Intracranial EEG evidence of semantic interference and phonological facilitation in spoken word production


Behavioral and classical electrophysiological methods (scalp EEG) provide important information about the timing of different processes involved in spoken word production, while functional Magnetic Resonance Imaging (fMRI) allows the localization of these mechanisms. However, these techniques do not allow for both spatial and temporal resolution simultaneously, and scalp EEG is sensitive to articulatory-motor artifacts. The present study aims to investigate the phenomena of semantic interference and phonological facilitation during spoken word production using intracranial electroencephalography (iEEG). This technique, which consists of recording EEG signal from electrodes implanted in several cortical areas belonging to the language network, is used in patients with medically intractable epilepsy as presurgical evaluation. We will present preliminary data collected from patients with electrodes implanted in the dominant hemisphere, who completed a Picture-Word Interference task (PWI). Participants named a set of target pictures while ignoring a written distractor word that was categorically (e.g., fox-HORSE) or phonologically (hose-HORSE) related or unrelated (violin-HORSE). This paradigm has shown, consistent results across the last 30 years for both semantic interference and phonological facilitation and has helped shape the two-step processing architecture of contemporary production models. In addition, competitive lexical selection models attribute the semantic interference effect to increased competition between related lemmas, delaying word production, while non-competitive selection accounts attribute this delay to a post-lexical mechanism that outweighs conceptual priming. By contrast, all models attribute the locus of phonological facilitation to the word-form retrieval stage. Our planned Local-Field Potentials (LFPs) and time-frequency analyses will interrogate high gamma band (80-120 Hz) and beta band (12-30 Hz) activity as a function of distractor type. The outcomes of this study will provide important insights into brain networks involved in speech production and on the timing of each stage. Additionally, the use of iEEG will provide insight into the role played by the Inferior Temporal Gyrus, the Temporal Pole and the Baso-temporal language areas in speech production that are not usually investigated with other neuroimaging techniques (e.g. fMRI) due to artifacts or signal dropouts. While recruitment has been impacted by COVID-19, we expect to collect data from ~15 patients across three sites (Mater Hospital – Brisbane – Australia; Cleveland Clinic - Ohio – USA; Pittsburgh University – Pennsylvania – USA) to complete the full study.

Poster presentation at SNL 2020, Sandbox Series