1.2 Research objectives and research questions

Given the above research gaps, we aspire to achieve the following research objectives:

Firstly, this study intends to verify that event categorization is dynamic and the lexical items utilized for event categorization are variant in their conceptual ranges. Specifically, lexical items keep adjusting themselves in their peripheral edge and their prototypical core in accordance with cognitive evolution and social-cultural development. Within-category conceptual fine-tuning leads first to conceptual variation and further to cross-category conceptual reorganization. Meanwhile, conceptual boundaries between semantically related lexical items are dynamically evolving through time. Such reorganization is reflected in the different segmentation patterns employed in the categorization of the same event categories across different chronological stages.

Secondly, through investigating the conceptual variation of break verbs from a diachronic perspective, this study aspires to establish a multivariate model of diachronic conceptual variation. For one thing, this model reflects the cognitive mechanisms that drive and contribute to conceptual variation. Further, it also displays the multiple pathways of conceptual variation. Most importantly,this model not only reveals the conceptual variation within the extensional domain, but also uncovers the conceptual variation across the extensional and intensional domains.

Finally, methodologically speaking, this study aims to make a bold attempt to employ multivariate statistical methods in a diachronic semantic investigation. Although corpus-based methods are increasingly employed in diachronic semantics, the feasibility of multivariate statistical methods like MCA, Ctree (Conditional Inference Tree), and random forest call for more attention, especially in diachronic studies of Chinese.

This study addresses the conceptual boundary of break verbs from a diachronic point view and a multivariate conceptual one. Our research target involves three break verbs, namely pò, qiē and kāi. It should be noted that these three verbs are explored in terms of their collocational context. It makes no sense to focus on a single verb without taking into consideration its context. These three verbs are dynamic in their syntactic forms. They were more frequently used in their monomorphemic forms in ancient Chinese but are more pervasively used in their compound or phrasal forms in Mandarin Chinese. Despite this, these three lexical items are consistent in their conceptual ranges all through the development of Chinese, though their syntactic forms have changed. Given this, we set their syntactic changes in this background, and focus on their conceptual variations. Since pò, qiē, and kāi behaved the same in ancient Chinese, we subsume them under the node break verbs for the time being and observe how they gradually diverge into different verb classes in Mandarin.

This study sets Chinese pò, qiē, and kāi as the research target for two reasons. For one thing, the categorization of their English counterparts break, cut, and open constitutes a controversial issue in previous studies. A Chinese story of these three lexical items will contribute to previous arguments on their categorization. For another, these three lexical items are all along frequently used to describe events of separation state change in the development of Chinese. This makes a diachronic Chinese story of these three lexical items possible. It should be noted that this does not mean Chinese pò, qiē, and kāi can be completely equalized with English break, cut and open. Historical corpora reveal that these three lexical items were used similarly as English break, cut, and open in ancient Chinese but have gradually changed in their syntactic patterns in the development of Mandarin Chinese (compare 1.2a and 1.2d, 1.2b and 1.2e, 1.2c and 1.2f). Although such a syntactic change is backgrounded in this study, it is not ignored in our analysis of conceptual variation. Although they have changed in their syntactic patterns, the conceptual ranges of their new syntactic forms are in line with those of their older syntactic forms. In this sense, it is still feasible to explore the diachronic variation of break verbs by targeting at pò, qiē, and kāi.

The specific research questions to be explored can be presented from two aspects. The first aspect focuses on the specific conceptual variation pathways displayed by break verbs. The second aspect delves into the cognitive mechanisms that drive and contribute to these conceptual variations. The corresponding research questions to be answered are as follows:

1) What diachronic conceptual variation pathways do Chinese break verbs display?

a. What diachronic conceptual variation pathway is exhibited in the extensional reference of pò?

b. What diachronic conceptual variation pathways are displayed in the conceptual boundaries between pò, qiē, and kāi?

c. What diachronic conceptual variation pathways are exhibited in the intensional readings of pò?

2) What cognitive mechanisms motivate break verbs' diachronic conceptual variation?

a. What cognitive mechanisms drive the referential range variation of pò?

b. What cognitive mechanisms motivate the conceptual boundary adjustment of pò, qiē, and kāi?

c. How did the multiple intensional readings of pò arise from its extensional references?

A number of theoretical frameworks have been put forward in cognitive linguistics to explore diachronic semantic change, including Invited Inference Theory of Semantic Change (Traugott 1985, 1989; Traugott and Dasher 2002), Diachronic Prototype Semantics (DPS) (Geeraerts 1997) as well as theories of grammaticalization (Traugott 1982, 1988). This study gears toward DPS because this theoretical approach is essentially a usage-based approach and it argues for the prototypically-organized conceptual structures of lexical items. Moreover, DPS does not exclusively focus on extensional usages or intensional readings but integrates these two domains into a united whole. With this approach, this study tracks the semantic change of break verbs based on their usage variations recorded in historical corpora and represents their conceptual patterns in a

prototypical conceptual structure. And in doing so, diachronic conceptual variation of break verbs is displayed through cross-stage prototypical structure comparison. Diachronic boundary variation among break verbs is concluded based on their segmentation of events of separation state change across different chronological stages. Meanwhile, not only literal usages of break verbs are analyzed in detail but also the semantic parallels between extensional usages and intensional readings are extensively mapped out.

To reach these ends, multivariate statistical methods, such as MCA, Ctree, as well as random forest, are employed to analyze the historical dataset of break verbs and their near-synonymous verb classes. Among these methods, MCA maps the conceptual structure of lexical items out in a prototypical structure, with a conceptual centroid and a conceptual boundary. The distribution of numerous individual usages reflects the conceptual ranges of lexical items. The distribution of differential variable levels indicates the underlying variables that drive the conceptual ranges of lexical items. Ctree and random forest demonstrate the weight hierarchy of different conceptual variables in a more fine-grained fashion.