mary?
b.Mary is more pregnant than Nancy (is).
c.Mary is too pregnant to lift that chair.
---However, it is still the case that what people have called non-gradable adjectives cannot appear in the case of subcomparatives.
(34)Examples provided by Hoji-sensei at GGES (October, 2005)
a.*Mary is more pregnant than Nancy is sad.
b.*Mary is more pregnant than Bill is happy.
---Can we use subcomparatives a diagnostic for gradable vs. non-gradable?
=>Probably No. The sentences in (34) are bad because DSR is not satisfied in these cases.
3.3.Two types of gradable adjectives
3.3.1.With absolute adjectives
---It seems the case that gradable adjectives can also be divided into two types regarding whether PR should be satisfied or not.[1]
(35)Type 1: PR does not have to be satisfied.(See (36))
E.g., tall, long, short, old, young, heavy, light, wide, narrow, clever, bright, dim, high, low, difficult, easy, expensive, cheap, fast, slow, dense etc.
Type 2: PR should be satisfied.(See (37))
E.g., happy, sad, lonesome, glad, tired, worried etc.
(36)a.John is tall.
b.John is 6feet tall.No presupposition that John should be tall.
c.How tall is John?No presupposition that John should be tall.
(37)a.John was sad yesterday.
b.*John was MP sad.
c.How sad was he?It is presupposed that he was sad.
3.3.2.Ordinal comparatives
(38)a.John is taller than Bill is.NeitherJohn nor Bill has to be tall.
b.John was sadder than Bill was.It is presupposed that John was sad.[2]
3.3.3.Subcomparatives
(39)Subcomparatives with type 1 adjectives
a.The desk is longer than the table is wide.(Kennedy 1997: chap 1, (40))
b.#My copy of The Brother Karamazov is heavier than my copy of The Idiot is old.(Kennedy 1997: chap 1, (104))
=>With type 1, DSR should be satisfied, while PR does not have to.
Cf.a.#The class was longer than this table is.(Kennedy 1997: chap 1, (17))
b.My copy of The Brothers Karamazov is higher on a scale of heaviness than my copy of TheIdiot is on a scale of age. (Kennedy 1997: chap 1, (107))
(40)Subcomparatives with type 2 adjectives
a.John is more lonesome than Mary is sad.
b.#Larry is more tired than Michael is clever.(Kennedy 1997: chap 1, (179)
=> With type 2, both DSR and PR should be satisfied.
(41)
Non gradable adj. / Gradable type 1 / Gradable type 2Availability of having a direct MP / * / OK (PR is irrelevant) / *
How question / OK/??/* (depends on adjectives), PR is relevant / PR is irrelevant / PR is relevant
Ordinal comparative / OK/??/* (depends on adjectives), PR is relevant / PR is irrelevant / PR is relevant
Subcomparative / * since SDC is never satisfied? / OK only if SDC is satisfied. (PR is irrelevant) / OK only if both SDC and PR are satisfied.
3.4.Japanese adjectives
---The counterparts of the English sentences with absolute adjectives:
(42)a.John-wa segatakai.
John-top tall
b.John-wa 1m segatakai.*with the reading of a direct MP.
John-top 1m tall
c.John-wa doredake segatakai no?It is presupposed that John is tall.
John-top how:much tall Q
---Ordinal comparatives:
(18)b.John-wa Bill-yori segatakaiNeitherJohn nor Bill has to be tall.
John-top Bill-yori tall
(19)b.John-wa [Bill-ga segatakai] yori segatakai.John and Bill should be tall.
John-top Bill-nom tallyori tall
---Subcomparatives:
(24)*Kono tana-wa [ano doa-ga hiroi (no) yori] takai.(Beck: (5a)/(74a))
This shelf-top that door-nom wide one. yori tall
(43)*Ano kyuujyoo-wa [kono hon-ga furui yori] hiroi
(27)Kono ido-wa [ano koosoo biru-ga takai (no) yori] fukai
This well-top that high-rise building-nom tall no yori deep
(28)*Kono ido-wa [ano somatuna inugoya-ga takai (no) yori] fukai
This well-top that humble kennel-nom tall no yori deep
(44)This shabby kennel is taller than that small hole is deep.
(45)The summary (Japanese adjectives are added to (41).)
Non gradable adj. / Gradable type 1 / Gradable type 2 / JP adjectivesAvailability of having a direct MP / * / OK (PR is irrelevant) / * / *
How question / OK/??/* (depends on adjectives), PR is relevant / PR is irrelevant / PR is relevant / PR is relevant
Ordinal comparative / OK/??/* (depends on adjectives), PR is relevant / PR is irrelevant / PR is relevant / PR is irrelevant only if yori takes the yori phrase.
Subcomparative / * since SDC is never satisfied? / OK only if SDC is satisfied. (PR is irrelevant) / OK only if both SDC and PR are satisfied. / PR is relevant
4.Proposal
4.1.In general
(3)a.Every adjective, regardless of whether it is gradable or not, is universally of type <e,t>. The difference between gradable adjectives and non-gradable adjectives is whether a standard (which varies depending on a context) is supplied or not; it is in the case of the former, but not in the case of the latter.
(46)[[Adjective]] = x. x is Adjective
4.2.English comparatives
---My analysis diverges from the standard analysis in the following two respects.
1.There is only one semantic type for Adjectives, which is of type <e,t>. (see (3))
2.The functional category,#, introduces a scale. It takes an adjective as its first argument and derivesa #-Aconstituent, which is of type <d,et>. (Originally suggested by Roumi Pancheva, lecture note 5, Spring 2005).
(47) <d,et> = d. x. x is Adj to d.
3
# A <e,t> = x. x is Adj.
<e,t>, <d,et>
The functional category then takes Degree phrase (DegP) as its second argument (in its specifier position). This idea is also suggested in the lecture by Roumi Pancheva (Spring 2005).
(48) #P(cf. Roumi Pancheva lecture note 5, (11))
3
DegP 3
# A
4.3.Japanese yori constructions
4.3.1.Syntax of yori constructions
---I adopt the thesis in (48).
(49)There are no 'active functional categories' in Japanese (in the terms of Fukui 1986).[3]
The semantic category of yori is P, which takes an NP as its complement, and a PP is an adjunct to an AP.
(50)a.kono hon-wa [YORILGB-yori]nagai
this book-top LGB-yorilong
(Intended meaning) 'This book is longer than LGB.'
b.IP
3
NP I'
|3
kono hon-wa AP I
3
PP AP
3 |
NP P A
| | |
LGB yori nagai
4.3.2.The properties ofyori
--Yori is used in a formal context as in (50a)or an archaic use as in (50b). There, yorihas the meaning of 'from', and can be replaced by kara 'from'.
(51)a.kore-yori/karakaigi-ohajimemasu.
this:time-yori/fromconference-accstart
(At the beginning of a formal business meeting) '(Lit.) We start the conference from now.'
b.tomo ari.enpoo-yori/karakitaru.
friend there:isfar:away-yori/fromcome
'There is a friend who comes from far away.'
However, yori in 'comparatives' in Japanese cannot be replaced by kara.
(52)Kono hon-waLGB-yori/*karanagai
this book-top LGB-yori/*fromlong
(Intended meaning) 'This book is longer than LGB.'
X-kara and X-yori can be a predicate (the latter is marginally allowed as such), but yori can only be interpreted as from, as shown in (52b).
(53)a.kore-waamerika-karadesu
This-topAmerica-fromcopula
'This is from America.'
b.??kore-waamerika-yoridesu
This-topAmerica- yoricopula
'This is from America.'
#'This is more than America.'
Cf. (53).
(54)a.This is a fake gun.
b.#This is a gun and this is fake.
Adjectives as attributive modifiers are of type <et,et> (Heim & Kratzer 1998: 66-73.)
(55)3 <e,t>
This e 3e
is3
a NP<e,t>
3
AP <et,et> NP<e,t>
| |
fakegun
I propose that there are two distinct entries in the Lexicon of Japanese that has the same phonological properties, corresponding to yori, one of which is of type <e,t>, and the other of which is <et,et>.
(56)a.[[yori]] = y. x. x is from y
b.[[yori]] = y. f. x. f(x) = 1,surpassing y.
(57)kono hon-wa LGB-yori nagai(=(49a))