Groove in drum patterns as a function of both rhythmic properties and listeners' attitudes

Abstract

Music psychology defines groove as humans' pleasureable urge to move their body in synchrony with music. Past research has found that rhythmic syncopation, event density, beat salience, and rhythmic variability are positively associated with groove. This exploratory study investigates the groove effect of 248 reconstructed drum patterns from different popular music styles (pop, rock, funk, heavy metal, rock'n'roll, hip hop, soul, R&B). It aims at identifying factors that might be relevant for groove and worth investigating in a controlled setting in the future. Drum patterns of eight bars duration, chosen from 248 popular music tracks, have been transcribed and audio reconstructions have been created on the basis of sound samples. During an online listening experiment, 665 participants rated the reconstructions a total of 8,329 times using a groove questionnaire. Results show that, among 15 tested variables, syncopation (R2 = 0.010) and event density (R2 = 0.011) were positively associated with the groove ratings. These effects were stronger in participants who were music professionals, compared to amateur musicians or mere listeners. A categorisation of the stimuli according to structural aspects was also associated with groove (R2 = 0.018). Beat salience, residual microtiming and rhythmic variability showed no effect on the groove ratings. Participants' familiarity with a drum pattern had a positive influence on the groove ratings (η2 = 0.051). The largest isolated effect was measured for participants' style bias (R2 = 0.123): groove ratings tended to be high if participants had the impression that the drum pattern belonged to a style they liked. Combined, the effects of style bias and familiarity (R2 = 0.152) exceeded the other effects as predictors for groove by a wide margin. We conclude that listeners' taste, musical biographies and expertise have a strong effect on their groove experience. This motivates groove research not to focus on the music alone, but to take the listeners into account as well.

Fig 1. Study flow diagram.

Collection and triage of survey data.

Fig 2. Style preference.

Participants’ mean Style Preference ratings (z-scores) for 21 styles, grouped by musical Expertise. Error bars show the standard error of the mean.

Fig 3. Stimuli.

Transcription of one typical example for each Pattern Category.

Fig 4. Mean groove rating vs. recording year.

Mean Groove ratings of all 248 stimuli as a function of Recording Year. Number keys are given in Table 3.

Fig 5. Syncopation, event density, style bias, familiarity.

Groove ratings as a function of Syncopation (A), Event Density (B), Style Bias and Familiarity (C): Small dots denote single observations, large symbols denote the mean Groove ratings of the 248 stimuli. The symbol shapes, colours, and sizes represent the Pattern Category (see Fig 4). Sloping lines represent linear regression models.

Fig 6. Pattern category × expertise interaction plot.

Groove ratings as a function of Pattern Category, grouped by Expertise. Error bars are the standard error of the mean. Mean Beat Recognition and Rhythmic Interest ratings are given below each category.