Stimuli For the stimuli, the same female native British English speaker used for Delle Luche et al. (2016) recorded 40 tokens of each word, “mummy” and “bummy”, in isolation and in child-directed speech, which were then narrowed down to 8 sound files. The tokens were manipulated for the ones that sounded the same with similar word patterns. All the files lasted approximately 24 seconds, and the tokens were normalised for amplitude with Praat (Weenink, 2010). Both “mummy” and “bummy” tokens were chosen on the basis of having approximately similar duration, intensity and pitch, as to select the ones with trochaic patterns. The average duration, intensity and pitch was totalled using Praat, and then compared between the experimenters, as presented in Table 1. Four different orders of 13 were created. With that, 8 conditions were put together, half starting with “mummy”, and the other half starting with “bummy”. All infants …show more content…
Inter-rater reliability was measured using an intra-class correlation coefficient, through which a high degree of reliability was found between their measurements of duration of the CP as a whole word (ICC = .975, 95% CI [.936, .992], F (12, 24) = 39.663 p < .001), and its initial phoneme (ICC = .985, 95% CI [.962, .995], F (12, 24) = 66.702, p < .001), the intensity of the CP as a whole word (ICC = .854, 95% CI [.929, .952], F (12, 24) = 6.854, p < .001) and its initial phoneme (ICC = .901, 95% CI [.749, .967], F (12, 24) = 10.106, p < .001), and the pitch of the CP (ICC = .910, 95% CI [.772, .970], F (12, 24) = 11.133, p < .001). Similarly, a high degree of reliability was also found between their measurements of duration of the MP as a whole word (ICC = .994, 95% CI [.985, .998], F (12, 24) = 169.92, p < .001), and its initial phoneme (ICC = .959, 95% CI [.896, .986], F (12, 24) = 24.334, p < .001), and the pitch of the MP (ICC = 1.000, 95% CI [.999, 1.000], F (12, 24) = 4212.1, p