1.1 Research background

Break verbs, exemplified by English verbs like break, snap and tear, refer to actions that bring about complete or incomplete separation in the material integrity of entities, as exemplified in (1.1) (Guerssel et al. 1985; Hale and Keyser 1987; Levin 1993: 241-242; Pye 1996). Such breaking events, or alternatively termed as events of separation state change, are recorded to have been central to hominid cognition for more than 2.5 million years (Toth and Schick 1993: 349), thus they are generally identified as decipherer of human cognitive evolution, especially in disciplines like biological anthropology, socio-cultural anthropology, developmental psychology, archeology and primatology (Gibson and Ingold 1993). Consistent with this pursuit, break verbs, particularly the conceptual categorization of break verbs, have also evoked long and animate discussions in the field of linguistics.

(1.1)

a. She broke the window.

b. He snapped the stick.

c. The lady tore the cloth with her hands.

Straightforwardly speaking, the categorization of break verbs has been widely addressed from perspectives of generative semantics, developmental psycholinguistics as well as cognitive linguistics. But whatever perspectives have been taken, since breaking events are universally accessible to everyone and require no specialized knowledge (Majid et al. 2004), these studies are broadly motivated with an objective to examine whether the conceptual boundary of break verbs is equally universal. And in this regard, there remain controversial arguments. To be specific, generative semantic studies (Fillmore 1967; Guerssel et al. 1985; Levin 1993; Levin and Rappaport Hovav 1995, 2008) probe into break verbs through their participation in causative alternation (e.g. breaktr - breakintr) and propose that the identification of break verbs as a verb class is innate and universal. But even so, different generative semantic approaches have reached different conclusions on the boundary between break verbs and cut verbs. Generative derivational studies (Guerssel et al. 1985; Hale and Keyser 1987) tell break verbs apart from cut verbs based on their syntactic behaviors across typologically different languages. They find that break verbs but not cut verbs (cut, slash, chop, grind, hit, shoot, strike) participate in causative alternation, cut verbs but not break verbs appear in conative constructions. Both cut verbs and break verbs show up in middle constructions. Guerssel et al. (1985) argue that such syntactic difference reflects different lexical conceptual structures (LCSs). The basic structure of break is monadic while the basic structure of cut is dyadic. Different from this view, Hale and Keyser (1987) propose that the basic LCS of break verbs is composite and complex, consisting of an “event of causation” and a “central event”. Nonetheless, the basic LCS of cut verbs is only a simple “central event”. In contrast to these binary divisions of break verbs and cut verbs, Levin and Rappaport Hovav (2008) and Rappaport Hovav and Levin (2010) support that both break verbs and cut verbs are subtypes of result verbs. And they hold that the basic LCSs of both break verbs and cut verbs are dyadic and causative. The reason why cut verbs do not allow causative alternation is they “require an animate intentional and volitional agent as subject”. But this is not the case for break verbs. Such diverse arguments in generative semantics indicate that the seemingly clear-cut syntactic diagnostic is far from uncontroversial to locate the boundary of break verbs.

Developmental psycholinguistic studies (Shaefer 1979, 1980; Pye et al. 1995; Bowerman et al. 2004; Bowerman 2005, 2012), with evidence from children's first language acquisition of break verbs, prove that the conceptual boundaries between break verbs, cut verbs, and open verbs are not sharp but indiscrete, language-specific and determined by multiple conceptual variables in a matrix-like combination. Different conceptual variables assume different weights in accounting for the conceptual boundary between break verbs and other verb classes. Since young children have not mastered the weight combination among different conceptual variables, they have difficulty in correctly categorizing breaking events, cutting events, and opening events. Pye et al. (1995) further demonstrate that linguistic specificity is also an important factor that influences semantic categorization. Children's overextension in event categorization displays language-specific patterns. This indicates that specific linguistic input rather than nonlinguistic cognition influences semantic categorization (Bowerman and Choir 2001, 2003). Generally speaking, developmental psycholinguistic studies have pointed out the importance of variable combination and linguistic specificity in determining verb and event categorization. But they fail to show the specific internal structure and the specific weight hierarchy demonstrated by different conceptual variables in differentiating break verbs from other verb classes.

Cognitive linguistic studies are characterized by their attention to languages of diverse typologies and cultural backgrounds. They address the conceptual boundary of break verbs through cross-linguistic comparison and aim to arrive at universal patterns or typological distributions. Such cognitive typological studies include cognitive extensionalist studies (Majid et al. 2007a), natural semantic metalanguage (NSM) studies (Goddard and Wierzbicka 2009), Frame Semantic studies (Fujii et al. 2013a) as well as Causal Chain studies (Croft 2015). Cognitive extensionalist studies (Majid et al. 2007a, 2007b; Enfield 2007; Essegbey 2007; Narasimhan 2007) compare event categorizations by utilizing elicitation tools of video clips. Based on event descriptions produced by participants speaking diverse languages, these studies try to explore typological universals in the linguistic categorization of these events of separation state change and meanwhile try to reveal the underlying dimensions that motivate such typological universals. Majid et al. (2007a) build an MCA (multiple correspondence analysis) plot based on a dataset of 28 languages and reveal that three conceptual variables (predictability of locus of separation, rigidity, and spatial configuration) are accountable for demarcating breaking events from other event types. But despite such universal distributions, they also show that there are always exceptions in event categorization in languages like Yélî Dnye, Sranan, and Mandarin (Chen 2007; Levinson 2007).

NSM studies (Wierzbicka 1996, 2006; Goddard and Wierzbicka 1994, 2002, 2009; Goddard 1998, 2005; Otomo and Torii 2005) approach the conceptual structure of lexical items by means of extended explanatory paraphrases. NSM argues against the view that the non-durative structure is basic (as in the case of Guerssel et al. 1985 and Levin and Rappaport Hovav 2008) and supports that imperfective structure is basic because imperfective denotes only the action in progress. In this way, break verbs and cut verbs are distinguished from each other because break verbs appear in perfective structures while cut verbs show up in imperfective structures. Frame Semantics treats lexical items in terms of the frames they evoke (Fillmore 1982, 1985; Fillmore and Atkins 1992). Those evoked frames demonstrate the variable combinations that differentiate break verbs from other verb classes. And the element profiled in the frame indicates a higher proportion of weight accorded to certain conceptual variables. Moreover, since Frame Semantics follows a gestalt principle, it puts forward more variables that cannot be neglected, like <functionality>, <intentionality>and <focus> etc. (Bouveret 2009; Bouveret and Sweetser 2009; Fujii et al. 2013b) But despite this, Frame Semantics remains an interpretative approach. It fails to uncover the weight assumed by different variables within a frame. Causal Chain studies (Croft 2009, 2012, 2015) address the categorization of break verbs by employing a three-dimensional event structure, which is composed of qualitative, aspectual, and causal dimensions. With such an event structure, they support the conclusion that cut verbs encode both manner and result (Goldberg 2010: 46-50; Beavers and Koontz-Garboden 2012) and state that cut verbs are akin to break verbs when they are particularly used to denote result.

In terms of Chinese break verbs, the majority of previous studies on break verbs are concerned with the transitivity of break verb pò. For instance, Li (1994) holds that pò is a pseudo autonomous verb. Jiang (1999) states that pò was originally a transitive verb. Wu (1999: 330) insists that pò was initially an autonomous verb. Wei (2000: 821) proves that pò can be used either as a causative verb or as a resultative verb. Xu (2005, 2006: 174) conducts a diachronic study and shows that pò has evolved from a monosyllabic transitive verb to an intransitive verbal complement as well as an adjective. In addition, a limited number of studies have investigated the categorization of break verbs based on causative alternation. For instance, Bohnemeyer (2007) insists that there are neither break verbs nor cut verbs in Mandarin because Mandarin verbs appear in compound forms and do not allow causative alternation. In contrast, Chen (2007) states that Mandarin does not support the binary division of break and cut verbs but supports a three-way division of break verbs, cut verbs, and resultative compounds. To conclude, previous studies have paid much attention to the syntactic structure of Chinese break verbs, but little attention has been shifted to the conceptual variation of Chinese break verbs. In view of this, it is necessary to further explore Chinese break verbs, particularly from a perspective of conceptual variation.

Above all, break verbs have aroused a wide range of discussion and have been examined from a great number of theoretical perspectives. But even so, some problems remain unresolved.

First of all, previous studies address the linguistic categorization of break verbs primarily based on the causative alternation structures where they appear and the conceptual variables relevant to the events of separation state change they represent. But the syntactic variable of causative alternation gives rise to controversial arguments and the numerous conceptual variables are addressed in a piecemeal fashion. More importantly, historical corpora reveal that causative alternation was not a sensitive issue at all in ancient Chinese as it is claimed in Mandarin. In examples (1.2a–1.2c), pò, qiē, and kāi all appeared in causative structures in EMC. It is the evolving conceptual range of break verbs that remains a catching but underestimated question. Comparing examples (1.2a–1.2c) with examples (1.2d–1.2f), we can find these three verbs vary in their conceptual ranges. In this sense, there is a necessity to revisit the conceptual range of break verbs by giving up causative alternation and by taking into consideration those conceptual variables systematically through corpus-based multivariate methods.

(1.2)

a. EMCpoe34： 内白滑，四破去之。

nèi báihuá, sì pò qù zhī

inside white smooth, four break remove it

“The fruit is white and smooth inside. Break it into four

parts and remove its seeds.”

b. EMCqie32： 猪肉一斤，合煮令熟，细切之。

zhūròu yījīn, hézhǔ lìnɡ shú, xì qiè zhī

pork  one-kilo,  together-boil make cooked,  fine cut it

“Take a kilo of pork, boil it cooked and cut it finely.”

c. EMCkai23： 开匮得书，见公之功。

kāi ɡuì dé shū, jiàn ɡōnɡ zhī ɡōnɡ

open case get book, see  him  DE contribution

“Opened the case, got the book and found his contribution.”

d. MdMandpoe1153： 一个不小心，把手指划破了。

yī ɡè bùxiǎoxīn, bǎ shǒuzhǐ huápò le

one-CL not careful,  BA  finger  scratch-wound-LE

“Uncarefully, he scratched wound his finger.”

e. MdMandqie79： 铁三爷拿起刀来……“唰”就切下一块肉来。

Tiě Sānyé náqǐ dāo lái … “shuā” jiù qiēxià yīkuài ròu lái

Tie Sanye hold-up  knife DEIXIS …‘shua' and cut-down a

piece meat DEIXIS

“Tie Sanye held up the knife …and cut down a piece of meat.”

f. MdMandkai73： 等到了跟前一瞧，把虎儿小子的衣裳撕开了。

děnɡ dào le ɡēnqián yī qiáo, bǎ Hǔér xiǎozi de yīshanɡsīkāi le

wait reach LE front  a look, BA Huer  boy DE clothes  tear-open-LE

“When came closer, they found Huer's clothes was torn apart.”

Secondly, most previous studies attend to break verbs from a synchronic perspective. Since synchronic data only constitute a snapshot of the history of a language, this study decides to trace the diachronic evolution of break verbs. Only with a diachronic track can we tell a full story of the conceptual boundary variation of break verbs. As shown in example (1.2), although qiē behaved exactly in the same way as pò did both syntactically and conceptually in ancient Chinese, they are claimed to be rather different in Mandarin [compare (1.2a) with (1.2b) and (1.2d) with (1.2e)].

Finally, it is the extensional reference of break verbs that has been emphasized while their intensional readings are seldom discussed. Thus, this study also diachronically traces the arising of the intensional readings of break verbs.

In view of these research gaps, this study intends to attend to the extensional and intensional usages of Chinese break verbs from both a diachronic and a multivariate conceptual perspective.