Summary
An experiment was set up to discover whether ten‐ to eleven‐year‐old children would perform better on an imaginative writing task if given an aural stimulus consisting of an imaginary dialogue rather than the more commonly used written titles. The main energy of the investigation is focussed on the viability of different forms of assessing the quality of the resulting compositions. The most useful criteria from a large battery proved to be vocabulary, weighted index of modifying clauses, pre‐verb length and sequence.