“…A rating in the middle of the line (at 50 mm, or ''neutral'') indicated that the speaker samples were equally severe. In this way, listeners were able to indicate their relative preference for one speaker sample compared to the other; the farther from the endpoint for a given speaker sample, the more ''preferred'' (i.e., less severe) it was compared to the other speaker sample (Searl & Small, 2002). Ninety sample pairs were created (i.e., every possible combination in AB and BA orders); ten samples were repeated to determine intrarater reliability.…”