Konopka Agnieszka E, Meyer Antje, Forest Tess A
University of Aberdeen, Scotland, UK; Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands.
Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands.
Cogn Psychol. 2018 May;102:72-104. doi: 10.1016/j.cogpsych.2017.12.003. Epub 2018 Mar 8.
The leading theories of sentence planning - Hierarchical Incrementality and Linear Incrementality - differ in their assumptions about the coordination of processes that map preverbal information onto language. Previous studies showed that, in native (L1) speakers, this coordination can vary with the ease of executing the message-level and sentence-level processes necessary to plan and produce an utterance. We report the first series of experiments to systematically examine how linguistic experience influences sentence planning in native (L1) speakers (i.e., speakers with life-long experience using the target language) and non-native (L2) speakers (i.e., speakers with less experience using the target language). In all experiments, speakers spontaneously generated one-sentence descriptions of simple events in Dutch (L1) and English (L2). Analyses of eye-movements across early and late time windows (pre- and post-400 ms) compared the extent of early message-level encoding and the onset of linguistic encoding. In Experiment 1, speakers were more likely to engage in extensive message-level encoding and to delay sentence-level encoding when using their L2. Experiments 2-4 selectively facilitated encoding of the preverbal message, encoding of the agent character (i.e., the first content word in active sentences), and encoding of the sentence verb (i.e., the second content word in active sentences) respectively. Experiment 2 showed that there is no delay in the onset of L2 linguistic encoding when speakers are familiar with the events. Experiments 3 and 4 showed that the delay in the onset of L2 linguistic encoding is not due to speakers delaying encoding of the agent, but due to a preference to encode information needed to select a suitable verb early in the formulation process. Overall, speakers prefer to temporally separate message-level from sentence-level encoding and to prioritize encoding of relational information when planning L2 sentences, consistent with Hierarchical Incrementality.
句子规划的主流理论——层次递增性和线性递增性——在将言语前信息映射到语言的过程协调假设上存在差异。先前的研究表明,对于母语(L1)使用者而言,这种协调会因规划和生成话语所需的信息层面和句子层面过程的执行难易程度而有所不同。我们报告了首个系列实验,旨在系统研究语言经验如何影响母语(L1)使用者(即使用目标语言有终身经验的使用者)和非母语(L2)使用者(即使用目标语言经验较少的使用者)的句子规划。在所有实验中,使用者自发地用荷兰语(L1)和英语(L2)生成对简单事件的单句描述。通过分析早期和晚期时间窗口(400毫秒之前和之后)的眼动,比较了早期信息层面编码的程度和语言编码的起始时间。在实验1中,使用者在使用L2时更有可能进行广泛的信息层面编码并延迟句子层面编码。实验2 - 4分别选择性地促进了言语前信息的编码、施事角色(即主动句中的第一个实义词)的编码和句子动词(即主动句中的第二个实义词)的编码。实验2表明,当使用者熟悉事件时,L2语言编码的起始时间没有延迟。实验3和实验4表明,L2语言编码起始时间的延迟并非由于使用者延迟施事编码,而是由于在表述过程早期倾向于对选择合适动词所需的信息进行编码。总体而言,在规划L2句子时,使用者倾向于在时间上分离信息层面和句子层面的编码,并优先对关系信息进行编码,这与层次递增性一致。