Xu Y
Department of Linguistics, University of Connecticut, Storrs 06269.
Mem Cognit. 1991 May;19(3):263-73. doi: 10.3758/bf03211150.
In testing the hypothesis that surface phonetic form is included in short-term memory (STM) representation, the tone sandhi phenomenon in Mandarin Chinese was exploited, and, as a prerequisite, the hypothesis that tonal similarity affects STM of verbal material in a tone language was also tested. In Experiment 1, subjects recalled visually presented sequences of seven monosyllabic Chinese morphemes having either the same tone or different tones. More errors were made on the monotonal sequences than on the multitonal sequences, confirming the effect of tonal similarity on STM. In Experiment 2, subjects recalled visually presented sequences of disyllabic nonsense words. The sequences were designed in such a way that half of them were subject to the tone sandhi rule in Mandarin Chinese, whereas the other half were not. The consequence of applying the tone sandhi rule, as designed, was to make all the first characters in the sequences identical in pronunciation, thus creating potential phonological confusion. More errors, indeed, occurred on the sequences subject to the tone sandhi rule than on those not subject to it, indicating the existence of a surface phonetic representation in STM. The findings in this study provide further insight into the phonological mechanism of STM. Different accounts for this mechanism are also discussed in the light of the new findings.
在检验表面语音形式包含在短期记忆（STM）表征这一假设时，研究利用了汉语普通话中的连读变调现象，并且，作为前提条件，还检验了声调相似性影响声调语言中言语材料的STM这一假设。在实验1中，受试者回忆视觉呈现的由七个单音节汉语语素组成的序列，这些语素要么具有相同的声调，要么具有不同的声调。与多声调序列相比，单声调序列的错误更多，这证实了声调相似性对STM的影响。在实验2中，受试者回忆视觉呈现的双音节无意义词序列。这些序列的设计方式是，其中一半遵循汉语普通话的连读变调规则，而另一半则不遵循。按照设计，应用连读变调规则的结果是使序列中所有的首字发音相同，从而造成潜在的语音混淆。实际上，遵循连读变调规则的序列比不遵循该规则的序列出现的错误更多，这表明STM中存在表面语音表征。本研究的结果为STM的语音机制提供了进一步的见解。还根据新发现讨论了对此机制的不同解释。
