The study reported here is an attempt to understand human verbal-gestural behavior in a multimodal bilingual setting. Specific questions addressed are: 1) What kind of deictic gestures people use in a machine-mediated condition, and how these differ from those used in a human-mediated condition, 2) How significant the use of gestures is in each condition, and 3) How verbal and gestural behaviors are interrelated, 4) What the implications of our findings are for a multimodal spoken language interpretation system.