Testing the Ecological Validity of Repetitive Speech

Can one trust experiments conducted with repeti- tive speech to represent normal language behaviour? We compare the spectra of repetitive productions of phrases with the same phrases read from a ran- domised list. We use a data-driven spectral distor- tion measure that is trained to respond to linguisti- cally relevant differences: it is based on a classifier that separates sounds into linguistically equivalent or not. We find that repetitive speech is not distinct from individually uttered speech. The difference be- tween these two sorts of speech is smaller than vari- ation within each. It is substantially smaller than typical differences between utterances produced by different subjects.