Track 2 – 09:30-13:00
Duration: 3 hours, 30min coffee break
Location: Conference 2+3
Presenters: Jonathan S. Brumberg1, Cara Stepp2, Edmund Lalor3, Adrian KC Lee4
1) Speech-Language-Hearing: Sciences and Disorders, University of Kansas, USA
2) Speech, Language and Hearing Sciences, Boston University, USA
3) Electrical Engineering, Trinity College, Dublin, Ireland
4) Institute for Learning and Brain Sciences, University of Washington, USA
Target Audience: New researchers to the field of speech electrophysiology, including neurophysiology and neuroimaging. The tutorial will be accessible to both research students and scientists. Specialists in modeling of speech production, automatic speech recognition and speech perception will also benefit from the tutorial.
This tutorial will present an overview of the current best practice approaches to the recording, analysis, and interpretation of surface electromyography (sEMG), electroencephalography (EEG), and magnetoencephalography (MEG) as applied to speech research. These non-invasive neurophysiological tools have incredible promise for accelerating understanding of the basis of normal processes, communication disorders as well as opening doors for advancement in speech technology, but also present unique opportunities for abuse due to a lack of knowledge of the factors affecting the signal, inherent technical limitations, and special considerations for studying speech. It is imperative that both investigators and consumers of speech research understand the appropriate methodologies and limitations of these tools: to avoid both wasting time with poor study design as well as misinterpreting data. We will begin with a tutorial on sEMG, followed by EEG for speech production and perception, and end with topics in MEG.
1) Surface electromyography (sEMG)
This portion of the seminar will focus on a review of the theory behind recording sEMG for speech systems. We will first present details about the generation of the underlying signal, we will then review signal detection, the signal composition, data analysis techniques, and possible forms of signal degradation. This discussion will include the physiology of muscle and motor units and the effect of individual physiology on sEMG detection, types of electrodes and electrode configurations and their impact on the signal recorded, and how to identify and avoid noise in recordings. We will close with a focus on special considerations for speech anatomy and will present recent clinical and translational applications of sEMG. The muscles of speech differ from those of the limbs in terms of size, amount of overlap in fibers of different muscles, and knowledge about innervation zones. These limitations and the subsequent impact on clinical and translational applications will be specifically discussed. The information provided in this section will form a foundation for the EEG and MEG sections.
2.1) EEG for speech production
This portion of the tutorial will focus on the suitability and application of EEG to speech research. We will first review the neurological basis for the EEG signal and its relationship to the orofacial sEMG and MEG signals. We will then discuss common event-related potentials (ERP) used to describe speech processing in the brain including a summary the motivations, techniques and anticipated responses. We will focus especially on EEG responses in preparation for speech production (readiness potential and contingent negative variation), the motor-induced suppression of the auditory N100, and speech potentials during imagined, or covert speech production. We will close with a discussion of the potential pitfalls of EEG during overt speech production and highlight methods for mitigating the effects of orofacial muscular activity on the EEG signal.
2.2) EEG speech perception & hearing
This portion of the tutorial will expand on the topics discussed in the previous section on speech production. In this portion we will focus especially on EEG responses elicited during receptive speech as a result of sensitivity to different discrete speech tokens (e.g., the mismatched-negativity). We will then go on to discuss the derivation and application of EEG indices of continuous speech processing. We will discuss recent theoretical perspectives on the meaning of these indices and recent findings that examine these theories. This will include consideration of whether or not these indices represent only obligatory responses to changes in the acoustic input, or whether they also reflect active processing of linguistic aspects of speech.
3) MEG for speech and hearing
This portion of the seminar will focus on a review of how scientists capitalize the fine temporal resolution of MEG to study speech and hearing sciences. We will first review what neurophysiological signals that can be captured using this technique. We will also discuss how, by incorporating cortical anatomy, we can estimate the origin of these MEG signals from the cortex. We will close by discussing the merits of combining EEG and MEG measurements together and how this approach would provide information about cortical dynamics associated with perceiving speech in a crowded environment.
The presenters of this tutorial are researchers in the fields of speech and hearing sciences with backgrounds in engineering, computer science, and neuroscience. Working at this interface, the authors are specially equipped to provide this tutorial for careful application of sEMG, EEG, and MEG to the study of speech and hearing.