Clustering model to identify biological signatures for English language anxiety

The high-level contribution of this paper is the development of a clustering model to identify biological signatures for language anxiety in non-native English speakers. We use the Gas Discharge Visualization (GDV)-based Electro-photonic impulse analyzer to collect electro-photonic emission of fingertips, called GDV-grams, of students belonging to three different categories: Native English speakers, Indians (Commonwealth country) and Confucian Heritage Cultures (CHC). The built-in GDV-software analyzes the GDV-grams of an individual and quantifies the activity status of the organs/organ systems in the form of energy coefficients (EC). The basic idea behind our clustering model is to first compute the average of the absolute difference in the EC values, Δ(EC), for each of the three categories of the students, before and after a language test. Using the average Δ(EC) values for native English speakers as the baseline, we compute the relative absolute difference, ΔΔ(EC), in the energy coefficient values for the CHC group and the Indians. We run the K-Means clustering algorithm on a ΔΔ superset comprising of ΔΔ(EC) values obtained for the different organs/organ systems for the CHC group and the Indian students and classify these values to three different clusters representing organs/organ systems that have low, moderate and high impact due to English language anxiety. The corresponding range of the ΔΔ(EC) values are the biological signatures for anxiety of non-native English speakers with respect to any particular language activity and can be used as benchmarks to classify a test subject as having low, moderate or high levels of English language anxiety.