The Chinese geng clausal comparative
Chen-Sheng Luther Liu
Department of Foreign Languages and Literatures, National Chiao Tung University, 1001 Ta Hsueh Road, Hsinchu 300, Taiwan
Received 15 January 2009; received in revised form 24 September 2009; accepted 28 September 2009 Available online 31 October 2009
Abstract
Based on the syntax and semantics of the Chinese geng ‘GENG’ clausal comparative, I first argue that Chinese does have degree comparison in syntax, contra the claim made byKennedy (2005, 2007), andLin (2009). I then discuss two further issues raised by the geng ‘GENG’ clausal comparative: first, to show how the geng ‘GENG’ clausal comparative fits into a larger system of Chinese comparatives, I dichotomize Chinese comparatives into presupposition and non-presupposition comparatives. Second, I argued that the Degree Abstraction Parameter proposed by Beck et al. (2004), though being empirically challenged by the Chinese geng ‘GENG’ clausal comparative, can be maintained if languages involving degree comparison in syntax are further divided into two types depending on how the degree variable is bound (Kennedy, 2007).
# 2009 Elsevier B.V. All rights reserved.
Keywords: Clausal comparatives; Degree comparison; Degree intervals; Degree points; Individual comparison; Movement versus non-movement parameter in degree abstraction; Phrasal comparatives
1. Introduction
Through studying the syntax and semantics of the Chinese comparative like (1c) (henceforth the Chinese geng ‘GENG’ clausal comparative), in which two difference degree intervals ((A) the difference between the degree value of Zhangsan’s happiness and the contextually determined standard degree value of happiness; and (B) the difference between the degree value of Lisi’s happiness and the contextually determined standard degree value of happiness) are compared, I claim that Chinese does have degree comparison in the syntax, contraKennedy (2005, 2007),Xiang (2005),Erlewine (2007), andLin (2009).1
www.elsevier.com/locate/lingua Available online at www.sciencedirect.com
Lingua 120 (2010) 1579–1606
E-mail address:csliu@faculty.nctu.edu.tw.
1To make the presentation smooth and not digress from the main theme, I simply assume the morpheme hen ‘HEN’ is a degree adverb, instead of
specifying whether it is the overt form of the positive morpheme or an intensifier marker, unless it is necessary (Kennedy, 2005; Liu, 2010). I will also use GENG to represent the degree adverb geng, which is better glossed as ‘even more’ in this paper. Abbreviations used in this paper are: ASP: aspect marker, CL: classifier, DE: the verbal suffix or the marker for modifying phrases like genitive phrases, relative clauses, and noun complement clauses, and SFP: sentence final particle.
0024-3841/$ – see front matter # 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.lingua.2009.09.005
Based on my interval-based analysis of the syntax and semantics of the Chinese geng ‘GENG’ clausal comparative, two further issues are discussed. First, to show how the Chinese geng ‘GENG’ clausal comparative fits into a larger system of Chinese comparatives, I assume that Chinese has a covert comparative morpheme (i.e., ) and the marker bi ‘than’ simply functions like the English than; I go on to dichotomize Chinese comparatives into presupposition comparatives like (1b–d) on the one hand and non-presupposition comparatives like (1a) on the other, using the occurrence of the comparative morpheme geng ‘GENG’ (henceforth geng ‘GENG’) as criterion.2 Second, although the Degree Abstraction Parameter proposed by Beck et al. (2004) is empirically challenged by the Chinese geng ‘GENG’ clausal comparative, this parameter still can be maintained if languages involving degree comparison in syntax are further divided into two types depending on how the degree variable is bound.
The organization of this paper is as follows. I shall start section 2 by discussing the syntactic and semantic properties of the Chinese geng ‘GENG’ clausal comparative, and end it by pointing out what questions this construction raises for the current syntactic and semantic theories of comparatives. In section3, I first introduce as preliminaryKennedy’s (2001a)interval-based account for the semantics of comparatives, and then an interval-based analysis of the semantics of the Chinese geng ‘GENG’ clausal comparative is proposed. Pushing this further, a new way of classifying Chinese comparatives is suggested in section4, and the question of how to accommodate the impact onBeck et al.’s (2004)Degree Abstraction Parameter from the Chinese geng ‘GENG’ clausal comparative is discussed in section5. Finally, the conclusion is reached in section 6.
2. The syntax and semantics of the Chinese geng clausal comparative
In this section, I discuss the syntactic and semantic properties of the Chinese geng ‘GENG’ clausal comparative from the following three perspectives: the semantics and syntax of the morpheme geng ‘GENG’, comparison of deviation and divergence, and degree comparison in syntax.
2Because of space limitations, I will not discuss two other types of Chinese comparatives, as shown by (i)–(ii):
2.1. The semantics and syntax of the morpheme geng
Studies on Chinese bi ‘than’ comparatives, includingXiang (2005),Erlewine (2007), andLin (2009), assume that the morpheme bi ‘than’ not only has the function of English than but also has the function of the English comparative morpheme -er. For instance, according toLin (2009:20), the morpheme bi ‘than’, as in (1b), is a dyadic comparative degree head that might take one or more than one (individual, time or location) argument, and the whole DegP-shell headed by the degree word bi ‘than’ is an adjunct adjoined to the predicate of comparison. In addition, the morpheme bi ‘than’, with a denotation like (2a), behaves like the English comparative morpheme -er. So, example (1b) has a syntactic structure like (2b) underLin’s (2009)analysis.
Lin (2009), like other previous studies on Chinese comparatives, does not discuss the morpheme geng ‘GENG’, as in (1b), leaving its syntactic and semantic properties unclear under his analysis. However, examples like (1c,d), which involve comparison but do not contain the morpheme bi ‘than’, imply that the morpheme geng ‘GENG’ should have a semantic function like that of the English comparative morpheme -er. Besides, both the semantic interpretation of (1b–d) and the fact that (3a,b) are true if and only if both Zhangsan and Lisi are tall indicate that the morpheme geng ‘GENG’ presupposes that the properties predicated of the compared ‘objects’ are true in the absolute sense.
Added to these, as (1b) and (1d) further indicate, it is not necessary for the degree adverb geng ‘GENG’ to correlatively occur with a degree adverb, for instance hen ‘HEN’. (See footnote(15)for the further discussion on (1d)). Hence, the morpheme geng ‘GENG’, as one anonymous reviewer suggests to me, must be an independent morpheme.3
3
Based on the following two facts, I argue that example (1c) should not be treated as the reduced form of a bi ‘than’ comparative like (i).
First, under the following scenario, example (ii), a Chinese geng ‘GENG’ clausal comparative, can be felicitously interpreted, but example (iii), which involves a construction like (i), cannot.
For these reasons, I analyze (1c) as a multi-clausal sentence rather than a correlative construction (i.e., a correlative comparative) because examples like (1c) show the following properties of the Chinese multi-clausal sentence listed by Liu et al. (2004:863–864), andXing (2004:191–193): (A) Clauses contained in a multi-clausal sentence are connected to each other in interpretation in order to express the meaning of the whole sentence completely. As I have argued, the morpheme geng ‘GENG’, having a semantic function like that of the English comparative morpheme -er, denotes a greater-than relation (e.g., x is greater than y along some dimension). In a Chinese geng ‘GENG’ clausal comparative like (1c), the standard of comparison (i.e., the difference between the degree value of Zhangsan’s happiness and the contextually determined standard degree value of happiness), being an argument of the morpheme geng ‘GENG’, is syntactically realized as the first clause of the multi-clausal sentence (i.e., Zhangsan hen kaixin ‘Zhangsan HEN happy’), which is semantically connected to the second clause (i.e., Lisi geng kaixin ‘Lisi GENG happy’), which provides the other argument (i.e., the other difference degree interval) for the morpheme geng ‘GENG’. In other words, the full meaning of the whole sentence can only be expressed completely by having these two clauses connected to each other in meaning. (B) Neither of the two clauses involved is contained in the other. (C) Like the clauses contained in a multi-clausal sentence, the clauses contained in (1c) can be separated from each other by a short pause.4
Suppose Zhangsan attended the entrance examination of medical school last year and Lisi this year. The score of Zhangsan is 530 points and the lowest enrollment score last year was 515 points. The score of Lisi is 520 points and the lowest enrollment score this year is 490 points. So, both Zhangsan’s score and Lisi’s are high because each of them has their score significantly higher than the (contextually determined) standard of comparison (i.e., the lowest enrollment score). (See footnote(12)for the details.) The contrast in felicitousness between (ii) and (iii) under the scenario above implies that what are compared in example (ii) are the difference between Zhangsan’s score and last year’s lowest enrollment score (i.e., 15 points) and that between Lisi’s score and this year’s lowest enrollment score (i.e., 30 points); however, what are compared in (iii) are Zhangsan’s score (i.e., 530 points) and Lisi’s score (i.e., 520 points). Hence, if example (1c) can be treated as the reduced form of a bi ‘than’ comparative like (i), then we would expect example (iii) to be felicitous under the scenario above, contrary to fact. Second, a geng ‘GENG’ clausal comparative like (iv) does not have a bi ‘than’ comparative counterpart, as the contrast between (iv) and (v) illustrates.
4Similarly, in a Chinese comparative like (i), the compared object y (i.e., the standard of comparison) is syntactically realized as the bi ‘than’
phrase bi Lisi ‘than Lisi’ and adjoined to the predicate geng gao ‘GENG tall’. This extended predicate then forms a complete sentence with the other argument of the morpheme geng ‘GENG’, namely the subject NP Zhangsan, to have the meaning completely expressed.
However, one might say that example (1c) can be analyzed as a sequence consisting of two independent sentences because the two clauses contained inside can be uttered by different speakers in a dialogue, as shown below.
This way of reasoning, at first sight, sounds convincing, but it is not without problem because it offers no way to prevent us from saying that the two clauses involved in the following dialogue are also two independent sentences.
Although both (iiiA) and (iiiB) can be uttered by different speakers in a dialogue, no one would say (iiiB) can form an independent sentence by itself because (iiiB), being a subordinate clause, can never occur as an independent sentence by itself.
AsXing (2001:345–363, 2004)further points out, the Chinese geng ‘GENG’ clausal comparative consists of two or more clauses which can be combined together by coordinators, for example budan ‘but’, danshi ‘but’ and erqie ‘and’, or (prepositional) subordinators like lian ‘even’, guran ‘although’ or yaoshi if, as sentences in (4) illustrate.
Semantically, the last clause of the Chinese geng ‘GENG’ clausal comparative has to contain a degree phrase headed by the degree adverb geng ‘GENG’, which might have gengjia ‘GENG-ADD’ as alternate. This degree phrase functions to provide the ‘comparee’ degree with which the ‘standard degree of comparison’ provided by the degree phrase headed by the degree adverb like hen ‘HEN’ in the other clause (or the other clauses if the whole sentence consists of more than two clauses) is compared. For example, in (4a) the degree of roundness of Zhangsan’s eyes denoted by the degree phrase hen yuan ‘HEN round’ in the first clause plus the degree of erection of Zhangsan’s nose denoted by the degree phrase hen ting ‘HEN erect’ in the second clause functions as the ‘standard degree of comparison’ with which the ‘comparee’ degree (i.e., the size of Zhangsan’s ears) denoted by the degree phrase headed by the degree adverb geng ‘GENG’ (i.e., geng da ‘GENG big’) is compared. For convenience of exposition, in the rest of this paper I call the clause involving the degree phrase that denotes the ‘standard degree of comparison’ the standard clause and the clause containing the degree adverb geng ‘GENG’ the ‘comparee’ clause.
In addition to this characteristic, the syntactic and semantic properties about the relations among the clauses contained in a geng ‘GENG’ clausal comparative still include: the order between the clauses involved cannot be switches, as shown by the contrast between (1c) and (5a), it is not necessary for the standard NP to have the same grammatical function as the comparee NP (e.g., in (5b) the comparee NP is the object NP geng duo ren ‘GENG many people’ while the standard NP is the subject NP hen duo ren ‘HEN many people’); and it is not necessary for the predicates involved to be antonymous or to show a positive-negative polarity, as shown by (5a–c), respectively.
These characteristics work together to exclude the possibility of analyzing the Chinese geng ‘GENG’’ clausal comparatives as conjoined comparatives, as defined byStassen (1985:44).5
Besides, whenever the two (or more) degree-denoting elements in the Chinese geng ‘GENG’ clausal comparative are not directly dominated by the clauses further directly dominated by the whole construction, all the degree adverbs involved must be overtly realized and, more importantly, the coordinator danshi ‘but’ is obligatorily required, as the contrast between (6) and (7a,b) shows.
I shall argue that the coordinator danshi ‘but’ in examples like (7a,b) functions to help retain and intensify the comparison relation between the two ‘degrees’ in the Chinese geng ‘GENG’ clausal comparatives. Semantically, the coordinator danshi ‘but’ presupposes a contrast relation between the two conjuncts connected by it. In the Chinese geng ‘GENG’ clausal comparative, the two difference degrees ‘denoted’ by the two corresponding degree adverbs are compared with each other. Whenever these two degree-denoting elements are not directly dominated by the clauses that are further directly dominated by the whole construction, it becomes difficult for one to ‘capture’ the comparison relation between these two degree-denoting elements (i.e., the degree adverbs). Since the notion of contrast can be considered a special type of comparison, the coordinator danshi ‘but’ in this instance functions as the last resort to cooperate with the overt realization of the degree terms involved to rescue the comparison relation between these two degree adverbs. Thus, not only the coordinator danshi ‘but’ is obligatorily required but the degree term also has to be overtly realized.
Another important property of the morpheme geng ‘GENG’ in the Chinese geng ‘GENG’ clausal comparative is: although the degree adverb geng ‘GENG’ heading the degree phrase that provides the ‘comparee degree’ is obligatorily required in the Chinese geng ‘GENG’ clausal comparative, the degree adverb that functions to provide the
5Stassen (1985:44)imbues the conjoined comparative with the following characteristics: first, it consists of two structurally independent
clauses: one containing the comparee NP, whereas the other the standard NP. Second, the two clauses in question show a structural parallelism, as the grammatical function of the comparee NP in one of the clauses is reduplicated by the grammatical function of the standard NP in the other clause (i.e., it is a type of derived-case comparatives). Third, the comparative predicates involved either form an antonymous pair or exhibit a positive-negative polarity.
‘standard degree of comparison’ (e.g., hen ‘HEN’) can be omitted from the standard clause without affecting the grammaticality of the sentence, as (8) shows.6,7
However, I simply assume that, in examples like (8), the optionality of the degree adverb in the standard clause is an epi-phenomenon, and making such an assumption in fact does not significantly impact on the main theme I eventually argue for in this paper. My assumption is simply based on the semantic interpretation of examples like (8). As the semantic interpretation of (8) indicates, although Zhangsan’s height exceeds the contextually determined standard height of human beings in a significant difference, the difference in which Lisi’s height exceeds the contextually determined standard height of human beings further exceeds that significant difference. This semantic property shown by a Chinese geng ‘GENG’ clausal comparative like (8) serves as evidence for us to assume that the standard clause of the Chinese geng ‘GENG’ clausal comparative provides a context where the occurrence of a covert degree term (perhaps the covert positive morpheme) is licensed because semantically degree terms, asKennedy and McNally (2005:367)suggest, function to regulate the relationship between the degree of the property denoted by an adjective that modifies an entity (or a concept) and the (inherent or contextually determined) standard degree value on the scale denoted by that property.8
In contrast with the optionality of the degree adverb in the standard clause, the obligatoriness of the degree adverb geng ‘GENG’ might result from the labeling function it plays. Since the degree adverb geng ‘GENG’ in examples like (8) occurs as a marker to label the whole construction as a geng ‘GENG’ clausal comparative, the omission of the degree adverb geng ‘GENG’ from the comparee clause is prohibited. (See footnote(16)for further discussion.)
More significant here is that not all degree adverbs can occur in the standard clause of the Chinese geng ‘GENG’ clausal comparative, as shown by the contrast between (9a–d) and (10a,b).
6
See example (6)–(7a,b) and discussion there for the optionality of degree adverbs in the standard clause.
7Example (i), in contrast with the English comparative deletion construction like (ii), further indicates that no comparative deletion applies to the
Chinese geng ‘GENG’ clausal comparative.
8Kennedy and McNally (2005:367)characterize the meanings of degree morphemes (terms) in terms of the template in (i).
What distinguishes different degree morphemes from each other is the value of R – the specific restriction (or the restriction relation) they impose on the adjective’s degree argument.
At this point, we immediately encounter the question of what kinds of degree adverbs can occur in the standard clause of the Chinese geng ‘GENG’ clausal comparative. According toMa (1992),Lu and Ma (1999)andZhang (2002), Chinese degree adverbs can be classified as three types, depending on their distribution in different types of constructions, including the bi ‘than’ comparative, the bi-qilai ‘compare-qilai/compared with’ construction, and the non-comparative construction, and each type further consists of a strong and a weak group. To state it more clearly, the geng type (i.e., the GENG type) includes those that can occur only in the bi ‘than’ comparative and the bi-qilai ‘compare-qilai/compared with’ construction, the hen type (i.e., the HEN type) consists of those that can only occur in the bi-qilai ‘compare-qilai/compared with’ construction and the non-comparative constructions, and the zui type (i.e., the most type) contains only those occurring in the superlative, as illustrated by examples in (11)–(14), respectively.
As I have argued, one of the semantic properties of the degree adverb geng ‘GENG’ that occurs in the Chinese geng ‘GENG’ clausal comparative is to require the difference between the degree of ‘X-ness’ denoted by the ‘comparee’ element and the contextually determined standard degree of ‘X-ness’ in the comparee clause to be greater than the difference between the degree of ‘X-ness (or Y-ness)’ denoted by the ‘standard’ element and the contextually determined standard degree of ‘X-ness (or Y-ness)’ in the standard clause. For example, in (8) the difference between Lisi’s height and the contextually determined standard height of human beings must be greater than the difference between Zhangsan’s height and the contextually determined standard height of human beings. So, it is this specific property of geng ‘GENG’ that excludes the zui ‘most’ type of degree adverbs, either the strong or the weak group, from occurring in the standard clause of the Chinese geng ‘GENG’ clausal comparative. Besides, this specific semantic property of the degree adverb geng ‘GENG’ also excludes degree adverbs that belong to the strong group of the geng ‘GENG’ type from occurring in the standard clause of the Chinese geng ‘GENG’ clausal comparative because the degree adverb geng ‘GENG’ is also a member of the strong group of the geng ‘GENG’ type of degree adverbs.
2.2. Comparison of deviation and divergence
As I have pointed out, what are compared in the Chinese geng ‘GENG’ clausal comparative are (A) the difference between the degree of ‘X-ness’ denoted by the ‘comparee’ element and the contextually determined standard degree of ‘X-ness’ in the comparee clause, and (B) the difference between the degree of ‘X-ness (or Y-ness)’ denoted by the ‘standard’ element and the contextually determined standard degree of ‘X-ness (or Y-Y-ness)’ in the standard clause. Additionally, it is not necessary for the predicates involved to be antonymous or show a positive-negative polarity. Given these observations, I expect that a Chinese geng ‘GENG’ clausal comparative formed out of a ‘positive’
and ‘negative’ pair of adjectives would differ from its English counterpart in not showing the cross-polar anomaly. The facts bear out this expectation, as the contrast between (15a,b) and (16)–(17) illustrates (Hale, 1970; Bierwisch, 1987; Kennedy, 2001a:36).
More interestingly, Chinese geng ‘GENG’ clausal comparatives constructed out of the ‘positive’ and ‘negative’ pair of adjectives can be further divided into two types, depending on their semantic interpretations: one is represented by cases like (16a–d) and the other by the examples in (17). The former involve a comparison of deviation (henceforth COD), which compares the relative extents to which the two objects deviate from some standard value associated with the adjective. For instance, the meaning of (16a) can be paraphrased as in (18).
(18) The degree to which the blackness of Zhao Min’s hair exceeds the contextually determined standard of blackness of female hair is greater than the degree to which the whiteness of Zhao Min’s skin exceeds the contextually determined standard of whiteness of female skin.
This type of geng ‘GENG’ clausal comparative further shows the specific properties delineated in (A) through (D): (A) In contrast with the meaning of the COD type of Chinese geng ‘GENG’ clausal comparatives, standard bi ‘than’ comparatives compare the absolute projections of two objects on a scale at the semantic level, as shown in (19):
(B) Unlike standard comparatives, Chinese COD-like geng ‘GENG’ clausal comparatives entail that the properties predicated of the compared ‘objects’ are true in the absolute sense. This is in a way similar to whatKennedy (2001a) points out for the English COD comparative, as is verified by the contrast below.
More precisely, the fact that (20a) entails that the hands of that guy are long and his legs are short makes (20b) contradictory but (20a) not. This property is clearly related to the interpretation of Chinese COD-like geng ‘GENG’ clausal comparatives. Since the truth of an expression of the form ‘x is w’ is determined by whether the degree to which x is w exceeds an appropriate standard value, the fact that comparison deviation constructions compare the degrees to which two objects exceed their respective standard values derives the observed entailment patterns naturally.
(C) Interpretations of this type of Chinese geng ‘GENG’ clausal comparative are not restricted to comparatives formed out of the antonymous pairs of adjectives (Kennedy, 2001a:43).
(D) In Chinese geng ‘GENG’ clausal comparatives constructed out of antonymous pairs of adjectives, the COD interpretation is the only interpretation available. For example, (20a) has only the reading that the degree to which the length of that guy’s legs falls behind the standard length of human legs is greater than the degree to which the length of that guy’s hands exceeds the standard length of human hands.
The second type of Chinese geng ‘GENG’ clausal comparative that is formed out of adjectives of opposite polarity but not semantically anomalous is represented by examples like (17a,b), which involve a comparison of divergence.
Although the pairs of adjectives in (17a,b) are clearly opposites in some sense, (17a,b) are not anomalous only on a very specific interpretation: one in which the adjectives measure divergence from some common reference point (which, as (22) shows, need not be the same for the two objects) – a conventionalized value (e.g., on time for (17a)) – in different directions.
Hence, what is unique about comparisons of divergence and deviation is that both types of comparatives compare the degrees to which two objects deviate from some reference point (which need not be the same for the two objects) – a conventionalized value in the former case and a contextually determined standard value in the latter; whereas, standard comparatives, for example (19), compare the absolute measures of two objects on a scale at the semantic level. That is, what are compared in constructions involving a comparison of deviation or a comparison of divergence are two difference degree intervals rather than two (absolute) degree points (or intervals).9
2.3. Degree comparison in syntax
AsBeck et al. (2004)argue, the availability of degree abstraction would allow the generation of attributive degree binding structures and comparative subdeletion structures. So, the grammaticality of (23a), which involves attributive comparison, and the grammaticality of (23b), where a comparison between quantities of different sorts of stuff is involved, imply that the Chinese geng ‘GENG’ clausal comparative allows binding of degree variables in the syntax.
9
More noteworthy here, the Chinese geng ‘GENG’ clausal comparatives can co-occur with a dynamic verbal predicate, as (i)–(ii) illustrate.
Examples of this kind make the Chinese geng ‘GENG’ clausal comparatives so peculiar in the following two aspects: first, the degree adverb geng ‘more’ in the Chinese geng ‘GENG’ clausal comparatives differs from other degree adverbs (e.g., hen ‘HEN’) in that the former can modify a dynamic event predicate while the latter, for example hen ‘HEN’, cannot. Second, this kind of examples raises the question of what are compared in this type of geng ‘GENG’ clausal comparatives if a dynamic verb such as da ‘hit’ or sha ‘kill’ is considered as being non-gradable. However interesting and significant this type of the Chinese geng ‘GENG’ clausal comparatives is, to keep the paper from being too lengthy, following one anonymous reviewer’s suggestion, I defer considering this type of the Chinese geng ‘GENG’ clausal comparative here.
However, the grammaticality of (24a), in which the comparee clause is embedded inside a syntactic island, as well as the grammaticality of (24b), which lacks the negative island effect, indicates thatBeck et al.’s (2004)implication is empirically challenged by the Chinese geng ‘GENG’ clausal comparative.
In other words, the grammaticality of (23a,b) implies that the Chinese geng ‘GENG’ clausal comparative involves degree comparison and requires the degree variable involved to be bound. However, the insensitivity to the island effect shown by (24a,b) indicates that the degree variable involved should not be bound by a moved degree operator at the syntactic level.
Additionally, based on the semantic properties of the comparative morpheme geng ‘GENG’ that I have pointed out, the most proper translation for a Chinese geng ‘GENG’ clausal comparative like (25a) is (25b), in which the standard of comparison, namely the degree operator (i.e., Op) in [Spec, CP], must be syntactically local to the head of the comparative (i.e., blacker) (Williams, 1977; Hazout, 1995; Kennedy, 1999; Kennedy, 2001a,b).
The degree value (i.e., the degree of whiteness) of the degree operator in (25b) then is further contextually specified by that of the predicate very white in the first clause of the translation. However, the ungrammaticality of (26) implies that, instead of using a bi ‘than’ phrase as the Chinese counterpart of the English than phrase to introduce the standard of comparison, the Chinese geng ‘GENG’ clausal comparative directly uses the ‘standard clause’ (i.e., pifu hen bai ‘skin very white’) to introduce the standard of comparison. (See footnote(18)for discussion on the ungrammaticality of examples like (26)).
However, in contrast to an English clausal comparative, which strictly requires the degree operator to be local to the head of the comparative at LF, the Chinese geng ‘GENG’ clausal comparative allows either the standard clause or the comparee clause to occur inside a syntactic island, as the contrast between (27a,b) and (28a,b) illustrates (Ross, 1967; Chomsky, 1977).
So in a Chinese geng ‘GENG’ clausal comparative, the standard of comparison seems not to be necessarily local to the ‘head’ of the comparative, for example geng hei ‘GENG black’ in (28b).
Before concluding this section, I briefly highlight the empirical and theoretical questions to the current syntactic and semantic analyses of comparatives raised by the Chinese geng ‘GENG’ clausal comparative. First, as I have pointed out, what are compared in the Chinese geng ‘GENG’ clausal comparative are two difference degree intervals rather than two degree points. So any analysis of the syntax and semantics of the Chinese geng ‘GENG’ clausal comparative must not only derive the difference reading naturally but also must account for the challenge to the point-based analysis to the semantics of comparatives raised by this construction (Kennedy, 2001a; Schwarzschild and Wilkinson, 2002).
Second, besides the geng ‘GENG’ clausal comparative, Chinese still has a structurally simpler mono-clausal comparative like (1a), repeated as (29a), which differs from the geng ‘GENG’ clausal comparative like (29c) in syntax and semantics.
Many previous studies of Chinese comparatives not only group examples (29a,b) together as one type dubbed as the bi ‘than’ comparative by using the occurrence of the marker bi ‘than’ as criterion for classifying Chinese comparatives, but also analyze the marker bi ‘than’ as a comparative morpheme (Chao, 1968; Fu, 1977; Li and Thompson, 1981; Tsao, 1989; Paul, 1993; Liu, 1996; Shi, 2001; Liu, 2004; Xiang, 2005; Erlewine, 2007; Xu, 2007;
Lin, 2009).10So whether the morpheme geng ‘GENG’ in the Chinese geng ‘GENG’ clausal comparative is the same as the one in (29b) immediately becomes a question I cannot evade. Additionally, as two anonymous reviewers remind me, it is not entirely innocuous if the question of how the structurally more complex geng ‘GENG’ clausal comparative fits into a larger system of Chinese comparatives is missed.
Third, although the Chinese geng ‘GENG’ clausal comparative allows the generation of attributive comparison and comparison between quantities of different sorts of stuff, this construction surprisingly does not show the island sensitivity effect as the English clausal comparative does. So, can the Degree Abstraction Parameter (i.e., a language {does, does not} have binding of degree variables in the syntax) proposed byBeck et al. (2004:325)be maintained? If it can be, how can it be?
3. The proposal
I introduce as preliminary Kennedy’s (2001a) interval-based analysis to the semantics of comparatives first, and then propose an interval-based analysis to the semantics of the Chinese geng ‘GENG’ clausal comparative. To forecast the main themes of my analysis: first, the morpheme geng ‘GENG’, being a comparative morpheme denoting a greater-than relation, takes two arguments and requires one of them to be greater than the other along some dimension. Second, the comparative morpheme geng ‘GENG’ presupposes that the properties predicated of the compared ‘objects’ must be true in the absolute sense. Third, the type of arguments that the comparative morpheme geng ‘GENG’ takes depends on whether the compared ‘object’ is introduced by the bi ‘than’ phrase or the degree adverb contained in the standard clause of the Chinese geng ‘GENG’ clausal comparative. It is the degree adverb in the standard clause of the geng ‘GENG’ clausal comparative that provides a difference degree interval as the standard of comparison. This ‘difference-degree-interval-like’ standard of comparison further causes the comparative morpheme geng ‘GENG’ to require the other argument which it takes to be a difference degree interval.
3.1. Preliminary:Kennedy (2001a)
According toHale (1970),Bierwisch (1987), andKennedy (2001a:36), English comparatives constructed out of the ‘positive’ and ‘negative’ pair of adjectives like (30) are semantically anomalous, an anomaly referred to as the cross-polar anomaly.
(30) ?Alice is shorter than Carmen is tall.
The cross-polar anomaly, asKennedy (2001a:37)argues, cannot be explained in terms of syntactic ill-formedness because the structurally identical examples of ‘comparative subdeletion’ (i.e., CSD) where both adjectives have the same polarity are perfectly well-formed, as shown in (31).
(31) My watch is faster than yours is slow.
Given this,Kennedy (2001a:37) suggests that the factors underlying the cross-polar anomaly should involve an interaction of the semantics of positive and negative adjectives and the semantics of the comparative construction (Seuren, 1978; Ladusaw, 1986; Linebarger, 1980).
10This dichotomy is immediately challenged by the semantic distinction between (29a) and (29b): (29a) does not entail that both Zhangsan
and Lisi are happy but (29b), which contains the morpheme geng ‘GENG’, does. Although this semantic distinction seems to imply that the occurrence of geng ‘GENG’ might be used as a criterion for classifying Chinese comparatives, the situation is not so straightforward. In particular, even though (29b,c) both contain the morpheme geng ‘GENG’, example (29b) still differs from (29c) in that what are compared at the semantic level in (29b) are two absolute degrees (or degree intervals) (i.e., the degree of Zhangsan’s happiness and that of Lisi’s happiness) but what are compared in (29c) are the difference between the degree value of Zhangsan’s happiness and the contextually determined standard degree value of happiness and the difference between the degree value of Lisi’s happiness and the contextually determined standard value of happiness.
AsKennedy (2001a:37)points out, particularly significant here is that a large class of antonymous adjectives makes the relation in (32) valid, as shown by (33).
(32) x is more wposthan y if and only if y is more wnegthan x.
(33) Alice is shorter than Carmen if and only if Carmen is taller than Alice.
Within a model of point-based analysis to the semantics of gradable adjectives and comparatives, this fact, as Kennedy (2001a:38)argues, can be directly explained by adopting the following three natural assumptions. First, gradable adjectives are characterized as expressions that map objects to abstract representations of measurement (i.e., scales), which are sets of points (i.e., degrees) that are totally ordered along a dimension determined by the adjective (e.g., height, weight, . . .) (Cresswell, 1976; Hellan, 1981; von Stechow, 1984; Rullmann, 1995).
Second, comparatives define ordering relations between degrees.
Third, assuming the point-based analysis to the semantics of gradable adjectives, antonymous pairs of adjectives such as tall and short map identical arguments onto the same degrees, but they introduce the opposite ordering relations. Namely, such pairs are duals: for all antonymous adjectives wpos, wnegthat map their arguments onto a shared
scale S, and for all d1, d22 S, the relation in (34) holds.
(34) d1> wposd2, d2>wnegd1
Given these three assumptions, the truth condition of (32), for instance, can be paraphrased as in (35). (35) the degree to which Carmen is tall>tallthe degree to which Alice is tall,
the degree to which Alice is short>shortthe degree to which Carmen is short
Suppose degrees correspond to points in an ordered set, and positive and negative adjectives map their arguments onto the same degrees – an assumption necessary to explain the validity of constructions with the form in (32) – then (36) is equivalent to (37).
(36) the degree to which Alice is short>short the degree to which Carmen is tall
(37) the degree to which Alice is short>short the degree to which Carmen is short
This line of reasoning, asKennedy (2001a:38)points out, happens to make the wrong prediction about the cross-polar anomaly because we would incorrectly predict that (30) is grammatical. So, the point-based analysis to the semantics of gradable adjectives and comparatives is challenged by cross-polar anomaly.
To account for cross-polar anomaly,Kennedy (2001a) proposes an interval-based analysis to the semantics of comparatives with the following two major assumptions. First, adopting Landman’s (1991:110) definition of an interval for a linearly ordered set of points,Kennedy (2001a:52–57) analyzes degrees as intervals on a scale, and makes a structural distinction between two sorts of degrees: positive and negative degrees. Positive degrees (i.e., POS(S)) are intervals that range from the lower end of a scale to some point, while negative degrees (i.e., NEG(S)) are intervals that range from some point to the upper end of a scale.
(38) a. POS(S) = {d S j9p12 d 8p22 S[p2 p1! p22 d]}
b. NEG(S) = {d S j9p12 d 8p22 S[p1 p2! p22 d]}
In all of these, asKennedy (2001a:53)argues, the result amounts to inferring that the positive and negative projections of an object x on a scale S are intervals in complementary distribution on the same scale, as the diagram in (39) shows. (39) S: 0 — poss(x) — — negs(x)! 1
The adjectival polarity, therefore, can be characterized as a difference in the ranges of the functions denoted by positive and negative adjectives: positive adjectives denote functions from individuals to positive degrees; negative adjectives denote functions from individuals to negative degrees.
Kennedy (2001a:61) further argues that it is possible to identify the difference between any two degrees as the interval that is contained in one but not the other. Seen in this light, ‘difference degrees’, like positive degrees, correspond to finite closed intervals. Hence, what we have to do in interpreting difference comparatives, asKennedy (2001a:61)suggests, is introduce a function that maps difference degrees onto degrees with minimal elements that correspond to the zero point of the scale in a structure-preserving way, as (40) shows.
(40) [[ZERO]] = {<d1, d2>j
(i) 8p1, p22 d1[p1< p2! 9p’1, p’22 d2[p’1< p’2]]^
(ii) 8p’1, p’22 d2[p’1< p’2! 9p1, p22 d1[p1< p2]]^
(iii) MIN(d2) = MIN(S)
Thereby, a difference comparative with the form like (41a), with the measure phrase 12 cm, can be assigned the interpretation in (41b).
(41) a. Alice is 12 cm shorter than Carmen (is).
b. short(a) > short(c)^ ZERO(short(a) short(c)) 12 cm
As (41b) shows, the difference between the two negative degrees (i.e., short(a) and short(c)) is a closed interval on the scale; therefore, a mapping to a degree in the set of degrees named by the measure phrase can be established by the ZERO function.
The second assumption Kennedy (2001a:44)makes in his interval-based account for cross-polar anomaly is as follows: comparatives are semantically well-formed only if they define ordering relations between the same sorts of degrees: between positive degrees, between negative degrees, or between degrees that measure divergence from a referent point.
On these assumptions, the cross-polar anomaly shown by examples like (30) can be explained naturally because an ordering relation, asKennedy (2001a:58)suggests, requires its arguments to be elements of the same sort. However, in (30) the adjective in the main clause (i.e., short) is negative while the adjective in the than clause (i.e., tall) is positive; therefore, (30) is ill-formed. In a difference comparative like (41a), the difference between the two negative degrees, being a closed interval on the scale, belongs to the same kind of set as the degree denoted by 12 cm does; therefore, (41a) is well-formed. Thus, difference degrees, either between positive or negative degrees, can be mapped to degrees in the set of intervals that begin at the zero point of a scale. Hence, comparisons between such kind of derived degrees should be semantically well-formed because orderings between them can be defined.
3.2. The semantics of the Chinese geng clausal comparative: an interval-based analysis
Assuming Kennedy’s (2001)interval-based analysis to the semantics of comparatives, in the following I first propose a preliminary analysis for the semantics of the comparative morpheme geng ‘GENG’ that will be further revised as I discuss how the Chinese geng ‘GENG’ clausal comparative fits into a larger system of Chinese comparatives. The main theme of the preliminary analysis is as follows: in the Chinese geng ‘GENG’ clausal comparative, the morpheme geng ‘GENG’, in addition to presupposing that the properties predicated of the compared objects are true in the absolute sense, is a comparative morpheme that takes two difference degree intervals as argument and requires one of them to be greater than the other along some dimension.11For example, in the standard clause of (42), the degree adverb hen ‘HEN’ not only requires the degree interval denoted by the whiteness of Zhao Min’s skin (i.e., white(zms)) to be greater than the degree interval denoted by the contextually determined standard 11One anonymous reviewer wonders whether there is an argument here that we really need to compare difference degree intervals. The strongest
evidence for the assumption that what are compared in the Chinese geng ‘GENG’ clausal comparative are two difference degree intervals comes from the fact that the Chinese geng ‘GENG’ clausal comparative, for example (i), does not show cross-polar anomaly.
degree value of whiteness of female skin (i.e., ds(white)), but also requires the difference between these two degree
intervals to be significant, as (43) shows (Graff, 2000; Kennedy, 2005, 2007; Liu, 2010).12
Since the semantics of the degree morpheme hen ‘HEN’ requires that the difference degree interval between the degree interval denoted by the whiteness of Zhao Min’s skin and that denoted by the contextually determined standard degree value of whiteness of female skin (i.e., ZERO(white(zms) ds(white))) has to be significant, I further suggest that this
‘difference degree interval’ component of the semantics of the degree morpheme hen ‘HEN’ is highlighted and selected as the standard of comparison by the comparative morpheme geng ‘GENG’ in the Chinese geng ‘GENG’ clausal comparative at the syntactic level. The comparative morpheme geng ‘GENG’ in the comparee clause of (42) then requires the comparee clause to generate a difference degree interval that is greater than the one provided by the degree adverb hen ‘HEN’ contained in the standard clause as follows: the morpheme geng ‘GENG’ requires the comparee clause of (42) to generate a degree interval denoted by the blackness of Zhao Min’s hair (i.e., black(zh))) first; next this degree interval is required to be greater than the degree interval denoted by the contextually determined standard degree value of blackness of female hair (i.e., ds(black)); and the difference between these two degree intervals must be significant also, as (44) shows.
(44) black(zmh) > ds(black)^ ZERO(black(zmh) ds(black)) > 0^ ZERO(black(zmh) ds(black)) is significant
The comparative morpheme geng ‘GENG’, due to its comparison function, further requires the difference degree interval provided by the comparee clause to be greater than the difference degree interval provided by the degree adverb in the standard clause. So, example (42) can be assigned an interpretation like (45):
(45) white(zms) > ds(white)^ ZERO(white(zms) ds(white)) > 0^ ZERO(white(zms) ds(white)) is significant ^
black(zmh) > ds(black)^ ZERO(black(zmh) ds(black)) > 0^ ZERO(black(zmh) ds(black)) is significant^
ZERO(black(zmh) ds(black)) > ZERO(white(zms) ds(white))
(45) has these properties: first, the ZERO function is invoked to compare two degree intervals. Second, both difference degree intervals are significant. Third, the difference degree interval provided by the comparee clause is greater than the one provided by the standard clause. Fourth, given the presupposition of the comparative morpheme geng ‘GENG’, the properties predicated of the compared ‘objects’ (i.e., the whiteness of Zhao Min’s skin and the blackness of her hair) are true in the absolute sense.
Taking a line similar toKennedy’s (2001a)analysis to English clausal comparatives involving a comparison of divergence, I suggest that the Chinese geng ‘GENG’ clausal comparative that involves a comparison of divergence, for example (46), can be analyzed in the same way as I have done for the Chinese geng ‘GENG’ clausal comparative that involves a comparison of deviation. However, the two differ in one important aspect: the properties of comparatives involving a comparison of divergence suggest that adjectives in this type of comparatives map their arguments directly onto difference degrees because both members of the ‘antonymous’ pair, for example kuai ‘fast’ and man ‘slow’ in (46), accept measure phrases, as (47a,b) illustrate.
12The requirement that in (42) the difference between the two degree intervals must be significant is clearly shown by the fact that example (i), which
contains a predicative adjective modified by the degree adverb hen ‘HEN’, is unacceptable in scenario A (i.e., this rope is 15 m long, and that one is 14 m long), which involves a crisp judgment, but acceptable in scenarios B (i.e., this rope is 15 m long, and that one is 10 m long), which does not.
Put more precisely, in example (46) and (47a,b), both the adjective kuai ‘fast’ and man ‘slow’ measure the degrees to which their arguments diverge from some arbitrary point (whatever counts as ‘on time’ in the context of utterance), as (48) shows.
The acceptability of (46), therefore, can be accounted for by making an additional assumption that, on their difference interpretation, these adjectives (i.e., kuai ‘fast’ and man ‘slow’) include the ZERO function as part of their meaning. So, the logical representation of (46) is simply like a standard interpretation except for the addition of the ZERO function, which, by hypothesis, is built into each adjective’s meaning, as shown by (49).
(49) ZERO(slows(minute hand)) > 0^ ZERO(slows(minute hand)) is significant^
ZERO( fasts(second hand)) > 0^ ZERO( fasts(second hand)) is significant^
ZERO(slows(minute hand)) > ZERO( fasts(second hand))
This analysis straightforwardly explains why an antonymous pair of adjectives in the Chinese geng ‘GENG’ clausal comparatives which involve a comparison of divergence fails to make valid the substitution instances of (32), repeated as (50), but the same does not happen in an ordinary comparative, as illustrated by the different uses of kuai ‘fast’ and man ‘slow’ in (51a,b).
The reason why (51a) is false can be briefly described as follows: if both of our watches exceed the actual time but yours is farther ahead of it than mine is, then neither of these two watches is behind the actual time. So, (51a) is invalid. Thus far, the main themes of my analysis of the geng ‘GENG’ clausal comparative like (42) and (46) can be summarized as follows: first, the morpheme geng ‘GENG’ is a comparative morpheme. Second, this comparative morpheme takes two difference degree intervals as arguments and requires one of them to be greater than the other along some dimension. Third, it is the degree adverb in the standard clause of the geng ‘GENG’ clausal comparative that provides the standard of comparison (i.e., the difference degree interval) at the syntactic level. Fourth, the comparative morpheme geng ‘GENG’ presupposes that the properties predicated of the compared ‘objects’ must are true in the absolute sense.
However, this analysis of the semantics of the comparative morpheme geng ‘GENG’, as I have already pointed out, is immediately challenged by the fact that the morpheme geng ‘GENG’ is also found in a bi ‘than’ comparative like (29b), repeated as (52), if the marker bi ‘than’, as Lin (2009) suggests, is considered as a comparative morpheme.
Moreover, two anonymous reviewers have reminded me that it is not entirely innocuous if the question of how the structurally more complex geng ‘GENG’ clausal comparative fits into a larger system of Chinese comparatives is missed. 4. A new dichotomy of Chinese comparatives
In this section, I propose that Chinese comparatives should be classified as the presupposition comparative on the one hand, and the non-presupposition comparative on the other by using the occurrence of the morpheme geng ‘GENG’ as criterion. Central to this proposal are the following assumptions: first, there is only one geng ‘GENG’ in Chinese and it is a comparative morpheme. The semantics of this comparative morpheme (the final version) are as follows: the comparative morpheme geng ‘GENG’, in addition to presupposing that the properties predicated of the compared objects must be true in the absolute sense, takes two arguments and requires one of them to be greater than the other along some dimension. The type of arguments that the comparative morpheme geng ‘GENG’ takes at the syntactic level depends on whether the compared ‘object’ is introduced by the bi ‘than’ phrase or by the degree adverb contained in the standard clause of the geng ‘GENG’ clausal comparative. On one hand, the bi ‘than’ phrase introduces (individual, time, or location) arguments for the comparative morpheme geng ‘GENG’, and these arguments are mapped onto absolute degree intervals at the semantic level by using Heim’s (1985) direct analysis to phrasal comparatives (Lin, 2009). On the other hand, the degree adverb contained in the standard clause of the geng ‘GENG’ clausal comparative introduces a difference degree interval as argument for the comparative morpheme geng ‘GENG’ in the geng ‘GENG’ clausal comparative.
Second, Chinese has a covert comparative morpheme (i.e., ), which occurs only in a comparative construction not containing the comparative morpheme geng ‘GENG’.13 The covert comparative morpheme differs from the comparative morpheme geng ‘GENG’ in that (A) it does not presuppose that the properties predicated of the compared ‘objects’ must be true in the absolute sense; and (B) it takes two individuals (or times or locations) as arguments and requires one of them to be greater than the other along some dimension. In other words, I analyze Chinese comparatives
13As noted in footnote(2), the types of Chinese comparatives like (i) and (ii) there, repeated as (i)–(ii) here, are not discussed in this paper.
containing this covert comparative morpheme as phrasal comparatives and use Heim’s (1985) ‘direct analysis’ to interpret them.
Third, instead of assuming that the morpheme bi ‘than’ not only has the function of English than but also has the function of the English comparative morpheme -er, I suggest that the morpheme bi ‘than’ simply functions like the English than in introducing the compared ‘object’ only.
Fourth, I assume a revised dyadic argument comparison analysis to the Chinese bi ‘than’ comparatives. (SeeLin (2009)for discussion of other analyses on the Chinese bi ‘than’ comparative, includingTsao (1989),Paul (1993),Liu (1996),Xiang (2005), andErlewine (2007).) According toLin (2009), a Chinese bi ‘than’ comparative like (53) is analyzed as a dyadic argument comparative construction with a structure like (54), in which the marker bi ‘than’, being a dyadic argument comparative degree head, carries the function of the English than as well as that of the English comparative morpheme -er.
Syntactically the morpheme bi ‘than’ which takes three arguments (i.e., the individual argument Lisi, the time argument jintian ‘today’, and the location argument zai jia li ‘at home inside’) and must be flanked by constituents of the same type; the whole DegP-shell headed by the marker bi ‘than’ is an adjunct adjoined to the predicate of comparison. Semantically, the morpheme bi ‘than’ has a denotation like (55), a semantic function similar to that of the English comparative morpheme -er.
(55) jjbijj = (ll)i(li)j(lw)k
lxlP<d, <(l), <(i), <(e), <e,t>>>>>(ll’)i(li’)j(lz)kly
[imax d [P(d)(l’)(i’)(z)(y)] > imax d [P(d)(l)(i)(w)(x)]]
So (53) is analyzed as a (multiple)-phrasal comparative by usingHeim’s (1985) direct analysis to interpret it. Here, I basically followLin’s (2009)analysis to the Chinese bi ‘than’ comparative except that I suggest that the marker bi ‘than’ is a dyadic argument preposition head which simply functions as the English than to introduce the compared ‘object’ only. It is the covert comparative morpheme that carries the function of the English comparative morpheme -er.14
With these assumptions, I suggest that Chinese comparatives like (56a,b) belong to the same type, namely the presupposition comparative, though (56b) contains the marker bi ‘than’ while (56a) does not.
On one hand, in a presupposition comparative like (56a) (i.e., the geng ‘GENG’ clausal comparative), the degree adverb hen ‘HEN’ in the standard clause provides a difference degree interval as the standard of comparison at the syntactic level, and causes the comparative morpheme geng ‘GENG’ to require the other argument which it takes to be a difference degree interval, too. Among these two difference degree intervals, the latter must be greater than the former. 14The purpose of my dichotomizing Chinese comparatives is to show how particular the Chinese geng ‘GENG’ clausal comparative is within a
larger system of Chinese comparatives, especially from the semantic properties of difference comparison and presupposition as well as its multi-clausal syntactic structure. So, it will not have much impact on the dichotomy I propose here even ifLin’s (2009)dyadic argument comparison analysis on the Chinese bi ‘than’ comparative is challenged.
Moreover, the presupposition requirement of the morpheme geng ‘GENG’ further requires both Zhangsan and Lisi to be happy.
On the other hand, in a presupposition comparative like (56b), the marker bi ‘than’ introduces the individual Lisi as the compared ‘object’; therefore, what are compared by the comparative morpheme geng ‘GENG’ at the syntactic level are two individuals (i.e., Zhangsan and Lisi). AssumingHeim’s (1985)direct analysis to the semantics of phrasal comparatives, the dimension in (56b), namely happiness, is a function from individuals x to degrees d; therefore, the comparative morpheme geng ‘GENG’ in (56b) has a denotation like (57), in which x and y are individuals, and the meaning of (56b) is equivalent to ‘The maximal degree interval denoted by Zhangsan’s degree of happiness is greater than the maximal degree interval denoted by Lisi’s degree of happiness.’
(57) lx.lP<d,<e,t>>.ly [imaxd [P(d)(y)] > imaxd [P(d)(x)]]^
the properties predicated of x and y are true in the absolute sense
In addition, the presupposition characteristic of the morpheme geng ‘GENG’ further requires both Zhangsan and Lisi to be happy.15
In contrast with the presupposition comparative, comparatives like (58a,b), which do not contain the comparative morpheme geng ‘GENG’, belong to the non-presupposition comparative and contain the covert comparative
morpheme .
Semantically, the covert comparative morpheme simply denotes a greater-than relation, and takes as arguments the compared object (i.e., the individual Lisi) introduced by the bi ‘than phrase and its correspondent in the main clause (i.e., the individual Zhangsan), but does not presuppose that the properties predicated of the compared ‘objects’ are true in the absolute sense. So the covert comparative morpheme in (58a,b) has a denotation like (59), in which x and y are individuals and, assumingHeim’s (1985)direct analysis to the semantics of phrasal comparatives, they are mapped onto two absolute degree intervals by the dimension function.
(59) lx.lP<d,<e,t>>.ly [imaxd [P(d)(y)] > imaxd [P(d)(x)]]
15
Assuming this way of classifying Chinese comparatives, I would say that example (1d), repeated as (i), is the reduced form of either (1b) or (1c), repeated as (ii) and (iii), respectively.
In other words, examples like (i) are ambiguous between a reading of comparing two difference degree intervals and a reading of comparing two absolute measures of two objects along some dimension. Thus, the actual meaning of (i) depends on what type of ‘standard of comparison’ the discourse provides.
Hence, what are compared at the syntactic level in a non-presupposition comparative like (58a,b) are two individuals; but what are compared at the semantic level are two corresponding absolute degree intervals.16
5. Implicational parameters of comparison
Taking as fundamental, my study on the syntax and semantics of the Chinese geng ‘GENG’ clausal comparative, in this section, I discuss the parameters of comparison across languages implied by my study on the syntax and semantics of the Chinese geng ‘GENG’ clausal comparative. As I have pointed out, the Chinese geng ‘GENG’ clausal comparative, on one hand, allows both attributive comparison and comparison between quantities of different sorts of stuff; on the other hand, either the standard clause or the comparee clause of the geng ‘GENG’ clausal comparative can be embedded in a syntactic island. So it is not implausible for us to say that the Chinese geng ‘GENG’ clausal comparative involves degree comparison but no degree operator movement at the syntactic level. These characteristics are illustrated by (60a–c), respectively.
16Since the Chinese geng ‘GENG’ clausal comparative not only allows degree comparison but also involves presupposition necessarily, one
anonymous reviewer wonders why the geng ‘GENG’ clausal comparative clusters together the properties of ‘degree comparison’ and ‘presupposi-tion’ in this way. As I have argued, the comparative morpheme geng ‘GENG’ takes two arguments and requires one of them to be greater than the other along some dimension. The type of arguments which the comparative morpheme geng ‘GENG’ takes at the syntactic level depends on whether the compared ‘object’ is introduced by a bi ‘than’ phrase or by the degree adverb contained in the standard clause of the geng ‘GENG’ clausal comparative. And it is the ‘difference degree interval’ component of the semantics of the degree adverb contained in the standard clause that is highlighted and selected as the standard of comparison. AsLiu (2010)argues, in Chinese a predicative gradable adjective must appear in a complex form (e.g., a form with adverbs of degree) unless it occurs in a predicate-accessible operator[-wh]domain, where the occurrence of the covert positive
morpheme is licensed. This specific characteristic of the Chinese predicative gradable adjective guarantees that the predicative gradable adjective in the standard clause of the geng ‘GENG’ clausal comparative is always modified by a degree adverb. This degree adverb then saturates one of the two arguments of the comparative morpheme geng ‘GENG’ by the ‘difference degree interval’ component of its semantics. In addition, the semantic meaning of the degree adverb geng ‘GENG’ (i.e., even more) further requires the properties predicated of the compared ‘objects’ to be true in the absolute sense. In other words, the cluster of ‘degree comparison’ and ‘presupposition’ in the Chinese geng ‘GENG’ clausal comparative in fact is a result derived from the interaction between the specific property of the Chinese predicative gradable adjective (i.e., a predicative gradable adjective must be modified by a degree adverb) and the semantic meaning of the degree adverb geng ‘GENG’ (i.e., even more). In this way, the cluster of ‘degree comparison’ and ‘presupposition’ in the Chinese geng ‘GENG’ clausal comparative gets a principled explanation. However, the covert comparative morpheme occurs only in comparatives where the standard of comparison is introduced by the bi ‘than’ phrase. Since the marker bi ‘than’ cannot takes a clausal complement syntactically, degree comparison (which is possible only if a gradable adjectival predicate is involved) can never be allowed in a comparative construction containing the covert comparative morpheme. Hence, the distribution of the covert comparative morpheme is restricted to comparative constructions involving the individual (or phrasal) comparison, as the ungrammaticality of (i) shows.
Among these three examples, the grammaticality of (60a,b), in particular, is reminiscent ofBeck et al.’s (2004:325) Degree Abstraction Parameter, as stated in (61).
(61) Degree Abstraction Parameter (DAP)
A language {does, does not} have binding of degree variables in the syntax.
AsBeck et al. (2004)argue, the availability of degree abstraction would allow the generation of attributive degree binding structures and subdeletion structures; that is, the well-formedness of (60a,b) implies that the Chinese geng ‘GENG’ clausal comparative has binding of degree variables in syntax. However, this implication is immediately challenged by the grammaticality of (60c), which does not show the negative island effect.
At this point, we seem to have reached an impasse; however, the well-formedness of (60c) implies thatBeck et al.’s (2004)Degree Abstraction Parameter can be maintained if the following sub-parameter of comparison for languages with degree abstraction is proposed:
(62) Movement versus Non-Movement Parameter in Degree Abstraction
The binding relation between the operator and the degree variable {is, is not} a movement one.
In other words, languages with degree comparison in syntax can be further divided into two subtypes, depending on how the degree variable is bound. One is represented by languages like English, in which the degree variable is bound by the moved compared constituent. According toKennedy (1999, 2001a), in English clausal comparatives the degree morpheme that heads the compared constituent and the head of the comparative can be assigned the interpretations in (63a,b), respectively, where G is a function from objects to degrees, Q is a function from properties to truth values (the semantic value of a clausal constituent with an extracted DegP), and max is a maximality operator that returns the maximal element of an ordered set of objects (Bartsch and Vennemaun, 1973; von Stechow, 1984).
(63) a. Deg = lGlQ.max{djQ(lx.G(x) d)} b. er/more = lGldlx.G(x) > d
Since the comparative clause supplies the ‘standard of comparison’ argument for the comparative morpheme -er, which establishes a relation between two degrees, we expect the interpretation ultimately assigned to (64a) to be truth-conditionally equivalent to the representation given in (64b).
With this analysis, the compared constituent in (64a) (i.e., [DegPDeg wide]) must move at LF and take scope over
the rest of the clause to generate the right interpretation of the comparative clause because the quantificational force of the comparative clause (i.e., the maximality operator) is introduced by the degree morphology on the compared constituent rather than by a higher operator (Baker, 1970).
The other subtype is one in which the in situ degree variable is spelled out as a degree adverb and is unselectively bound by an operator base-generated in [Spec, CP] of the standard clause or of the clause that contains the standard clause but is directly dominated by the whole construction. For example, in the Chinese geng ‘GENG’ clausal comparative, the operator that unselectively binds the degree variable is base-generated in [Spec, CP] of the standard clause or of the clause that contains the standard clause but is directly dominated by the whole construction, as shown by (65a,b), respectively (Huang, 1982; Tsai, 1994; Tsai, 1999).17
17
BehindTsai’s (1999:61)analysis to wh-movement in Chinese is that only nouns, in contrast with adverbs that are intrinsic operators and must move to create variables, may enter into unselective binding because they are the sole providers of non-pronominal variables. This might prompt us to assume
As I have argued, in the standard clause of a Chinese geng ‘GENG’ clausal comparative like (65a), the degree adverb hen ‘HEN’ not only requires the degree interval denoted by Zhangsan’s height (i.e., tall(zs)) to be greater than the degree interval denoted by the contextually determined standard height of human beings (i.e., ds(tall)), but also
requires the difference degree interval between these two degree intervals (i.e., ZERO(tall(zs) ds(tall))) to be
significant, as (66) shows.
(66) tall(zs) > ds(tall)^ ZERO(tall(zs) ds(tall)) > 0^ ZERO(tall(zs) – ds(tall)) is significant
This ‘difference degree interval’ component of the semantics of the degree morpheme hen ‘HEN’ is further highlighted and selected as the standard of comparison by the comparative morpheme geng ‘GENG’ in the Chinese geng ‘GENG’ clausal comparative. To make the comparison possible, a correspondent difference degree interval greater than the one provided by the standard clause must be generated in the comparee clause. So, the morpheme geng ‘GENG’ in the comparee clause of (65a) not only requires the degree interval denoted by Lisi’s height (i.e., tall(ls)) to be greater than the degree interval denoted by the contextually determined standard height of human beings (i.e., ds(tall)), but also requires the difference degree interval between these two degree intervals to be
significant, as (67) shows.
(67) tall(ls) > ds(tall)^ ZERO(tall(ls) ds(tall)) > 0^ ZERO(tall(ls) – ds(tall)) is significant
In addition, the comparative morpheme geng ‘GENG’ further requires the difference degree interval provided by the comparee clause to be greater than the difference degree interval provided by the standard clause. So example (67a) can be assigned an interpretation like (68).
(68) tall(zs) > ds(tall)^ ZERO(tall(zs) ds(tall)) > 0^ ZERO(tall(zs) ds(tall)) is significant^
tall(ls) > ds(tall)^ ZERO(tall(ls) ds(tall)) > 0^ ZERO(tall(ls) ds(tall)) is significant^
ZERO(tall(ls) ds(tall)) > ZERO(tall(ls) ds(tall))
that the degree variable bound by the base-generated operator in the Chinese geng ‘GENG’ clausal comparative is a noun. This assumption in fact is not impossible.Izvorski (1995)argues for a movement account in which the terms targeted by movement in the English CSD (and possibly CD as well) construction are the amount adjuncts in what quantity and to what degree, as shown by (i)–(ii), rather than the degree heads how (many/much) in (iii)–(iv).
(i) In what quantity does Dennis have tattoos? (ii) To what degree were the shapes thick?
(iii) *How many does Dennis have [DPhow may tattoos]]?
(iv) *[CPHow are the shapes [DegPhow thick]]?
Further evidence for the assumption that the degree variable in the Chinese geng ‘GENG’ clausal comparative is a noun comes fromHuang’s (1982)treatment of the non-sensitivity to island effects shown by (zai) nali ‘(at) where’ and (zai) shenmeshihou ‘(at) when’ in Chinese. AsHuang (1982)argues, (zai) nali ‘at where’ and (zai) shenmeshihou ‘at when’, though always analyzed as adjuncts, actually are nouns. So, it is not implausible for us to say that the degree variable in the Chinese geng ‘GENG’ clausal comparative, though it looks like an adverbial adjunct, in fact can be considered a noun.
Since the operator based-generated in [Spec, CP] of the standard clause unselectively binds the degree adverb hen ‘HEN’, it is not unreasonable for us to say that the degree operator (i.e., Op) ‘inherits’ all the semantic properties of the degree adverb hen ‘HEN’. As a result, the ‘exceed the standard’ part of the meaning of the geng ‘GENG’ clausal comparative (65a) as well as the presupposition that both Zhangsan and Lisi are tall is naturally derived under my analysis.18
Put simply, although English clausal comparatives and the Chinese geng ‘GENG’ clausal comparative both involve degree comparison in syntax, they differ from each other in the way of binding the degree variable. The sub-parameter proposed (i.e., Movement versus Non-Movement Parameter in Degree Abstraction) further explains why the proper English translation of the Chinese geng ‘GENG’ clausal comparative (28b), repeated as (69a), is (69b), in which the standard of comparison introduced by the than phrase (i.e., the degree operator Opi) must be local to the head of the comparative (i.e., blacker), but this kind of local relation is not necessarily
required in the Chinese geng ‘GENG’ clausal comparative, as shown by the contrast between (69a) and (69b).19
Thus far, the discussion amounts to saying that Chinese does have degree comparison in syntax, contra the claim made byXiang (2005),Kennedy (2007), andLin (2009)that Chinese has only individual comparison in syntax. So we might want to say that both Chinese and English have individual and degree comparison in syntax. However, English uses the same type of comparative construction, namely the than particle comparative, to express degree and individual comparison, depending on whether the complement of than is a clause or a noun phrase, whereas Chinese uses the geng ‘GENG’ clausal comparative to express degree comparison but the bi ‘than’ comparative to express individual comparison at the syntactic level.
18One anonymous reviewer points out that if Zhangsan hen gao ‘Zhangsan HEN tall’ has an analysis as a degree and if Chinese has degree
comparison, then why is (i) ungrammatical?
The ungrammaticality of examples like (i) in fact is not related to my proposal that the Chinese geng ‘GENG’ clausal comparative involves degree comparison in syntax. AsBeck et al. (2004),Kennedy (2005, 2007),Xiang (2005)andLin (2009)argue, the Chinese bi ‘than’ comparative can only be analyzed as a phrasal comparative; in other words, the marker bi ‘than’ cannot take a clausal complement. So, (i), in which the marker bi ‘than’ takes a clause as complement, is ungrammatical.
19
Given that the modified English translation for (69a) provided by one anonymous reviewer (i.e., (i)) is grammatical and does not show island sensitivity, the reviewer suggests that the proposed parameter (i.e., the binding relation between the degree operator and the degree variable is achieved via movement or not) might not differ according to language but rather according to the syntax of the construction.
(i) The statement that the skin of Zhao Min is very white is absolutely true, but I wonder whether her hair is even blacker.
However, I argue that this alternative cannot hold because example (i) can be considered as one involving an elided than phrase that is local to the head of the comparative (i.e., blacker), as (ii) shows.
(ii) The statement that the skin of Zhao Min is very white is absolutely true, but I wonder whether her hair is even blacker than her skin is white. But (69a) cannot be analyzed as one containing an elided bi ‘than’ phrase that is local to the head of the comparative (i.e., geng hei ‘GENG black’), as the ungrammaticality of (iii) illustrates.
In other words, (69a) and the English example (i) cannot be considered as the same type of construction. So the alternative suggested by the reviewer cannot be maintained.
