Date of Award

Spring 1988

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Psychology

Program/Concentration

Industrial/Organizational Psychology

Committee Director

Glynn D. Coates

Committee Member

Terry L. Dickinson

Committee Member

Randall L. Harris

Committee Member

Raymond H. Kirby

Committee Member

William Muto

Abstract

The present study was conducted to determine design principles for menu-based natural language (MBNL) interfaces and to provide evidence for the nature of visual search processes with menu-based systems. The effects of window size, window activity, and query length were investigated. Window size was manipulated as a between-subjects variable with three levels representing a sixteen-item window size, an eight-item window size, and a four-item window size. Window activity was manipulated as a within-subjects variable with two levels representing single active and multiple active windows. Query length was manipulated as a within-subjects variable with three levels representing one-, two-, and three-item query lengths. Thirty six subjects randomly assigned to three groups, based on the window size factor, performed queries with the three query lengths in both window activity conditions in counterbalanced order. It was found that two- and three-item queries were performed faster with single active windows. However, subjects rated multiple active windows as more 'natural'. Query times also increased with query length and errors were most likely to occur on the longest query. Longer eye fixation durations were observed with the four-item window size. Fixation frequencies, fixation durations, dwell times, and relative dwell times all varied as a function of query length. Visual behavior also depended on which 'area of interest' subjects were viewing, and this effect interacted with window activity and query length. Finally, it was found that menus were not scanned randomly. However, scanpaths were less deterministic with multiple active windows and tended to become less constrained as query length increased. Based on the findings, human factors design principles were derived for application to MBNL interfaces.

DOI

10.25777/94hk-pg28

Share

COinS