Slide Slam

< Slide Slam Sessions

Slide Slam G9

Web-based Language Production Experiments: Semantic Interference Assessment is Robust for Spoken and Typed Response Modalities

Slide Slam Session G, Wednesday, October 6, 2021, 6:00 - 8:00 am PDT Log In to set Timezone

Kirsten Stark1,2,3, Cornelia van Scherpenberg2,4,5, Hellmuth Obrig2,4,5, Rasha Abdel Rahman2,3; 1Einstein Center for Neurosciences, Charité Berlin, 2Berlin School of Mind and Brain, Humboldt-Universität zu Berlin, 3Department of Neurocognitive Psychology, Humboldt-Universität zu Berlin, 4Department of Neurology, Max Planck Institute for Human Cognitive and Brain Sciences, 5Clinic for Cognitive Neurology, University Hospital and Faculty of Medicine Leipzig

For experimental research on speech production, temporal precision and high quality of the recorded audio-files are mandatory. These requirements are a considerable challenge if speech production is to be investigated online. However, besides the current situation, online research has a huge potential regarding efficiency, ecological validity and diversity of study-populations in psycholinguistic and related research. Here, we supply confirmatory evidence that language production can be investigated online and demonstrate that written naming responses (using the computer keyboard) are a reliable and efficient alternative to typical overt spoken responses. To assess semantic interference effects in both modalities we performed two pre-registered experiments (n=30 each, sample sizes estimated using power analyses) in online settings using the participants’ web-browsers. A cumulative semantic interference (CSI) paradigm was employed that required naming of several exemplars of semantic categories within a seemingly unrelated sequence of objects. Reaction time is expected to increase linearly for each additional exemplar of a category being named (Howard et al., 2006). In experiment 1, cumulative semantic interference effects in naming times described in lab-based studies were replicated. In experiment 2, the responses were typed on participants’ computer keyboards and the first correct key press was used for reaction time analysis. This novel response assessment yielded a qualitatively identical, very robust CSI effect. We additionally compared automated data processing procedures, including accuracy assessment using string-matching metrices, and manual preprocessing procedures. Thereby we provide evidence that automated assessment of participants’ accuracy in their typewritten answers can increase inter- and intra-rater replicability while considerably reducing the time needed to process such data prior to data analysis. Thus, besides technical ease of application, collecting typewritten responses and automated data preprocessing can reduce the work load for language production research. Results of both experiments open new perspectives for research on reaction time-sensitive effects in language experiments across a wide range of contexts, including cross-sectional or longitudinal studies which may have limited practicability in in-person, lab-based settings. Perspectively, employing speech production experiments in web-based settings may also open new possibilities to test participants with an acquired language disorder (most notably, post-stroke aphasia), for which long-term follow-up, especially regarding scientifically motivated questions, is often hampered by the efforts related to re-inviting and transporting the patient to the respective institution. We highlight important technical and conceptual considerations for the planning stages, response time assessment, and data analysis. Hereby, we hope to provide recommendations for an easy access to studying both typewritten and spoken language production online. JavaScript- and R-based implementations for reaction time assessment and data processing are available for download.

< Slide Slam Sessions

SNL Account Login

Forgot Password?
Create an Account