Word Order Communicates User Intent in Search Queries
Anastasia Smirnova
Abstract
One of the most intriguing features of the human-computer interaction is that human users rarely use full-fledged natural language utterances when communicating with computers. I investigate properties of linguistic expressions that arise in the context of human-computer interaction, focusing specifically on word order in search queries. I report results of two experimental studies that suggest that word order is meaningful and communicates user intent: the attribute that is more important or desirable to the user tends to be mentioned first. Study 1A and Study 1B show that these results hold for typed and spoken queries. Study 2 suggests that users do not have meta-awareness of this strategy. These findings can contribute to a better detection of user intent. Incorporation of word order sensitivity into technology design can lead to more efficient and satisfactory human-computer interactions.