Segmental speech error data elicited at prosodically-defined locations in tongue twisters

Tongue twister error data described here were collected from 44 unimpaired speakers producing six repetitions each of 40 tongue-twisters that manipulated the position of error-prone segments within two prosodic domains: intonational phrases and utterances. Data are counts of perseveration and anticipation errors on the initial segments of phrase-initial words or phrase-final words. The location of the phrase-level prominence and other factors associated with speech errors were carefully controlled. For more details about the design, materials and methods, and for interpretation and discussion, see Beirne and Croot (2018) [1].