Sense Differentiation And Description

A problem that lexicographers always have to face is to decide whether to "lump or split" in determining how many senses a polysemous word actually has. This dissertation comes to grips with this thorny problem directly, under the theoretical framework of cognitive semantics, whose investigation into polysemy in the past two decades has made it possible to approach it in a more empirical fashion. It is oriented to establish a practical model for identifying and differentiating the senses of polysemous words, which will help lexicographer reduce their arbitrariness in and make more consistent judgments on sense demarcation.The questions, such as "what is a word sense?" and "what is polysemy?" motivate us to explore meaning theories, to look into the nature of word meaning, and then to work out a working procedure for identifying idiosyncratic polysemy in contrast to monosemy, homonymy, and regular polysemy of different kinds.As lexicographers, we nail down arbitrariness, senses proliferation, and feature priority misplaced as the target problems to solve and we further argue that these problems are intrinsic to the traditional practice, and that any attempts in the traditional fashion are doomed to a failure.The elusiveness of sense identification in practical lexicography forces us to turn to the semantic networks of polysemous words as a stepping-stone. We are most interested in the cognitive semantic approaches, such as frame semantics, idealized cognitive model, and cognitive grammar. To look into the microstructure of the semantic networks, we also take into account of the mechanisms of sense generation and generating polysemy, in addition to the regularities in semantic change either diachronical or synchronical.With these theoretical underpinnings, we start to establish our model. The model consists of principles for identifying the protosense and for distinguishing the distinct senses, which can be further specified as a set of operational criteria:1. The distinct senses of a polysemous word should first be identified on the basic concept level in that they are embodied ones;2. These senses should be identified by studying their corresponding cognitive semantic frames respectively;3. Among these senses one (or more than one in some rare cases) is primary or central, owing to its cognitive importance in acquisition, its frequency in use, and its derivational (not necessarily diachronically true) relationships with other senses;4. The protosense, if there exists one, should be identified on the super-ordinate level to capture the invariant meaning components among the distinct senses on the basic level;5. A sense on the basic level, under some cases, can further differentiate into or instantiate as several subsenses on the subordinate level;6. The number of senses a definer can distinguish is indefinite unless sense granularity is set according to factors, such as how close a definer examines the citations collected, what (linguistic and encyclopedic) knowledge s/he expectsthe intended dictionary users to have.The advantages of the model lie in its capturing of the dynamic, bi-level, and multi-dimensional nature of the semantic networks of polysemous words.As a test for the usefulness and effectiveness of the model, a case study of da, a super polysemous verb in modern and contemporary Chinese, is carried out. To verify the results of our case study against corpus evidence, we extract all the sentences bearing da from A Dream of Red Mansions (the first eighty chapters) by Cao Xueqin and Four Generations Under One Roof (the first eighty-seven chapters) by Lao She. We construct two semantic networks of da as what is represented in their minds respectively. Through comparing these two semantic networks of da, we find that the sense identification and distinction of da in our case study is confirmed in that it captures both the basic structures of those two semantic networks and the majority of their senses, especially those core senses, which correlate to their corresponding high frequency in the citation...