Two experiments investigated the relationship between the words used to des
cribe event actions and estimates of event duration, and whether this relat
ionship might account for some of the variation found in duration estimates
. Experiment 1 manipulated sets of action words within an account of a robb
ery, and replicated the Burt and Popple (1996, Experiment 1) finding that e
stimated duration decreased as the implied action speed of the robbery incr
eased. Experiment 2 found considerable variation across individuals in the
action words they used to describe a robbery, and that the number of action
words used was negatively correlated with estimated duration. Taken togeth
er the experiments suggest that the interaction between individual differen
ces in action categorisation and the use of inferences about the relationsh
ip between action speed and event duration, may account for some of the bet
ween-subject variation found in estimates of event duration.