We investigate the effect of script-based (Schank and Abelson 1977)
extralinguistic context on the omission of words in fragments like
“Another slice?”, as compared to full sentences (“Would you like another
slice of pizza?”). At the case of a data set elicited with a production
task, we show that predictable words are more often omitted than
unpredictable ones, as the Uniform Information Density hypothesis (Levy
&Jaeger 2007) predicts. We take both effects of linguistic and
extralinguistic context on predictability into account and discuss two
ways of estimating the likelihood of words in the presence of ellipses.