孤獨字效果：以語義分類作業及字彙判斷作業探討中文心理字典的組織形式

Abstract

鄰群效果（neighborhood effect）在探討中文心理字典的組織形式上扮演著重要角色，然而過去拼音文字研究卻難以得到穩定一致的結果。Bowers、Davis及 Hanley （2005）曾以孤獨字（hermit words）檢視鄰群效果，並主張心理字典的組織形式是以抑制型連結為主，然而Qiao、Forster及Witzel （2009）卻發現孤獨字效果可能並非字彙效果。由於Bowers等人（2005）與Qiao等人（2009）的研究均以形音無法獨立操弄的拼音文字作為研究標的，實驗一遂以形音較為獨立的中文文字系統重新探究孤獨字效果，檢視當參與者透過練習中文假字的形音義訊息，進而讓中文假字進入到心理字典中後，孤獨字和練習過而成為非孤獨字的刺激字，兩者在語義分類作業上的反應時間和正確百分比差異。實驗一結果符合預期，參與者在經過練習後，其於語義分類作業上練習過的中文孤獨字的鄰群假字比沒練習過的假字有較高的正確百分比，且非孤獨字會比孤獨字的反應時間更長而正確百分比更低。此結果可說是文字辨識歷程中字彙表徵間相互競爭的證據之一，支持心理字典的組織形式是如Bowers等人（2005）所主張的以抑制型連結為主。若中文和拼音文字有同樣的促發字字彙效果的話，則按照Qiao等人（2009）的邏輯，使用實驗一改良後的練習方法學習假字的參與者做屏蔽式字形促發字彙判斷作業，應當也能得到不同於Qiao等人（2009）的結果。實驗二為了和Qiao等人（2009）的研究做對照，除了使用的刺激材料是中文以及假字練習方法改良過外，其餘皆盡量和Qiao等人（2009）相同。而實驗二結果和Qiao等人（2009）的結果不同，促發字字彙效果沒有在實驗二被重複。可能是由於Qiao等人（2009）實驗設計和程序導致參與者產生因應的策略行為。

Neighborhood effects play a critical role in examining the organization of Chinese mental lexicon. In alphabetic writing systems, there was no consensus on whether the inhibitory effect of hermit word (a word without neighbors), supporting the nature of competition in mental lexicon, was actually a lexical effect or not (Bowers, Davis, & Hanley, 2005; Qiao, Forster, & Witzel, 2009). In the present study, we applied Chinese writing system, in which the orthography and the pronunciation could be disentangled, to re-examine the hermit word effect. The original experimental design of Bowers’ et al. (2005) was adopted with some further improvements. The experiment 1 was a 2 (target type: hermit character vs. non-hermit character) x 2 (pseudo-character type: learned vs. unlearned) within-participant design. Participants were required to learn Chinese pseudo-characters. Each Chinese pseudo-character was created as a new orthographic neighbor of a specific Chinese hermit character. To examine the effect of neighbors newly learned on hermit characters, the semantic categorization task was conducted. We obtained inhibitory orthographic neighborhood effects in both analyses of reaction times and accuracies of all three administrations, supporting the nature of competitions among lexical representations in the mental lexicon as claimed by Bowers et al. (2005). In order to compare to the results of Qiao et al. (2009), we also used masked form priming lexical decision task in the experiment 2. However, we didn’ t repeat the prime lexicality effect in the experiment 2. It might be because the experimental design and the procedural of Qiaos’ et al. (2009) resulted in strategic behaviors of participants.