To appear in the Proceedings of the 39th Annual Meeting of the North East Linguistic Society (NELS 39, 2008). GLSA, University of Massachusetts, Amherst.

Clausal Comparatives and Cross-linguistic Variation* Junko Shimoyama McGill University

1.

Variation in than-Phrases and Clauses

Cross-linguistic variation in comparative constructions has attracted much attention in recent years. There has been much discussion, for instance, on how phrasal comparatives as in (1) should be analyzed. (1)

a. Yuya is taller [than Sota]. b. John [Bill-se] zyaadaa lambaa hai. John Bill-than more tall is ‘John is taller than Bill.’ (Hindi-Urdu, Bhatt & Takahashi 2007)

Phrasal complements of than could be base-generated as phrases, or derived from full clauses (such as Sota is tall for (1a)) by some reduction operation. Whether a language adopts a phrasal analysis or clausal analysis is a point of cross-linguistic variation. The phrasal analysis is argued to be the right one for Hindi-Urdu and Japanese by Bhatt and Takahashi (2007, 2008a,b). It is argued by Lechner (2001), on the other hand, that phrasal comparatives in English are underlyingly clausal. Slavic phrasal comparatives are claimed to lie somewhere in between, being derived from small clauses (Pancheva 2006).

*

For invaluable comments and discussions, I would like to thank the audiences at NELS 39 at Cornell University, McGill University, MIT, and University of Tuebingen, in particular, Sigrid Beck, Irene Heim, Bernhard Schwarz, Roger Schwarzschild, Arnim von Stechow, Yasutada Sudo, Lisa Travis, Lydia White, and John Whitman. Students in my Syntax 4 (Winter 2008) and Structure of Japanese (Winter 2009) at McGill also deserve special thanks. I have not been able to incorporate all the comments into this paper, unfortunately. The work reported here was supported in part by a startup grant from McGill University and by a grant from FQRSC, Établissement de nouveaux professeurs-chercheurs, to which I am grateful. All errors are of course my own.

Junko Shimoyama This paper looks at a lesser-known area of potential cross-linguistic variation in comparatives. It has to do with clausal comparatives as in (2). (2a) is a case of predicative adjectival comparatives and (2b) is a case of attributive adjectival comparatives. (2)

a. Yuya is taller [than Sota is]. b. Hanako bought a more expensive book [than Taro did].

In a standard view, these clausal comparatives have full clauses in the complement of than at LF, as shown in (3). (3)

a. [-er [op1 than Sota is t1-tall]]2 [Yuya is t2-tall] b. [-er [op1 than Taro did buy a t1-expensive book]]2 [Hanako bought a t2-expensive book]

(4)

[[-er]] = λPdt.λQdt. max(Q) > max(P)

The Japanese counterpart of (2b) is given in (5).1 It has been recently claimed that unlike in English, an LF like (3b) is not available for this Japanese sentence (Beck, Oda and Sugisaki 2004, Oda 2009, Kennedy, to appear, Sudo 2009). (5)

Hanako-wa [Taro-ga katta]-yori takai hon-o katta. Hanako-TOP Taro-NOM bought-than expensive book-ACC bought ‘Hanako bought a more expensive book than Taro did.’

Two questions arise. First, why is it, then, that sentence (5) seems to have the interpretation that its English translation has? According to Beck et al. (2004) and Kennedy (to appear), what looks like a clausal complement of yori ‘than’ in (5), Taro-ga katta ‘Taro bought’, is in fact a DP in the form of a free relative, and it denotes an individual, but not a degree or degrees. Thus a better translation of the sentence is (6). (6)

Hanako bought a more expensive book than what Taro bought.

The complement then involves maximalization of individuals, max(λx.Taro bought x), but not maximalization of degrees, max(λd.Taro bought a d-expensive book) (Jacobson 1995). According to this view, what appears to be a case of clausal comparatives in Japanese is in fact a case of (base-generated) phrasal comparatives. Another question that arises is about the source of the contrast between English and Japanese. One view says that Japanese belongs to a group of languages that do not allow movement of degree operators (Beck et al. 2004), while the other view makes a weaker claim, that Japanese belongs to a group of languages that do not allow movement of degree operators in the complement of than (Kennedy, to appear). These studies propose parameters that are responsible for the variation: a parameter that allows or disallows abstraction over degrees in the syntax in (7) (Beck et al.); or a parameter that 1

The Japanese counterpart of (2a) is not grammatical. See (10).

Clausal Comparatives and Crosslinguistic Variation restricts some languages to have only individual comparison, but no degree comparison in (8) (Kennedy). (7)

Degree abstraction parameter A language {does/does not} have binding of degree variables in the syntax.

(8)

Comparison type parameter A language has both individual and degree comparison, or has only individual comparison. (A language has both 2-place and 3-place comparative morphemes, or has only the latter.)

The goal of this paper is to present evidence from several empirical domains that genuine clausal comparatives with degree abstraction structures do exist in Japanese. Specifically, we will look at predicative adjectival comparatives (section 2.1), intensional contexts (section 2.2), negative islands (section 2.3), and a problem of overgeneration (section 2.4). As the above cross-linguistic claims on parametric variation are mainly based on data from Japanese, and how they appear to differ from English, they are not well supported and we need to look for evidence from other languages. In what follows, clause-looking yori-complements such as the one in (5) will be referred to as plain clausal complements, and clearly nominal complements as seen in (5)’ below as -no DP complements. (5)’

Hanako-wa [[Taro-ga katta]-no]-yori takai hon-o katta. Hanako-TOP Taro-NOM bought-NO-than expensive book-ACC bought ‘Hanako bought a more expensive book than {what/the one(s)} Taro bought.’

The individual-denoting DP analysis of yori-complements introduced above may receive initial plausibility because the noun-forming morpheme -no appears to optionally show up in yori-complements without change in meaning or grammaticality, as seen in (5) and (5)’. We will see below, however, that the interchangeability between a plain clausal complement as in (5) and a -no DP complement as in (5)’ does not always hold when we look at further data. Importantly, differences in meaning or grammaticality would follow if we assume that plain clausal complements involve degree abstraction structures that are familiar from analyses of their English counterparts. 2. 2.1

Degrees in the Syntax and Semantics Predicative Adjectival Comparatives

One consequence of the individual-denoting DP analysis is that we seem to have an explanation for why subcomparatives of degree as in (9) do not exist in Japanese (as observed by Snyder, Wexler and Das 1995), as they require degree comparison.

Junko Shimoyama (9)

* Hanako-no te-wa [Taro-no asi-ga nagai]-yori habahiroi. Hanako-GEN hand-WA Taro-GEN foot-NOM long-than wide ‘Hanako’s hands are wider than Taro’s feet are long.’

Note also that Japanese does not allow predicative adjectival comparatives as in (10), either. The unavailability of degree abstraction in the yori-clause also explains this fact, although there may well be other reasons, too.2 (10)

* [Hanako-no te]-wa [[Taro-no asi]-ga (ookii)]-yori ookii. Hanako-GEN hand-WA Taro-GEN foot-NOM big-than big ‘Hanako’s hands are bigger than Taro’s feet are (*big).’

Though the above account seems simple and elegant, immediate questions arise regarding the claim that there is no abstraction over degrees in the complement of yori. For example, certain predicative adjectival comparatives are allowed as shown in (11) and (12). (11)

[Hanako-no te]-wa [Taro-ga omotteita]-yori ookii. Hanako-GEN hand-WA Taro-NOM thought-than big ‘Hanako’s hands are bigger than Taro thought (they were).’

(12)

a. Kono musi-wa [me-de mi-eru]-yori tiisai. this bug-WA eye-with see-can]-than small ‘This bug is smaller than the eye can see.’ b. Kono kagu-wa [natu-no boonasu-de ka-eru]-yori takai. this furniture-WA summer-GEN bonus-with buy-can-than expensive ‘(lit.)This furniture is more expensive than (I) can buy with my summer bonus.’ (cf. This furniture is more expensive than I can afford.)

The standard degree analysis of the sentence in (11) gives us the intuitively correct meaning of the sentence in (13b), while this is not easily achieved in the individualdenoting DP analysis, whether we assume ellipsis or not, as in (14) and (15). In (14) and (15), the comparative morpheme -er that takes three arguments in (16) is used (see, e.g., Heim 1985, Bhatt & Takahashi 2007, 2008a,b, Kennedy, to appear, Nissenbaum 2000, Barker 2007, Kennedy and Stanley, to appear). (13)

a. [-er [op1 than Taro thought they were t1,d big]]2 [ Hanako’s hands are t2,d-big] b. max(λd.H’s hands are d-big) > max(λd.Taro thought they were d-big) 2

The explanation, in fact, is not as straightforward as it seems at first. There is no individual gap in the yori-complement in (9) and (10), and one cannot easily form a free relative-like DP. Beck et al. (2004: 310, 312-314), however, rely on an analysis where some yori-complements are internally headed relative clauses. At this point, there is no principled account of why the yori-copmlement in (9), for example, cannot receive interpretations such as Taro’s feet that are long or Taro’s long feet.

Clausal Comparatives and Crosslinguistic Variation (14)

a. [Hanako’s hands] [-er3-place [than [wh3 Taro thought t3,e]]] [2 [1 are [t2,d-big]]] b. max(λd.H’s hands are d-big) > max(λd. max(λx.Taro thought x) are d-big) c. Hanako’s hands are bigger than the thing(s) Taro thought (of).

(15)

a. [Hanako’s hands] [-er3-place [than [wh3 Taro thought t3,e were big]]] [2 [1 are [t2,d-big]]] b. max(λd.H’s hands are d-big) > max(λd. max(λx.T thought x were big) are d-big) c. Hanako’s hands are bigger than the thing(s) that Taro thought was/were big.

(16)

[[-er3-place]] = λxe.λPd,et.λye. max(λd.P(d)(y)) > max(λd.P(d)(x))

For the sentences in (12), one might be able to paraphrase the contributions of the yoricomplements by saying those things that one can see with naked eyes, or those things that I can buy with my summer bonus, but degree comparison seems intuitively more appropriate. Furthermore, unlike the sentences in (11) and (12), the predicative adjectival comparatives in (17a) and (18a) are ungrammatical (unless a -no DP complement of yori is used). This is unexpected if the yori-complement was a free relative DP, as paraphrased in (17c) and (18c). (17)

a.* Kono hon-wa [Hanako-ga katta]-yori takai. this book-TOP Hanako-NOM bought-than expensive b.* This book is more expensive than Hanako bought. c. This book is more expensive than what Hanako bought.

(18)

a. * Hanako-wa [Taro-ga yatotta]-yori kasikoi. Hanako-TOP Taro-NOM hired-than smart b.* Hanako is smarter than Taro hired. c. Hanako is smarter than the one(s) Taro hired.

As acknowledged in Beck et al. (2004: 308-309), the proposed individualdenoting free relative DP cannot occur freely in positions where DPs are expected to occur. (19) * [Taro-ga katta]-ga omosiroi. Taro-NOM bought-NOM interesting Intended: ‘What Taro bought is interesting.’ A stipulation must be added that the proposed free relative DP can only occur as the complement of yori. The examples in (17a) and (18a) above, however, show that adding this stipulation is not sufficient, and one is required to look for a more restricted theory of when plain clausal complements can and cannot receive an individual-denoting free relative DP analysis.

Junko Shimoyama In the standard degree analysis, on the other hand, an account for the ungrammaticality of (17a) and (18a) is likely to lie in the kind of ellipsis involved in these examples. Using sentence (5), a grammatical case of attributive adjectival comparative from above, we can see that while in the degree analysis, there must be an antecedent-ellipsis relation between the matrix clause and the yori-clause, this is not the case in the individual analysis. (5)

Hanako-wa [Taro-ga katta]-yori takai hon-o katta. Hanako-TOP Taro-NOM bought-than expensive book-ACC bought ‘Hanako bought a more expensive book than Taro bought.’

(20)

a. Hanako bought a more expensive book [op1 than Taro bought a t1,d expensive book] (degree) b. Hanako bought a more expensive book [op1 than Taro bought t1,e] (individual)

It is this communication between the matrix clause and the yori-clause that fails in (17a) and (18a), as well as in their English counterparts (17b) and (18b). As shown in (21a) and (22a), the degree analysis can attribute the ungrammaticality of (17a,b) and (18a,b) to unlicensed ellipsis.3 (21)

a. * This book is more expensive [op1 than Hanako bought a t1,d-expensive book] (degree) b. This book is more expensive [op1 than Hanako bought t1,e] (individual)

(22)

a. * Hanako is smarter [op1 than Taro hired a t1,d-smart person] (degree) b. Hanako is smarter [op1 than Taro hired t1,e] (individual)

2.2

Intensional Context

As we just saw in (20), the yori-constituent of example (5) looks like (23a) in the standard degree analysis, and like (23b) in the individual-denoting DP analysis. (5)

Hanako-wa [Taro-ga katta]-yori takai hon-o katta. Hanako-TOP Taro-NOM bought-YORI expensive book-ACC bought ‘Hanako bought a more expensive book than Taro did.’

(23)

a. [op1 than Taro bought a t1,d expensive book] b. [op1 than Taro bought t1,e]

3

Sudo (2009) puts forth a very interesting proposal that what I have been calling plain clausal complements are in fact relative clauses with invisible degree-denoting nominal heads, such as takasa ‘expensiveness’ and kasikosa ‘smartness’. Unlike in Beck et al.’s analysis, there is presumably a new type of ellipsis involved in this analysis. Its predictions need to be examined and contrasted with predictions made by the standard degree analysis (see also Hayashishita 2008).

Clausal Comparatives and Crosslinguistic Variation In the standard degree analysis, it is expected that the elided DP in (23a), takai hon ‘a t1,dexpensive book’, could interact scopally with other scope bearing elements that may occur in the yori-clause. On the other hand, the individual-denoting DP analysis predicts no such scope interactions. Thus, the individual-denoting DP analysis predicts that sentence (24), for instance, lacks a de dicto reading of takusan-no onigiri ‘t1,d-many rice balls’, while in the standard degree analysis such a reading would result from the yoriclause structure in (25a). In this structure, takusan-no onigiri ‘t1,d-many rice balls’ takes narrow scope with respect to hosigatteita ‘wanted’. This gives rise to a reading that is true when, for instance, Hanako wanted three rice balls, but not any specific rice balls, and Taro bought five rice balls. (24)

Taro-wa [[Hanako-ga hosigatteita]-yori] takusan-no onigiri-o katta. Taro-TOP Hanako-NOM wanted -than many-GEN rice.ball-ACC bought ‘Taro bought more rice balls than Hanako wanted.’

(25)

a. [op1 than Hanako wanted t1,d many rice balls] b. [op1 than Hanako wanted t1,e]

The de dicto interpretation is in fact the only interpretation available in this sentence.4 The structure of the yori-complement according to the individual-denoting DP analysis in (25b) does not predict this reading, as the complement denotes the maximal individual x such that Hanako wanted x. When a -no DP complement is used as in (26), on the other hand, only the de re reading is available, expectedly.5 (26)

Taro-wa [[Hanako-ga hosigatteita]-no]-yori takusan-no onigiri-o katta. Taro-TOP [Hanako-NOM wanted -NO-than many-GEN rice.ball-ACC bought ‘Taro bought more rice balls than {what/the ones} Hanako wanted.’

I add a few more similar examples here. They only allow de dicto interpretations. (27)

Taro-wa [[Hanako-ga yomi-tagatteiru] yori] nagai hon-o yonda. Taro-WA Hanako-NOM read-want than long book-ACC read ‘Taro read a longer book than Hanako wants to read.’

4

A de re reading is not available in (24), and the same holds for its English translation. I speculate that the structure in (i) for a de re reading is not available due to violation of a parallelism constraint (Fox 2000). (i) [-er [op1 than [t1,d-many rice balls]2 Hanako wanted t2 yori]]3 [Taro bought t3,d-many rice balls] 5 The sentence may not be the most natural-sounding one, as was pointed out to me for the English translation by Arnim von Stechow. Similar oddness is felt in the Japanese and English sentences in (i). The problem, however, disappears when we look at, for example, the -no DP counterpart of (27), which does not involve amount comparison. (i) Taro-wa [(Hanako-ga hosigatteita) korera-no onigiri]-yori takusan-no onigiri-o katta. Taro-TOP Hanako-NOM wanted these-GEN rice.ball-than many-GEN rice.ball-ACC bought ‘Taro bought more rice balls than these rice balls (that Hanako wanted).’

Junko Shimoyama (28)

Taro-wa [[pro tyuumonsi-nakerebanaranai] yori] Taro-WA order-must than sukunai kazu-no kyookasyo-o tyuumonsita. few number-GEN textbook-ACC ordered ‘Taro ordered a smaller number of textbooks than he had to order.’

The following data on free relatives in English suggest that there is something more than what is expected in the standard maximal individual analysis (Jacobson 1995). The free relatives in (29) do not always have a simple individual interpretation.6 (29)

a. What Hanako is looking for is a rice spatula. b. Hanako wanted three rice spatulas for her birthday, and Taro indeed bought what Hanako wanted.

It would be interesting to explore the question of whether an appropriate analysis of this type of cases is extendable to yori-complements. A challenge for the individual-denoting DP analysis would then be to account for the de dicto only interpretation of, for example, (24), without dropping its central claim that yori-complements in Japanese denote individuals, and without also allowing a de dicto interpretation for examples like (26).7 2.3

Negative Islands

Japanese comparatives exhibit so-called negative island effects as shown in (30) and (31).8 (30)

a. *John bought a more expensive book [than nobody did]. b. *John bought a more expensive book [than Mary didn’t buy]. 6

Thanks to Sigrid Beck and Roger Schwarzschild for bringing these types of examples to my attention. It should be noted that example (29a) is a copular sentence, and the apparent de dicto reading of the pre-copular part may be due to connectedness. Further, the relevance of kind reference in (29b) should be explored, as suggested to me by Bernhard Schwarz. 7 English free relatives in the complement of than (e.g., in the English translation of (26)) generally disallow de dicto interpretation, as pointed out to me by John Whitman. Example (i), however, seems to be ok when Yumi is not looking for any specific house. Here again, reference to kinds may be relevant. (i) This house is smaller than what Yumi is looking for. It should also be noted that unlike the English examples in (29), it is only (ii), but not (iii), that can be used in a de dicto context. For (iii) to be used in a de dicto context, -no must be replaced with mono ‘thing’, which possibly refers to kinds. As far as Japanese is concerned, then, one could speculate that the nonspecific reading in (ii) arises from the copular construction. (ii) [[Hanako-ga sagasiteiru]-no]-wa syamozi-da. Hanako-NOM looking.for-NO-WA rice.spatula-COP.PRES ‘What Hanako is looking for is a rice spatula.’ (iii) Taro-wa [[Hanako-ga hosigatteita]-no]-o katta. Taro-WA Hanako-NOM wanted-NO-ACC bought ‘Taro bought {what/the one(s)} Hanako wanted.’ 8 Though the judgment reported for (31) in Beck et al. (2004: 315) is ??, the sentence is incomprehensible and unsalvageable to my ear.

Clausal Comparatives and Crosslinguistic Variation (31) * John-wa [dare-mo/Mary-ga kawanakatta]-yori John-TOP anybody/Mary-NOM didn’t.buy -than takai hon-o katta. expensive book-ACC bought ‘John bought a more expensive book than nobody did/Mary didn’t buy.’ (Beck et al. 2004: 315) If plain clausal complements of yori received a standard degree analysis, we can let the ungrammaticality of English (30) and Japanese (31) follow from the same source, for example, from undefinedness of the maximal degrees in (32) (von Stechow 1984, Rullmann 1995; see also Gajewski, to appear and Schwarzschild 2008). (32)

a. max(λd. nobody bought d-expensive books) b. max(λd. Mary didn’t buy d-expensive books)

As the next example shows, if we replace the plain clausal complement in (31) with a -no DP complement, the sentence is fine. (33)

John-wa [[dare-mo/Mary-ga kawanakatta]-no]-yori John-TOP anybody/Mary-NOM didn’t.buy -NO -than takai hon-o katta. expensive book-ACC bought ‘John bought a book that is more expensive than {what/the one(s)} nobody bought/Mary didn’t buy.’ (Beck et al. 2004: 314)

In the individual-denoting DP analysis, both the ungrammatical (31) and the grammatical (33) involve maximalization of individuals. One thus wants to find a principled way of distinguishing the maximalization of individuals involved in these two cases. It is suggested in Beck et al. (2004: 315) that the maximal individual that the yoricomplement in (31) is supposed to denote, for example, max(λx. nobody bought x), is generally undefined. We do not want the same analysis to extend to the -no DP complement in (33), though, because the sentence is fine. One idea, suggested to me by Sigrid Beck, is to say that domains can be more easily restricted in (33) than in (31), for some reason. However, while adding overt restrictions to English (negative) free relatives may make them more usable (e.g., what nobody bought in this store in the last year, what Mary didn’t buy from her reading list), it has no effect in improving the Japanese example in (31). 2.4

Unattested Reading

In general, phrasal comparatives give rise to a potential ambiguity. Example (34) can be interpreted as (35a) or as (35b). (34) (35)

Taro found a smarter person than Ziro. a. Taro found a smarter person than Ziro did. b. Taro found a person who is smarter than Ziro is.

Junko Shimoyama The same holds for Japanese phrasal comparatives. Example (36) is ambiguous in the same way (34) is. (36)

Taro-wa Ziro-yori kasikoi hito-o mituketa. Taro-TOP Ziro-than smart person-ACC found ‘Taro found a smarter person than Ziro.’

Assuming with Bhatt and Takahashi (2008a,b) that the phrasal comparatives in Japanese are best analyzed in terms of the phrasal (or direct) analysis, as opposed to the clausal (or reduction) analysis, the ambiguity is captured by (i) using the 3-place -er in (16), repeated from above; and (ii) placing the yori-phrase in two different positions, high (37) and in situ (38). (16)

[[-er3-place]] = λxe.λPd,et.λye. max(λd.P(d)(y)) > max(λd.P(d)(x))

(37)

a. [Taro] [DegP [-er [than Ziro]] [2 1[t1,e found a t2,d-smart person]]]] (High yori-phrase) Taro

-er3place than Zriro

2

1 t1 found a t2-smart person

b. [[-er]] (Ziro) (λdλx. x found a d-smart person) (Taro) = max(λd.Taro found a d-smart person) > max(λd.Ziro found a d-smart person) (38)

a. Taro found [a [AP [DegP -er [than Ziro]] smart] person] (yori-phrase in situ) b. [[AP]] = [[-er]] (Ziro) (λdλx. x is d-smart) = λx. max(λd. x is d-smart)) > max(λd. Ziro is d-smart) c. Taro found a person whose smartness exceeds the smartness of Ziro.

Recall that in the individual-denoting DP analysis, plain clausal complements of yori are individual-denoting DPs, so what appears to be clausal comparatives are in fact cases of phrasal comparatives. A prediction made by the individual-denoting DP analysis, then, is that examples like (39) should in principle be ambiguous. (39)

Taro-wa [Hanako-ga tukutta]-yori ii kuruma-o tukutta. Taro-TOP Hanako-NOM made-than good car-ACC made a. Taro made a better car than Hanako did. (degree) b. Taro made a better car than what Hanako made. (individual)

(39a) is the only reading expected in the standard degree analysis of this sentence. In the individual-denoting DP analysis, this reading is paraphrased as (39b), or more precisely,

Clausal Comparatives and Crosslinguistic Variation a version of (39b) in which the yori-constituent (or the than-constituent in English) is interpreted in situ, namely, Taro made a car that is better than what Hanako made. In addition to this reading, it is expected in the individual-denoting DP analysis that, if a context is set up appropriately, an LF is available where the yori-constituent is high, just as was the case in (37) above. (40)

a. Taro

2

-er3place than what H made

1 t1 made a t2-good car

b. Taro made a better car than [what Hanako made] did. Imagine that we are discussing whether robots can make better cars than humans can. In particular, we are discussing Taro, who makes cars, and robots made by Hanako that build cars. 9 In this scenario, the English example in (39b), as well as the no-DP counterpart of (39) in (41) below, allow the high reading in (40). (41)

Taro-wa [[Hanako-ga tukutta]-no]-yori ii Taro-TOP Hanako-NOM made-no-than good ‘Taro made a better car than what Hanako made.’

kuruma-o tukutta. car-ACC made

The Japanese sentence in (39), on the other hand, cannot be interpreted in this way. This means that the individual-denoting DP analysis overgenerates. 3.

Concluding Remarks

This paper has shown that a closer look at clausal comparatives in Japanese reveals that genuine clausal comparatives with degree abstraction structures do exist in the language.10 The recent claims on typological variation are thus not well supported by Japanese data and require evidence from other languages. Recent work suggests that Mandarin Chinese is a good candidate (Xiang 2003, 2005, Krasikova 2008, Lin 2009). The conclusion reached in this paper is also supported by the observation that there is a syntactic island effect in clausal comparatives in Japanese (Kikuchi 1987, Ishii 1991; see also Beck et al. 2004, section 4.5 for discussion). Comparing the patterns of locality found in plain clausal complements and in relatives reveals a significant contrast. In the individual-denoting DP analysis, one would have to stipulate a special type of relative clause formation that is not subject to the same range of constraints as other relatives. Both relative clause formation (42) and -no DP formation (43)/(44) are fine 9

Thanks to Omer Preminger for suggesting to me this type of context where the high reading is forced. Thanks also to Jonathan Abramsohn for suggesting to imagine a golem created by Hanako that builds cars. 10 See also Aihara (2007) for evidence that superlatives in Japanese involve degree abstraction.

Junko Shimoyama across wh-islands. Clausal comparative formation across wh-islands, on the other hand, results in ungrammaticality as seen in (45). (42)

Kono hon-wa [Taro-ga [dare-ga _ karita ka] this book-TOP Taro-NOM who-NOM borrowed Q siritagatteiru hon]-to niteiru. want.to.know book-with look.like ‘This book is similar to [the book Taro wonders [who checked out _]]’

(43)

Kono hon-wa [Taro-ga [dare-ga _ karita ka] this book-TOP Taro-NOM who-NOM borrowed Q siritagatteiru-no]-to niteiru. want.to.know-NO-with look.like ‘This book is similar to [the one Taro wonders [who checked out _]]’

(44)

Hanako-wa [Taro-ga [dare-ga _ karita ka] this book-TOP Taro-NOM who-NOM borrowed Q siritagatteiru-no]-yori atui hon-o karita. want.to.know-NO-than thick book-ACC borrowed ‘Hanako checked out a thicker book than [the one [Taro wonders [who checked out _]]]’

(45) * Hanako-wa [Taro-ga [dare-ga _ karita ka] this book-TOP Taro-NOM who-NOM borrowed Q siritagatteiru]-yori atui hon-o karita. want.to.know-than thick book-ACC borrowed ‘Hanako checked out a thicker book than [Taro wonders [who checked out _]]’ Finally, the lack of subcomparatives of degree remains as a puzzle. This is certainly the case in the standard degree analysis, but also in the individual-denoting DP analysis (see footnote 2, as well as Bhatt and Takahashi 2008a: section 9). Furthermore, as discussed in Ishii 1991 and Beck et al. (2004: 312), attributive amount (sub)comparatives such as (46) are fine.11 (46)

Taro-wa [Hanako-ga hon-o katta]-yori ooku-no zassi-o katta. Taro-TOP Hanako-NOM book-ACC bought-than many-GEN magazine-ACC bought ‘Taro bought more magazines than Hanako bought books.’ (Bhatt and Takahashi 2008a)

It is noted in Bhatt and Takahashi (2008a) that a similar asymmetry surfaces in English in the attributive domain.

11

Such cases are analyzed as involving internally headed relative clauses in the complement of yori in Beck et al. (2004).

Clausal Comparatives and Crosslinguistic Variation (47)

a. b. c.

Michael Jordan has more scoring titles than Denis Rodman has tattoos. *Pico wrote a more interesting novel than Brio wrote a play. *Anna read a longer article than Roxani read a book. (Kennedy and Merchant 2000) (48) * Taro-wa [Hanako-ga syoosetu-o kaita]-yori omosiroi ronbun-o kaita. Taro-TOP Hanako-NOM novel-ACC wrote-than interesting paper-ACC wrote ‘Taro wrote a more interesting paper than Hanako wrote a novel.’ (Bhatt and Takahashi 2008a) The presence or absence of degree abstraction structure in than/yori-complements does not by itself account for the contrast in (47) or the one in (46) and (48). More work is needed. References Aihara, Masahiko. 2007. Japanese superlative constructions: evidence for ‘est’movement. In Proceedings of NELS 37. GLSA, University of Massachusetts, Amherst. Barker, Chris. 2007. Parasitic scope. Linguistics and Philosophy 30:407-444. Beck, Sigrid, Toshiko Oda and Koji Sugisaki. 2004. Parametric variation in the semantics of comparison: Japanese vs. English. Journal of East Asian Linguistics, 13:289344. Bhatt, Rajesh, and Shoichi Takahashi. 2007. Direct comparisons: resurrecting the direct analysis of phrasal comparatives. In Proceedings of SALT 17, 19-36. CLC Publications, Cornell University, Ithaca, NY. Bhatt, Rajesh, and Shoichi Takahashi. 2008a. When to reduce and when not to: crosslinguistic variation in phrasal comparatives, handout for talk at GLOW XXXI, Newcastle University, March 2008. Bhatt, Rajesh. and Shoichi Takahashi. 2008b. Reduced and unreduced phrasal comparatives. Ms., University of Massachusetts, Amherst and University of Tokyo. Fox, Danny. 2000. Economy and semantic interpretation. Cambridge, Mass.: MIT Press. Gajewski, Jon. to appear. More on quantifiers in comparative clauses. In Proceedings of SALT 18. CLC Publications, Cornell University, Ithaca, NY. Hayashishita, J.-R. 2008. Yori-comparative: comments on Beck et al. (2004). Ms., University of Otago. Heim, Irene. 1985. Notes on comparatives and related matters. Ms., University of Texas, Austin. Ishii, Yasuo. 1991. Operators and empty categories in Japanese. Doctoral dissertation, University of Connecticut, Storrs. Jacobson, Pauline. 1995. On the quantificational force of English free relatives. In Quantification in Natural Languages, ed. Emmon Bach, Eloise Jelinek, Angelika Kratzer, and Barbara H. Partee, 451-486. Dordrecht: Kluwer. Kennedy, Christopher. to appear. Modes of comparison. In Proceedings of CLS 43. University of Chicago, Chicago, Ill.

Junko Shimoyama Kennedy, Christopher. and Jason Merchant. 2000. Attributive comparative deletion. Natural Language and Linguistic Theory 18:89-146. Kennedy, Christopher. and Jason Stanley. to appear. What an average semantics needs. In Proceedings of SALT 18. CLC Publications, Cornell University, Ithaca, NY. Kikuchi, Akira. 1987. Comparative deletion in Japanese. Ms., Yamagata University. Krasikova, Sveta. 2008. Comparison in Chinese, In Empirical issues in formal syntax and semantics 7: papers from CSSP 2007, ed. Olivier Bonami & Patricia Cabredo Hofherr, 263–281. Lechner, Winfried. 2001. Reduced and phrasal comparatives. Natural Language and Linguistic Theory 19:683-735. Lin, Jo-wang. 2009. Chinese comparatives and their implicational parameters. Natural Language Semantics 17:1-27. Nissenbaum, Jon. 2000. Investigations of covert phrase movement. Doctoral dissertation, MIT, Cambridge, Mass. Oda, Toshiko. 2008. Degree constructions in Japanese. Doctoral dissertation, University of Connecticut, Storrs. Pancheva, Roumyana. 2006. Phrasal and clausal comparatives in Slavic. In Formal approaches to Slavic linguistics 14: the Princeton meeting, ed. J. Lavine, S. Franks, M. Tasseva-Kurktchieva and H. Filip, 236-257. Rullmann, Hotze. 1995. Maximality in the semantics of wh-constructions. Doctoral dissertation, University of Massachusetts, Amherst. Schwarzschild, Roger. 2008. The semantics of comparatives and other degree constructions. Language and Linguistics Compass 2:308-331. Snyder, William, Kenneth Wexler and Dolon Das. 1995. The syntactic representation of degree and quantity: perspectives from Japanese and Child English. In Proceedings of WCCFL 13, CSLI, Stanford University, Stanford, Calif. von Stechow, Arnim. 1984. Comparing semantic theories of comparison, Journal of Semantics 3:1-77. Sudo, Yasutada. 2009. Invisible degree nominals in Japanese clausal comparatives. In Proceedings of the 5th Workshop on Altaic Formal Linguistics, ed. Reiko Vermeulen and Ryosuke Shibagaki. MITWPL, Cambridge, Mass. Xiang, Ming. 2003. A phrasal analysis of Chinese comparatives. In Proceedings of CLS 39, 739-754. Xiang, Ming. 2005. Some topics in comparative constructions. Doctoral dissertation, Michigan State University, East Lansing. Department of Linguistics McGill University 1085 Dr. Penfield Avenue Montreal, Quebec Canada H3A 1A7 [email protected]