ATR is constructing a Dialogue Database called ADD (ATR Dialogue Database). It consists of simulated and actual telephone or keyboard dialogues. This report shows the linguistic items recorded in ADD and conversational subjects of dialogues. ADD data have many linguistic items. Especially, correspondence data between Japanese and English are added, then, differences of expressions between the two languages can be extracted. We use the conversational topics for each tasks to gather unbiased data. Each dialogue has some keywords by which we can determine the outline of the dialogue.