In this paper, we choose about one hour expressive speech and make a pilot study on how to use the breath segments to get more natural and expressive speech. We focus on exploring when the breath segments occur and how their acoustic features are affected by the speaker’s emotional states: valence and activation. The statistic analysis has been carried out to find the relationship between the length and intensity of the breath segments and the two state parameters. Finally, a perceptual experiment is done by employing the analysis results to the synthesized discourses, the results imply that breath segment insertion can help to improve the expressiveness and naturalness of the synthesized speech. Keywords: breath segment, expressive speech, emotion, valence, activation