Tonal priming beyond tonics

Marmel, F. & Tillmann, B.

Université Claude Bernard Lyon 1, Université de Lyon, laboratoire Neurosciences Sensorielles Comportement Cognition, CNRS UMR 5020, IFR 19, France

Music Perception (in revision)

Correspondence concerning this article should be addressed to

Frédéric Marmel

Université Claude Bernard - Lyon I

CNRS UMR 5020

Neurosciences Sensorielles Comportement Cognition

50 Av. Tony Garnier

F-69366 Lyon Cedex 07

France

Tel : +33 (0) 4 37 28 74 90

Fax : +33 (0) 4 37 28 76 01

frederic.marmel@olfac.univ-lyon1.fr

The musical priming paradigm allows investigating listeners’ expectations based on their implicit knowledge of tonal stability. Up to now, priming data are limited to report facilitated processing of tonic over non-tonic events. The special status of the tonic as a cognitive reference point brings into question the subtlety of listeners’ tonal knowledge: Is the facilitated processing observed in priming studies limited to tonic events, or is tone processing influenced by subtler tonal contrasts? Our present study investigated tonal priming for mediants over leading tones presented in melodic contexts. Experiment 1 used a timbre discrimination task and Experiment 2 an intonation task. Facilitated processing was observed for mediants over leading tones, thus showing that priming effects are not limited to pairs of tonal degrees including the tonic. This finding emphasizes the subtlety of non-expert listeners’ tonal knowledge.

One contribution of music cognition research to our understanding of music has been to investigate how listeners process musical structures. The Generative Theory of Tonal Music (GTTM) proposed a set of rules describing how listeners could “organize all the pitch-events of a piece into a single coherent structure” (Lerdahl & Jackendoff, 1983, p.106). The formalized rules lead to empirically testable predictions of listeners’ music perception (for a review, see Bigand, 1993). The GTTM predicts that listeners organize musical events into a hierarchical tree structure following reduction processes. These processes require - among others – listeners’ knowledge of the contextual tonal stability of the musical events. Our present study investigated this knowledge by assessing the influence of a subtle difference in tonal stability on the speed of tone processing. Listeners have been reported to have implicit knowledge of tonal stability since Francès (1958). This knowledge has been investigated in numerous empirical studies using ratings of completion (Krumhansl & Kessler, 1982) or of similarity (Krumhansl, 1979, Experiment 1), recognition memory tasks (Bartlett & Dowling, 1980; Dewar, Cuddy, & Mewhort, 1977; Dowling, 1978; Krumhansl, 1979, Experiment 2 and 3), or the harmonic priming paradigm (Bharucha & Stoeckig, 1987; Bigand & Pineau, 1997; Bigand, Poulin, Tillmann, Madurell, & D’Adamo, 2003). Among these approaches, the priming paradigm is a tool for the indirect investigation of a context’s influence on event processing and allows studying musical expectations based on listeners’ tonal knowledge. In this paradigm, participants are not required to make direct judgments on the relation between the prime context and the target or on the target’s expectedness. The task focuses on a perceptual (non structural) feature of the target chord, such as intonation judgments (Bharucha & Stoeckig, 1987; Bigand et al., 2003), phoneme identification for sung music (Bigand, Tillmann, Poulin, D’Adamo, & Madurell, 2001), and timbre discrimination (Tillmann, Bigand, Escoffier, & Lalitte, 2006). The priming paradigm has provided evidence for the influence of tonal stability on the processing speed of musical events, notably with facilitated processing for tonic targets over subdominant targets (Bigand & Pineau, 1997; Bigand et al., 2003). An advantage of the priming paradigm is to investigate listeners’ knowledge implicitly. Explicit investigations had suggested that children differentiated harmonic functions by the age of ten (Imberty, 1981), but the harmonic priming paradigm has revealed children’s sensitivity to harmonic function by the age of six (Schellenberg, Bigand, Poulin-Charronnat, Garnier, & Stevens, 2005). The advantage of implicit approaches has also been highlighted by the finding that a brain-damaged patient exhibiting severe amusia showed sensitivity to tonal stability differences in priming experiments, but not in an explicit task requiring completion judgments on the same musical sequences (Tillmann, Peretz, Bigand, & Gosselin, 2007).

A difficulty for studying listeners’ knowledge of tonal stability is that expectations elicited by this knowledge are entwined with expectations elicited by lower-level factors. Tonal stability partly correlates with tones’ frequencies of occurrence (Francès, 1958; Krumhansl, 1990; Huron, 2006), thus knowledge-driven expectations linked to tonal stability can be confounded with bottom-up expectations elicited by tone repetition. Additional musical expectations are elicited by intervals and melodic contour, mostly following Gestalt principles (Narmour, 1990; Schellenberg, Adachi, Purdy, & McKinnon, 2002). These expectations have thus to be controlled in the experimental material aiming to investigate cognitive expectations. After single-chord primes, Bharucha and Stoeckig (1987) observed priming even when prime and target chords did not share any component tones. After chord sequence prime contexts, facilitated processing was observed for tonic target chords over subdominant target chords, even when the context shared more tones with the subdominant chord than with the tonic chord (Bigand et al., 2003). Facilitated processing for tonics over subdominants was also observed for target tones at the end of melodic contexts, in which the change of the targets’ tonal function was instilled by a single note change in the beginning of the melodic primes (Marmel, Tillmann, & Dowling, In Press).

The implicit nature of the priming paradigm is an opportunity to assess cognitive representations of tonal stability that may be more difficult to assess with explicit paradigms. Yet, the musical priming experiments have only manipulated a few tonal degrees and always for targets being the last event of the musical context. Priming experiments contrasting within-key tonal degrees have contrasted tonic targets to subdominant or dominant targets. Testing these comparisons for targets being the last event of the context might favor the tonic in addition to its tonal stability. The tonic is the most common degree at the end of a musical phrase (Huron, 2006) and it provides a feeling of closure. It might thus be expected more strongly than other tonal degrees at the final position. This confound between musical closure and tonal stability has been recently addressed by varying the position of the target (i.e., a tonic or a dominant) within chord sequences (Tillmann & Marmel, submitted). Our present study addresses this confound by contrasting two tonal degrees that are different from the tonic. We adapted the priming paradigm to melodic sequences ending on either the mediant (defining the related condition) or the leading tone (defining the less-related condition). This choice of tonal degrees also allowed us to investigate tonal knowledge on a more fine-grained level: In the tonal stability profiles obtained with the probe-tone judgments (Krumhansl & Kessler, 1982), the difference in mean ratings between mediant and leading tone is smaller than the difference between tonic and subdominant (1.5 versus 2.3). Thus, a priming effect for mediants over leading tones would highlight the subtlety of listeners’ tonal knowledge and, more specifically, its influence on the speed of tone processing.

Two priming experiments were conducted. Experiment 1 used a timbre task and Experiment 2 an intonation task. The tonal degree of target tones was manipulated with minimal changes in the melodic contexts to focus on expectations linked to listeners’ tonal knowledge. Related and less-related contexts differed only by their first half being transposed (Experiment 1a), or by one (possibly repeated) tone in this first half (Experiment 1b and 2). This material construction controlled for tone repetition and kept constant the melodic contour preceding the target tone. Our hypothesis was that processing should be facilitated for mediant targets over leading tone targets. The experimental methods were very similar, and so we first provide a description of general features.