Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

The Tonal-Metric Hierarchy

2014, Music Perception: An Interdisciplinary Journal

254 Jon B. Prince & Mark A. Schmuckler T H E T O NA L -M E T R I C H I E R A R C H Y: A C O R P U S A NA LYS I S J O N B. P R I N C E Murdoch University, Perth, Australia M A R K A. S C H M U C K L E R University of Toronto Scarborough, Toronto, Canada DESPITE THE PLETHORA OF RESEARCH ON THE ROLE of tonality and meter in music perception, there is little work on how these fundamental properties function together. The most basic question is whether the two hierarchical structures are correlated – that is, do metrically stable positions in the measure preferentially feature tonally stable pitches, and do tonally stable pitches occur more often than not at metrically stable locations? To answer this question, we analyzed a corpus of compositions by Bach, Mozart, Beethoven, and Chopin, tabulating the frequency of occurrence of each of the 12 pitch classes at all possible temporal positions in the bar. There was a reliable relation between the tonal and metric hierarchies, such that tonally stable pitch classes and metrically stable temporal positions co-occurred beyond their simple joint probability. Further, the pitch class distribution at stable metric temporal positions agreed more with the tonal hierarchy than at less metrically stable locations. This tonal-metric hierarchy was largely consistent across composers, time signatures, and modes. The existence, profile, and constancy of the tonal-metric hierarchy is relevant to several areas of music cognition research, including pitch-time integration, statistical learning, and global effects of tonality. Received: October 18, 2012, accepted March 17, 2013. Key words: pitch, time, tonality, meter, hierarchy A NEAR- UNIVERSAL CHARACTERISTIC OF Western music is the use of tonality (Krumhansl, 1990) and meter (Lerdahl & Jackendoff, 1983). Accordingly, there is abundant research in music cognition on these topics (cf. Krumhansl & Cuddy, 2010; London, 2004). For instance, tonality influences memory for musical materials (Dowling, 1978, 1991; Krumhansl, 1991), discrimination of musical chords (Beal, 1985), change detection (Cohen, Trehub, & Thorpe, 1989; Lebrun-Guillaud & Tillmann, 2007; Trainor & Trehub, Music Perception, VOLUM E 31, ISSU E 3, R IG HTS RES E RV E D . PLEASE DIR E CT ALL PP. 254–270, IS S N 0730-7829, 1994; Trehub, Cohen, Thorpe, & Morrongiello, 1986), ratings of completion and goodness (Boltz, 1989a, 1989b; Cuddy, 1991; Prince, 2011), melodic recognition (Bigand & Pineau, 1996; Schmuckler, 1997), recall (Boltz, 1991), and expectation (Schmuckler, 1989, 1990; Tillmann, Janata, Birk, & Bharucha, 2008). Similarly, meter plays a pivotal role in the perception of musical sequences, guiding attention and providing a framework for organizing the events in a melody (Jones & Boltz, 1989; Large & Jones, 1999; Palmer & Krumhansl, 1990). Its importance as a central organizational factor is evident in its influence on recognition memory (Jones & Ralston, 1991; Keller & Burnham, 2005), complexity (Povel & Essens, 1985), categorization (Bigand, 1990; Desain & Honing, 2003; Hannon & Johnson, 2005), similarity (Creel, 2012; Monahan & Carterette, 1985), change detection (Bergeson & Trehub, 2006; Hannon & Trehub, 2005; Jones, Johnston, & Puente, 2006; Repp, 2010; Smith & Cuddy, 1989), and expectancy (Huron, 2006; Large & Palmer, 2002; Repp, 1992). The role of meter in performance deserves its own volume (including Drake & Botte, 1993; Essens & Povel, 1985; Palmer, 1997; Povel & Essens, 1985; Repp, Iversen, & Patel, 2008; Shaffer, Clarke, & Todd, 1985; Sloboda, 1983), but is beyond the purview of the current study. Literature on how tonality and meter work together, however, remains surprisingly small. Experimental work manipulating the alignment of the tonal and metric hierarchies supports the notion that listeners are sensitive to both hierarchies when listening to melodies (Bigand, 1997; Palmer & Krumhansl, 1987a, 1987b; Serafine, Glassman, & Overbeeke, 1989). Interestingly, these authors reported that the tonal and metric hierarchies functioned independently, implying that these hierarchies need not necessarily co-occur. Other researchers report interactive relations between tonality and meter (Boltz, 1989b, 1991; Dawe, Platt, & Racine, 1993, 1994, 1995; Schmuckler, 1990; Schmuckler & Boltz, 1994). A substantial related literature has examined the independence versus interaction of pitch and time (see for reviews, Ellis & Jones, 2009; Krumhansl, 2000; Prince, Thompson, & Schmuckler, 2009), although these investigations have been confined to aspects of pitch and temporal information other than tonality and meter per se (but see Prince, Thompson, et al., 2009, for an exception). Deutsch (1980) found ELEC TR ONIC I SSN 1533-8312.  2014 BY T HE REG ENT S O F THE U NIV E RSI T Y OF CA LI FOR NIA A LL REQ UEST S F OR PER MISSION T O PHOT O COPY OR R EPRO DUC E A RTI CLE CONT ENT T HRO UGH T HE UNI VE R S IT Y OF CALI FO RNIA P R E SS ’ S R IGHTS AND PER MISSION S W EBSIT E , HT TP :// W W W. U CPRESSJ OUR NAL S . CO M / REPR IN TIN FO . A S P. DOI: 10.1525/ M P.2014.31.3.254 Tonal-Metric Hierarchy that using temporal grouping in musical sequences influences the encoding of tonal information (see also Laden, 1994). She found that recall accuracy (melodic dictation) declined when temporal segmentation cues conflicted with pitch sequencing cues, which decreased further when sequences did not conform to a simple hierarchical structure. However, more recent work in the perception of probe events (Prince, Thompson, et al., 2009) found mixed evidence, specifically an asymmetric interaction between pitch and time. When reporting if probe events occurred on or off the beat, tonally stable pitches were more likely to be reported as being on the beat, showing an interference of pitch structure (tonality) on temporal perception. However there was no complementary effect of temporal position when reporting if the probe was in or out of the key. The music theory literature also explores the relation between tonality and meter. Combining concepts from cognitive psychology and Schenkerian analysis, Lerdahl and Jackendoff (1983) created a model that uses wellformedness and preference rules to describe how listeners apprehend the structure in a musical sequence. Wellformedness rules provide a number of structural descriptions of the musical piece in question, whereas the preference rules function to determine which description most accurately corresponds to the listener’s experience. Their model includes rules for deriving metrical structure from the musical surface; this information then feeds into a time-span reduction that focuses on the pitch structure. Interestingly, in this model pitch and temporal structures are not treated entirely independently – for example, metric preference rule 7 specifically incorporates tonal function into deriving the metrical structure (‘‘Strongly prefer a metrical structure in which cadences are metrically stable,’’ p. 88). Similarly, time-span reduction preference rules incorporate rhythmic structure (rule 1: ‘‘prefer a choice that is in a relatively strong metrical position,’’ p. 160) and meter (rule 5: ‘‘prefer a choice that results in more stable choice of metrical structure,’’ p. 165). Benjamin (1984) also acknowledges the importance of tonal structure in his theory of musical meter, arguing that tonal structure influences the derivation of the meter at several levels of metrical analysis. Temperley (2001) discusses a number of possible factors in key identification, but does not consider metrical stability; the success of his model notwithstanding casts doubt on the idea that metrical strength and tonal stability are aligned in any consistent way. There are several algorithms that derive musical meter based on coding melodic (pitch) information, although these models rely on pitch repetition rather 255 than tonal stability. For instance, Collard, Vos, and Leeuwenberg (1981) determine the meter by compiling distance between pitch pattern repetitions. The approach of Steedman (1977) is similar, in that he takes repetitions in melodies to indicate points of metric accent (expanding on Longuet-Higgins, 1976). The period of these accents then forms the metric grouping and thus indicates the metrical framework. It is noteworthy that Steedman asserts that key identification is independent from knowledge of meter, and that scale has no influence on metric structure. The success of these techniques suggests that composers (at least Bach) use melodic repetition – but not necessarily tonality – and metric hierarchies in a congruent manner. Other algorithms extract metrical structure with purely temporal factors (particularly onsets and grouping), without including any parameters related to pitch (Desain, 1992; Desain & Honing, 2003; Povel & Essens, 1985). Similarly, the most common keyfinding algorithm does remarkably well despite ignoring all temporal information (Krumhansl, 1990; Krumhansl & Schmuckler, 1986; Schmuckler & Tomovski, 2005; but see Schmuckler, 2009); however, others contend that note order, recency, and possibly other temporal factors are critically important (Brown, 1988; Brown, Butler, & Jones, 1994; Huron & Parncutt, 1993; Temperley & Marvin, 2008). For example, Aarden (2003), using reaction times to the seven diatonic tones of the major scale, found that stability of these tones varied as a function of placement in a musical passage. Specifically, he observed that the diatonic major tonal hierarchy values as defined by Krumhansl and Kessler (1982) best characterized listeners’ responses at points of closure (i.e., cadences), but were less descriptive of tonal percepts at other points in a passage. Overall, the body of theoretical literature that specifically examines the relation between tonality and meter is miniscule. Similarly, the extant empirical literature on how tonality and meter combine focuses principally on perceptual processes, and not on determining whether tonal and metric structures actually co-occur in music. Thus, there appears to be virtually no information as to whether or not tonality and meter are systematically aligned. The only related research on this topic comes from Järvinen’s (1995) work examining 18 examples of bebop style jazz improvisation. He calculated a weighted average of the number of times each pitch class occurred on a given temporal position (down to the eighth note level). Järvinen found that tonally stable pitches occurred more often at metrically stable temporal positions, and tonally unstable pitches (i.e., nondiatonic) occurred more often at less stable temporal positions. 256 Jon B. Prince & Mark A. Schmuckler In this study we examine this issue with regard to the common practice period of Western music by asking the following questions. First and foremost, is there an alignment of tonal and metric information in typical tonal music, such that tonally stable notes occur on metrically stable temporal positions, and vice versa? Alternatively, tonally important pitches might be equally prevalent across temporal position (i.e., not selective for metric stability), or similarly, metrically stable positions may not favor any particular pitch. To investigate this question, the most obvious starting point is to examine the tonal and metric hierarchies in corpus data. With occasional deviations, perceptual studies of both hierarchies concur with frequency of usage and theoretical predictions (Krumhansl, 1990; Palmer & Krumhansl, 1990, but see Aarden, 2003); similar hierarchies also arise in music production studies (Palmer & Pfordresher, 2003; Schmuckler, 1989, 1990). Accordingly, it is of interest to determine whether the frequency of cooccurrence of each pitch class and temporal position (within a measure) conforms to perceptual studies of the tonal-metric hierarchy (e.g., Prince, Thompson, et al., 2009). The next most fundamental question is whether the relationship between tonality and meter varies across musical genre, musical mode, and/or time signature. For instance, this hypothetical relationship may vary across musical genre, particularly between styles that are largely diatonic (e.g., Baroque music) relative to styles containing increasing levels of chromaticism (e.g., Classical to Romantic and beyond). Similarly, musical mode is of interest, given the evidence that minor keys are somewhat more perceptually ambiguous than major keys (Delzell, Rohwer, & Ballard, 1999; Harris, 1985; Vuvan, Prince, & Schmuckler, 2011; Vuvan & Schmuckler, 2011). Lastly, research examining in detail the frequency of occurrence of tones across temporal positions within different time signatures is virtually nonexistent; therefore, it is unknown whether any existing alignment between tonal and metric information varies for different metric structures. Answering these questions is the main goal of the present research. Towards this end, we quantified the frequency of occurrence of each scale degree at each temporal position in 365 works from representative composers of tonal Western music (Bach, Mozart, Beethoven, and Chopin), totaling 721,293 notes. Within a given piece, we assumed one metric hierarchy and one tonal hierarchy (i.e., ignoring modulation) throughout; we return to the implications of this latter assumption in the General Discussion. Method MATERIALS Sequenced (i.e., not performed) MIDI files of a selection of works by Bach, Mozart, Beethoven, and Chopin were downloaded from http://kern.ccarh.org/ (CCARH, 2001) and http://www.piano-midi.de/ (maintained by Bernd Krueger; see Appendix A for full list of pieces). The corpus consists of works for piano, with the exception of Beethoven’s Symphony 5, 1st movement. All pieces were transposed to have a tonic of C major or A minor (as appropriate), and sorted into time signature categories of 2/4, 3/4, 4/4, 6/8, 9/8, and 12/8. The key indicated in the title of each piece was used to classify the mode and key signature. For pieces whose title provided no key information, the key signature of the score, in addition to a visual inspection of the first and last eight measures of the piece, were used to determine the key (tonic and mode). Again, one key was assigned for the entire piece, ignoring modulations. Note onset times were used to determine the position of each note within the metric hierarchy of the appropriate time signature, correcting for any pickup beats. Custom scripts written in MATLAB 7.0 were used to analyze the corpus. PROCEDURE The first step in analyzing the tonal-metric hierarchy was to examine the independent tonal and metric hierarchies of these pieces. Accordingly, we tabulated how often each scale degree occurred, separately for the major and minor mode pieces.1 Quantification of the metric hierarchy was accomplished by counting how often a note was sounded at each temporal position; this count was done separately for different time signatures. The tonal and metric counts were then compared to perceptual measures of tonal and metric stability, taken from Krumhansl and Kessler (1982) and Palmer and Krumhansl (1990), respectively. The purpose of these comparisons was to determine if the tonal and metric information of these pieces conformed to the perceptually derived profiles for these two dimensions. The second step of the analysis was to determine the frequency of occurrence of each pitch class (after transposition) at each temporal position, separately for all pieces in the corpus. These tonal-metric distributions were grouped within composer, time signature, and 1 In addition to measuring frequency of occurrence, the cumulative duration of each pitch class was also calculated. These two calculations gave identical results (r > .99). For continuity of calculation techniques across dimensions, frequency of occurrence was used. Tonal-Metric Hierarchy 257 TABLE 1. Correlations Between the Pitch Class Distributions of the Current Corpus and the Tonal Hierarchy Values of Krumhansl and Kessler (1982). Major modality K&K 1982 Bach Mozart Beethoven MEAN Minor modality Bach Mozart Beethoven Chopin Bach Mozart Beethoven Chopin .83 .91 .96 .91 .92 .99 .82 .87 .95 .86 .94 .99 .90 .96 .95 .95 .90 .98 .99 .95 .92 .95 .95 .84 .96 .99 .99 .95 TABLE 2. Mean Intercorrelation of the Metric Distributions of the Current Corpus (for Each Composer) and the Matching Metric Hierarchy Values of Palmer and Krumhansl (1990). 2/4 3/4 4/4 6/8 9/8 12/8 MEAN P&K 1990 Bach Mozart Beethoven Chopin .83 .95 .88 .82 .86 .94 .95 .92 .96 .88 .92 .89 .94 .97 .93 .88 .94 .97 .93 .96 .88 .93 .79 .90 .96 .90 .87 .93 .89 Note: All hierarchies are restricted to the sixteenth note level. mode, then compared with the perceptual profiles and across composers. Results DISTRIBUTION OF PITCH CLASSES AND TEMPORAL POSITIONS Composers’ use of the 12 pitch classes in the corpus correlated highly with the perceptually derived major and minor tonal hierarchies (see Table 1). However, correlations between composers were higher (mean r ¼ .94) than those with the tonal hierarchies (mean r ¼ .87). Similarly, the observed metric distributions within the measure correlated well with the PalmerKrumhansl metric hierarchies (mean r ¼ .87), but again were lower than the inter-composer correlations (mean r ¼ .92, see Table 2). Overall, these tables reveal that there was a remarkable agreement across composers in their use of the tonal and metric hierarchies, despite differences in mode, musical genre, and time signature. Figure 1 shows three different versions of both the major and minor tonal hierarchies – the frequency of occurrence observed in the corpus (limited to the downbeat metric position), the overall total frequency of occurrence (summing across all metric positions), and the Krumhansl-Kessler (1982) goodness of fit ratings. Figure 2 compares the metric hierarchies in the corpus to the Palmer and Krumhansl (1990) goodness of fit ratings (for time signatures 2/4, 3/4, 4/4, and 6/8). TONAL-METRIC DISTRIBUTION ANALYSIS For ease of depiction, the tonal-metric hierarchy counts present pitch classes and temporal positions by decreasing levels of frequency of occurrence instead of the conventional method of sorting these dimensions chromatically and chronologically, respectively (as in Figures 1 and 2). For example, Table 3 depicts the tonalmetric frequency of occurrence values for the corpus pieces in a major tonality (all transposed to C major) in the most common time signature of 4/4. The pitch classes (rows) are sorted as C, G, E, D, F, A, B, G#, D#, F#, A#, and C#, an ordering based on the observed pitch class distribution occurring on the downbeat (not across all metric positions) for all major-mode pieces in the corpus.2 2 It is important to note that this ordering of pitch classes does not correspond exactly with their perceived psychological stability as indicated by the classic tonal hierarchy values of Krumhansl and Kessler (1982). Probably the most significant point of divergence in this regard arises in the ordering of the diatonic scale tones, which ranked D, F, A, and B in frequency of occurrence in our corpus, as opposed to F, A, D, and B in the Krumhansl and Kessler tonal hierarchy values. Of lesser note, the non-diatonic tones in our corpus appear in the order G#, D#, F#, A#, and C#, as opposed to F#, G#, D#, A#, and C#, in Krumhansl and Kessler. Although the convergence with the Krumhansl and Kessler major profile is strong (r ¼ .96), our data also mirror Aarden’s (2003) rank ordering of the seven diatonic major scale tones (based on reaction times) more strongly at points of closure (cf. p. 52; r ¼ .71) than for continuations (cf. p. 75; r ¼ .29). Separately, for the minor pieces the pitch class distribution on the downbeat correlates with the Krumhansl and Kessler minor tonal hierarchy at r ¼ .91. 258 Jon B. Prince & Mark A. Schmuckler FIGURE 1. Tonal hierarchy values: frequency of occurrence (on downbeat, and all temporal positions) and Krumhansl-Kessler (1982) goodness of fit ratings, all standardized to maximum of 1. (a) Major mode hierarchy, (b) Minor mode. The columns of Table 3 represent temporal positions within the measure, sorted as beats 1 (downbeat), 3, 2þ4, eighth notes, sixteenth notes, and Other, grouped in accordance with theoretical predictions of beat strength (e.g., Lerdahl & Jackendoff, 1983). That is, the frequency of occurrence of temporal positions is averaged (not summed) within a category of equal metric strength, such that beats 2 and 4 are averaged, becoming 2þ4, then the remaining 4 eighth note (off-beat) positions, followed by the next lower metric hierarchy level (off-beat sixteenth notes), then ‘‘Other’’ (designating thirty-second notes, sixteenth note triplets, and any non-quantized metric positions). Finally, Table 4 provides the same data for pieces in the minor mode (4/4 time signature, all composers), also with the rows sorted according to the pitch class distribution of the minor mode on the downbeat. The full data set of all time signatures, modalities, and composers is available upon request from the first author. Figures 3-8 depict the tonal-metric hierarchies of each time signature as 3-dimensional histograms, separately for major and minor (collapsing across composer). The y axis shows the frequency of occurrence of each combination of pitch class (x axis) and temporal position (z axis). These graphs use the same arrangement as Tables 3 and 4; namely, they sort the axes by frequency of occurrence on the downbeat, as well as averaging across temporal positions with equal metric stability. These graphs demonstrate that the most common occurrence is a tonally stable pitch (e.g., the tonic) at a metrically stable temporal position (e.g., the downbeat), and that the frequency of occurrence decreases at lower levels of tonal and metric stability. At first glance, the overall tonal-metric hierarchy looks like a replication of the tonal hierarchy at different temporal positions. Closer inspection, however, reveals that the favoring of tonally stable pitch classes is more discernible at the downbeat than at lower levels of metric stability. Indeed, the correlation of the KrumhanslKessler major tonal hierarchy with the pitch class distribution decreases across metric category stability: r ¼ .96 for the downbeat, then r ¼ .92, .90, .88, and .87 down to the ‘‘Other’’ level. For the same metric categories in the minor mode, the pattern is r ¼ .91 for the downbeat, then r ¼ .83, .84, .85, and .79. This change is also evident in Figure 1, as it compares the tonal hierarchy for only downbeat occurrences to the overall total (summing all metric positions). These figures also demonstrate the remarkable consistency of the tonal-metric hierarchy across time signature and mode. Despite these changes in the metrical framework and major/minor modality, the profile of the tonal-metric hierarchy remains largely the same. This shape is slightly less reliable for Figures 7 and 8, likely due to the small sample size (note the markedly lower scale of the y axis compared to other figures). Because Figures 3-8 show that notes rarely occurred at fine subdivisions, and in the interest of simplicity, further analyses of the tonal-metric hierarchy used the 16 most common temporal positions (e.g., sixteenth note Tonal-Metric Hierarchy 259 FIGURE 2. Metric hierarchy values: frequency of occurrence and Palmer-Krumhansl (1990) goodness of fit ratings, all standardized to maximum of 1. (a) 2/4 time signature, (b) 3/4 time signature, (c) 4/4 time signature, (d) 6/8 time signature. level for 4/4).3 Note that the exclusion of the least common temporal positions represents a more conservative analysis – the uniformly rare use of the positions of lower metric stability might otherwise artificially inflate correlations between the tonal and metric hierarchies. Table 5 shows the results of correlating the resulting 12 (pitch class) by 16 (metric stability) tonal-metric matrices across composer, separately for each time 3 We used the 16 most common temporal positions from all time signatures even if it did not correspond to the sixteenth note level (e.g., 3/4) because we were sorting by stability as indexed by frequency of occurrence rather than beat strength. signature and modality. Values are the average correlation coefficients of each composer with all other composers. All Table 5 values below .70 are the result of cells with the minimum possible small sample size (N pieces ¼ 2). The strikingly high inter-composer tonal-metric correlations of Table 5 motivated including additional composers, including some from more modern compositional periods; we chose Schubert, Brahms, Liszt, and Scriabin (see Appendix B for the list of included pieces of these composers). These data are not included in the earlier analyses because our corpus had too few pieces of these composers for valid inter-composer comparisons. 260 Jon B. Prince & Mark A. Schmuckler TABLE 3. Tonal-metric Hierarchy of Major Tonality 4/4 Pieces, Summed Across Composer. Major mode, Time signature 4/4, all composers Pitch C G E D F A B G# D# F# A# C# Downbeat Beat 3 Beats 2þ4 8th level 16th level Other 3356 3365 2149 1859 1588 1586 1598 517 401 531 307 320 2562 2681 1877 1978 1475 1474 1468 500 359 601 376 374 2154 2161 1650 1723 1358 1248 1302 451 334 614 319 339 1322 1528 1140 1123 813 895 819 254 176 384 215 214 530 612 423 481 308 344 356 77 75 149 80 70 36 41 30 31 22 26 26 13 12 13 11 10 Note: The four rightmost columns represent the average frequency of occurrence at the relevant temporal positions. Pitches are sorted by frequency of occurrence at the downbeat across all major mode pieces in the corpus, transposed to C major. TABLE 4. Tonal-metric Hierarchy of Minor Tonality 4/4 Pieces, Summed Across Composer. Minor mode, Time signature 4/4, all composers Pitch A E C D B G F G# C# F# D# A# Downbeat Beat 3 Beats 2þ4 8th level 16th level Other 2134 2280 1462 1079 1108 1000 942 521 255 290 367 230 1591 1804 1239 1106 1169 775 922 633 261 314 354 256 1489 1721 1280 1084 1097 870 922 524 259 325 340 269 1020 1160 738 716 721 567 529 313 184 220 172 150 333 422 257 279 291 164 194 105 60 84 63 66 22 29 20 17 19 14 15 11 5 8 7 4 Note: The four rightmost columns represent the average frequency of occurrence at the relevant temporal positions. Pitches are sorted by frequency of occurrence at the downbeat across all minor mode pieces in the corpus, transposed to A minor. Smaller sample size can lead to unstable relationships in the data (Knopoff & Hutchinson, 1983), as observed for isolated cases in Table 5. However, collapsing across time signature before testing inter-composer correlations allows a more reliable comparison, and is justified given the similarity of the tonal-metric hierarchy across time signature (cf. Figures 3-8). Table 6 shows the average inter-composer correlations of these eight composers, after collapsing across time signatures before correlating. Major and minor modes remain separate. Although intriguing, none of the aforementioned analyses establish definitively whether the tonal and metric hierarchies are truly correlated. Combining any pitch class distribution with a set of temporal positions would result in the commonest pitch class occurring most frequently at the commonest temporal location (i.e., approximately the shape of Figures 3-8), regardless of if there is any true connection between the two. One way to establish quantitative proof of a link between the two hierarchies is to compare the average metric stability of each pitch class with the tonal hierarchy, and compare the average tonal stability of each temporal position with the metric hierarchy. The first procedure tests if tonally stable pitches are more likely to have high metric stability (and tonally unstable pitches have lower metric stability). The second approach tests if metrically stable positions have high tonal stability (and metrically unstable positions are more likely to have lower tonal stability). Thus, although these two procedures are Tonal-Metric Hierarchy 261 FIGURE 3. Tonal-metric hierarchy for 2/4 time signature, for major (a) and minor (b) mode. FIGURE 4. Tonal-metric hierarchy for 3/4 time signature, for major (a) and minor (b) mode. conceptually related, they test separate hypotheses. More importantly, by using the average tonal or metric stability, these calculations are not biased by the simple joint probability of the two independent distributions. The average metric stability of each pitch class has a strong positive correlation with the tonal hierarchy (r ¼ .77 when collapsing across mode; r ¼ .68 for major pieces, and .58 for minor). There is a similar relationship between the average tonal stability of each temporal position and the metric hierarchy (r ¼ .68; .70 for major pieces, and .55 for minor). Thus, the tonal and metric hierarchies are indeed connected beyond the extent predicted by their joint probability, yet slightly less so for minor than major. Additional inter-composer analyses repeated these calculations (comparing the metric stability of each pitch class and the tonal stability of each metric position) for each composer separately, collapsing further across mode (see Table 7). Even though the correlations were always positive, they varied across composer. In 262 Jon B. Prince & Mark A. Schmuckler FIGURE 5. Tonal-metric hierarchy for 4/4 time signature, for major (a) and minor (b) mode. FIGURE 6. Tonal-metric hierarchy for 6/8 time signature, for major (a) and minor (b) mode. Bach’s compositions, all pitch classes had similar levels of metric stability, but metrically stable temporal positions had higher tonal stability. One way to conceptualize this result is to say that in these pieces, tonally stable pitch classes could occur at any time in the measure, but on the downbeat, only pitches with high tonal stability occurred. The Mozart pieces showed the opposite pattern – metrically stable locations could have pitches of any tonal stability, while tonally stable pitches were more limited to stable metric locations. Chopin moved further in this direction, whereas Beethoven respected both relationships more equally. Discussion The current study analyzed how central composers from a variety of Western tonal music periods used pitch classes and temporal positions, and in particular, if this information was used congruently (i.e., were the tonal and metric hierarchies positively correlated). In Tonal-Metric Hierarchy 263 FIGURE 7. Tonal-metric hierarchy for 9/8 time signature, for major (a) and minor (b) mode. FIGURE 8. Tonal-metric hierarchy for 12/8 time signature, for major (a) and minor (b) mode. general, both distributions were strongly – but not perfectly – related to the perceptually derived tonal and metric hierarchies of Krumhansl and Kessler (1982) and Palmer and Krumhansl (1990). Although this finding is hardly surprising on one level, it is notable in that it emerged consistently in a corpus of music that covers a wide range of musical styles, including styles that contain high levels of chromaticism and rhythmic deviation. As such, the demonstration that the distribution of tonal and metric information in such works nevertheless conforms to a prototypical hierarchy of stability is a striking finding. The primary goal of this study was to test if tonal and metric information aligned in Western tonal music. Indeed, they correlated strongly throughout this corpus, such that tonally stable pitch classes were more likely to occur on metrically stable temporal positions. Moreover, and mirroring the findings with the independent tonal and metric distributions, this relation was relatively stable across time signature and modality. 264 Jon B. Prince & Mark A. Schmuckler TABLE 5. Average Inter-composer Correlation. Major modality 2/4 3/4 4/4 6/8 9/8 12/8 MEAN Minor modality Bach Mozart Beethoven Chopin .86 .82 .86 .69 .82 .91 .91 .92 .87 .92 .91 .90 .84 .82 .85 .83 .83 .81 .90 .88 .84 Bach Mozart Beethoven Chopin .72 .79 .88 .74 .78 .72 .77 .80 .87 .90 .65 .79 .88 .89 .75 .78 .72 .80 .63 .85 .89 .74 .80 .78 Note: Each value represents the average correlation of that composer with all other composers for that time signature and modality. Missing values mean no data were available. TABLE 6. Inter-composer Correlations Including More Modern Composers, Collapsed Across Time Signature and Mode. Major Minor Bach Mozart Beethoven Schubert Chopin Brahms Liszt Scriabin .84 .86 .92 .93 .93 .93 .92 .89 .88 .87 .91 .90 .80 .86 .86 .85 TABLE 7. Variations Across Composer in the Correlations of Pitch Class Average Metric Stability with the Tonal Hierarchy, and Temporal Position Average Tonal Stability with the Metric Hierarchy. Calculation Bach Mozart Beethoven Chopin r(pitch class metric stability, tonal hierarchy) r(temporal position tonal stability, metric hierarchy) .35 .67 .66 .37 .71 .67 .78 .44 The reality of a correlated tonal-metric hierarchy may not be much of a revelation to some, yet it is far from a trivial finding. By themselves, tonally strong pitches and metrically stable temporal positions function as important cognitive reference points for the listener when constructing a mental representation of the music they are hearing (Krumhansl, 1990; Palmer & Krumhansl, 1990). But regardless of its intuitive nature, demonstrating this alignment between hierarchies is significant, particularly given the frequent assumptions of its existence. For example, Serafine et al. (1989) explicitly state that ‘‘all theorists would agree’’ (p. 404) with the notion that tonally stable pitches occur on metrically strong positions. Similarly, in the related (yet arguably separate) context of harmony in melodic anchoring, Bharucha (1984) simply asserts that ‘‘chord tones tend to occur in metrically stressed positions more often than nonchord tones’’ (p. 492). Both of these authors make these assumptions explicitly, yet based on no actual evidence for this relation. More cautious approaches cite an alignment between tonal and metric hierarchies as a ‘‘generally held belief ’’ (Palmer & Krumhansl, 1987a). The unique profile of the tonal-metric hierarchy derives in part from changes in the use of pitch classes across temporal position. Indeed, pitch class distributions correlated more highly across composers than they did with the perceptually derived tonal hierarchy (particularly for the minor mode, likely because of our treating all three minor modes as one). The fact that the tonal hierarchy did best at metrically stable locations concurs with Aarden’s (2003) findings – that the diatonic members of the original major tonal hierarchy are most representative of pitch expectations at points of closure. Because the most expected closure is a cadence on a metrically stable location, it follows that the shape of the tonal hierarchy would change slightly based on metric position. This interpretation accords with Huron’s (2006) proposal of four pitch schemas in Western music (major/minor modes both for cadences as well as continuations). Additionally, the fact that composers deliberately violate schematic expectations (Meyer, 1957) means that their use of pitch classes and temporal positions should not be a perfect match to the schematic hierarchies. The current findings also converge with the only (to our knowledge) other existing corpus analysis of the relation between the tonal and metric information: Järvinen (1995). The extension and generalization provided Tonal-Metric Hierarchy by the current findings is critical given that the metric hierarchies in jazz may function differently than in other musical genres. Indeed, Palmer and Pfordresher’s (2003) distribution analysis of 1930’s swing jazz (well before the bebop period analyzed by Järvinen) showed that in binary meter, events occurring on beat 4 were most common, followed by beats 2 and 1. One of the more intriguing aspects of Järvinen’s study is that his weighting technique found that non-diatonic (i.e., tonally unstable) pitches were actually more common off the beat than on the beat, thus observing a crossover effect. His weighting technique involved multiplying the frequency of occurrence of notes at the quarter-note level by 2, those at the half-note level by 4, and the whole-note level by 8. This procedure corrected the imbalance in how often a given metric position could occur within one bar (i.e., eighthnotes could occur at 8 different points, whole notes only at one). Using Järvinen’s weighting technique with the current data did not yield a similar crossover, possibly indicating how a major stylistic difference (classical music to bebop jazz) may actually change the tonalmetric hierarchy. Nevertheless, the advantage of tonally stable pitches decreased at lower levels of metric stability, thus the pattern in the current findings is qualitatively similar to Järvinen’s, only not to the point of observing a full crossover. Comparisons of the tonal-metric hierarchy across composer yielded particularly intriguing results. The overall shape of the tonal-metric hierarchy is surprisingly consistent across composer, even when including more modern composers still within the tonal tradition (Brahms, Liszt, Schubert, Scriabin). Despite all intercomposer correlations remaining above .8, the more modern composers tended to show relatively lower coefficients, but the fact that Bach is comparable to Scriabin in this analysis undermines this interpretation. Despite this commonality, there were variations in how the composers used the tonal-metric hierarchy. Specifically, correlations of the average metric stability of pitch classes with the tonal hierarchy, as well as the average tonal stability of temporal positions with the metric hierarchy, did not remain entirely constant across composer. Table 7 suggests that Bach placed tonally stable pitches in multiple places throughout the measure, while reserving metrically stable positions for tonally stable pitches. Conversely, Chopin was more likely to limit tonally stable pitches to metrically stable points, but was less restrictive on what pitch classes occurred at these times (i.e., tonally unstable pitches could also occur at strong metric locations). One possible mechanism for this shift in the use of the tonal-metric hierarchy could involve changes in composers’ use of appoggiatura and 265 grace notes, but such interpretations are speculative at this point. The fact that composers correlate the tonal and metric hierarchies in their compositions raises the question of why they would do so in the first place. One possibility is that a correlated tonal-metric hierarchy offers a way to maximize the strength of musical expectancies, which are of central importance to communicating musical emotion (Huron, 2006; Meyer, 1957). By themselves, the tonal hierarchy influences expectations of what (pitch) event might occur, and the metric hierarchy concerns when. Used independently (uncorrelated), they both offer a means of creating and fulfilling/denying expectancies. But when used in a correlated manner, the strength of the resulting expectancies go beyond the simple joint probability of what and when towards exactly what happens at exactly when. In other words, the tonalmetric hierarchy may provide a mechanism of strengthening the communicated musical emotion, analogous to making the whole (expectancy) greater than the sum of the parts (tonal and metric hierarchies). Demonstrating the existence of a tonal-metric hierarchy is relevant to research on how pitch and time combine in music perception. This question has received considerable treatment over the years, yet there has been little consensus (for reviews, see Ellis & Jones, 2009; Jones & Boltz, 1989; Krumhansl, 2000; Prince, 2011). Perceptual work examining the relation between tonality and meter (as opposed to pitch and time) found an asymmetric relationship: tonality affected metric judgments, but meter had no influence on key membership judgments (Prince, Thompson, et al., 2009). In fact, an effect of rhythmic regularity on pitch height comparisons only occurred in the context of atonal sequences (Prince, Schmuckler, & Thompson, 2009). Moreover, Prince, Thompson, et al. (2009) found that goodness of fit ratings for probe events consisting of any one of 12 pitch classes at any one of eight temporal positions (eighth note level) were predictable based on a linear combination of both the tonal and metric hierarchies. These hierarchies explained 89% of the variance in one experiment, and 86% in another, but a multiplicative interaction term never accounted for any additional variance. In contrast, the shape of the tonal hierarchy varied across levels of metric stability in the current study. The most obvious explanation of the divergence between these findings and earlier research is that it involves comparing a corpus analyses to perceptually derived results (goodness of fit ratings to probe events) from perceptual experiments. Accordingly, this result might represent a divergence between perceptual and theoretical levels of analysis. However, humans demonstrate exquisite 266 Jon B. Prince & Mark A. Schmuckler sensitivity to the statistical regularities of the environment across an array of domains and modalities (Goldstone, 1998), with music being no exception (for examples specifically in pitch and time, see Loui, Wessel, & Kam, 2010; Tillmann, Stevens, & Keller, 2011). Consequently, this discrepancy between theoretical and perceptual findings on the tonal-metric hierarchy is unexpected. Future research should explore this issue, as it provides a possible exception to this principle of perceptual learning of statistical information. Another interpretation of this discrepancy between perceptual results and our corpus analysis stems from the largest limitation of this study; namely, assigning one tonal centre to an entire piece ignores the fact that pieces often modulate to new keys. Previous corpus analyses take the same approach (Knopoff & Hutchinson, 1983; Krumhansl, 1990; Youngblood, 1958), with the notable exception of Temperley (2007). Despite this limitation, it is worth noting that composers typically modulate to tonally proximate keys, and then return to the original key. Given the similarity in the use of pitch classes between related keys, short-term modulations should not influence the overall shape of the tonal hierarchy when compiled across the entire piece. In fact, effects of modulation would primarily serve to flatten the distribution of the tonal hierarchy; if this flattening has occurred in our data it would only reduce the changes of the tonal hierarchy across metric position. Overall, the consistent shape of the tonal-metric hierarchy demonstrates that on the aggregate, modulation is a local effect, and does not radically influence the distribution of frequency of occurrences. Nevertheless, determining how modulation may affect this corpus is the most urgent next step in this research. Methodological approaches could be to examine only pieces that do not modulate, or extract excerpts that stay in the original key, or interpret modulated sections in terms of their new key (such that during a modulation from C to G, all G pitches would count as instances of the tonic). We are currently exploring these approaches; in tandem we are planning behavioral work to test if see if listeners’ perceptions of modulation are similarly local. For shorter chorale excerpts, at least, listeners are indeed sensitive to the extent of modulation (Thompson & Cuddy, 1989, 1992); thus, once established, the tonal hierarchy may function as a robust schema that persists through even longer time spans. Overall, the analyses reported in this paper substantiate the correlation of tonal and metric information in typical tonal music, thus providing evidence for the existence of a tonal-metric hierarchy, similar to the previously identified and well-studied tonal and metric hierarchies. The observed tonal-metric hierarchy remained remarkably consistent across compositional period, time signature, and mode. Thus it is indeed true that these composers used tonally stable pitches at points of high metric stability. Importantly, the tonalmetric hierarchy is not a simple linear combination of the separate hierarchies of tonality and meter. Instead, it represents an interactive relationship between the two hierarchies: the tonal hierarchy is most pronounced on metrically stable temporal positions, and less distinct at positions of lower metric stability. The current findings therefore offer several useful contributions to the music cognition literature. We have validated and elaborated on a critical yet tacit assumption regarding the use of the tonal and metric hierarchies, extended theoretical research on hierarchical organization in musical pitch and time, and revealed an interesting discrepancy between perceptual and theoretical research on tonality and meter. Hopefully these findings will stimulate future research on the complex issue of pitch, time, and structural hierarchies in music. Author Notes Preparation of this manuscript was supported by the Walter Murdoch Distinguished Collaborator Scheme and a grant from the Natural Sciences and Engineering Research Council of Canada to the second author. The authors wish to thank Carol Krumhansl for her encouragement of this line of inquiry, and Davy Temperley for his insightful comments. Correspondence concerning this article should be addressed to Jon B. Prince, School of Psychology, Murdoch University, 90 South Street, Murdoch, WA 6150. E-mail: j.prince@murdoch.edu.au References A ARDEN , B. J. (2003). Dynamic melodic expectancy. (Unpublished doctoral dissertation). Ohio State University. Retrieved from http://rave.ohiolink.edu/etdc/view?acc_ num¼osu1060969388 B EAL , A. L. (1985). The skill of recognizing musical structures. Memory and Cognition, 13, 405-412. B ENJAMIN , W. E. (1984). A theory of musical meter. Music Perception, 1, 355-413. Tonal-Metric Hierarchy B ERGESON , T. R., & T REHUB , S. E. (2006). Infants’ perception of rhythmic patterns. Music Perception, 23, 345-360. B HARUCHA , J. J. (1984). Anchoring effects in music - The resolution of dissonance. Cognitive Psychology, 16, 485-518. B IGAND, E. (1990). Abstraction of two forms of underlying structure in a tonal melody. Psychology of Music, 18, 45-59. B IGAND, E. (1997). Perceiving musical stability: The effect of tonal structure, rhythm, and musical expertise. Journal of Experimental Psychology: Human Perception and Performance, 23, 808-822. B IGAND, E., & P INEAU, M. (1996). Context effects on melody recognition: A dynamic interpretation. Cahiers de Psychologie Cognitive/Current Psychology of Cognition, 15, 121-134. B OLTZ , M. G. (1989a). Perceiving the end: Effects of tonal relationships on melodic completion. Journal of Experimental Psychology: Human Perception and Performance, 15, 749-761. B OLTZ , M. G. (1989b). Rhythm and good endings - Effects of temporal structure on tonality judgments. Perception and Psychophysics, 46, 9-17. B OLTZ , M. G. (1991). Some structural determinants of melody recall. Memory and Cognition, 19, 239-251. B ROWN , H. (1988). The interplay of set content and temporal context in a functional theory of tonality perception. Music Perception, 5, 219-249. B ROWN , H., B UTLER , D., & J ONES , M. R. (1994). Musical and temporal influences on key discovery. Music Perception, 11, 371-407. CCARH. (2001). T HE M USE DATA DATABASE (http://www. Musedata.Org). Stanford, CA: Center for Computer Assisted Research in the Humanities. C OHEN , A. J., T REHUB , S. E., & T HORPE , L. A. (1989). Effects of uncertainty on melodic information-processing. Perception and Psychophysics, 46, 18-28. doi: 10.3758/bf03208070 C OLLARD, R., VOS , P., & L EEUWENBERG , E. (1981). What melody tells about meter in music. Zeitschrift Fur Psychologie, 189, 25-33. C REEL , S. C. (2012). Similarity-based restoration of metrical information: Different listening experiences result in different perceptual inferences. Cognitive Psychology, 65, 321-351. C UDDY, L. L. (1991). Melodic patterns and tonal structure: Converging evidence. Psychomusicology, 10, 107-126. DAWE , L. A., P LATT, J. R., & R ACINE , R. J. (1993). Harmonic accents in inference of metrical structure and perception of rhythm patterns. Perception and Psychophysics, 54, 794-807. DAWE , L. A., P LATT, J. R., & R ACINE , R. J. (1994). Inference of metrical structure from perception of iterative pulses within time spans defined by chord changes. Music Perception, 12, 57-76. DAWE , L. A., P LATT, J. R., & R ACINE , R. J. (1995). Rhythm perception and differences in accent weights for musicians and nonmusicians. Perception and Psychophysics, 57, 905-914. 267 D ELZELL , J. K., R OHWER , D. A., & B ALLARD, D. E. (1999). Effects of melodic pattern difficulty and performance experience on ability to play by ear. Journal of Research in Music Education, 47, 53-63. doi: 10.2307/3345828 D ESAIN , P. (1992). A (de)composable theory of rhythm perception. Music Perception, 9, 439-454. D ESAIN , P., & H ONING , H. (2003). The formation of rhythmic categories and metric priming. Perception, 32, 341-365. doi: 10.1068/p3370 D EUTSCH , D. (1980). The processing of structured and unstructured tonal sequences. Perception and Psychophysics, 28, 381-389. D OWLING , W. J. (1978). Scale and contour: Two components of a theory of memory for melodies. Psychological Review, 85, 341-354. D OWLING , W. J. (1991). Tonal strength and melody recognition after long and short delays. Perception and Psychophysics, 50, 305-313. D RAKE , C., & B OTTE , M.-C. (1993). Tempo sensitivity in auditory sequences: Evidence for a multiple-look model. Attention, Perception, and Psychophysics, 54, 277-286. doi: 10.3758/ bf03205262 E LLIS , R. J., & J ONES , M. R. (2009). The role of accent salience and joint accent structure in meter perception. Journal of Experimental Psychology: Human Perception and Performance, 35, 264-280. E SSENS , P. J., & P OVEL , D. J. (1985). Metrical and nonmetrical representations of temporal patterns. Perception and Psychophysics, 37, 1-7. G OLDSTONE , R. L. (1998). Perceptual learning. Annual Review of Psychology, 49, 585-612. doi: 10.1146/annurev.psych.49.1.585 HANNON , E. E., & J OHNSON , S. P. (2005). Infants use meter to categorize rhythms and melodies: Implications for musical structure learning. Cognitive Psychology, 50, 354-377. HANNON , E. E., & T REHUB , S. E. (2005). Metrical categories in infancy and adulthood. Psychological Science, 16, 48-55. HARRIS , R. W. (1985). Perceived relatedness of musical tones in major and minor tonal contexts. American Journal of Psychology, 98, 605-623. doi: 10.2307/1422513 H URON , D. (2006). Sweet anticipation: Music and the psychology of expectation. Cambridge, MA: MIT Press. H URON , D., & PARNCUTT, R. (1993). An improved model of tonality perception incorporating pitch salience and echoic memory. Psychomusicology: Music, Mind and Brain, 12, 154171. doi: 10.1037/h0094110 JÄRVINEN , T. (1995). Tonal hierarchies in jazz improvisation. Music Perception, 12, 415-437. J ONES , M. R., & B OLTZ , M. G. (1989). Dynamic attending and responses to time. Psychological Review, 96, 459-491. J ONES , M. R., J OHNSTON , H. M., & P UENTE , J. (2006). Effects of auditory pattern structure on anticipatory and reactive attending. Cognitive Psychology, 53, 59-96. 268 Jon B. Prince & Mark A. Schmuckler J ONES , M. R., & R ALSTON , J. T. (1991). Some influences of accent structure on melody recognition. Memory and Cognition, 19, 8-20. K ELLER , P. E., & B URNHAM , D. K. (2005). Musical meter in attention to multipart rhythm. Music Perception, 22, 629-661. doi: 10.1525/mp.2005.22.4.629 K NOPOFF, L., & H UTCHINSON , W. (1983). Entropy as a measure of style: The influence of sample length. Journal of Music Theory, 27, 75-97. K RUMHANSL , C. L. (1990). Cognitive foundations of musical pitch. New York: Oxford University Press. K RUMHANSL , C. L. (1991). Memory for musical surface. Memory and Cognition, 19, 401-411. K RUMHANSL , C. L. (2000). Rhythm and pitch in music cognition. Psychological Bulletin, 126, 159-179. K RUMHANSL , C. L., & C UDDY, L. L. (2010). A theory of tonal hierarchies in music. In M. R. Jones, R. R. Fay, & A. N. Popper (Eds.), Music perception. Springer handbook of auditory research. New York: Springer Science. K RUMHANSL , C. L., & K ESSLER , E. J. (1982). Tracing the dynamic changes in perceived tonal organization in a spatial representation of musical keys. Psychological Review, 89, 334-368. K RUMHANSL , C. L., & S CHMUCKLER , M. A. (1986, July). Keyfinding in music: An algorithm based on pattern matching to tonal hierarchies. Paper presented at the 19th Annual Meeting of the Society of Mathematical Psychology, Cambridge, MA. L ADEN , B. (1994). Melodic anchoring and tone duration. Music Perception, 12, 199-212. L ARGE , E. W., & J ONES , M. R. (1999). The dynamics of attending: How people track time-varying events. Psychological Review, 106, 119-159. L ARGE , E. W., & PALMER , C. (2002). Perceiving temporal regularity in music. Cognitive Science, 26, 1-37. L EBRUN -G UILLAUD, G., & T ILLMANN , B. (2007). Influence of a tone’s tonal function on temporal change detection. Perception and Psychophysics, 69, 1450-1459. L ERDAHL , F., & J ACKENDOFF, R. (1983). A generative theory of tonal music. Cambridge, MA: MIT Press. L ONDON , J. (2004). Hearing in time: Psychological aspects of musical meter. Oxford, UK: Oxford University Press. L ONGUET-H IGGINS , H. C. (1976). Perception of melodies. Nature, 263, 646-653. L OUI , P., W ESSEL , D., & K AM , C. (2010). Humans rapidly learn grammatical structure in a new musical scale. Music Perception, 27, 377-388. M EYER , L. B. (1957). Emotion and meaning in music. Chicago, IL: University of Chicago Press. M ONAHAN , C. B., & C ARTERETTE , E. C. (1985). Pitch and duration as determinants of musical space. Music Perception, 3, 1-32. PALMER , C. (1997). Music performance. Annual Review of Psychology, 48, 115-138. PALMER , C., & K RUMHANSL , C. L. (1987a). Independent temporal and pitch structures in determination of musical phrases. Journal of Experimental Psychology: Human Perception and Performance, 13, 116-126. PALMER , C., & K RUMHANSL , C. L. (1987b). Pitch and temporal contributions to musical phrase perception - Effects of harmony, performance timing, and familiarity. Perception and Psychophysics, 41, 505-518. PALMER , C., & K RUMHANSL , C. L. (1990). Mental representations for musical meter. Journal of Experimental Psychology: Human Perception and Performance, 16, 728-741. PALMER , C., & P FORDRESHER , P. Q. (2003). Incremental planning in sequence production. Psychological Review, 110, 683-712. doi: 10.1037/0033-295x.110.4.683 P OVEL , D. J., & E SSENS , P. (1985). Perception of temporal patterns. Music Perception, 2, 411-440. P RINCE , J. B. (2011). The integration of stimulus dimensions in the perception of music. Quarterly Journal of Experimental Psychology, 64, 2125-2152. doi: 10.1080/ 17470218.2011.573080 P RINCE , J. B., S CHMUCKLER , M. A., & T HOMPSON , W. F. (2009). The effect of task and pitch structure on pitch-time interactions in music. Memory and Cognition, 37, 368-381. doi: 10.3758/MC.37.3.368 P RINCE , J. B., T HOMPSON , W. F., & S CHMUCKLER , M. A. (2009). Pitch and time, tonality and meter: How do musical dimensions combine? Journal of Experimental Psychology: Human Perception and Performance, 35, 1598-1617. doi: 10.1037/ a0016456 R EPP, B. H. (1992). Probing the cognitive representation of musical time - Structural constraints on the perception of timing perturbations. Cognition, 44, 241-281. R EPP, B. H. (2010). Do metrical accents create illusory phenomenal accents? Attention, Perception, and Psychophysics, 72, 1390-1403. doi: 10.3758/app.72.5.1390 R EPP, B. H., I VERSEN , J. R., & PATEL , A. D. (2008). Tracking an imposed beat within a metrical grid. Music Perception, 26, 1-18. doi: 10.1525/mp.2008.26.1.1 S CHMUCKLER , M. A. (1989). Expectation in music - Investigation of melodic and harmonic processes. Music Perception, 7, 109-150. S CHMUCKLER , M. A. (1990). The performance of global expectations. Psychomusicology, 9, 122. S CHMUCKLER , M. A. (1997). Expectancy effects in memory for melodies. Canadian Journal of Experimental Psychology, 51, 292. S CHMUCKLER , M. A. (2009). Components of melodic processing. In S. Hallam, I. Cross, & M. Thaut (Eds.), Oxford handbook of music psychology (pp. 93-106). Oxford, UK: Oxford University Press. S CHMUCKLER , M. A., & B OLTZ , M. G. (1994). Harmonic and rhythmic influences on musical expectancy. Perception and Psychophysics, 56, 313-325. Tonal-Metric Hierarchy S CHMUCKLER , M. A., & T OMOVSKI , R. (2005). Perceptual tests of an algorithm for musical key-finding. Journal of Experimental Psychology: Human Perception and Performance, 31, 1124-1149. S ERAFINE , M. L., G LASSMAN , N., & O VERBEEKE , C. (1989). The cognitive reality of hierarchic structure in music. Music Perception, 6, 397-430. S HAFFER , L. H., C LARKE , E. F., & T ODD, N. P. (1985). Meter and rhythm in piano playing. Cognition, 20, 61-77. S LOBODA , J. A. (1983). The communication of musical meter in piano performance. Quarterly Journal of Experimental Psychology, 35, 377-396. S MITH , K. C., & C UDDY, L. L. (1989). Effects of metric and harmonic rhythm on the detection of pitch alterations in melodic sequences. Journal of Experimental Psychology: Human Perception and Performance, 15, 457-471. S TEEDMAN , M. J. (1977). Perception of musical rhythm and meter. Perception, 6, 555-569. T EMPERLEY, D. (2001). The cognition of basic musical structures. Cambridge, MA: MIT Press. T EMPERLEY, D. (2007). Music and probability. Cambridge, MA: MIT Press. T EMPERLEY, D., & M ARVIN , E. W. (2008). Pitch-class distribution and the identification of key. Music Perception, 25, 193-212. T HOMPSON , W. F., & C UDDY, L. L. (1989). Sensitivity to key change in chorale sequences: A comparison of single voices and four-voice harmony. Music Perception, 7, 151-168. 269 T HOMPSON , W. F., & C UDDY, L. L. (1992). Perceived key movement in four-voice harmony and single voices. Music Perception, 9, 427-438. T ILLMANN , B., JANATA , P., B IRK , J. L., & B HARUCHA , J. J. (2008). Tonal centers and expectancy: Facilitation or inhibition of chords at the top of the harmonic hierarchy? Journal of Experimental Psychology: Human Perception and Performance, 34, 1031-1043. doi: 10.1037/0096-1523.34.4.1031 T ILLMANN , B., S TEVENS , C., & K ELLER , P. E. (2011). Learning of timing patterns and the development of temporal expectations. Psychological Research, 75, 243-258. T RAINOR , L. J., & T REHUB , S. E. (1994). Key membership and implied harmony in Western tonal music: Developmental perspectives. Perception and Psychophysics, 56, 125-132. T REHUB , S. E., C OHEN , A. J., T HORPE , L. A., & M ORRONGIELLO, B. A. (1986). Development of the perception of musical relations - Semitone and diatonic structure. Journal of Experimental Psychology: Human Perception and Performance, 12, 295-301. V UVAN , D. T., P RINCE , J. B., & S CHMUCKLER , M. A. (2011). Probing the minor tonal hierarchy. Music Perception, 28, 461-472. doi: 10.1525/mp.2011.28.5.461 V UVAN , D. T., & S CHMUCKLER , M. A. (2011). Tonal hierarchy representations in auditory imagery. Memory and Cognition, 39, 477-490. YOUNGBLOOD, J. E. (1958). Style as information. Journal of Music Theory, 2, 24-35. doi: 10.2307/842928 270 Jon B. Prince & Mark A. Schmuckler Appendix A Bach Well-Tempered Clavier Mozart Piano sonatas Beethoven Piano Sonatas Chopin Preludes, Waltzes, Mazurkas 24 Preludes and Fugues, Book 1 24 Preludes and Fugues, Book 2 K.279 K.280 K.281 K.282 K.283 K.284 K.309 K.310 K.311 K.330 K.331 K.332 K.333 K.357 K.545 K.570 K.576 Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. Op. 24 Preludes Op. 28, No. 1-24 Op. 6, No. 1-4 Op. 7, No. 1-5 Op. 10, No. 2, 5, 9 Op. 17, No. 1-4 Op. 24, No. 1-4 Op. 30, No. 1-4 Op. 33, No. 1-4 Op. 41, No. 1-4 Op. 50, No. 1-3 Op. 56, No. 1-3 Op. 59, No. 1-3 Op. 63, No. 1-3 Op. 64, No. 1-2 Op. 67, No. 1-4 Op. 68, No. 1-4 Op. 69, No. 2 B. 82 B. 134 B. 140 B. 150 2, No. 1-3 7 10, No. 1-3 13 14, No. 1-2 22 26 27, No. 1-2 28 31, No. 1-3 49, No. 1-2 53 54 57 78 79 81 90 101 106 109 110 111 Scherzo, Op. 81 Ballade, Op. 52, No. 4 Piano Rondo Op. 129 Symphony 5, Op. 67, 1st Mvmt. Appendix B Schubert Brahms Liszt Scriabin 13 Variations on a Theme by Anselm Hüttenbrenne String Quartet No. 10 in Eb Major, Op. 125, No. 1, Mvmt. 1-4 Trio for violin, viola and cello, Sept. 1816 String Quartet D 804, Mvmt. 2 Ballade No. 1 in D minor, Op. 10 Hungarian Rhapsodies, No. 12, 15 Grandes Etudes de Paganini, No. 3, 4 Liebestraume Notturno, No. 3 Transcendental Etudes, No. 5 Feux Follets Etude, Op. 2, No. 1 Impromptu in Gb major View publication stats String Quartet in C minor, Op. 51, No. 1 (1873) Waltz No. 1 in B major Waltz No. 2 in E major Waltz No. 8 in Bb Minor Waltz No. 9 in D minor Scherzo in Eb minor for piano, Op. 4 Piano Sonata No. 1 in C major, Op. 1 (2.) Andante Fantasia, Op. 116, No. 2, 5, 6 Intermezzi, Op. 117, No. 1, 2 Rhapsodie from 4 Piano pieces, Op. 119 12 Etudes, Op. 8, No. 1-6 24 Preludes, Op. 11, No. 4, 15