To determine whether an “expectation-of-learning” effect was operative in judging a child’s phonetic imitation, two techniques for scoring were used to evaluate sequentially one child’s phonetic imitations. A tape of this child attempting, in 20 trials, to imitate a novel syllable [′e ⊣ Ʒ ə] was played forward (Trials 1 to 20) to six judges. Of these six, one group of three judges used a seven-point rating scale to evaluate the adequacy of the child’s imitation of the syllable, while the other three used a phonemic-classification system. A reverse tape (Trials 20 to 1) also was evaluated similarly by two different groups of three judges. Judges assigned lower scores to “early” rather than “late” responses only if initial responses were heard first. The phonetic learning of the child in this experiment was real, but judgmental bias radically enhanced it in forward-order and nearly destroyed it in reverse-order listening.