Lengths of utterances in an utterance length-limited medium

Abstract

Understanding how a society communicates is important in understanding the society itself. We studied the distribution of lengths of utterances, the basic units of speech, in a medium having a maximum utterance length. We considered the messages of Twitter, a social network and microblogging platform, as utterances and derived its length distribution. The derived distribution has two characteristic peaks: one at 33 characters and another at 140 characters, the maximum message length. We showed that the second peak is due to folding of a natural message length distribution about the maximum message length.