Brysbaert Marc, Stevens Michaël, Mandera Paweł, Keuleers Emmanuel
Department of Experimental Psychology, Ghent University.
J Exp Psychol Hum Percept Perform. 2016 Mar;42(3):441-58. doi: 10.1037/xhp0000159. Epub 2015 Oct 26.
Keuleers, Stevens, Mandera, and Brysbaert (2015) presented a new variable, word prevalence, defined as word knowledge in the population. Some words are known to more people than other. This is particularly true for low-frequency words (e.g., screenshot vs. scourage). In the present study, we examined the impact of the measure by collecting lexical decision times for 30,000 Dutch word lemmas of various lengths (the Dutch Lexicon Project 2). Word prevalence had the second highest correlation with lexical decision times (after word frequency): Words known by everyone in the population were responded to 100 ms faster than words known to only half of the population, even after controlling for word frequency, word length, age of acquisition, similarity to other words, and concreteness. Because word prevalence has rather low correlations with the existing measures (including word frequency), the unique variance it contributes to lexical decision times is higher than that of the other variables. We consider the reasons why word prevalence has an impact on word processing times and we argue that it is likely to be the most important new variable protecting researchers against experimenter bias in selecting stimulus materials.
凯勒斯、史蒂文斯、曼德拉和布里斯巴特（2015年）提出了一个新变量——词频，定义为人群中的词汇知识。有些词比其他词为更多人所知。对于低频词来说尤其如此（例如，“屏幕截图”与“鞭笞”）。在本研究中，我们通过收集30000个不同长度的荷兰语词元（荷兰语词汇项目2）的词汇判断时间，来检验该测量方法的影响。词频与词汇判断时间的相关性仅次于词频：即使在控制了词频、词长、习得年龄、与其他词的相似度和具体性之后，人群中每个人都知道的词的反应速度比只有一半人知道的词快100毫秒。由于词频与现有测量方法（包括词频）的相关性相当低，它对词汇判断时间的独特方差贡献高于其他变量。我们思考了词频对词汇处理时间产生影响的原因，并认为它很可能是保护研究人员在选择刺激材料时免受实验者偏差影响的最重要的新变量。
