Introduction: combining formal theories and statistical data in natural language processing

The papers in this issue address, or illustrate, the relation between symbo lic and numeric approaches to text and speech processing. This is currently an exciting and productive area of research and development in natural lan guage processing research. This introduction summarizes the background, lis ts important questions to be addressed, indicates how the papers relate to these, and draws out major lessons to be learnt from this state-of-the-art collection.