Neural decoding of single vowels during covert articulation using electrocorticography

Abstract

The human brain has important abilities for manipulating phonemes, the basic building blocks of speech; these abilities represent phonological processing. Previous studies have shown change in the activation levels of broad cortical areas such as the premotor cortex, the inferior frontal gyrus, and the superior temporal gyrus during phonological processing. However, whether these areas actually convey signals to representations related to individual phonemes remains unclear. This study focused on single vowels and investigated cortical areas important for representing single vowels using electrocorticography (ECoG) during covert articulation. To identify such cortical areas, we used a neural decoding approach in which machine learning models identify vowels. A decoding model was trained on the ECoG signals from individual electrodes placed on the subjects' cortices. We then statistically evaluated whether each decoding model showed accurate identification of vowels, and we found cortical areas such as the premotor cortex and the superior temporal gyrus. These cortical areas were consistent with previous findings. On the other hand, no electrodes over Broca's area showed significant decoding accuracies. This was inconsistent with findings from a previous study showing that vowels within the phonemic sequence of words can be decoded using ECoG signals from Broca's area. Our results therefore suggest that Broca's area is involved in the processing of vowels within phonemic sequences, but not in the processing of single vowels.

Acoustic contamination of electrophysiological brain signals during speech production and sound perception

“…It is very important to underline that this report does not question the existence of relevant physiological neural information in high-gamma frequency signals underlying speech production or sound perception. Indeed, it has been shown by several groups that spectral features of imagined speech or silent articulation can be predicted from low or high-gamma signals recorded in participants not overtly speaking (Pei et al, 2011;Ikeda et al, 2014;Martin et al, 2014Martin et al, , 2016Bocquelet et al, 2016a;Anumanchipalli et al, 2019;Gehrig et al, 2019). Future developments of speech prostheses should thus build upon these findings.…”

The Potential for a Speech Brain–Computer Interface Using Chronic Electrocorticography

“…In these studies, areas within the STG, including the planum temporale within the A1 (BA41/42) and Wernicke's area (BA22), showed more pronounced activation during covert speech than in the primary motor cortex, with some disagreement about the premotor cortex [53,54]. However, the disagreement about the premotor cortex was cleared in ECoG studies by Pei et al [56] and Ikeda et al [57] studying phonemes in isolation, in which high gamma activity in the premotor cortex and the STG contributed most to decoding performance for covert speech. Of particular note is that unlike previous studies, Ikeda et al [57] did not find significant activation in Broca's area, which led them to postulate an involvement of Broca's area in sequencing phonemes [57].…”

“…However, the disagreement about the premotor cortex was cleared in ECoG studies by Pei et al [56] and Ikeda et al [57] studying phonemes in isolation, in which high gamma activity in the premotor cortex and the STG contributed most to decoding performance for covert speech. Of particular note is that unlike previous studies, Ikeda et al [57] did not find significant activation in Broca's area, which led them to postulate an involvement of Broca's area in sequencing phonemes [57]. Together, these studies suggest that a tightly controlled covert speech paradigm is necessary to ensure consistent results and that auditory approaches may perhaps be more readily adaptable to covert neural speech decoding.…”

The use of intracranial recordings to decode human language: Challenges and opportunities

“…A number of studies have provided evidence for decoding of neural activity associated with imagined speech features into categorical representations – i.e., covertly articulated isolated vowels (Ikeda et al 2014), vowels and consonants during covert word production (Pei et al 2011) and intended phonemes (J.S. Brumberg et al 2011).…”