詞義相關性在詞彙歧義理解上的效應: 以中文動詞為例

Abstract

本論文以心理語言學的研究途徑，探討不同種類的詞彙歧義在語句理解中的處理過程。過去有關詞彙歧義理解的研究並無一致的發現：有些研究持模組觀點(modular)，認為不同語言層次是獨立運作的，句子語境在詞彙觸接初期並不會影響詞義激發(Ahrens, 1998, 2001; Onifer& Swinney, 1981; Swinney, 1979)；有些研究則支持交互觀點(interactive)，認為語境的效應會直接影響詞彙語意的提取(Li & Yip, 1996, 1998; Tabossi, Colombo, & Job, 1987; Tabossi & Zardon, 1993)。然而，過去文獻並未區分不同類型的歧義詞進行檢視，進而可能影響實驗結果。

根據多重語意的相關性程度，詞彙歧義可分為同形異義(homonymy)與一詞多義(polysemy)。過去許多實證研究已發現兩者在認知處理過程與心理語意表徵的不同(Azuma & Van Orden, 1997; Frazier & Rayner, 1990; Klepousniotou, 2002)。因此本研究旨在以跨模式詞彙促發實驗(cross-modal lexical priming)，探討兩種不同的詞彙歧義，同形異義詞與多義詞，在語句理解中是否有不同模式的處理過程。

實驗結果顯示，不同的詞彙歧義確實有不同的理解過程。同形異義詞在語句理解過程中只有與語境相符的意義被激發，而多義詞的所有詞義在初期都會被激發。因此，本研究建議過去研究的爭論可能受到不同詞彙歧義的影響，也表示詞義相關性的效應在詞彙歧義理解的研究是重要的。此外，本研究也指出不同類型歧義詞的處理過程之所以相異，歸因於他們詞義表徵的不同：同形異義詞的多重意義是分別且獨立儲存於心理詞彙，而多義詞的多重語意是儲存於單一核心意義的心理表徵之下。This thesis adopts a psycholinguistic approach to examine how different types of lexical ambiguity are resolved and processed during sentence comprehension. Previous findings on lexical ambiguity resolution were contradictory: some studies took a modular position, which suggested that different levels of processing system operate independently and thus contextual information from the discourse level would not immediately influence the access of a word’s meaning at the lexical level (e.g. Ahrens, 1998, 2001; Onifer& Swinney, 1981; Swinney, 1979); others took an interactive position, which argued that sentential context could have an immediate effect on lexical access (e.g. Li & Yip, 1996, 1998; Tabossi, Colombo, & Job, 1987; Tabossi & Zardon, 1993). However, it was observed that most research to date has not distinguished different types of ambiguity in their experiments, which may hence influence the results.

Lexical ambiguity has been traditionally differentiated into homonymy and polysemy, with respect to the dimension of relatedness of its multiple meanings. A number of psycholinguistic studies concerning the semantics of ambiguous words did provide empirical evidence for the processing distinction between homonymy and polysemy, and some of the studies further pointed out their underlying differences on meaning representations in the mental lexicon (e.g. Azuma& Van Orden, 1997; Frazier & Rayner, 1990; Klepousniotou, 2002). The main purpose of this thesis, therefore, is to investigate whether two types of ambiguity, homonymy and polysemy, have different processing mechanisms during on-line sentence comprehension, by conducting a cross-modal lexical priming experiment.

The results of the present study showed that homonymy and polysemy did perform distinct processing patterns on the resolution of lexical ambiguity. For a homonymous word, only the contextually appropriate meaning was accessed, whereas both meanings of a polysemous word were activated at an early stage of semantic access. Our findings suggest that inconsistent results in previous literature might be confounded by the relatedness of multiple meanings, and thus the effect of sense relatedness should not be overlooked in the issue of lexical ambiguity resolution. Furthermore, the distinct processing patterns between homonymy and polysemy could be explained in terms of their different lexical representations: while meanings of a homonymous word are stored distinctly and separately, multiple related meanings of a polysemous word are stored and listed under a single core sense.