Are Degree Achievements Really Achievements?

Transcription

1 Are Degree Achievements Really Achievements? Jimmy Lin College of Information Studies University of Maryland, College Park Abstract This paper, which builds of t work of Hay, Kennedy, and vin (1999), examines t puzzling aspectual behavior of so-called degree achievements. Drawing evidence from Mandarin Chinese, I argue that degree achievements without difference values denote punctual events, i.e., ty are true achievements. I present an analysis that is consistent with facts from both English and Mandarin; to explain t complex aspectual behavior of degree achievements, my account appeals to coercion operators that are licensed to resolve type class. It differs from previous tories in that complex aspectual behavior arises from interactions between predicates and otr sentential elements, and not from properties inrent to t predicates tmselves. Cross-linguistic differences can be attributed to t availability of tse operators, which is a parameter of Universal Grammar. 1 Introduction So-called degree achievements, a class of verbs that includes widen, cool, warm, and dry, display interesting distributional characteristics with respect to telicity and present challenges for tories of argument structure based on aspectual properties, e.g., (Dowty, 1979, 1991; Tenny, 1992, 1994). In this paper, I extend t work of Hay, Kennedy, and vin (1999), which draws interesting connections between boundedness of t difference value in a degree achievement and t telicity of t denoted event. While I agree with tir basic claim that t event is telic if t difference value is bounded, and atelic if not, t facts are more complicated than has been suggested. Drawing evidence from both Mandarin Chinese and English, I will argue that degree achievements without explicit difference values are punctual events, i.e., ty are true achievements, and different aspectual readings arise through coercion effects. Crosslinguistic differences can be captured by t availability of different coercion operators, which is a parameter of Universal Grammar. To begin, consider English degree achievements with explicit difference values: (1) a. T gap widened 3% in a year/*for a year. b. T soup cooled 5 degrees in an hour/*for a hour. 1

2 T aspectual behavior of tse sentences are straightforwardly explained by t tory of Hay, Kennedy, and vin. Degree achievements impose a homomorphism between t temporal progress of t event and t gradable property underlying t verb; if t difference value is bounded, t event is correspondingly bounded, nce telic. As a historic note, t term degree achievement, attributed to Dowty (1979), is actually somewhat of a misnomer t examples above suggest that such verbs are actually accomplishments. For consistency, however, I will continue to employ t label degree achievement to refer to this class of verbs. In Mandarin Chinese, degree achievements arise from t addition of t particle le to a stative verb (note that since Mandarin lacks t adjective/verb distinction, t stative verb is t analog of t English adjective): (2) a. shu 4 tree tall shi 2 ten gung 1 fen 1 centimeter T tree is ten centimeters tall. b. shu 4 tree tall shi 2 ten gung 1 fen 1 centimeter T tree grew ten centimeters. Example (2a) asserts a state, i.e., that of t tree being ten centimeters tall. In contrast, example (2b) describes a change of state, i.e., t event of t tree becoming by ten centimeters. Since tse examples represent a minimal pair differing only in t absence or presence of t particle le, it must be t source of t inchoative reading. For more in-depth discussions about t particle le and its inchoative function, please refer to (Lin, 2004). In t presence of a difference value, Mandarin degree achievements exhibit t same aspectual behavior as tir English counterparts: (3) a. ta 1 zai 4 at nian 2 year nei 4 in He grew three centimeters in a year. san 2 three gung 1 fen 1 centimeter b. *ta 1 san 2 three gung 1 fen 1 centimeter He grew three centimeters for a year. nian 2 year In Chinese, a bare temporal adverbial, e.g., ge 5 zhong 1 tou 2 hour, is interpreted in a durative manner. Verb reduplication is necessary, however, in many circumstances to render t sentence grammatical. T construction zai 4 X nei 4, literally at X in, is t Chinese equivalent of t frame adverbial. 1 Just as in English, Mandarin degree achievements with difference values are telic. Turning our attention to degree achievements without explicit difference values, t tory of Hay, Kennedy, and vin predicts that degree achievements based on closed-range adjectives will be telic, and those based on open-range adjectives will be atelic. Acceptability with t adverbial completely is a well-known diagnostic for t open-/closed-range distinction: (4) a. completely straight/empty/dry (closed-range) 1 See (Liu, 1997) for discussions about tse diagnostics. 2

3 b.??completely long/wide/short (open-range) It has long been noted that atelic predicates are entailed by tir progressive forms, while telic predicates are not (Vendler, 1957; Dowty, 1979): (5) a. John is singing. John has sung. b. John is baking a cake. John has baked a cake. It appears that t predictions are borne out: (6) a. Ty are straightening t pipe. Ty have straightened t pipe. b. T clots are drying. T clots have dried. (7) a. Ty are lengtning t beam. Ty have lengtned t beam. b. T snow is slowing. T slow has slowed. Since t scale associated with adjectives such as straight and dry have maximal values, Hay, Kennedy, and vin claim that a bound on t difference value can be identified. T measure of change takes t affected argument to t end of t scale. In contrast, adjectives such as long and slow lack maximal values, and nce tre is no basis for determining t bound on t difference value. If this analysis is correct, tn why are all degree achievements without difference values compatible with both durative and frame adverbials? (8) a. T sun dried t clots in a hour/for an hour. b. T soup cooled in an hour/for an hour. Example (8a) is predicted to be telic because t adjective dry is a closed-scale adjective, yet it displays atelic behavior. Example (8b) is predicted to be atelic because t adjective cool is an open-scale adjective, yet it displays atelic behavior. Hay, Kennedy, and vin explain t second example by appealing to contextual influences. Ty attribute t telic reading to a conversational implicature: t soup cooled to room temperature, which represents a bounded difference value. However, such an explanation significantly weakens t predictive force of t tory. If degree achievements without difference values can exhibit dual behaviors, how does distinguish t effect of t open-/closed-range distinction from t influence of context? Furtrmore, turning to Mandarin, we see an even more complex picture. Degree achievements without difference values sound odd with both durative and frame adverbials: (9) a.??ta 1 zai 4 at nian 2 year He grew in a year. b. *ta 1 He grew for a year. nei 4 in nian 2 year T only way to rescue t above sentences is t following: (10) ta 1 shun 4 jian 1 moment He suddenly became. 3

4 This behavior is puzzling. Why do t aspectual profiles of degree achievements without difference values differ cross-linguistically? Does this imply that degree achievements do not form a corent class of verbs across languages? In t following sections, I present a unified account of degree achievements consistent with evidence from both English and Mandarin Chinese. I will argue that degree achievements without difference values are punctual, i.e., ty are true achievements. Different aspectual readings arise from coercion effects and cross-linguistic differences can be attributed to t availability of coercion operators in different languages. 2 Accomplishments or Achievements? Consider t following Mandarin sentences involving change of state predicates: (11) a. Li 3 si 4 Lisi pang 4 fat liang 3 two Lisi gained two kilograms. b. bo 1 li 2 glass sui 4 shatter man 3 whole gong 1 jing 1 kilograms di 4 floor T glass shattered all over t floor. In (Lin, 2004), I argue extensively that all Mandarin change of state predicates (and nce all degree achievements) are achievements. In t next section, I will extend this tory to degree achievements in English, but first, I will review t evidence supporting this claim for Chinese. T progressive test has frequently been cited as a diagnostic for distinguishing stative and non-stative verbs in English (Lakoff, 1966); only non-statives can occur in t progressive: (12) a. *John is knowing t answer. (state) b. John is dancing. (activity) c. John is painting a picture. (accomplishment) Achievements, however, present a more complex story: (13) a. John is winning t game. (achievement) b.??john is noticing t sign. (achievement) Since achievements are punctual, Smith (1991) argues that t progressive refers to t preliminary stages of t event leading up to t change of state ratr than t change of state itself (a point that will become important later on). For now, I will simply assume that compatibility of English achievements with t progressive varies by verb. Anotr useful test for separating achievements from accomplishments is t acceptability of t verb as t complement of stop: (14) a. Mary stopped knowing t answer. (state) b. Mary stopped dancing. (activity) c. Mary stopped painting a picture. (accomplishment) d.??mary stopped reaching t top. (achievement) 4

5 Achievements sound odd as t complement of stop, expect perhaps in a habitual interpretation, e.g., John stopped noticing t noise after a while. T results of tse two tests are summarized below: (15) compatible with prog.? complement of stop state no ok activity yes ok accomplishment yes ok achievement maybe bad We can now apply tse diagnostics to Mandarin sentences. As it turns out, t sentences in (11) are incompatible with t progressive and as t complement of stop, as shown in (16) and (17). From tse results, we can conclude that t sentences denote achievements. (16) a. *Li 3 si 4 Lisi zng 4 zai 4 in.process.of pang 4 fat liang 3 two intended: Lisi is gaining two kilograms. gong 1 jing 1 kilograms b. *bo 1 li 2 glass (17) a. *Li 3 si 4 Lisi zng 4 zai 4 in.process.of sui 4 shatter man 3 whole di 4 floor intended: T glass is shattering all over t floor. ting 2 zhi 3 stop pang 4 fat liang 3 two gong 1 jing 1 kilograms intended: Lisi stopped gaining two kilograms. b. *bo 1 li 2 glass ting 2 zhi 3 stop sui 4 shatter man 3 whole di 4 floor intended: T glass stopped shattering all over t floor. In Mandarin, accomplishments are typically formed through t process of resultative verb compounding. In such a compound, t first verb denotes an activity and t second verb denotes t result of that activity: (18) a. Zhang 1 san 1 Zhangsan kan 3 chop dao 3 fall shu 4 tree Zhangsan chopped t tree down. b. Zhang 1 san 1 Zhangsan da 3 hit sui 4 shatter Zhangsan shattered t glass. bo 1 li 2 glass Such constructions are accomplishments, as demonstrated by tir compatibility with t progressive (although t sentences sound much more natural in t ba construction): (19) a. Zhang 1 san 1 Zhangsan zng 4 zai 4 in.process.of ba 3 Ba shu 4 tree Zhangsan is chopping t tree down. kan 3 chop dao 3 fall b. Zhang 1 san 1 Zhangsan zng 4 zai 4 in.process.of ba 3 Ba Zhangsan is shattering t glass. bo 1 li 2 glass 5 da 3 hit sui 4 shatter

6 Summarizing so far, Mandarin degree achievements (and in fact, all change of state predicates in general) appear to be punctual, i.e., ty denote an instantaneous change of state. In t next section, I will extend this analysis to account for t variable aspectual behavior of English degree achievements. 3 T Semantics of Degree Achievements Following t standard account of adjectives presented in t literature, I analyze tm as expressions that map tir arguments onto abstract representations of measurement, i.e., scales that contains degrees (Cresswell, 1998; von Stechow, 1984): (20) wide(x)(t) = t degree to which x is wide at time t Beyond this, I essentially employ Hay, Kennedy, and vin s underlying semantic machinery to analyze degree achievements. T function Increase, contributed by t verbalizing affix or -en in English, takes a gradable adjective meaning φ and returns a description of an event of some object undergoing a change in t degree to which it is φ: (21) increase(φ)(x)(d)(e) = 1 iff φ(x)(spo(e)) + d = φ(x)(epo(e)) T functions Spo and Epo pick out t starting and ending points of an event, respectively. In prose, Increase (φ)(x)(d)(e) is true of an event e if and only if t degree to which x is φ at t end of t event equals d plus t degree to which x is φ at t beginning of t event, i.e., just in t case that x increases in φ-ness by d during t event. This measure of change d is t difference value. With tse definitions, Hay, Kennedy and vin assign t following logical representations to degree achievements: (22) a. T road widened. e, d[increase(wide(road))(d)(e)] b. T road widened 5m. e[increase(wide(road))(5cm)(e)] Example (22a) is true just in t case that t road increased in width by an unspecified (i.e., existentially quantified) amount. Since t adjective wide lacks a maximal value, tre is no basis for determining a bound on d; nce, t event is atelic. On t otr hand, example (22b) is true just in t case that t road increased in width by exactly five meters. Since t difference value is bounded, t event is telic. My proposal that degree achievements without difference values are punctual can be implemented by replacing t existentially quantified d with a variable called δ, which I define as t margin of comparison (t analysis of degree achievements with difference values remains t same): (23) Li 3 si 4 Lisi pang 4 fat Lisi gained weight. e[increase(weight(lisi))(δ)(e)] 6

7 T variable δ represents t minimum difference in degree such that a comparative judgment can be made in (23), Lisi can be said to have gained weight only if weighs more now than did before. T truth condition in this example can only be determined with t aid of some measuring device such as a scale. T δ is thus dependent on t resolution of t measuring device, i.e., t smallest weight increment that t scale can measure. Under my implementation, a degree achievement without a difference value becomes true t moment tre is a measurable change in t underlying property (weight in this case). This change is instantaneous, and nce t event is interpreted as an achievement. This analysis appears to capture t facts observed in Mandarin. However, t puzzling aspectual behavior of degree achievements without difference values in English remains unexplained. If my claims are correct, what accounts for t grammaticality of t following sentence? (24) T soup cooled in an hour. If cooling is punctual (i.e., t statement becomes true as soon as tre is a measurable drop in temperature), tn why is cool compatible with a frame adverbial? I believe that sentences such as (24) parallel t following examples, all involving unambiguous achievements: (25) a. John reacd t top in an hour. b. Mary won t css match in ten minutes. For example, t transition from not having reacd t top to having reacd t top is instantaneous, as well as t change of state from not having won to having won a css match. Yet, tse achievements are compatible with frame adverbials. I believe that examples such as (24) and examples such as (25) form a corent class of pnomenon. T solution to this mystery involves t addition of a preparatory phase to an instantaneous change, which renders an achievement compatible with a frame adverbial. T duration of an hour in winning t css match in a hour refers to t period of time leading up to t actual attainment of victory. It is clear that t actual achievement (i.e., t transition from not having won to having won) occurs precisely at t end of t interval specified by t frame adverbial. Thus, if t css match concluded in an hour with Mary as t victor, t following sentence would be pragmatically odd: (26) #Mary won t css match in two hours. To furtr illustrate t fact that t frame adverbial refers to t time leading up to t achievement as opposed to t achievement itself, consider t felicity of t following utterance, given t context: (27) Context: T css match lasted an hour. Up until t last second, Mary was actually in a losing situation, but in t end, s pulled a gambit that paid off. Mary won t css match in an hour. I believe that degree achievements behave exactly t same way. Because ty are punctual, t frame adverbial refers to t preparatory phase leading up to t change in degree. 7

8 To highlight this, consider t following scenario. A bowl soup sits on t kitcn table; its initial temperature is 120. At precisely nine minutes, fifty-nine seconds, t temperature drops to (assume that our trmometer can only measure temperature to an accuracy of 0.1 ). Here are possible statements can make about t situation: (28) a. T soup cooled in five minutes. b. T soup cooled in ten minutes. c. T soup cooled in fifteen minutes. T statement in (28a) is obviously false, while t statement in (28b) is true. However, (28c) sounds pragmatically odd, in t same way that (26) is infelicitous. This demonstrates that wnever an achievement appears with a frame adverbial, a preparatory phase is added before t actual event, placing t change of state at t end of t denoted interval. More formally, I believe that this process can be capture by aspectual coercion, e.g., (Moens and Steedman, 1988; de Swart, 1998, 2000). According to de Swart, coercion operators come into play in t semantic composition wn and only wn a type clash licenses tm: Coercion... is governed by implicitly contextual reinterpretation mechanisms triggered by t need to resolve aspectual conflicts (de Swart, 1998:360). In this case, achievements have t feature [ durative] and frame adverbials have t feature [+durative]. To resolve this type clash, t operator Add-Prep is introduced, adding a preparatory phase to t event. T same coercion operator allows many English achievements to appear in t progressive. In tse cases, t progressive does not apply to t actual achievement, but ratr to t unspecified event leading up to it: (29) a. Mary is winning t css match. b. T training is arriving. A coercion account can be similarly applied to explain t grammaticality of t following sentence: (30) T soup cooled for an hour. Here, t iteration operator Iter is applied to resolve t type clash. In otr words, (30) refers to a successive series of small cooling events, each of which results in t temperature of t soup dropping by δ. Why, tn, are degree achievements in Mandarin incompatible with both durative and frame adverbials? Recall tse facts, repeated from (9): (31) a.??ta 1 zai 4 at He grew in a year. nian 2 year nei 4 in b. *ta 1 He grew for a year. nian 2 year 8

9 Quite simply, t coercion operators Add-Prep and Iter are not available in Mandarin. I posit that t availability of tse coercion operators is governed by a parameter of Universal Grammar. Thus, t only temporal adverbials allowed with achievements are s that have t [ durative] property, as in t following, repeated from (10): (32) ta 1 shun 4 jian 1 moment He suddenly became. A prediction of this claim is that achievements cannot appear in t progressive in Mandarin, because t relevant coercion operators are unavailable. This is indeed borne out: (33) a. *Zhang 1 san 1 Zhangsan zng 4 zai 4 in.process.of ying 2 win na 4 that pan 2 intended: Zhangsan is winning that css game. Cl qi 2 css.game b. *huo 3 c 1 train zng 4 zai 4 in.process.of dao 4 arrive intended: T train is arriving. To summarize, I have presented a semantic analysis of degree achievements as punctual events (i.e., true achievements), and shown how this account is consistent with t aspectual behavior of tse verbs in English and Mandarin Chinese. T differences between t two languages can be explained by t availability of coercion operators that resolve type class. 4 T Scalar Structure of t Base Adjective Although degree achievements are true achievements, t scalar structure of tir underlying adjectives nevertless influences tir aspectual behavior. T difference between degree achievements based on closed-range adjectives and those based on open-range adjectives remains unexplained. T following sentences are repeated from (6) and (7): (34) a. Ty are straightening t pipe. Ty have straightened t pipe. b. T clots are drying. T clots have dried. (35) a. Ty are lengtning t beam. Ty have lengtned t beam. b. T snow is slowing. T slow has slowed. An interesting observation is that for degree achievements based on closed-range adjectives without explicit difference values, t change of state takes t affected argument to t maximal value of t scale. (36) a. #Ty have straightened t pipe, but it s still crooked. b. #T clots have dried, but ty re still wet. 9

10 For example, straightening t pipe implies that t pipe has attained its maximal straightness, and thus denying t maximal value results in a contradiction. Asserting that t clots have dried cannot be followed up by a claim that t clots remain wet (i.e., or that t clots are in any state otr than t maximal value of dryness). In contrast, corresponding sentences for degree achievements based on open-range adjectives are perfectly acceptable: (37) a. Ty have widened t road, but it s still very narrow. b. T soup has cooled, but it s still really hot. My explanation is that δ for degree achievements based on closed-range adjectives is always t difference in degree between t state of t entity and t maximal value on t relevant scale (i.e., δ spans t maximum range of t scale). In otr words, such expressions are true only if t entity attains t maximum allowed degree value. If this analysis is indeed correct, how can we explain t following behavior? (38) a. T sun dried t clots in an hour. b. T sun dried t clots for an hour. T interpretation of (38a) is so far consistent with my analysis. T closed-range nature of dryness sets δ to be t difference between t current dryness of t clots and t maximal value on t dryness scale (i.e., zero water content). T sentence is only true in t case wre t maximal dryness is attained at t end of t hour. Upon closer examination, we can see that (38b) does not refer many successive smaller drying events, wreby t water content is reduced. Otrwise, t following sentence would be perfectly acceptable: (39) #T sun dried t clots for an hour, but ty re still wet. Instead, t sentence indicates that t clots remained in t dry state under t sun for an hour. T durative adverbial is modifying t resulting state, not t actual change of state. 5 Conclusion By considering evidence from Mandarin Chinese, I have presented an analysis that explains t puzzling aspectual behavior of degree achievements cross-linguistically. My analysis differs from previous accounts in that complex aspectual behavior arises from interactions between predicates and otr sentential elements, and not from properties inrent to t predicates tmselves. However, tre remain many interesting unresolved issues. For, it appears that Mandarin stative verbs are comparative in tir unmarked form: (40) a. ta 1 /pang 4 /lao 3 /fatter/older He is /fatter/older. (compared to a person determined by context) b. si 2 who /pang 4 /lao 3? /fatter/older ta 1 Who s /fatter/older? He is. /pang 4 /lao 3 /fatter/older 10

11 It is important to note that (40a) does not mean is tall. For a non-comparative reading, t positive degree marker n 3 very (counterintuitively) is required. In tse contexts, n is typically unstressed and does not serve as a intensifier (cf. Sybesma, 1997): (41) ta 1 n 3 very He is tall/fat/old. /pang 4 /lao 3 /fatter/older To furtr complicate t situation, a stative verb with a measure phrase is ambiguous between a comparative reading and an absolute reading: (42) shu 4 tree shi 2 ten gung 1 fen 1 centimeter i. T tree is ten centimeters tall. ii. T tree is ten centimeters. (than something determined by context) And to add even more complexity, a degree achievement (i.e., stative predicate + le) in Mandarin can have a comparative reading in addition to t inchoative reading: (43) shu 4 tree le 4 shi 2 ten gung 1 fen 1 centimeter i. T tree grew ten centimeters. ii. T tree is ten centimeters. (than something determined by context) Consider anotr example: (44) z 4 this fang 2 jian 1 room xiao 5 smaller dian 3 bit i. This room is a bit smaller (than expected). ii. T room grew smaller. In an Alice-in-Wonderland setting wre rooms magically grow and shrink in size, t change of state reading would be felicitous. Under normal circumstances, however, only t comparative reading is available. From tse facts in Mandarin, tre appears to be a deeper connection between comparatives, degree achievements, and underlying states. I have only begun to scratch t surface in this work, but furtr study of tse issues promises to advance our understanding of lexical semantics. References Cresswell, Max J T semantics of degree. In Montague grammar, ed. Barbara Partee, New York: Academic Press. de Swart, Hendriëtte Aspect shift and coercion. Natural Language and Linguistic Tory 16: de Swart, Hendriëtte Tense, aspect and coercion in a cross-linguistic perspective. In Proceedings of t Berkeley Formal Grammar Conference. 11

12 Dowty, David Word meaning and Montague Grammar. Dordrecht, T Netrlands: D. Reidel Publishing Company. Dowty, David Tmatic proto-roles and argument selection. Language 67: Hay, Jennifer, Christopr Kennedy, and Beth vin Scalar structure underlies telicity in degree achievements. In Proceedings of SALT 9. Lakoff, George Stative adjectives and verbs in English. NSF-Report 17, Harvard Computational Laboratory. Lin, Jimmy Event structure and t encoding of arguments: T syntax of t Mandarin and English verb phrase. Doctoral Dissertation, Massachusetts Institute of Technology. Liu, Feng-Hsi An aspectual analysis of ba. Journal of East Asian Linguistics 6: Moens, Marc, and Mark Steedman Temporal ontology and temporal reference. Computational Linguistics 14: Smith, Carlota S T parameter of aspect. Dordrecht, T Netrlands: Kluwer Academic Publisrs. Sybesma, Rint Why Chinese verb-le is a resultative predicate. Journal of East Asian Linguistics 6: Tenny, Carol T Aspectual Interface Hypotsis. In xical matters, ed. Ivan A. Sag and Anna Szabolcsi. Stanford, California: CSLI Publications. Tenny, Carol Aspectual roles and t syntax-semantics interface. Dordrecht, T Netrlands: Kluwer Academic Publisrs. Vendler, Zeno Verbs and times. Philosophical Review 56: von Stechow, Arnim Comparing semantic tories of comparison. Journal of Semantics 3: