ATR is constructing a Dialogue Database named ADD (ATR Dialogue Database). It is built from simulated telephone and keyboard conversations. This report shows the token count statistics of several linguistic phenomena of ADD and makes a comparison between telephone and keyboard conversations. The sample sizes are 54,931 words from the telephone conversations and 39,606 words from the keyboard conversations. The following computations have been included in this report counts of words, word categories, interjections, restatements, phrases, phrase patterns, clause patterns, sentence patterns, syntactic dependencies, semantic dependencies, word corespondences between Japanese and English, phrase corespondences, clause corespondences, sentence corespondences, utterance corespondences, word bigrams and word trigrams. From the statistical data, we conclude the following. Words used in telephone conversations are a superset of those used in keyboard conversations. The phrase patterns of telephone conversations are nearly the same as those of keyboard conversations. Telephone conversations tend to be more varied in sentence length than keyboard conversations. We were unable to extract the statistics for sentence patterns because the sample size was insufficient.