Human motion modelling is essential in understanding and enhancing human-human collaboration, particularly during the collaborative handling of rigid objects. This paper presents a data-driven approach to developing a new model for human motion in collaborative object-handling scenarios, aiming to enhance the effectiveness and adaptability of cooperative tasks. This study also addresses the accurate analysis of captured data from human-human collaboration. Through an experimental setup utilizing a motion capture system, human motion data is collected and preprocessed for a collaborative handling task to create a model that can generate motion styles. The proposed model is trained using experimental data based on Gaussian Mixture Models (GMM) and functional Principal Component Analysis (FPCA). The performance of the proposed model is evaluated in terms of its spatial accuracy. The results demonstrate that the model is capable of capturing the dynamics of human-human collaboration and motion styles, which makes it feasible for collaborative processes and potential applications in human-robot collaboration (HRC).