Rachna Raman | University of Texas at Dallas (original) (raw)

Papers by Rachna Raman

Modulation, a shift in mode (rāgam), is important in South Indian classical (Carnātic) music. Her... more Modulation, a shift in mode (rāgam), is important in South Indian classical (Carnātic) music. Here we investigate the sensitivity of Carnātic and Western listeners to such shifts. Carnātic music has two kinds of shifts: rāgamālikā (retaining tonal center, resembling a shift from C major to C minor in Western music) and grahabēdham (shifting tonal center, resembling a shift from C major to A minor). Listeners heard modulating pieces of music and indicated the point of modulation, and were measured for accuracy and latency. Indians were more accurate than Westerners with both types of modulation but Westerners were faster with grahabēdhams. Cues could explain performance differences between nationalities: Indians were more familiar with rāgamālikā-type modulations whereas Westerners' culture made them more familiar with grahabēdham-type modulations. Increased caution toward the less familiar grahabēdhams for Indians could explain their slower response time compared to rāgamālikās. With grahabēdhams, hit rates for both groups were comparably high, but Westerners’ lower level of accuracy was due to higher false-alarm rates to lures that were superficially similar to actual modulations. This indicated their dependence on surface-level cues in the absence of familiarity and culture-specific information. Music training helped teachers in both groups make fewer errors when compared to students. Older listeners’ performance was comparable to that of younger listeners.

Bookmarks Related papers MentionsView impact

We used Toiviainen and Krumhansl's (2003) concurrent probe-tone technique to track Indian and Wes... more We used Toiviainen and Krumhansl's (2003) concurrent probe-tone technique to track Indian and Western musicians' tonal-hierarchy profiles through modulations in Carnātic (South Indian classical) music. Changes of mode (rāgam) are particularly interesting in Carnātic music because of the large number of modes (more than 300) in its tonal system. We first had musicians generate profiles to establish a baseline for each of four rāgams in isolation. Then we obtained dynamic profiles of two modulating excerpts, each of which incorporated two of the four baseline rāgams. The two excerpts used the two techniques of modulation in Carnātic music: grahabēdham (analogous to a Western shift from C major to A minor), and rāgamālikā (analogous to a shift from C major to C minor). We assessed listeners' tracking of the modulations by plotting the correlations of their response profiles with the baseline profiles. In general, the correlation to the original rāgam declined and the correlation to the new rāgam increased with the modulation, and then followed the reverse pattern when the original rāgam returned. Westerners' responses matched those of the Indians on rāgams with structures similar to Western scales, but differed when rāgams were less familiar, and surprisingly, they registered the shifts more strongly than Indian musicians. These findings converged with previous research in identifying three types of cues: 1) culture-specific cues—schematic and veridical knowledge—employed by Indians, 2) tone-distribution cues—duration and frequency of note occurrence—employed by both Indians and Westerners, and 3) transference of schematic knowledge of Western music by Western participants.

Bookmarks Related papers MentionsView impact

Conference Presentations by Rachna Raman

We present three experiments using Toiviainen and Krumhansl's (2003) continuous probe-tone method... more We present three experiments using Toiviainen and Krumhansl's (2003) continuous probe-tone method to track listeners' perception of tonal modulations. Listeners hear music in one ear and a steady probe in the other, continuously rating how well the probe goes with the music. The excerpt repeats 12 times, using each of the 12 possible semitones. We construct tonal hierarchy profiles for 10-s segments throughout the piece, and correlate them with the profiles for the keys involved. Shifts in key are reflected in the pattern of those correlations. In Study 1 Indian and Western music teachers heard South Indian classical songs. Only Indian teachers were familiar with the songs. Surprisingly, differentiation of modes in modulations was clearer for the Westerners. We thought perhaps greater familiarity led to more global perception, in which a piece is heard not as sharply modulating but as blending a cluster of different keys throughout. In Study 2 highly trained Western musicians heard 2-min Haydn quartet excerpts. We compared performance during the first three trials in which they were unfamiliar with the pieces with the last three trials where they were more familiar. Differentiation of the keys was greater at first, and later became more global. In Study 3 we manipulated familiarity strongly with student orchestra members in three sessions: before encountering the piece they were to learn, in the middle of rehearsals, and after performing the piece. However, there was little change in their responses from beginning to end, perhaps because their active involvement maintained their analytic mode. In conclusion, there is some tendency for musicians to hear a piece more and more as a unified whole with increasing familiarity, but not always. In some cases they form a holistic representation of the piece as a cluster of related keys, rather than as a sharply differentiated sequence of individual episodes.

Bookmarks Related papers MentionsView impact

We investigated listeners’ responses to modulations in two Haydn string quartets from op.76 in D ... more We investigated listeners’ responses to modulations in two Haydn string quartets from op.76 in D minor (no. 2) and C major (no. 3). Listeners at two levels of musical expertise (untrained, N = 60, vs. 6 or more years of training, N = 60) heard the 2-min exposition sections of the quartets, continuously rating how well the 12 possible probe tones fit the music, for 12 trials with each quartet. We assessed tonal hierarchy profiles for 11 5-s samples at approximately equal time intervals throughout the excerpt, with particular attention to points at which the music modulated, and correlated those profiles with the standard profiles of the keys in question. In general, more experienced listeners reflected the key changes in their ratings, whereas inexperienced listeners responded more globally, producing profiles that captured the principal keys of the excerpts as a whole. Here we compared ratings produced for the first three trials with a given excerpt, when it was relatively unfamiliar, with those for the last three trials, when they had heard it 9 or more times. As the more experienced listeners became more familiar with an excerpt, they tended to respond more globally, flattening out the shifts in their responses at points of modulation. In contrast, the inexperienced listeners became more attentive to the details with greater familiarity, registering the modulations more sharply. Both groups showed similar latencies of response to the modulations; in particular, in the C-major quartet they responded very slowly to the modulation from G major to G minor, and very quickly to the shift from G minor to Eb major.

Bookmarks Related papers MentionsView impact

Previous investigations showed that Western participants' perception of wrong notes in familiar W... more Previous investigations showed that Western participants' perception of wrong notes in familiar Western melodies was influenced primarily by key membership (diatonic or nondiatonic) and to a lesser extent by interval size (1 or 2 semitones away from the original note). These results were supported by a cross-cultural study between South Indian classical (Carnātic) and Western musicians and nonmusicians with highly familiar Western melodies. However, Indian participants were slower and less accurate with Carnātic than with Western melodies, presumably due to the complexity of the Carnātic music system. In this study, we examined the effect of song familiarity in Westerners' perception of wrong notes. We chose 32 Western melodies previously rated as familiar, and 32 highly unfamiliar melodies, similar in style to the familiar melodies. Participants heard each melody twice, each time with one wrong note that was determined from one of eight possible types of wrong note based on key membership, interval size, and direction (up vs. down). Participants identified the wrong note by pressing a key. The results indicated an effect of music experience with unfamiliar melodies only, with musicians detecting wrong notes faster than nonmusicians. All groups were faster with familiar melodies than with unfamiliar melodies. Key membership influenced perception of wrong notes in both familiar and unfamiliar melodies: Participants were slower at recognizing wrong notes that were diatonic and faster when they were nondiatonic. Interval size influenced perception of wrong notes only in unfamiliar melodies: Participants were slower at recognizing wrong notes that were 2 semitones away than when they were 1 semitone away, which was the opposite of our previous studies with familiar melodies. We take these results as converging evidence that people remember the pitches in melodies in terms of steps of an overlearned modal scale, and not as successive intervals.

Bookmarks Related papers MentionsView impact

Our theory (Music Perception, 2014) holds that early processing of a melody depends on binding co... more Our theory (Music Perception, 2014) holds that early processing of a melody depends on binding contour to scale at a particular pitch level, which requires a representation of the scale. Here we used the continuous probe-tone method to track the formation of this representation following a change of key. Listeners at three expertise levels heard two 2-min excerpts from Haydn string quartets, rating how well the 12 possible probe tones fit in relation to the music, for 12 trials with each quartet. We correlated these tonal-hierarchyprofiles with baseline profiles for the keys for the 10 s before and after each shift of tonality. We also inspected the time course of responses for 30 s following each modulation. Experienced musicians registered all the modulations; nonmusicians registered just the closely-related keys. Musicians responded within 10 s to shifts among the closely-related keys, but took up to 30 s to respond to distant modulations.

Bookmarks Related papers MentionsView impact

Sorting tasks can reveal the underlying intuitive structure of a collection of items, in this cas... more Sorting tasks can reveal the underlying intuitive structure of a collection of items, in this case musical excerpts. Sorting tasks can be used to compare experts and non-experts without relying on specialized vocabulary, and they tend not to fatigue participants (Chollet, Valentin, & Abdi, 2014). Here, we used the sorting technique with excerpts
from the piano music of Bach, Mozart, and Beethoven. Experiment 1 involved sorting 21 MIDI-generated stimuli. Experiment 2 utilized 36 excerpts from recorded performances of four pianists (Arrau, Barenboim, Pirès, and Richter). Each experiment involved two parts: In Part 1, participants sorted excerpts freely into any number of clusters. In Part 2, participants sorted excerpts into three clusters according to whether a single composer could have written the pieces in the group. We divided participants into three groups based on music training. We investigated the effects of composer, pianist, and music training on sorting. To analyze the data, we applied DiSTATIS (Abdi, Williams, Valentin, & Bennani-Dosse, 2012), a recent adaptation of multi-dimensional scaling specifically adapted to reveal the perceived dissimilarity among items, as well as to investigate group differences. The results showed an effect of composer in both experiments; participants were able to strongly differentiate Mozart’s excerpts from
Beethoven’s, with Bach falling in between those two. In Experiment 2, participants’ sorting decisions were strongly influenced by pianists. Richter’s performances of the three composers were clustered relatively close to the Mozart region of the solution, indicating their clarity and balance; in contrast, those of Barenboim were clustered in the Beethoven region, indicating their sumptuousness and passion. Experience effects were not strong and the highest and lowest expertise groups were differentiated only in the free sorting task of Experiment 2.

Bookmarks Related papers MentionsView impact

In previous investigations, Western listeners were quicker to respond to wrong notes in familiar ... more In previous investigations, Western listeners were quicker to respond to wrong notes in familiar melodies that were out‐of‐key (vs. in‐key) and 2 (vs. 1) semitones away from the original note (APCAM, 2008). Here we examined responses of South Indian classical (Carnātic) music teachers, students, and aficionados to wrong notes in familiar Carnātic and Western melodies. Carnātic music provides a fertile ground for studying phenomena associated with key membership, given the
complexity of its tonal system, which employs around 350 distinct modes. We chose 6 Carnātic and 6 Western melodies on the basis of Carnātic listeners’ high familiarity ratings. Each melody was presented approximately 8 times—since some of the Carnātic melodies were not amenable to all conditions—during 2 blocks of 48 randomly‐ordered trials, with a different wrong note each time, defined by key membership, interval size, and direction (up or down from the original note).
Participants responded as quickly as possible to the wrong note by pressing a key. In general, teachers recognized wrong notes faster than students, who were faster than aficionados. All groups were faster and more accurate with Western than Carnātic melodies, reflecting the complexity of the Carnātic system. Carnātic listeners’ response times followed qualitatively the same pattern with both types of melodies, a pattern similar to the earlier Western results: out‐of‐key notes and notes more distant from the original notes were easier to notice. Data on accuracy followed generally the same pattern, but also suggested that 1‐semitone in‐key alterations in the Carnātic melodies are often perceived as permissible deviations rather than “wrong.” Control groups of Western listeners with three levels of musical training also responded to the 6 Western melodies in a qualitatively similar manner to the Carnātic listeners.

Bookmarks Related papers MentionsView impact

Problem: Sorting tasks are frequently used to study the psychology of sensory perception and cons... more Problem: Sorting tasks are frequently used to study the psychology of sensory perception and consumer preference. In a sorting task, participants sort items into groups based on perceived similarity. Sorting usually reveals an intuitive structure amongst the items of interest. Sorting tasks are advantageous because they require little-to-no training and do not quickly fatigue participants (Cartier et al., 2006; Chollet, Valentin, & Abdi, 2014). Furthermore, sorting tasks can be used to compare experts and non-experts: populations that may use different vocabulary, but may or may not report the same relationships between items of interest. Sorting tasks have been analyzed by various univariate and multivariate approaches (for review see Ares ’14), but most often—especially in the food sciences—they are analyzed by metric multidimensional scaling (MDS; Abdi, 2007). However, MDS loses data from individual assessors, and so cannot investigate effects of assessor expertise.
Methods: A recent adaptation of MDS called DiSTATIS (Abdi, Williams, Valentin, & BennaniDosse, 2012; Valentin, Chollet, & Chrea, 2007) can investigate effects of assessor expertise. MDS and DiSTATIS both convert distance (dissimilarity) matrices into cross-product matrices
so they can be analyzed with the eigen-decomposition (which provides features such as explained variance, orthogonal components, and component maps). Unlike MDS, DiSTATIS is specifically adapted to reveal the perceived dissimilarity among items of interest (as does MDS), but also allows us to investigate individual (or group) differences.
Results: Here we present DiSTATIS—and its discriminant extension—applied to two studies in music cognition. In one experiment, assessors sorted into 3 categories 21 sound clips of classical music written by 1 of 3 composers (Bach, Beethoven, or Mozart) as played on a midisynthesizer. In the other experiment, assessors sorted 36 sound clips of classical music written by 1 of (the same) 3 composers but played by 1 of 4 pianists (Arrau, Barenboim, Pirès, or Richter). Assessors were binned into low and high musical experience (i.e., expertise). In addition to investigating effects of composer, effects of pianist, and composer-pianist interactions, DiSTATIS (unlike MDS) revealed effects of musical expertise and differences between individuals.
Conclusions: This study was selected because music cognition involves both implicit and (for expert assessors in particular) explicit knowledge, and so a sorting task (analyzed by DiSTATIS) could reveal effects of musical expertise (i.e., between-assessor effects). We highlight how
sorting tasks are useful in studies on perception, and how DiSTATIS and its discriminant extension provide insight into the intuitive structures within sorting data.

Bookmarks Related papers MentionsView impact

We examined differences between musicians’ and nonmusicians’ perception of modulations as they un... more We examined differences between musicians’ and nonmusicians’ perception of modulations as they unfolded in time in Carnātic music using Toiviainen and Krumhansl’s (2003) concurrent probe-tone technique. Previous investigations showed that with culturally familiar music listeners use culture-specific and psychophysical cues, whereas with music from another culture they rely more on psychophysical cues and schematic knowledge imported from their own culture. We compared baseline profiles of four rāgams (modes) with profiles of modulating excerpts containing the same rāgams. Indian musicians’ tonal hierarchy profiles tracked the modulations in and out of the new rāgam. Western musicians’ profiles tended to match those of Indian musicians, indicating their employment of psychophysical cues as well as western schematic cues. Indian and western nonmusicians’ profiles did not reflect the modulations. Indian nonmusicians’ inability to identify modulations in melodies from their own culture suggests that musical training, irrespective of culture, facilitated performance on the continuous probe-tone task.

Bookmarks Related papers MentionsView impact

Previously we used Toiviainen and Krumhansl’s (2003) concurrent probe-tone technique to study lis... more Previously we used Toiviainen and Krumhansl’s (2003) concurrent probe-tone technique to study listeners’ tracking of modulations in Carnatic music (APCAM, 2013). With modulation, the correlation of probe ratings and the tonal hierarchy profile for the initial key declines, and correlation with the new key profile increases. This reverses when the piece modulates back. Here we apply that technique to the investigation of modulations in classical minuets. We used minuets in studies finding an improvement effect in recognition memory, in which the rejection of same-contour lures improved during the first 12-15 s of hearing a novel melody because the memory system has time to bind the melodic contour to the musical scale (Dowling & Tillmann, 2014; Tillmann et al., 2013). This led us to investigate the time course of formation of the tonal hierarchy representation involved. In Study 1, 84 musicians and 84 nonmusicians rated one of the 12 probe tones continuously while hearing the first eight measures of one of two minuets (one modulating and the other not). A Latin square to control the order of presentation of the probe tones, so that each probe was rated an equal number of times on each of the 12 trials with a minuet. Thus, we could look at the development of the profiles across trials. The tonal profile of the initial key emerged clearly during the first three trials, as assessed by correlation. There was some tendency for the correlation of responses with the profile for the dominant to increase following the modulation, but there was no tendency for the correlation with the initial tonic to decrease. This agrees with earlier results suggesting that a conventional modulation to the dominant does not strongly alter the sense of tonal center very much. In Study 2 we are currently replicating Study 1 with six minuets.

Bookmarks Related papers MentionsView impact

We examined the differences between musicians’ and nonmusicians’ perception of tonality in famili... more We examined the differences between musicians’ and nonmusicians’ perception of tonality in familiar and unfamiliar melodies. Previous investigations show that: (a) with familiar music listeners use culture-specific and psychophysical cues, whereas with unfamiliar music they use psychophysical cues and schematic knowledge imported from their own culture; (b) listeners access their mental representations of the hierarchy of notes in musical scales of their culture when listening to familiar and unfamiliar melodies. Indian and western musicians and nonmusicians rated brief South Indian classical excerpts in four modes (rāgams) using Toiviainen and Krumhansl’s (2003) concurrent probe-tone technique. Indian and western musicians’ responses were similar on rāgams resembling western modes but differed with rāgams unfamiliar to westerners. Our findings supported previous research and identified three types of cues that musicians used: 1) culture-specific cues—theoretical knowledge of the rāgams and familiarity with the excerpts in the study—employed by Indian musicians, 2) psychophysical cues—note duration and frequency of note occurrence—employed by all musicians, and 3) transference of western schematic knowledge by western musicians. Nonmusicians’ responses differed from the musicians’; their partial use of these cues suggests that musical training facilitated performance on the binaural probe-tone task and in applying these cues.

Bookmarks Related papers MentionsView impact

A fundamental purpose of this study was to test whether theories and principles of music cognitio... more A fundamental purpose of this study was to test whether theories and principles of music cognition and perception derived using western music can be applied to other styles of music, in this case, Carnātic music. Previous investigations showed that listeners use culture-specific and low-level sensory cues to comprehend familiar music, whereas they relied mainly on sensory cues and schematic knowledge gained from listening to music of their own culture to understand unfamiliar music. Research also shows that listeners hold mental representations of the hierarchical ordering of notes in musical scales of one’s culture which they readily access when listening to familiar and unfamiliar melodies. The existence of such representations has been demonstrated in real time using behavioral and neuroimaging techniques. The goals of the present study were to 1) behaviorally track listeners’ responses to two types of modulation as they developed over time in Carnātic music and 2) identify the various cues that listeners, familiar and unfamiliar with the music, utilized in order to perceive the modulations. The purpose of Experiment 1 was to obtain baseline profiles of four rāgams (modes) and compare these against profiles of modulating excerpts containing the same rāgams in Experiment 2. Carnātic and western music teachers heard brief Carnātic excerpts in one ear while in the other ear they heard a continuously sounded probe-tone. The probe-tone was one of the twelve chromatic notes of an octave. Participants judged continuously how well the probe-tone fitted with the melody. Results showed that western teachers’ responses matched those of the Indians on rāgams that had structures similar to the western scales but differed considerably when the rāgams were unfamiliar. Also, there were differences in the two groups’ responses to modulations. These discrepancies were primarily due to the influence of culture. The findings supported previous research and identified three types of cues: 1) culture-specific cues, which included theoretical knowledge of the rāgams and familiarity with the excerpts in the study, employed by Indian teachers, 2) basic psychophysical cues of duration and frequency of note occurrence employed by both Indian and western participants, and 3) transference of schematic knowledge of western music by western participants.

Bookmarks Related papers MentionsView impact

The study was aimed at (1) identifying cues that help listeners perceive tonality changes, (2) in... more The study was aimed at (1) identifying cues that help listeners perceive tonality changes, (2) investigating if cues learnt from one culture help toward understanding music across cultures, and (3) understanding if musical training is advantageous for cross-cultural perception. Carnātic music has two kinds of tonality shifts: the popular rāgamālikā (shifts of rāgam, retaining tonal center; e.g., C to C minor), and the controversial grahabēdham (shifts of rāgam and tonal center; e.g., C to A minor). Stimuli consisted of songs containing 45 rāgamālikā and 46 grahabēdham shifts. South Indian and American teachers and students were further divided by age (older or younger than 60 yr), and served in either the rāgamālikā or grahabēdham condition. Participants indicated the point at which a modulation occurred which we measured in terms of accuracy and latency. Results showed that Indians were more accurate and faster with rāgamālikās while westerners performed better with grahabēdhams. Cues could explain performance differences between nationalities: Indians performed better than westerners with rāgamālikās presumably because of their familiarity with it; westerners, on the other hand, performed better with grahabēdhams because they were probably able to apply cues from their culture to a type of modulation that was familiar to them. Indians and westerners had similar hit rates with grahabēdhams. Increased caution toward the less familiar grahabēdhams for Indians could explain their slower response time compared to rāgamālikās. Musical training was advantageous to teachers across both conditions and ages: they had more hits and fewer errors than students. This could be attributed to an enhanced representation for systems of pitches and modalities.

Bookmarks Related papers MentionsView impact