82
Chapter Six Conclusion
6.1 Summary of the Findings
Most previous studies on Hakka focus on the phonological rules (Huang 2003, Lai 2004, Peng 2004). This study, however, approaches Hakka from the semantic perspective, especially its classifier system. It deals not only with the general classification of Hakka classifiers, but concerns the semantic principles underlying the classification of Hakka individual classifiers, which reveal most about the inherent properties of the associated objects. In the present work, prototype theory serves as a framework to analyze individual classifiers in Hakka. We find that each classifier category has its own categorical structure, and the result provides abundant evidence for human categorization. We may understand how Hakka people view things in this world.
The first part of this thesis investigates the classification of the Hakka classifier system. We have arrived at a general classification of it in order to get a whole picture of the system. Eight types of classifiers are identified; they are standard classifies, container classifiers, kind classifiers, extrinsic collective classifiers, intrinsic collective classifiers, event classifiers, individual classifiers, and partitive classifiers.
Though these types also exist in Mandarin, several classifiers are closely related to Hakka culture and only appear in the Hakka classifier system. For example, the individual classifier sa (儕) is used to classify human beings, and siong24 (廂) is used for farmland in square. Both of them are only found in the Hakka classifier system,
The second part of the present study examines the semantic features proposed
by Tai (1992, 1994) and Allan (1977) to be the underlying criteria for classification of
Hakka individual classifiers, including animacy, shape, size, consistency, and part of
83
object. We contend that Hakka individual classifiers are systematically categorized.
The major semantic category for classification is animacy. Both the animate and the inanimate categories are further subcategorized. The animate category includes two groups--human and animals. The inanimate category is subcategorized into shape and non-shape. In the shape category, longness, flatness, and roundness are common categories (All1n 1977, Tai 1994,). Size and consistency serve as the secondary features to make further distinctions in the shape domain. Besides those above, the lumpy category is also included in Hakka. Entities in lumpy shape may be rectangles or irregular in shape; for example, yit2 liau5 zu24 ngiuk11 (一廖豬肉) ‘ a lump of pork,’
yit2 kut5 tong11 (一木骨糖) ‘a grain of sugar, ’yit2 de5 teu55 fu55 (一迭豆腐)
‘ a piece of tofu.’ The result of the study clearly demonstrates that the Hakka classifier system correlates to human perception, in which similar objects are categorized into the same category. Therefore, speakers do not pick up a certain classifier randomly to classify an object in question, but selecting a proper one for it.
The third part of this research starts with taking a scrutiny of the categorical structure of zak2 in light of the prototype theory. It is demonstrated that zak2 has two major semantic domains: animate and inanimate. The inanimate category even has different subcategories, making its categorical structure presented like a tree structure with hierarchical levels.
The numerous semantic domains that zak2 possesses may explain its high frequency in usage. However, it also causes a difficulty in predicting its members.
For instance, we have no problem to understand that zak2 is used for animals, based on animacy. However, zak2 in Hakka also collocates with abstract referents such as idea, thoughts, conditions, or hope. Compared with animals, those abstract entities are disparate.
At last, the present study investigates the categorical structures of functional
84
classifiers in Hakka, including gien24 (間), dung5(棟), and zak2(座). The result is interesting. All three classifiers are building classifiers, and their prototypical members are different from one another. Both gien24and dung5 are used for buildings with enclosed space, especially for human dwelling. Gien24 is typically used for buildings in average size, and dung5 is used for those either are large or tall in height.
Different from gien24 and dung5 essentially, zak2 is used for constructions with a wide base, and is usually used for buildings for non-human dwelling such as animals or objects.
6.2 Further Research
While previous studies primarily focus on the phonological rules of Hakka (Huang 2003, Lai 2004, Peng 2004), the present study aims to provide a study of numeral classifiers in Hakka with respect to the semantic principles involved in the classification. We have shown that prototype theory of categorization provides enough explanation for the categorical structure of Hakka classifiers, especially individual classifiers. Prototype theory helps language users define the central member and the less typical one. The result confirms that people view this world with conceptual classification, according to which, similar entities fall into the same category in people’ s mind. In addition to the present work, there are still some research directions we may take in the future.
First of all, the categorical structure of each individual classifiers is worthy of further research. It seems that each individual classifier has its own categorical structure. Though it is time-consuming to complete this work, the result may offer a wealth data-base for Hakka learners.
Second, we have known that prototype theory can serve as a tool to analyze the
categorical structure of individual classifiers. Can it also provide enough explanations
85
