Auer Edward T
Department of Communication Neuroscience, House Ear Institute, Los Angeles, CA 90057, USA.
Psychon Bull Rev. 2002 Jun;9(2):341-7. doi: 10.3758/bf03196291.
The neighborhood activation model (NAM; P. A. Luce & Pisoni, 1998) of spoken word recognition was applied to the problem of predicting accuracy of visual spoken word identification. One hundred fifty-three spoken consonant-vowel-consonant words were identified by a group of 12 college-educated adults with normal hearing and a group of 12 college-educated deaf adults. In both groups, item identification accuracy was correlated with the computed NAM output values. Analysis of subsets of the stimulus set demonstrated that when stimulus intelligibility was controlled, words with fewer neighbors were easier to identify than words with many neighbors. However, when neighborhood density was controlled, variation in segmental intelligibility was minimally related to identification accuracy. The present study provides evidence of a common spoken word recognition system for both auditory and visual speech that retains sensitivity to the phonetic properties of the input.
言语识别的邻域激活模型(NAM;P.A.卢斯和皮索尼,1998)被应用于预测视觉言语识别准确性的问题。一组12名听力正常的受过大学教育的成年人和一组12名受过大学教育的成年聋人识别了153个口语中的辅音-元音-辅音单词。在两组中,项目识别准确性与计算出的NAM输出值相关。对刺激集子集的分析表明,当刺激清晰度得到控制时,邻域较少的单词比邻域较多的单词更容易识别。然而,当邻域密度得到控制时,片段清晰度的变化与识别准确性的关系最小。本研究提供了证据,表明听觉和视觉言语存在一个共同的言语识别系统,该系统对输入的语音特性保持敏感。