Word onset phonetic properties and motor artifacts in speech production EEG recordings

Abstract

Electrophysiological research using verbal response paradigms faces the problem of muscle artifacts that occur during speech production or in the period preceding articulation. In this context, this paper has two related aims. The first is to show how the nature of the first phoneme influences the alignment of the ERPs. The second is to further characterize the EEG signal around the onset of articulation, both in temporal and frequency domains. Participants were asked to name aloud pictures of common objects. We applied microstate analyses and time-frequency transformations of ERPs locked to vocal onset to compare the EEG signal between voiced and unvoiced labial plosive word onset consonants. We found a delay of about 40 ms in the set of stable topographic patterns for /b/ relative to /p/ onset words. A similar shift was observed in the power increase of gamma oscillations (30-50 Hz), which had an earlier onset for /p/ trials (∼150 ms before vocal onset). This 40-ms shift is consistent with the length of the voiced proportion of the acoustic signal prior to the release of the closure in the vocal responses. These results demonstrate that phonetic features are an important parameter affecting response-locked ERPs, and hence that the onset of the acoustic energy may not be an optimal trigger for synchronizing the EEG activity to the response in vocal paradigms. The indexes explored in this study provide a step forward in the characterization of muscle-related artifacts in electrophysiological studies of speech and language production.

The Effect of Lexical Cohort Size Is Independent of Semantic Context Effects in a Picture–Word Interference Task: A Combined ERP and sLORETA Study

“…Ouyang et al (2016) provided evidence that articulation-related artifacts could start up to 300 ms or more prior to articulation. Fargier et al (2017) found an earlier component onset around 150 ms before vocal onset when comparing the different movements of voice sounds. Wong et al (2017) observed an averaged naming time of 832 ms, and the onset of articulation was estimated to occur around 687 ms after pictures onset.…”

“…Studies regarding speech production have shown that stimulusaligned Event-Related Potentials (ERPs) could be well employed without significant contamination to the signals before articulation (Dell'Acqua et al, 2010;Aristei et al, 2011;Janssen et al, 2015;Rose and Abdel Rahman, 2017). Since articulation-related artifacts may emerge around 150 ms (Fargier et al, 2017) or 300 ms before voice onset (Ouyang et al, 2016), it is effective to analyze the semantic context effects occurring in the stage of word planning rather than articulatory buffering (Roelofs and Piai, 2015).…”

Variation in the speech signal as a window into the cognitive architecture of language production

“…The electromyographic activity resulting from movement of the articulators contaminates the EEG signal during and before the articulation (e.g., De Vos et al, 2010;Ganushchak & Schiller, 2008;Goncharova, McFarland, Vaughan, & Wolpaw, 2003;Ouyang et al, 2016;Porcaro, Medaglia, & Krott, 2015). How long before the articulation the signal is contaminated is a matter of discussion (e.g., Fargier, Bürki, Pinet, Alario, & Laganaro, 2017;Ouyang et al, 2016;Porcaro et al, 2015), but several studies have suggested that the phonological encoding process and at least part of the phonetic encoding processes can be investigated with EEG (see Bürki, Pellet-Cheneval, & Laganaro, 2015, or Laganaro, Python, & Toepel, 2013. Important efforts have also been put into developing and testing ways of removing artifacts from the EEG signal of interest (James & Hesse, 2005;Ouyang et al, 2016;Pham, Fine, & Semrud-Clikeman, 2011;Urigüen & Garcia-Zapirain, 2015;Vigario & Oja, 2008).…”

When a second language hits a native language. What ERPs (do and do not) tell us about language retrieval difficulty in bilingual language production. Preprint.

“…Because language production is associated with strong facial muscle and glosso-kinetic artifacts, we restricted the ERP analysis to uncontaminated time-windows. Speech production related artifacts occur up to 300 ms before the speech onset (Fargier, Bürki, Pinet, Alario, & Laganaro, 2018;Ouyang et al, 2016). Because the analyzed component occurred in the 250-350ms time-window, we restricted the analysis to trials with reaction times longer than 650ms (90% of trials).…”