Enochson Kelly, Culbertson Jennifer
Linguistics Program, Department of English, George Mason University, Fairfax, Virginia, United States of America.
Linguistics Program, Department of English, George Mason University, Fairfax, Virginia, United States of America; School of Philosophy, Psychology and Language Sciences, University of Edinburgh, Edinburgh, United Kingdom.
PLoS One. 2015 Mar 30;10(3):e0116946. doi: 10.1371/journal.pone.0116946. eCollection 2015.
Researchers in linguistics and related fields have recently begun exploiting online crowd-sourcing tools, like Amazon Mechanical Turk (AMT), to gather behavioral data. While this method has been successfully validated for various offline measures--grammaticality judgment or other forced-choice tasks--its use for mainstream psycholinguistic research remains limited. This is because psycholinguistic effects are often dependent on relatively small differences in response times, and there remains some doubt as to whether precise timing measurements can be gathered over the web. Here we show that three classic psycholinguistic effects can in fact be replicated using AMT in combination with open-source software for gathering response times client-side. Specifically, we find reliable effects of subject definiteness, filler-gap dependency processing, and agreement attraction in self-paced reading tasks using approximately the same numbers of participants and/or trials as similar laboratory studies. Our results suggest that psycholinguists can and should be taking advantage of AMT and similar online crowd-sourcing marketplaces as a fast, low-resource alternative to traditional laboratory research.
语言学及相关领域的研究人员最近开始利用在线众包工具，如亚马逊土耳其机器人（AMT），来收集行为数据。虽然这种方法已成功应用于各种离线测量——语法判断或其他强制选择任务——但其在主流心理语言学研究中的应用仍然有限。这是因为心理语言学效应通常依赖于反应时间上相对较小的差异，而且对于能否通过网络收集精确的时间测量结果仍存在一些疑问。在此，我们表明，结合用于在客户端收集反应时间的开源软件，利用AMT实际上可以复制三种经典的心理语言学效应。具体而言，我们在自定步速阅读任务中发现了主语确定性、填充语-空位依存关系处理和一致性吸引的可靠效应，参与者数量和/或试验次数与类似的实验室研究大致相同。我们的结果表明，心理语言学家能够且应该利用AMT和类似的在线众包市场，作为传统实验室研究的一种快速、低资源的替代方式。
