NIRS-Based Language Learning BCI System

Abstract

This paper describes a non-invasive, less restrictive, stable measurement system for a brain-computer interface (BCI) for second-language (L2) learning. The system outputs the arousal of the Yerkes-Dodson law. We employ non-invasive near-infrared spectroscopy (NIRS) as a basic device to measure the blood volume. However, the blood volume measured by NIRS includes base-line drift and is not stable. Here, we introduce a new drift-free variable defined as blood flow, which is the time derivative of the blood volume. Problems to be considered are: 1) Can the blood flow represent brain activity? 2) Where are the fewest brain areas strongly influenced by the language listening? 3) What parameter expresses arousal? We also present a measurement system. To verify the system, we carried out experiments with 40 listeners (10 advanced, 15 intermediate, and 15 novice listeners). When advanced L2 listeners were listening to the first and second languages, the distribution patterns of the root mean squares of the blood flow in the prefrontal regions were close to the correlation coefficient of 0.89, which shows that blood flow can represent brain activity in language processing. The center of BA10 and the right and left BA46 in the prefrontal regions were sufficient to detect language processing. The root mean squares of the differences of the left and right BA46 from BA10 peaked at a certain L2 readability level for all L2 listeners; they can be the parameter that expresses arousal. Thus, the measurement system can function as an input measurement device for BCI.