英文二語學習者重音偏好：非中性後綴字之驗證

Abstract

英文後綴字重音的位置由眾多因素決定，且會造成第二外語學習者的學習困難，而非中性後綴有各種不同的重音衍生方式，且還會導致重音轉移，對二語學習者更添難度，且遇到重音轉移時犯錯機率也相當高，例如：demon'strative。然而不同非中性後綴的重音衍生機制對學習者的影響仍尚未被系統性地研究，本研究將針對非中性後綴的次類別對學習者重音位置正確率的影響加以探討，我們邀請了三十九位母語為國語的英語學習者來參加包含真字與假字的實驗，我們選擇了後綴前一音節重音(Pre 1)及後綴前一或前二音節重音(Pre 1/2)的衍生字，在有重音轉移及無重音轉移的情形加以測試。Pre 1衍生字的重音固定都在後綴前一音節，而Pre 1/2衍生字的重音有兩個可能位置：後綴前一音節為重音節時將獲得重音(1L)，後綴前一音節為輕音節時，後綴前第二音節將獲得重音(2L)。實驗結果發現學習者的正確率在Pre 1/2衍生字、假字、及有重音轉移的表現都顯著的低。Pre 1/2衍生字的結果中，後綴前第二音節重音(2L)的低正確率及錯誤分析更指出，學習者可能有重音放在後綴前一音節的偏好。我們推測此偏好是來自學習者從學習環境中，把各個衍生字及各種衍生字的重音分布機率推衍出自己的規則，並類推到新的衍生字上所導致。

Stress position assigned by suffixes is parametered by multi-dimensional factors, which can cause imperfect L2 learning. Non-neutral suffixes, in particular, have different mechanisms in deriving stress and may result in stress shift, making it even more difficult for L2 learners. Errors in stress shift have also been widely seen in derived forms such as *demon'strative. Yet little of the stress-deriving natures in different non-neutral subtypes has been systematically compared and accounted for. The present study aims to investigate how different subtypes of non-neutral suffixes affect stress-assignment accuracy in L2 production. Thirty-nine Mandarin speakers participated in producing real and pseudo derivatives. Stress-shifted vs non-shifted derivatives from the Pre-stress 1 and Pre-stress 1/2 categories were chosen. While stress is always on presuffixal syllable in Pre-stress 1 derivatives, stress in Pre-stress 1/2 can be either on heavy presuffixal syllables (1L) or on the second syllable left of the suffix (2L) when the presuffixal syllable is light. The results showed significantly lower accuracy in the Pre-stress 1/2 words, pseudo words, and in cases of stress shift. The lower accuracy of 2L stress, as well as error analysis from Pre-stress 1/2 derivatives, further reveal the preference for presuffixal stress in the absence of native-like knowledge of suffixation. Analogies overgeneralized through the distribution of each suffix and suffix type are suspected to be the cause of L2 errors.