In this work, we present our recent efforts to find an effective framework for the interaction between lexical hypothesization modules and syntactical parsers of an Automatic Continuous Speech Recognition. In this perspective the main problem is the uncertainty processing. We consider the chart a suitable blackboard to represent and to work on the unperfectly recognized linguistic data and structures. Our approach to the uncertainty processing is based on the asynchronous interaction between several modules in which we implement the basic functionalities of the word hypothesization and of the syntactical parsing with different running modalities (top-down and bottom-up knowledge activation, left-to-rigth and bidirectional input processing, breadth-first and depth-first search, best-first and beam use of the scoring). We describe the developed modules and the interaction preliminary framework then we present a brief discussion on the use of the system. Some ideas for the further activity are presented.