Interrogative pronouns such as what in English, shenme in Mandarin Chinese, and mwe/mwusun in Korean all have developed extended uses beyond interrogation. Such uses may include filling a gap in conversation, softening a speaker’s epistemic stance, and indicating strong emotions such as surprises or incredulity. Yet there is little research dealing with crosslinguistic patterns with large corpora of interactive discourse data. This paper investigates the extended uses based on corpora of multiple telephone calls from the three languages. It shows that eight categories of extended use can be identified in the corpora and that most of the extended uses tend to fall in the negative territory. The paper provides a pragmatic interactive account for this phenomenon and hopes that the taxonomy and coding scheme developed here can serve as a starting point for future crosslinguistic and corpus-based comparative studies of what-like tokens as well as of the discourse pragmatic uses of other interrogative forms.