Variability and recognition memory: Are there analogous indexical effects in music and speech?

Abstract

Indexical effects refer to the influence of surface variability of the to-be-remembered items, such as different voices speaking the same words or different timbres (musical instruments) playing the same melodies, on recognition memory performance. The nature of timbre effects in melody recognition was investigated in two experiments. Experiment 1 showed that melodies that remained in the same timbre from study to test were discriminated better than melodies presented in a previously studied but different, or unstudied timbre at test. Timbre effects are attributed solely to instance-specific matching, rather than timbre-specific familiarity. In Experiment 2, when a previously unstudied timbre was similar to the original timbre and it played the melodies at test, performance was comparable to the condition when the exact same timbre was repeated at test. The use of a similar timbre at test enabled the listener to discriminate old from new melodies reliably. Overall, our data suggest that timbre-specific information is encoded and stored in long-term memory. Analogous indexical effects arising from timbre (nonmusical) and voice (nonlexical) attributes in music and speech processing respectively are implied and discussed.

Keywords:

• Melody recognition memory
• Spoken word recognition
• Talker-variability effects
• Timbre-repetition effects

Notes

¹The main analyses were preceded by an examination of potential effects of musical training for the sample, although there was no systematic attempt to control the number of participants with or without formal musical training as this was not a primary goal of the study. A two-way mixed design analysis of variance (ANOVA), with music training as the between-subjects factor (participants who had at least 4 years of formal training, n=14, vs. those without or with less than 4 years of training, n=38) and timbre context as the within-subjects factor, revealed no reliable interaction, F<1.60. This shows that music training did not influence the timbre effects. A main effect of musical training was marginally significant, F(1, 50) = 3.94, MSE=0.66, p=.053. Participants with at least 4 years of training (M=1.27, SD=0.77) tended to discriminate better than those without or with less than 4 years of training (M=0.98, SD=0.60). Since training did not interact with timbre context, the main findings for the timbre-context conditions in the main analyses can be generalised across all participants within the sample, and all tabulations of results are collapsed across music training.

²No reliable interaction between music training (participants who had at least 4 years of formal music training, n=10, vs. those without or with less than 4 years of music training, n=32) and timbre context was observed, F<2.43. Again, this shows that music training did not influence timbre effects. A main effect of musical training was significant, F(1, 40) = 8.17, MSE=0.36, p<.01. Participants with at least 4 years of training (M=1.36, SD=0.66) discriminated better than those without or with less than 4 years of training (M=0.93, SD=0.66). As in Experiment 1, the main findings for the timbre-context conditions can be generalised across all participants.