Résumé: © 2017 The Author(s). Published by S. Hirzel Verlag EAA.Ultrasound is widely used in speech production research. In this paper, we explore the feasibility of speckle tracking in ultrasound tongue image sequences to obtain tissue point correspondence. The speckle tracking approaches explored include: deformation registration; optical flow; and local invariant feature based methods. Moreover, an image similarity-based automatic tracking re-initialization method is incorporated into the speckle tracking to improve robustness. Results demonstrate that deformation registration provides the best performance, and that for short-duration sequences, speckle tracking is feasible and can provide point correspondence with millimeter accuracy, which is useful for practical applications in speech production research such as 3D tongue dynamic modeling and strain analysis of the tongue tissue.