Huettig Falk
Max Planck Institute for Psycholinguistics, P.O. Box 310, 6500 AH Nijmegen, The Netherlands; Donders Institute for Brain, Cognition, and Behaviour, Radboud University, Nijmegen, The Netherlands.
Brain Res. 2015 Nov 11;1626:118-35. doi: 10.1016/j.brainres.2015.02.014. Epub 2015 Feb 20.
The notion that prediction is a fundamental principle of human information processing has been en vogue over recent years. The investigation of language processing may be particularly illuminating for testing this claim. Linguists traditionally have argued prediction plays only a minor role during language understanding because of the vast possibilities available to the language user as each word is encountered. In the present review I consider four central questions of anticipatory language processing: Why (i.e. what is the function of prediction in language processing)? What (i.e. what are the cues used to predict up-coming linguistic information and what type of representations are predicted)? How (what mechanisms are involved in predictive language processing and what is the role of possible mediating factors such as working memory)? When (i.e. do individuals always predict up-coming input during language processing)? I propose that prediction occurs via a set of diverse PACS (production-, association-, combinatorial-, and simulation-based prediction) mechanisms which are minimally required for a comprehensive account of predictive language processing. Models of anticipatory language processing must be revised to take multiple mechanisms, mediating factors, and situational context into account. Finally, I conjecture that the evidence considered here is consistent with the notion that prediction is an important aspect but not a fundamental principle of language processing. This article is part of a Special Issue entitled SI: Prediction and Attention.
近年来，预测是人类信息处理的一项基本原则这一观点颇为流行。对语言处理的研究可能对检验这一说法特别具有启发性。传统上，语言学家认为，由于语言使用者在遇到每个单词时都有大量可能性，预测在语言理解过程中只起次要作用。在本综述中，我考虑了预期语言处理的四个核心问题：为什么（即预测在语言处理中的功能是什么）？什么（即用于预测即将到来的语言信息的线索是什么，以及预测的是何种类型的表征）？如何（预测性语言处理涉及哪些机制，以及诸如工作记忆等可能的中介因素的作用是什么）？何时（即个体在语言处理过程中是否总是预测即将到来的输入）？我提出，预测是通过一组多样的PACS（基于生成、联想、组合和模拟的预测）机制发生的，这些机制是全面解释预测性语言处理所必需的最低要求。预期语言处理模型必须进行修订，以考虑多种机制、中介因素和情境背景。最后，我推测，这里所考虑的证据与预测是语言处理的一个重要方面但不是基本原则这一观点相一致。本文是名为“SI：预测与注意力”的特刊的一部分。
