SPEAK aims to establish the intonation unit as a basic unit of organization of spoken discourse. It does so by exploring the nature of the relationship between intonation units, syntactic constituents, and communicative units in Kazakh and Italian conversation. To achieve this goal, the project combines corpus analysis and experimental methods that approach the main question from the point of view of both production and perception. This project is innovative in that it represents the first study of the intonation units that systematically addresses the issue of written language bias by employing data from the only genre that is cross-linguistically available, namely conversation, and building an explicit comparison between two languages with different traditions of written production (Kazakh and Italian). It is timely given the increasing amount of economic, human, and environmental resources devoted to the development of AI products for the treatment of spoken language. It is impactful insofar as its insights can contribute to the understanding of grammar representation in the human mind and to the technical application that can be leveraged for the development of speech technologies.