Racine Isabelle, Grosjean François
Laboratoire de traitement du langage et de la parole, Université de Neuchâtel, Suisse.
Can J Exp Psychol. 2005 Dec;59(4):240-54. doi: 10.1037/h0088052.
We studied the processing of two word strings in French made up of a determiner and a noun which contains a schwa (mute e). Depending on the noun, schwa deletion is present ("la tir'lire"), optional ("le ch(e)min") or absent ("la cornemuse"). In a production study, we show that schwa deletion, and the category of the noun, have a large impact on the duration of the strings. We take this into account in two perception studies, which use word repetition and lexical decision, and which show that words in which the schwa has been deleted usually take longer to recognize than words that retain the schwa, but that this depends also on the category of the word. We explain these results by examining the influence of orthography. Based on the model proposed by Grainger and Ferrand (1996), which integrates the written dimension, we suggest that two sources of information, phonological and orthographic, interact during spoken word recognition.
我们研究了由限定词和包含弱读音节（哑音e）的名词组成的两个法语单词串的处理过程。根据名词的不同，弱读音节的删除情况有所不同，有的存在删除现象（“la tir'lire”），有的是可选择删除（“le ch(e)min”），还有的不存在删除（“la cornemuse”）。在一项产出研究中，我们发现弱读音节的删除以及名词的类别，对单词串的时长有很大影响。在两项感知研究中我们考虑了这一点，这两项研究使用了单词重复和词汇判断任务，结果表明，弱读音节被删除的单词通常比保留弱读音节的单词识别时间更长，但这也取决于单词的类别。我们通过考察拼写的影响来解释这些结果。基于格兰杰和费朗德（1996）提出的整合了书写维度的模型，我们认为在口语单词识别过程中，语音和拼写这两种信息来源会相互作用。
