Abstract
Several studies have found learning of biconditional grammars only under intentional rule-search conditions (e.g., Johnstone & Shanks, 2001). Memorization of strings merely led to the learning of chunks. We used a musical grammar, a diatonic inversion, that is a type of biconditional grammar. Participants either were required to memorize a set of grammatical tunes (incidental learning), or were asked to search for the underlying rule whilst being given feedback about their performance (intentional learning). The results showed that participants in the incidental-learning condition did not learn the inversion rule and merely acquired explicit knowledge about chunks. However, participants in the intentional-learning condition learnt both the inversion rule and chunks.
Notes
1Bigram is a term use to define two adjacent elements, trigrams are three adjacent elements, n-grams are n adjacent elements.
2The letter strings were generated from six different letters and were six elements long. There was a rule that linked the letter in Position 1 with the letter in Position 5, 2 with 6, and so on.
3That is, intervals between two adjacent notes were defined not in terms of numbers of semitones between them but in terms of number of steps in the scale of C major that separated them.
4C3 indicates Middle C.
5 There was no difference in musical experience between participants in the control group and those in the experimental group, χ2 = 1.56, p = .48 (memorization group, 5 musically experienced subjects; rule-searching group, 6 musically experienced subjects; control group, 10 musically experienced subjects). Musical experience was defined in terms of whether participants had more than 3 years of formal music education. Furthermore, having musical education as an extra variable led to no significant main effects or interactions.
6Contour bigram was defined in terms of two adjacent contours (e.g., an up followed by a down is defined as a + − bigram). In the contour and the chunk sets, the ungrammatical tunes violated the inversion rule in terms of their contour, thus leading to differences in contours between the grammatical and the ungrammatical items. In the magnitude set the inversion rule was only violated in terms of interval magnitude, not contour, and thus the grammatical and the ungrammatical tunes had identical contours.
7For the training and the test materials and some audio samples see http://www.lifesci.sussex.ac.uk/home/Gustav_Kuhn/QJEP2006
8Each training block consisted of 24 trials.
9There was no difference in musical experience between participants in the control group and those in the experimental group, χ2 = 1.63, p = .44 (memorization group, 5 musically experienced subjects; rule-searching group, 3 musically experienced subjects; control group, 6 musically experienced subjects). Furthermore, having musical education as an extra variable led to no significant main effects of interactions.
10 Data from 2 participants were lost due to a problem in the presentation program.
11 Each training block consisted of 24 trials.
12 These nonsignificant results should be treated with some caution, as the confidence intervals were very large, reflecting a lack of statistical power for detecting metaknowledge.
13There was no significant difference between the number of subjects with musical experience between the memorization group in Experiment 3 (6 musically experienced subjects) and the memorization group in Experiment 2 (5 musically experienced subjects), χ2 = 0.19, p = .68.