In search of tonal grounds for short term melody recognition

Abstract

What information is retained after the first hearing of a note sequence? Taylor and Pembrook (1983) investigated for this question with melodies in C major scale using a singing back and a dictation task. They observed that: (1) recall performance for the last notes was better, (2) contour complexity (more direction changes) reduced correct contour recall, (3) fewer errors are made when the melodies started with C4 instead of C5, (4) recall for a note of direction change is better if the preceding interval is larger. Our study reassessed their results with a larger and more controlled stimulus set and a recognition paradigm with non-transposed and contour preserving lures. We found a full J-shaped serial position curve (recency > primacy > center). Participants’ correct recognition of melodic intervals decreased with contour complexity. Also, recognition performance was greater when a note of direction change (Ù or Ú) was changed, especially when the preceding interval was larger. These findings indicate that non-transposed comparison melodies did not turn our study into a simple pitch memory test. Furthermore, when filling the retention interval with verbal nonsense stimuli as opposed to melodic stimuli, less interference was observed. Secondly, the higher the similarity between interference and test items, the lower the performances (diatonic > nondiatonic > verbal interference). Finally, we collected memorability and complexity ratings for our note sequences. These ratings were found to be consistent with the recognition rates. Results are discussed with regards to Baddeley’s (2000) and Berz’ (1995) models of short-term memory.