The emotion in speech is expressed both by universal and specifically linguistic means. Cases of miscommunication between native and nonnative speakers in terms of expressing and interpretation this emotional information occurs mostly due these cultural and linguistic peculiarities. In addition to lexical means, the role of intonation in such cases is enormous. The present paper describes an experiment to discriminate the universal and linguistic features of expressing the emotional information in Japanese spontaneous speech by native and nonnative (Russians not speaking Japanese) informants in the perception level. A multi-dimensional model of emotion is used for marking the perceived emotional information. The results demonstrate considerable differences in defining the emotional information for native and nonnative informants, especially for such parameters as the basic atmosphere of an utterance and the attitude in interrogative utterances. This experiment also proved to be a good method for eliciting utterances where the linguistic features for expressing the emotion dominate which is particularly useful for studying the intonation in a given language.