Agreement attraction error and timing profiles in continuous speech

Abstract

Until recently, most web-based psycholinguistic experiments have focused on simple responses such as button presses or typed text. In this paper, we compare two sentence production experiments carried out in the lab and in an unsupervised web-based setting in order to test the robustness of error patterns and speech timing effects in web-based experimental settings. The open-ended spoken responses generated in this task could elicit more variable and noisier recordings when conducted outside a controlled lab environment. The experiments investigate patterns of well-documented attraction effects in the production of subject–verb agreement, examining error rates as well as speech timing effects in correctly produced sentences. In both the error and timing analyses, the results are strikingly similar across the experimental settings. Furthermore, the two experiments challenge common assumptions about a strong plural markedness effect in agreement attraction. Through this replication and comparison, we have found that we can do production experiments via the internet with a high level of similarity to those done in the lab. These results will allow for future production research to be conducted online, which will provide more flexibility and efficiency to this type of experimentation.

Number attraction in verb and anaphor production

“…The reduced markedness effect we observe thus appears to be consistent with prior literature. In fact, we have since replicated the presence of increased singular attraction using materials from Experiment 1 in a web-based experiment (Kandel et al, 2021): we again observed a high number of PS errors and observed an even more reduced markedness effect, with little difference between singular and plural attraction (the attraction effect was 2% larger for singular attractors). The presence of singular attraction in our verb agreement experiments challenges the traditional interpretation of the markedness effect.…”

“…We investigated whether subject phrases containing nouns of different number would lead to verb agreement errors and to delays in verb articulation, even when agreement was produced correctly. Experiment 1 is also reported as Experiment 1 in Kandel et al (2021), where it serves as a point of reference for a methodological comparison of in-lab and web-based speech data collection.…”