The goal of the study is cross-linguistic research of 8-12 year old children's emotional speech classification by experts and automatically using the material of the Russian and Tamil languages. In the proposed project, this objective will be solved by a complex analysis of the emotional speech of informants with typical development in the age range of 8–12 years, using four emotional states “sadness, joy, anger and neutral”, to analyze what is relevant and new in connection with the lack of analogous data. Comparison of data on recognizing the emotional speech of children on the material of two languages belonging to different language groups will be made. Two approaches to emotional speech recognition are planned to use – by man and automatically, moreover, for automatic recognition, different analysis algorithms will be applied. The use of different algorithms for automated assessment of emotional states according to speech features that were previously used in analyzing a specific language is new and relevant. This will make it possible to determine their universality or specificity and, perhaps, based on them to develop an entirely new approach to analyzing emotional speech of children. The standardized approach to the collection of speech material will provide valid data on the recognition of child's emotions on the base of their speech characteristics. In this paper, we describe the methodology of our study and preliminary results of a perceptual experiment.