ABSTRACT
In the Dhrupad genre of Hindustani vocal improvisation, singers often engage with melodic ideas by manipulating intangible, imaginary objects with their hands. Despite no real object being present, vocalists execute repeated bi-manual effortful gestures, comprising gripping, exerting, and releasing phases. They stretch, pull, push, throw, and perform other movements, appearing to combat or yield to some imaginary resistive force. This observation suggests that certain patterns of change in acoustic features of the voice hint at the effortful interactions these imaginary objects afford through their physical properties, such as viscosity, elasticity, weight, and friction.
This paper focuses on these types of gestures (referred to as 'MIIOs' or Manual Interactions with Imaginary Objects) and examines whether hand movements during MIIOs relate to the action of imagined forces, and if so, what type of forces and on what types of materials. Such voluntary imitations of real object interactions are of interest to embodied music cognition research, suggesting that musical thinking is grounded in our ecological knowledge of interacting with real-world objects. Despite the specificity of the genre, this study addresses concerns in the gesture-sound relationship study relevant to a broader research community and extends to other musical lineages.
I. INTRODUCTION
Dhrupad singing offers a distinctive case for studying links between sound and physically inspired concepts. The deliberate and easily identifiable imitation of real-world interactions in Dhrupad makes it easier to study robust gesture-sound links that vocalists have established over years of practice. Dhrupad is particularly interesting for studying the non-discrete nature of gestures and mechanical phenomena, as smooth melodic glides (mīṇḍ) abound, and notes are approached with a strong sense of pitch continuum. Additionally, Dhrupad performances, starting at a strikingly low pace, exhibit slow melodic development and rigorous precision in intonation. As an oral tradition, knowledge is transmitted through direct imitation from teacher to disciple, including body movements, supporting the deliberate choice of collecting material from a single musical lineage, namely students of Zia Fariduddin Dagar.
Singing gestures have drawn little attention in research, although there is a growing body of published work on Hindustani music. None has studied the Dhrupad genre or MIIOs. MIIOs present a unique case where the hands, although free to move, are deliberately constrained by the conception of an imaginary object. Familiar interactions from the real world, employed spontaneously by Hindustani vocalists, may offer unique insights into embodied music cognition. Despite the absence of a real mediator, these gestures may reveal significant cognitive processes associated with fundamental concepts rather than unequivocal mechanical cross-modal couplings of a particular instrument.
Dhrupad relies heavily on rule-based improvisation conforming to the ‘rāga’ (modal) system. The notion of a rāga as a movement in melodic pitch ‘space’ is also accompanied by smooth hand movements in real space. According to Rahaim and personal observations, Hindustani singers engage with melodic content in two main ways, reflecting different relationships between body and voice: (a) by keeping the hand(s) open and tracing curves in an effortless way as trajectories in space, and (b) by closing the hands (forming a grip) and manipulating an imaginary object, acting against an imagined opposing force associated with changing its shape, size, and position.
Dhrupad vocalists do not merely project melographic content by the absolute position of the hands and their trajectories in space; the dynamic aspect of movement also conveys various types of related sonic information. This paper is concerned with the second type of gestures, where notes can be manipulated as smooth pitch glides (mīṇḍ) through gestures resembling stretching a rubber band, moving a heavy object, or bouncing a ball. These gestures suggest a sensation of resistance that the performer needs to defy to move the imaginary object. This interaction mediates the sound through imagined material, showing an observable match between the voice and the manipulative gestures in terms of synchronization and temporal congruency of features like melody, dynamics, and timbre.
Drawing on theories that consider human cognition founded on repeated patterns of interactions with the real world, and extending Gibson’s ecological theory of affordances to cover the imaginary objects of MIIOs, this work examines the assumption that Dhrupad singers' interactions with imaginary objects are linked to the sound through the interaction possibilities the object affords.
II. METHODOLOGY
This study employs a combined methodological approach, using qualitative and quantitative methods on original recordings of interviews, audio-visual material, and 3D-movement data (using an Optitrack passive marker system) of vocal improvisations. Recordings of fourteen vocalists from the same musical lineage (students of Zia Fariduddin Dagar, including the maestro himself) were collected in domestic spaces in India (2010-2011).
The qualitative analysis includes a thematic analysis of interview material and a video observation analysis of performance recordings. The quantitative analysis involves developing a formalized description for classifying MIIOs by fitting general logistic models to a set of movement and sonic features.
Thematic analysis of interview material aimed to unfold common underlying cross-modal concepts embedded in explicit and implicit knowledge by the interviewees. The analysis focused on the conception of melody as movement and the importance of visual and motor imagery in conceiving sound. The analysis concentrated on the melodic function of imagined objects, aiming to understand how the extension of the body through imaginary objects and materials facilitates engagement with melodic ideas.
Video observation analysis aimed to identify and classify MIIOs and melodic movements, finding recurrent gesture-sound associations. It involved visually identifying, segmenting, labeling, and classifying repeated patterns of manual gestures that allude to MIIOs for each performer, and annotating the effort level required for each gesture. The coding scheme emerged from the analysis of interview material and multiple viewings of the video footage. An inter-coder validation of the visual material was carried out by two professional dancers/choreographers for one performer's performance.
Logistic regression models were developed to validate and augment qualitative findings, using exact measurements to classify gestures based on acoustic and movement features. Two variations of models were presented: one fitting each individual performer and one overlapping between performers to describe more generic cross-performer behaviors.
III. RESULTS
During interviews, performers displayed a high visual element in conceptualizing music and frequently used physically inspired linguistic descriptors. Some vocalists made explicit statements about gesture-sound associations, such as:
mīṇḍ – stretching a rubber band
gamak – applying pressure
hudak – throwing
Musicians acknowledged the role of body transmission and similarities between their gestures and their teacher’s, not always as exact replicas but in shared movement qualities. Motor-based metaphors alluding to a resistive force, such as elasticity/stretching, pressure/pushing, weight/lifting, and friction/scratching, were recurrent descriptors.
Video observation analysis identified recurrent gesture classes for each performer: stretching or compressing an elastic object, pulling, collecting, pushing away, or throwing a rigid object. Gesture-sound association analysis revealed consistent links shared between performers, reflecting shared cross-modal morphologies. For instance, double-sloped pitch glides were associated with gestures comprising two opposing spatial directions, while monotonic ascending melodic glides were associated with single-direction gestures in space.
Quantitative analysis rejected the null hypothesis that voice and gesture are unrelated. Using a small set of movement and sound features, a good part of the variance in gesture classification was estimated. The logistic models showed that gesture classes are more dependent on acoustic rather than movement features, with a stronger idiosyncratic factor in manual performance of MIIOs.
IV. CONCLUSIONS
Despite the flexibility in hand use by Dhrupad vocalists, the high degree of association between classes of virtual interactions and the voice provides evidence for non-arbitrariness. This association may be grounded in shared cross-modal morphologies and, in some cases, in the melodic organization of the ālāp improvisation. Although there is higher consistency within an individual performance rather than across performers, the type of imagined object and the nature of the imagined opposing forces reflect qualities describing melodic aspects of the voice.
A larger dataset is needed to make a strong claim about MIIOs in Dhrupad vocal music. However, this study has reported on the first study of gesture-sound associations during MIIOs in the Dhrupad genre. It has contributed to understanding the role of voluntary imitations of real object interactions by using interview material and recordings of real performances in India and by proposing a combined methodological approach comprising both qualitative and quantitative methods.
By taking an embodied approach and mapping effort to features from both auditory and movement domains, this work can enhance mapping strategies in empty-handed artificial interactions, making gesture-sound links more physically plausible. Novel interaction paradigms inspired by our real-world interactions can be developed.
Commentaires