TOWARD DEFINING TRADITION: A STATISTICAL AND NETWORK ANALYSIS OF GIACOMO PUCCINI’S TURANDOT AT THE METROPOLITAN OPERA

By

JOSHUA ORIN NEUMANN

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2016

© 2016 Joshua Orin Neumann

In memory of Professor Scott Nygren, and in honor of all who made this possible. Soli Deo Gloria.

ACKNOWLEDGMENTS

My personal interests in this dissertation stem primarily from the juxtaposition of my love of opera, an undying interest in “how” things come to be, my desire to form relationships with the performers who create beauty in opera houses and concert halls, and an ironic and idiosyncratic penchant for numbers and data. I thank my wife Lindsey, whose constant support and encouragement lit the many late nights necessary for completing this dissertation. I also thank my parents and siblings, from whom I learned the values of patience, dedication, and perseverance, for their support and curiosity, which has been a source of inspiration for clarity of explanation.

I am indebted to Sara Vellutini, Paolo Nuti, Carmen da Prato, and Simonetta

Puccini for enhancing my understanding of Puccini’s life and the culture in which and for which he composed so many beautiful melodies. This dissertation would not be possible without assistance from the staff at the New York Public Library’s Rodgers and

Hammerstein of Recorded Sound, and the Archives of the Metropolitan Opera. I am also thankful to Marcello Giordani, Andréa Gruber, Stephen Costello, Kristin Lewis, and

Alberto Veronesi for granting interviews that have helped foster a more nuanced understanding of the creative process and its relationship to tradition. I am grateful for the support and insights of my friends and colleagues, especially Matthew Franke,

Morgan Rich, Christy Thomas, Michael Vincent, and Oren Vinogradov.

My committee members Mark Katz, Scott Nygren, Tony Offerle, Jane

Southworth, Jennifer Thomas, and David Waybright have all greatly influenced me along this journey. Each has offered unique insight and pushed me to think about my work in a variety of ways and to make it widely accessible. Finally, I offer my most hearty thanks to my advisor, Margaret Butler, without whose unyielding support, insight,

4

guidance, and constant encouragement of multivalence in analytical technique and communication this document would not exist.

5

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 10

LIST OF OBJECTS ...... 15

LIST OF ABBREVIATIONS ...... 16

ABSTRACT ...... 17

CHAPTER

1 INTRODUCTION: WHAT IS TRADITION? ...... 19

Aims and Objectives ...... 19 Tradition’s Origins ...... 25 Tradition’s Characteristics ...... 29 Challenges to Understanding Tradition ...... 40

2 CASE STUDY PARAMETERS AND METHODS ...... 45

Case Study Parameters ...... 46 Turandot: A Brief Overview ...... 46 Turandot’s Technoshpere...... 47 Place and Time ...... 58 Musical Moments for Investigation ...... 65 Musical Phenomena for Investigation ...... 70 Case Study Methods...... 75 Conclusion ...... 88

3 PROFILING PERFORMANCE VIA TEMPO HIERARCHY ANALYSIS ...... 90

Rethinking Musical Objects ...... 90 Scores, Performances, and Works ...... 91 New Musical Objects for Discourse ...... 100 Reading an Operatic Time Scape ...... 101 Analysis of Tempo Hierarchies From 4 March, 1961 ...... 104 Analysis of Tempo Hierarchies From 7 November 2009 ...... 119 Further Examples in Textualizing Performance ...... 135 “Signore, ascolta!” ...... 135 “Non Piangere, Liù!” ...... 137 “In questa reggia” ...... 140

6

“Straniero, ascolta!” ...... 143 “Nessun dorma!” ...... 146 “Tu che di gel sei cinta” ...... 150 Conclusions ...... 152

4 QUANTITATIVE AND CHRONOLOGICALLY-BASED MODELS OF TRADITION ...... 156

“Signore, ascolta!” ...... 158 1960s...... 158 1970s...... 165 1980s...... 168 1990s...... 172 2000s...... 177 “Non piangere, Liù” ...... 182 “Tu che di gel sei cinta” ...... 196 Conclusions ...... 201

5 QUANTITATIVE AND DEMOGRAPHIC-BASED MODELS OF TRADITION ...... 203

“In Questa Reggia” in Two Productions ...... 204 The Beaton Production ...... 205 The Zeffirelli Production...... 209 “Straniero, ascolta!” ...... 214 “Nessun dorma!” ...... 220 Conclusions ...... 225

6 CORRELATION NETWORK ANALYSIS OF PERFORMANCE AND TRADITION ...... 228

Overview ...... 228 “Signore, ascolta!” ...... 230 “Non piangere, Liù” ...... 236 “In questa Reggia” ...... 241 “Straniero, ascolta!” ...... 246 “Nessun dorma” ...... 252 “Tu che di gel sei cinta” ...... 255 Conclusions ...... 258

7 CONCLUSIONS AND FURTHER CONSIDERATIONS ...... 261

Conclusions ...... 261 Further Considerations ...... 269

APPENDIX

A TIME SCAPES OF “SIGNORE, ASCOLTA!” ...... 275

7

B TIME SCAPES OF “NON PIANGERE, LIÙ” ...... 291

C TIME SCAPES OF “IN QUESTA REGGIA” ...... 307

D TIME SCAPES OF “STRANIERO, ASCOLTA!” ...... 323

E TIME SCAPES OF “NESSUN DORMA” ...... 339

F TIME SCAPES OF “TU CHE DI GEL SEI CINTA” ...... 355

G “SIGNORE, ASCOLTA!” CORRELATION MATRIX ...... 371

H “NON PIANGERE, LIÙ” CORRELATION MATRIX ...... 381

I “IN QUESTA REGGIA” CORRELATION MATRIX ...... 391

J “STRANIERO, ASCOLTA!” (RIDDLE SCENE) CORRELATION MATRIX ...... 401

K “NESSUN DORMA” CORRELATION MATRIX ...... 411

LIST OF REFERENCES ...... 431

BIOGRAPHICAL SKETCH ...... 440

8

LIST OF TABLES

Table page

2-1 Historical Overview of Opera and Technology...... 49

2-2 Operas broadcast from The Metropolitan Opera more than 50 times ...... 59

2-3 Illustrative method for calculating correlation with respect to performance chronology and to resulting evolving arithmetical mean ...... 84

9

LIST OF FIGURES

Figure page

1-1 Traditional folksong “Hot cross buns!” ...... 41

2-1 Emile Berliner’s direct, hand powered device for making sound recordings...... 48

2-2 Three-spring (two in the larger spring barrel, one in the smaller) drive mechanism for powering the rotation of a disc or cylinder. This required the turning of a winding key to store energy in the springs...... 51

2-3 Broadcast history of Turandot at the Met ...... 65

2-4 Text and translations of arias in Turandot ...... 70

2-5 Screen capture of time instants layer with adjusted beat placements (middle), and Harmonic Spectrogram (bottom) of Chopin Op. 17, No. 4, m. 29...... 79

2-6 Scape plotting domain structure (left) and example using mathematical average (right)...... 82

2-7 Prototypical tempo similarity network...... 88

3-1 Time Scape, text, and tempo contour graph (with linear trend line) for Anna Moffo 04 March 1961 "Signore, ascolta!"...... 106

3-2 Time Scape, text, and tempo contour graph (with linear trend line) for Franco Corelli 04 March 1961 "Non piangere, Liù." ...... 108

3-3 Time Scape, text, and tempo contour graph (with linear trend line) for Birgit Nilsson 04 March 1961 “In questa Reggia” ...... 110

3-4 Time Scape, text, and tempo contour graph (with linear trend line) for 04 March 1961 riddle scene ...... 113

3-5 Time Scape, text, and tempo contour graph (with linear trend line) for Franco Corelli 04 March 1961 "Nessun dorma!" ...... 115

3-6 Time Scape, text, and tempo contour graph (with linear trend line) for Anna Moffo 04 March 1961 "Tu che di gel sei cinta" ...... 117

3-7 Time Scape, text, and tempo contour graph (with linear trend line) for Marina Poplavskaya 07 November 2009 “Signore, ascolta!” ...... 121

3-8 Time Scape, text, and tempo contour graph (with linear trend line) for Marcello Giordani 07 November 2009 “Non piangere, Liù” ...... 122

10

3-9 Time Scape, text, and tempo contour graph (with linear trend line) for Maria Guleghina 07 November 2009 “In questa Reggia” ...... 125

3-10 Time Scape, text, and tempo contour graph (with linear trend line) for 07 November 2009 riddle scene ...... 129

3-11 Time Scape, text, and tempo contour graph (with linear trend line) for Marcello Giordani 07 November 2009 “Nessun dorma” ...... 130

3-12 Time Scape, text, and tempo contour graph (with linear trend line) for Marina Poplavskaya 07 November 2009 “Tu che di gel sei cinta” ...... 134

3-13 Time Scape, text, and tempo contour graph (with linear trend line) for Hei- Kyung Hong 14 April 2007 “Signore, ascolta!” ...... 136

3-14 Time Scape, text, and tempo contour graph (with linear trend line) Plácido Domingo 21 February 1970 “Non piangere, Liù” ...... 139

3-15 Time Scape, text, and tempo contour graph (with linear trend line) for Andréa Gruber 29 January 2005 “In questa Reggia” ...... 142

3-16 Time Scape, text, and tempo contour graph (with linear trend line) for 01 February 1992 riddle scene...... 145

3-17 Time Scape, text, and tempo contour graph (with linear trend line) for Nicola Martinucci 13 February 1988 “Nessun dorma” ...... 149

3-18 Time Scape, text, and tempo contour graph (with linear trend line) for Licia Albanese 24 February 1962 “Tu che di gel sei cinta” ...... 150

4-1 Time Scape of “Signore, ascolta!” 04 March 1961 – Performance 1 (Anna Moffo as Tradition originator) ...... 159

4-2 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 2 (A1-2) ...... 160

4-3 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 3 (A1-3) ...... 161

4-4 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 4 (A1-4) ...... 162

4-5 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 5 (A1-5) ...... 163

4-6 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 6 (A1-6) ...... 165

11

4-7 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 7 (A1-7) ...... 166

4-8 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 8 (A1-8) ...... 167

4-9 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 9 (A1-9) ...... 168

4-10 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 10 (A1-10)...... 169

4-11 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 11 (A1-11)...... 171

4-12 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 12 (A1-12)...... 172

4-13 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 13 (A1-13)...... 173

4-14 Tempo line comparison Average 1-12 and Performance 13 (Teresa Stratas, 01 Feb 1992) ...... 174

4-15 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 14 (A1-14)...... 175

4-16 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 15 (A1-15)...... 176

4-17 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 16 (A1-16)...... 177

4-18 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 17 (A1-17)...... 178

4-19 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 18 (A1-18)...... 179

4-20 Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1- 19 (A1-19)...... 180

4-21 Tempo contour map for entire recorded and available performance history of “Signore, ascolta!” at the Metropolitan Opera, 1961 – 2009...... 182

4-22 Time Scapes for “Non piangere, Liù” tradition origin and evolution 1960s (Top row R-L: P1, A1-2, A1-3; Bottom row R-L: A1-4, A1-5) ...... 185

12

4-23 Time Scapes for or “Non piangere, Liù” tradition evolution 1970s (L-R: A1-6, A1-7, A1-8) ...... 186

4-24 Time Scapes for or “Non piangere, Liù” tradition evolution 1980s (L-R: A1-9, A1-10, A1-11) ...... 189

4-25 Time Scapes for or “Non piangere, Liù” tradition evolution 1990s (Top Row, L-R: A1-12, A1-13, A1-14, Bottom Row: A1-15)...... 191

4-26 Time Scapes for or “Non piangere, Liù” tradition evolution 2000s (Top Row, L-R: A1-16, A1-17, A1-18, Bottom Row: A1-19)...... 194

4-27 Tempo contour maps for performances (top) and moments in the tradition (bottom) for all recorded and accessible performances of “Non piangere, Liù” at the Metropolitan Opera, 1961-2009...... 195

4-28 Time Scapes representing tradition of “Tu che di gel sei cinta,” 1961-87 ...... 197

4-29 Time Scapes representing tradition of “Tu che di gel sei cinta,” 1988-2009 ..... 198

4-30 Tempo contour map for performances and moments in the tradition for all recorded and accessible performances of “Tu che di gel sei cinta” at the Metropolitan Opera, 1961-2009...... 200

5-1 Detail of headdress, dress, and set as designed by Cecil Beaton for Act II, scene 2. This production was first in use beginning 1960-61 season. Images used with permission by the Metropolitan Opera Archives...... 206

5-2 Detail of headdress, dress, and set as designed by Zeffirelli for Act II, scene 2. This production was first in use beginning 1986-87 season. Images used with permission by the Metropolitan Opera Archives...... 207

5-3 Tempo tradition Time Scapes for “In questa Reggia” in Cecil Beaton’s production...... 208

5-4 Tempo tradition Time Scapes for “In questa Reggia” in Franco Zeffirelli’s production...... 212

5-5 Tempo tradition Time Scapes for Zeffirelli production performances, inclusive of tempo data from Beaton production...... 213

5-6 Tempo tradition Time Scapes for Turandot’s riddle scene, “Straniero, ascolta!”, part 1 (1961-1987) ...... 217

5-7 Tempo tradition Time Scapes for Turandot’s riddle scene, “Straniero, ascolta!”, part 1 (1988-2009) ...... 218

13

5-8 Tempi across performances conducted by Marco Armiliato in 2003 (dark blue) and 2007 (light blue), in comparison to tempo relationships as evidenced by the tradition before and after these performances...... 219

5-9 Time Scapes representing tradition of “Nessun dorma!”, 1961-1987...... 223

5-10 Time Scapes representing tradition of “Nessun dorma!”, 1988-2009...... 224

6-1 Prefuse Force Directed Layout of first order correlations amongst performances and points in the performance tradition of “Signore, ascolta!”. ... 232

6-2 Variation on a yFiles hierarchical layout for first, second, and third order correlations among performances and the tradition for “Non piangere, Liù.”. ... 238

6-3 Edge-weighted spring embedded layout on the first and second order correlations for “In questa Reggia” showing a clear delineation between performances and traditions associated with production history...... 242

6-4 y-Files Orthogonal layout for 1st and 2nd order connections from each node to a performance and a point in the tradition’s evolution for “Straniero, ascolta!”...... 247

6-5 yFiles Orthogonal layout for 1st and 2nd correlation orders of performances only for “Straniero, ascolta!”...... 249

6-6 Chronologically arranged yFiles hierarchical layout with only performances of “Nessun dorma” as target nodes for first, second, and third order correlations...... 252

6-7 yFiles orthogonal layout of 1st, 2nd, and 3rd order correlations for “Tu che di gel sei cinta.”...... 256

7-1 Map of Performance Features Ontology. Line types reflect different kinds of interactions between points...... 271

7-2 Revised map of the musical work ontology, emphasizing the role of performance as a realization made by artists, and of the score as an encoding of the work...... 272

14

LIST OF OBJECTS

Object page

4-1 Dynamic Evolution of tradition for "Tu che di gel sei cinta." (.mp4 file 8.1MB) .. 196

15

LIST OF ABBREVIATIONS

AHRC Arts and Humanities Research Council

CHARM Centre for the History and Analysis of Recorded Music

LoC Library of Congress

MIREX Music Information Retrieval Evaluation eXchange

NYPL New York Public Library

OERC Oxford e-Research Centre

RIAA Recording Industry Association of America

16

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

TOWARD DEFINING TRADITION: A STATISTICAL AND NETWORK ANALYSIS OF GIACOMO PUCCINI’S TURANDOT AT THE METROPOLITAN OPERA

By

Joshua Orin Neumann

August 2016

Chair: Margaret Butler Major: Music

Tradition is an integral part of musical life, especially with regard to Italian opera.

At its core, a performance tradition connects all behaviors by all performers in a given work through time and across boundaries. Analytical attempts to examine tradition have historically been language-based and qualitative. This reality simultaneously mirrors the wide interpretive spectrum of performances that create traditions, and results in significant ambiguity. In constructing a statistical and network analysis of performances of six excerpts from Giacomo Puccini’s Turandot, this study establishes performances as historical texts and uses them to interrogate the concept of tradition; posits that tradition results from performance rather than shaping it; uncovers a link between production history and performance; and reveals the unique network each excerpt’s performance history embodies. The evidence shows that conductor Nello Santi and singers Birgit Nilsson, Franco Corelli, and Plácido Domingo exerted by far the strongest influence in shaping the tradition of Puccini’s final work at the Metropolitan Opera.

In this digitally based time-series analysis of performances of Turandot at the

Met, I extract data describing the temporal features for six excerpts from each of

17

nineteen performances spanning fifty seasons. I then use techniques developed by the

Centre for the History and Analysis of Recorded Music’s “Mazurka Project” to create visualizations of each performance’s tempo structure of each excerpt. These documents, called Time Scapes, allow me to interrogate the creative process in new ways. Statistical averages share nearly identical behavioral properties with traditions, making them useful for constructing quantitative models of traditions. Constructing Time

Scapes of these averages and comparing them across performances provides a means of visualizing points in a tradition’s incremental and continual development and evolution. The inherent connectivity of these performances of the same work at

America’s preeminent opera house embodies a network. The strengths of the relationships herein are statistically measurable in several ways, each of which reveals unique features of a network’s configuration. This study offers a new knowledge structure of the relationship among performances, works, and genres, and of how human behavior interacts with them all in the creation of a performance tradition.

18

CHAPTER 1 INTRODUCTION: WHAT IS TRADITION?

We are always situated within tradition…It is always part of us.

Hans Georg Gadamer Truth and Method1 Aims and Objectives

Performance tradition is an integral component of musical life. Individual performances can often provide delightful entertainment and can sometimes facilitate transcendental experiences for audiences. An important part of the human experience is the perception of the present, of which the past, and, perhaps more strongly, memory of the past are primary framing devices. How a person, whether a performer or consumer, perceives and thus engages his or her tenth experience of a work depends on his or her memory of the preceding nine encounters, and the memory of his or her reaction to them. When a performance meets expectations, audiences generally speak well of it. When a performance does not meet expectations, audiences generally offer lukewarm responses at best. This sense of anticipation is evidence of conventions, which set up expectations about how performers present a work. Conventions result from the application of collective memory pertaining to “how” a work “should” be, based on how it has been in the past. “How” a work’s performances have been in the past, and unadulterated by conversion into comparison rubric, constitute tradition.

Because performance tradition deals with the past, it exists in purely conceptual terms and therefore impedes our efforts to identify and analyze it. Scholars can trace musical practices that construe various traditions, though until the late nineteenth, these

1 Hans Georg Gadamer, Truth and Method (New York: Bloomsbury Academic, 2013), 294.

19

remnants have been only in the forms of textual or notational encoding, owing to the absence of recording technology. Scholars have thus lacked tangible resources and the means to engage performances and their resultant traditions at the level of their physical properties. Language and musical notation, with their respective inherent opacities stemming from requisite nuanced interpretation and therefore diversity of meaning, have heretofore been the fora for engaging performance tradition.

The aim of this study is to develop and demonstrate a means of quantifying the concept of musical performance tradition in order to supplement long standing qualitative descriptions of that concept. This enhancement will facilitate analysis that is broad in its scope yet simultaneously sensitive to the nuances of a musical performance’s minutiae. One result of this dissertation is the establishment of new documents of texts of opera performance and its traditions, thus “textualizing” each for analysis in ways that have heretofore been elusive. Another result is that the difference between notions of tradition and convention becomes clearer, with tradition embodying a resultant amalgamation of musical practices with convention implying a sense expectation.

Realistically, any work of music could serve as a case study for this dissertation.

Opera is an ideal as an entry point because perhaps of its longstanding link with the notion of tradition. Julian Budden asserts, “Over the centuries, Italian opera has always thrived on tradition, each composer building on the achievement of his predecessors.”2

Despite the emphasis on tradition in musicological discourse, current understanding of it remains incomplete. Social scientist Craig Calhoun highlights the extant knowledge gap

2 Julian Budden, Puccini: His Life and Works (Oxford: Oxford University Press, 2002), 477.

20

regarding tradition, “The very notion of tradition needs further interrogation,” and “opera is a terrific site for the interrogation of what tradition means and how its different discrete dimensions interrelate.”3 To date, most scholarship on tradition consists of qualitative or theoretical analyses.4 Such studies include reception histories and acknowledgement of differences between notated and realized music, among others. Even in instances where scholars have addressed the music performers make in an opera house, they have thus far engaged analyzed these practices with notation or written language, both of which require interpretation to understand the sonic events that an author describes.

However, these analyses have been unable to precisely and definitively identify what constitutes a particular tradition for performers and audiences.5 Naturally, this reality elucidates questions surrounding tradition’s objective identification and, in turn, highlights the need for a means of effectively answering those questions. Chapter 1 explores current scholarship on tradition in order to demonstrate the current level of complexity and opacity of discourse on the concept. This chapter is thus an object lesson in the limitations of contemporary language- and score- based approaches.

Chapter 2 begins by establishing why Giacomo Puccini’s Turandot is ideal for exploring these questions. First of all, Turandot is the only major opera whose performance history exists entirely in the era of electronically regulated and captured

3 Craig Calhoun, foreword to Opera and Society in Italy and France from Monteverdi to Bourdieu, edited by Victoria Johnson, et al. (Cambridge: Cambridge University Press, 2007), xxix.

4 This is only an indication that most current scholarship engages tradition on the conceptual level and is absent of precise, definite indications of what constitutes tradition, not that such inquiries are expendable. The winter issue of Telos 94 (Winter 1993-94) is devoted to the concept of tradition. No article in this edition engages tradition in a quantitative method, but, in keeping with this journal’s nature as a critical theory publication, every article discusses tradition from a conceptual standpoint.

5 For the purposes of this study, operatic practitioners and performers are interchangeable, and include singers and conductors, and, to a lesser extent, stage directors and orchestra members.

21

sound recording. The analysis of recorded music in its history is therefore far more straightforward than works whose recorded performance history incorporates processes of sonic preservation predating electronically-regulated recording (i.e. – the power source for the turning motor). Furthermore, recordings of performances of Turandot at

The Metropolitan Opera through the New York Public Library’s Rodgers and

Hammerstein Archive of Recorded Sound make such an investigation possible. Finally, the Metropolitan Opera is the preeminent opera house in the United States, so to focus on performances here is to focus on performances sanctioned by this country’s primary arbiter of operatic taste. For these reasons the performance corpus of Turandot as recorded at New York’s Metropolitan Opera is the case study content for this dissertation’s analysis.

Any tradition ultimately results from patterns of human behavior occurring in and over time; they are thus quantifiable. Contextualizing performance practices as such renders them readily analyzable by statistical means. Quantitative analysis allows for precise identification of these behaviors based on their sonic phenomena, which in turn allows for interrogating their evolution. The most appropriate tools to analyze these sonic phenomena—musical sound’s physical properties of tempo, pitch, and volume— are digital. Sonic Visualiser is an open source “application for the viewing and analysing of music audio files.”6 Its primary use is to identify and extract data from digital audio files. As solo passages are more easily analyzable than full ensemble textures, and given Sonic Visualiser’s current limits that inhibit analyzing the thickest operatic

6 Chris Cannam, Christian Landone, and Mark Sandler, Sonic Visualiser: An Open Source Application for Viewing, Analysing, and Annotating Music Audio Files, in Proceedings of the ACM Multimedia 2010 International Conference. http://sonicvisualiser.org.

22

textures, this dissertation examines six solo passages from Turandot. Highlighting the work of the AHRC’s Centre for the History and Analysis for Recorded Music (CHARM) as a model, Chapter 2 then describes how CHARM’s scape plot generator creates a color-coded graph of this data as a new kind of text for use in performance analysis.

Furthermore, mathematical conceptualizations of statistical averages of the data very closely mirror the principles of a tradition, especially if arranged chronologically. Finally,

Craig Sapp’s correlation network of performances of Chopin’s Mazurka in c# minor, Op.

63, no. 3 demonstrates how this approach can be useful for determining similarity groupings not immediately apparent either in listening or in comparing scape plots.

These methods—graphical, statistical, and network—are applicable because human activity are a wide-ranging set of behaviors that one can quantify. As part of music, which is fundamentally a chronological art, performance practices are ideally suited for such analyses, both theoretically and practically. Chapter 2 thus outlines the selection of a case study and the methods for use in analyzing performance in its own right, as a contributor to tradition, and in relationship to other performances and the evolutionary trajectory of its tradition.

Chapter 3 exhibits new modes of analysis and communication about performance. It focuses on the numerical data of performances, rendering them identifiable as facts, composed of varying numbers and degrees of sometimes highly technical parts. Visualization of this data creates a hierarchical graph, analysis of which reveals more expediently and more precisely information about the musical events of a given performance than any amount of close listening can provide. Chapter 3 examines the measurable tradition’s origin point—04 March 1961—and currently measurable

23

terminus—07 November 2009 as bookends of the tradition of performing Turandot at the Met. Chapter 3 concludes with an analysis of each of the six excerpts from various points in Turandot’s recorded performance history.

Chapters 4 offers Time Scape analysis of quantitative models of a performance tradition in relationship to its chronological progression from 1961 – 2009 for three excerpts. It begins by tracing a performance-by-performance evolution of a resulting tradition for one excerpt, then by following the decade-by-decade progression for another excerpt, and ends with a survey of the entire tradition’s evolution for a third excerpt. These analyses combine to affirm tradition as an incrementally dynamic entity whose sometimes seemingly minor variances from one performance to the next can significantly shape it in the future.

Chapter 5 also uses Time Scape analysis to consider tradition, but does so with respect to individual singers, conductors, and the production history of Turandot for extant recorded performances, respectively for each of the three excerpts not addressed in Chapter 4. This approach affirms that the more times a performer (singer or conductor) appears in a performance history, the more likely they are to shape a performance tradition. Furthermore, it also suggests the emergence of a sub-tradition resulting from Nello Santi’s nearly ten-year reign as the de facto conductor for Turandot at the Met. Finally, it suggests that production practices might influence the way a tradition evolves.

Chapter 6 demonstrates how network analysis can further express the nature of tradition, especially in how it differs from convention. Furthermore, network analysis offers a variety of ways to characterize how performances and the performers who

24

make them relate to each other and to the traditions they create. This approach also illustrates how the resulting revelations of multi-tiered data analysis can newly and succinctly illuminate a performance history and its most significant participants. The data-rich and data-driven analyses in Chapter 6 underscore the premises of tradition that Chapter 1 establishes below. In sum, tradition is a central component of a performance history, and it functions like a reflective repository of memories for every performance that creates it and is also thus contained in it. Furthermore, Chapter 6 confirms that Nello Santi, Franco Corelli, Plácido Domingo, Gwyneth Jones, Cecil

Beaton, and Franco Zeffirelli were the most influential members of the operatic art world in terms of shaping the ways in which Turandot’s performance tradition evolved over fifty seasons. Chapter 7 considers how the methods demonstrated within the preceding chapters enhance knowledge structures of musical works and poses questions for future consideration. Some of these questions will address other aspects of musical expression in operatic performance, while others will investigate links between other aspects of a performance history—such as place, world events, and so on—and onstage behaviors. All avenues for future inquiry that this dissertation highlights or creates anew will benefit from the use and continued development of the digital methods at its core.

Tradition’s Origins

Philosopher Hans Georg Gadamer situates tradition as the result of human activity. “Tradition is not simply a permanent precondition; rather, we produce it ourselves inasmuch as we understand, participate in the evolution of tradition, and

25

hence further determine it ourselves.”7 Tradition is synchronically pre-extant and continually emergent, and the result of human endeavor. This duality of its nature helps elucidate how it occupies a seemingly unilateral place in creative, cultural, or technological enterprise. As Hans Georg Gadamer stipulates, musicians, dancers, painters, sculptors, actors alike all interact with something predating them, but whose ongoing identity depends upon their continued creative practices. Thus, tradition naturally implies a historical consciousness and yet inevitably derives from human behaviors as they accrue over time. Moreover, the fact that traditions evolve over time intimates that they are relational in nature.

In musicology, the understanding of tradition is most evident in the evolution of compositional styles with respect to eras, nationalities, or other cultural affiliations.

Composers’ activities have traditionally been the primary focus of interrogations of musical traditions because the score has been the privileged element in those inquiries.8 Unlike scores, performances could not be preserved before the advent of recording technology. Yet understanding a performance and the traditions attending to it is necessary for achieving a comprehensive understanding of a musical work. This reality is in large part because the score is the text of the work (and not the work itself), and what performers do with that text contributes to the work’s identity.9 Thus,

7 Gadamer, 305.

8 As Chapters 2 and 3 will demonstrate, the means to document a performance has only emerged alongside the technology to record music and then to analyze the inscribed or encoded contents.

9 Peter Kivy, Authenticities: Philosophical Reflections on Musical Performance (Ithaca, NY: Cornell University Press, 1995), 262.

26

understanding any given tradition regarding performance requires the understanding of performance itself.

Tradition is more than a mere amalgamation of a series of performances. For some it means that “[Performances] are all subject to the supreme criterion of “right” representation.”10 Given the non-universality of human experience, this seems an untenable perspective. More realistic is Gadamer’s acknowledgment that tradition

“stems from the production, the creation of a role, or the practice of a musical performance. Here there is no random succession, a mere variety of conceptions; rather, by constantly following models and developing them, a tradition is formed with which every new attempt must come to terms.”11 Notions of performance tradition in musicology have shared a long association with the ever-troublesome idea of authenticity. In line with Richard Taruskin’s admonition that, “The appeal to intentions is an evasion of the performer’s obligation to understand what he [or she] is performing,” performance tradition is not a set of strictures designed to control the minutiae of performance behaviors.12 If such strict accountability to the written text representing a musical work is the aim, then, the result is the reduction of “performance practice to a lottery” over which the “performer can exercise no control.”13 Coming to terms with an established set of conventions necessitates, as Gadamer postulates, assessing the

“rightness” or “wrongness” of a performance. However, convention is unlike tradition in

10 Gadamer, Truth and Method, 122.

11 Ibid.

12 Richard Taruskin, Text and Act: Essays on Music and Performance (Oxford: Oxford University Press, 1995), 98.

13 Ibid, 4-5.

27

that the former implies a sense of authenticity, whereas this document will demonstrate in chapters 4 and 5 that the latter results from performances as they relate to each other through time.

In the same way that traditional music theory allows for the objective examination of a score’s contents (harmony, meter, motive, and so on), analysis of performance increasingly emphasizes an objective approach. This development is recent. As Chapters 2 and 3 postulate, the comparative latency in examining performances in a parallel fashion to scores stems from the dilemma of each one’s essential nature. While performances have a permanent ephemerality, scores exist in respective stasis. Scores’ relative fixity comes from the existence of multiple versions— autographs, various drafts, first edition, Urtext edition—and thus none can exhaustively or definitively represent a musical work. The same is true for performances; none is the absolute representation of a musical work. Even so, a score is a tangible object whose contents facilitate analysis, provided one has the tools to employ to derive a meaningful result. Score study has led to the establishment of a canon of celebrated works, which attain such status by, among other means, the acclamation that they contain elements more worthwhile than others. Focusing on the ‘what’ of a musical performance, makes the work something one communicates, and the canon of works a static body of things to communicate. Performers must then strive to represent that body of things in the

“right” way.

Nicholas Cook offers an alternative to this idea, stating “Perhaps the canon might be defined as a set of works so familiar that they function more as medium than

28

message.”14 This view allows a parallel view of performances, in which they are various means of communicating a work, rather than any being the work itself. As Chapter 3 elucidates, neither the score nor the performance alone is the work, but both convey certain elements about it. As communicative modes then, both scores and performances become sorts of technologies. Those using or studying both of them should be aware of the necessity for openness to multi-modal expression and analysis.

Singularity of characterization can easily result in what Cook describes as the “false belief that language can circumscribe and contain reality, from which it follows that what cannot be said does not exist.”15 Thus, like the multivalent nature of a work, the traditions surrounding it are also multivalent, and no single approach to analyzing or describing either is entirely adequate. For this reason, as Chapters 3 and 4 exhibit, non- language based, visual representations provide an innovative medium for examining performances. Better understanding of what a work’s performance is, together with its constituent parts, allows more accurate identification and analyses of its traditions.

Tradition’s Characteristics

The following survey of tradition’s many characteristics serves as a useful framework not only for understanding its linguistically derived properties, but also for providing qualitative lenses through which viewing large data sets becomes less daunting. Theodor Adorno suggested that “tradition is…not consciousness but the pregiven, unreflected and binding existence of social forms – the actuality of the

14 Nicholas Cook, “Analysing Performance, Performing Analysis,” Rethinking Music (Oxford: Oxford University Press), 245.

15 Ibid, 256.

29

past…”16 He elsewhere stated that “by the very fact of its existence, tradition claims that temporal succession sustains and transmits meaning.”17 Thus, tradition connects behaviors in the past to those in the present and future, and by its very existence provides a context for the practices in each era. Also, because time progresses steadily, filmmaker Hollis Frampton’s characterization of the evolution of eras and their attending behaviors is a useful reminder that tradition is a process that informs modern perceptions of an inherited past:

As one era slowly dissolves into the next, some individuals metabolize the former means for physical survival into new means for psychic survival. These latter we call art. They promote the life of human consciousness by nourishing our affections, by reincarnating our perceptual substance, by affirming, imitating, reifying the process of consciousness itself.18

For Frampton, tradition links artistic behaviors, connecting the very nature of consciousness and humanity from generation to generation.

Tradition, is perhaps first and foremost, historical, thus existing as a kind of afterthought. For this reason, Bernard Stiegler argues is interchangeable with

ēpimētheia. This term is an etymological derivative of the mythological character of

Epimetheus, whose name can mean hindsight or after-thinker.19 Ēpimētheia is “always already there,” “faithful to traditional historiality,” and therefore “It is its past, whether

16 Theodore Adorno, Essays on Music, trans. Richard Leppert (Berkeley, CA: University of California Press, 2002), 78.

17 Theodor Adorno, “On Tradition,” in Telos, 94 (Winter 1993-94), New York: Telos Press, 81.

18 Hollis Frampton and Bruce Jenkins, On the Camera Arts and Consecutive Matters: The Writings of Hollis Frampton (Cambridge, MA: MIT Press, 2009), 135.

19 Epimetheus’s brother was Prometheus, who stole fire from Mount Olympus to give to humans, and whose name can mean foresight or fore-thinker.

30

explicitly or not.”20 Moritz Epple clarifies the distinction between historiality and history or historicism: “The kind of temporality that appears here is not based on ‘historical’ or

‘objective’ time. It is the plural temporality of individual research episodes, a kind of temporality that Rheinberger described by invoking Jacques Derrida's term

‘historiality.’”21 For Stiegler, tradition simply exists given the forward trajectory of chronology and events that occur along its path. Igor Stravinsky alluded to this progressive nature, asserting that “Far from implying the repetition of what has been, tradition presupposes the reality of what endures.”22

Somewhat less philosophically, anthropologist Allan Hanson suggests that while tradition is historical, it is a social construct with pointed intentionality. “Tradition is now

‘understood quite literally to be an invention designed to serve contemporary purposes’.”23 Richard Taruskin summarized contemporary misuse of tradition as “an attempt to read the present in terms of the past by writing the past in terms of the present.”24 While this characterization alludes to Taruskin’s primary concern with rectifying the erroneous conflation of tradition with authenticity, it also reveals tradition’s simultaneous presence in past and present. Inasmuch as a tradition is historical, Igor

Stravinsky asserted that it also “is the relic of a past that is irretrievably gone; it is a

20 Bernard Stiegler, Time and Technics 1: The Fault of Epimetheus (Stanford, CA: Stanford University Press, 1998), 207.

21 Moritz Epple, "Between Timelessness and Historiality: On the Dynamics of the Epistemic Objects of Mathematics," Isis 102 (September 2011), Chicago: University of Chicago Press, 487. 22 Igor Stravinsky, Poetics of Music in the Form of Six Lessons (Cambridge, MA: Harvard University Press, 1947), 57.

23 Taruskin, 178-9.

24 Ibid, 179.

31

living force that animates and informs the present.”25 Furthermore, Adorno’s conceptualization of tradition’s historically relational nature stipulates that it “keeps music from being ‘utterly impoverished, abstract, and in the most real sense lacking in essence’.”26 Thus, tradition’s historiality provides a portion of a work’s richness, realness, and even worth.

Because of its historiality, tradition links the past to the present and both to the future. This connection occurs through transmission, as Craig Stuart Sapp indicated:

“The unwritten rules of a composition are transmitted aurally between performers as well as passed down from teacher to student. These performance conventions can apply to specific pieces, genres, composers or entire time periods.” 27 Sapp’s conceptualization echoes Stravinsky’s in that tradition connects past and present as a gift and responsibility from former generations to present and future ones: “It appears as an heirloom, a heritage that one receives on condition of making it bear fruit before passing it on to one’s descendants.”28 For Will Crutchfield, it is “a means for keeping

[things, practices, and so on.] connected while they change.”29 Gadamer qualifies this connection by also acknowledging how tradition enables the maintenance of a unique identity within an expanding network or artistic creations: “Not only does a work of art never completely lose the trace of its original function which enables an expert to

25 Stravinsky, 56-7.

26 Adorno, Essays on Music, 98.

27 Craig Stuart Sapp, “Computational Methods for Analyzing Musical Structure” (Ph.D. Diss – Stanford University, 2011), 111-112.

28 Stravinsky, 57.

29 Will Crutchfield, “What is Tradition?”, in Fashions and Legacies of Nineteenth-Century Opera (Cambridge: Cambridge University Press, 2010), 244.

32

reconstruct it, but the work of art that has its place next to others in a gallery is still its own origin.”30 Tradition’s connectivity links more than diverse temporalities; it links the people in each to each other and to those across time. As Adorno observed, it “has a kind of social-aesthetic dimension to the extent that it binds person to person; it acts as a force of reconciliation.”31 This observation mirrors Stravinsky’s: “Tradition thus assures the continuity of creation.”32

Tradition’s connective property comes through its resultant development as an inclusive, accrued aggregate. In line with Walter Benjamin’s attack on historicism,

Adorno admits that tradition is not exclusionary, remembering “only that which responds to the requirements of the elite and powerful,” who seek to use it “to affirm the here and now.”33 Instead, the tradition of a work accounts for all creative acts associated with it.

In the realm of performance, tradition is an aggregate resulting from performers. Craig

Sapp adopts a broad view of performers and their interpretive inputs, claiming that their processes “may involve combining interpretations from several sources, such as teachers or other admired [interpreters].”34 Even if a particular performer reacts against a convention or set of conventions, in so doing, he or she still acknowledges the reality of the amalgamated behaviors of his or her artistic predecessors. Adorno claims that the

30 Gadamer, 124.

31 Adorno, Essays on Music 78.

32 Stravinsky, 57.

33 Adorno, Essays on Music, 40.

34 Sapp, 111-112.

33

antithesis of (false) tradition as an exclusionary entity “is to keep faith with true universality.”35

Tradition’s embodiment at any time or place depends also on the context of the activities that create it. The ability to identify, reify, and analyze tradition likewise depends on the context of its creating events’ technosphere—the realm of human technological activity and the technologically modified environment. What scholars can know about traditions of performing Mozart’s operas is, in some senses, limited by the lack of technology to mechanically record and reproduce sounds of Mozart’s time from the Burgtheater or the Theater auf der Wieden. Conversely, what scholars can know about traditions of performing Jake Heggie’s operas might be in some ways far greater.

However, the performance history of Die Zauberflöte or Le nozze di Figaro outweighs that of Dead Man Walking or Moby Dick. Thus, what scholars can know about each has more to do with difference than with volume, because diverse eras, locations, and technologies all affect the composition and performance of musical works.

Much of the extant literature on performance tradition classifies it according to varying mixtures of composer, periodicity, nationality, and gender. Moreover, most of this work does not engage the specific sonic events of a given performance beyond what notation and written language can convey. Many studies exemplify this paradigm for Italian opera in the long nineteenth century. Scholars have explored Verdi’s works in relationship to his compositional process and conception of music and drama with regard to performance (Martin Chusid) or relevance for modern audiences (Alison

35 Adorno, Essays on Music, 82.

34

Latham and Roger Parker).36 Lorenzo Bianconi and Giorgio Pestelli’s three-volume investigation of the history of Italian opera provides an overview of the people responsible for bringing an opera to the stage, the evolution of staging and spectacle, and opera’s role in Italy’s popular and national culture.37 Evan Baker adopts a similar scope to focus strictly on the visual style and set design and the behind-the-scenes workers whose efforts bring opera to the stage, focusing far less on the sonic results of the performers’ realizations.38 Francesco Izzo’s study of Italian comic opera at the height of the Risorgimento proffers the importance of this genre’s ongoing tradition to

Italian national identity during Italy’s emergence as a unified political state.39

Accounting for singers’ cultural or sociological roles and their musical behaviors onstage, Philip Gossett provides a view of performance traditions that covers vocal style, ornamentation, transposition, translation, and adaptation. His musical concerns relate to which scores or tuning systems performers used, but in respect to notation, not sonic events.40 Hilary Poriss’s study of insertion arias treats the music that singers sang as interchangeable blocks, examining it at a broader level than the note-to-note or beat-

36 Martin Chusid, ed. Verdi’s Middle Period: Source Studies, Analysis, and Performance Practice (Chicago: University of Chicago Press, 1997); Alison Latham and Roger Parker, eds. Verdi in Performance (Oxford: Oxford University Press, 2001).

37 Lorenzo Bianconi and Giorgio Pestelli, Opera in Theory and Practice, Image and Myth (Chicago: University of Chicago Press, 2003); Lorenzo Bianconi and Giorgio Pestelli, Opera on Stage (Chicago: University of Chicago Press, 2002); Lorenzo Bianconi and Giorgio Pestelli, Opera Production and Its Resources (Chicago: University of Chicago Press, 1998).

38 Evan Baker, From the Score to the Stage: An Illustrated History of Continental Opera Production and Singing (Chicago: University of Chicago Press, 2013).

39 Francesco Izzo Laughter Between Two Revolutions: Opera Buffa in Italy, 1831-1848 (Rochester, NY: University of Rochester Press, 2013).

40 Philip Gossett, Divas and Scholars: Performing Italian Opera (Chicago: University of Chicago Press, 2006).

35

to-beat relationships embedded within each aria.41 Karen Henson’s analysis of four singers from the fin-de-siècle situates them amidst the cultural phenomenon of fandom based on what they sang, but does not delve into specifics of the sonic events of any of their performances.42 Similarly, Rachel Cowgill and Hilary Poriss trace the reception history of several nineteenth-century singers as a means of examining the development of the prima donna culture.43

Two studies mark a turn toward investigating the unique sonic events in a performance history. Roberta Marvin and Hilary Poriss’s edited volume seeks to examine transformations of musical works from their premiere, endeavoring to identify causes and effects with respect to geographic, performative, temporal, technological contexts.44 A contribution from Will Crutchfield examines the very notion of tradition, using the elongation of the highest note in Turandot’s “Nessun dorma” as an example of performers ignoring a composer’s intent in order to cater to the tastes of a different era.45 Roger Freitas has also investigated singers’ realizations in the context of establishing composer intention and cultural norms associated with performing Verdi’s operas in different times from either their creation or even the composer’s lifetime.46

41 Hilary Poriss, Changing the Score: Arias, Prima Donnas, and the Authority of Performance (Oxford: Oxford University Press, 2009).

42 Karen Henson, Opera Acts: Singers and Performance in the Late Nineteenth Century (Cambridge: Cambridge University Press, 2015).

43 Rachel Cowgill and Hilary Poriss, eds., The Arts of the Prima Donna in the Long Nineteenth Century (Oxford: Oxford University Press, 2012).

44 Roberta Montemorra Marvin and Hilary Poriss, Fashions and Legacies of Nineteenth-Century Italian Opera (Cambridge: Cambridge University Press, 2010).

45 Crutchfield. See footnote 26.

46 Roger Freitas, “Towards a Verdian Ideal of Singing: Emancipation from Modern Orthodoxy, Journal of the Royal Musical Association 127 (2002): 226-57.

36

As this corpus of studies demonstrates, performers play an active role in the cultivation and development of a tradition, and the ways in which they do depends on productions, directors, cast members, the acoustics of the theater, and the social milieu of the opera-going public which patronizes them. Each of these studies underscores, even if indirectly, Gadamer’s explanation of the context-dependent nature of traditions and the events that create them.

Every age has to understand a transmitted text in its own way, for the text belongs to the whole tradition whose content interests the age and in which it seeks to understand itself. The real meaning of a text, as it speaks to the interpreter, does not depend on the contingencies of the author and his original audiences. It certainly is not identical with them, for it is always co-determined also by the historical situation of the interpret and hence by the totality of the objective course of history.47

Like human nature, tradition is enigmatic because it results from and accounts for all of the individuality and uniqueness of a performer or a group of performers. Adorno stipulated this property thus: “Sustainable tradition is hardly ever a straight, unbroken, self-assured continuation or sequel.”48 This component of tradition or convention exists according to Gadamer because “…In tradition there is always an element of freedom and of history itself…It needs to be affirmed, embraced, cultivated.”49 As such, it “cannot be denied, only deplored.”50 Gustav Mahler famously declared that “Tradition ist

Schlamperei!,”51 while Richard Strauss lamented that tradition is merely “the last bad

47 Gadamer, 307.

48 Adorno, Essays on Music, 155.

49 Gadamer, 293.

50 Taruskin, 183.

51 Crutchfield, 240. “Tradition is sloppiness!”

37

performance.”52 Arturo Toscanini famously derided tradition, highlighting its non- conformity by characterizing it for the NBC Symphony thus, “The first asino, the first jackass, did it that way, and everyone follow [sic] him!”53 Mahler, Strauss, and Toscanini all despised what they viewed as capricious interpretations. For Igor Stravinsky, such instances are akin to “habit, even…an excellent one, since habit is by definition unconscious acquisition and tends to become mechanical.”54 Interpretations deemed capricious, aberrant, or not in keeping with the composer’s wishes or intent are, in fact, part of the tradition of that work. These instantiations are those which have “been ‘left along the way,’ [those] which [have] been forgotten or dismissed as outdated,” what

Theodor Adorno “elsewhere names as ‘scars’.”55

Performer views of performance tradition, especially with respect to nineteenth- century Italian opera, reflect much of the foregoing analysis while also offering unique perspective on its practical implications. These perspectives are a vital component of the current analysis because performers’ onstage behaviors are the performances that create a tradition. For tenor Stephen Costello, “some Bel Canto cadenzas have become more standardized over time, marking a departure from their more improvisatory roots.”56 For soprano Andréa Gruber, “Tradition is there for a reason. It’s kind of like an

52 Taruskin, 183.

53 Joseph Horowitz, Understanding Toscanini: A Social History of American Concert Life (Berkeley, CA: University of California Press, 1994), 254.

54 Stravinsky, 56.

55 Adorno, Essays on Music, 81.

56 Stephen Costello, interview with the author on 19 January 2015. He estimates that the cadenza in Edgardo’s concluding aria in Lucia di Lammermoor is approximately 90-95% the same among contemporary singers.

38

old blanket,” but it “is not just doing what I’ve heard others doing; it has to have a purpose.”57 Tenor Marcello Giordani acknowledges that tradition can be influential as a standard of expectation, but that performers must “always try to find new things, even small details.”58 Soprano Kristin Lewis alludes to technology’s role in perpetuating notions of authenticity as tradition and echoes Giordani’s analysis that tradition, not authenticity, ensures continual creation: “People tend to rely on recordings that have become 'standards,' but not the best way, because they allow for an easier and more time efficient 'development' of an interpretation. Time has prevented continued

'development' and has allowed 'tradition' to stand.”59 All of this affects how performers interact with their onstage colleagues, conductors, the work itself, and the audience, for whom the balance of convention and innovation can be notoriously perilous to maintain.

As each of these singers alluded, tradition is not prescriptive. Taruskin argues emphatically against assimilating tradition into the quest for authenticity, thereby contradicting Gadamer’s notion that tradition carries with it the necessity to account for present actions with past. When this happens, according to Taruskin, whatever one elects as a standard of authenticity “visits us now like the ghost of Jacob Marley, weighted down by generations of accrued tradition…(misdeeds).”60 Adorno also attacks the use of tradition as a prescriptive force: “The appeal by artists to the past is fraught with danger precisely to the extent that modernity has rendered tradition a thing, ore to

57 Andrea Gruber, interview with the author on 29 January 2015.

58 Marcello Giordani, interview with the author on 8 July 2013.

59 Kristin Lewis, interview with the author on 9 July 2013.

60 Taruskin, 106.

39

be mined for future product.”61 Such a view immediately stratifies the instantiations of a musical work, which in turn facilitates disregarding performances outside of one’s realm of preference. It results, as Friedrich Nietzsche asserts, “a new past from which we would should have liked to descend instead of the past from which we actually descended.”62

Challenges to Understanding Tradition

The foregoing discussion provides one possibility for a linguistically characterized understanding of tradition, but it remains language-bound and theoretical rather than tangible. As such, it embodies the opacity and complexity of discourse about tradition.

This discourse reveals how current understanding of tradition resists reification, thus requiring the development of an analytical apparatus capable of responding to Stiegler’s exhortation to “provide an accounting for the causes of…diversity” that contribute to and reflect tradition at a fundamental level.63 Since tradition emerges from an accrual of performances, one must acknowledge the impossibility of understanding tradition without reference to what musicians are performing. Alternatively, it is impossible to understand tradition absent from the events that create it.

A longstanding and fruitful option is to use score-based analysis, as Will

Crutchfield, Roger Freitas, and Hilary Poriss have done with opera. Yet these analyses lack the specificity necessary to engage the analytical objects (performed musical works and their traditions) in a straightforwardly comprehensible manner. This inability lies in a

61 Adorno, Essays on Music, 79-80.

62 Ibid, 178.

63 Stiegler, 47.

40

score’s nature as a medium of representing or communicating musical thought rather than as being an object to communicate.64 In other words, the required interpretation inherent in understanding a score and what it expresses creates the same set of obstacles as purely language-based approaches—indefinite identification of what the object actually is. Both scores and performances fall under the Heideggerian category of technē, and thus serve to reveal or point towards the musical work. For example, any two people will perform (reify) the folksong “Hot cross buns!” differently, even if their interpretation of its notated (conceptualized) textual and musical elements is identical

(see figure 1-1).

Figure 1-1. Traditional folksong “Hot cross buns!”

Even so, both representations and the notation of Figure 1-1 of this folksong point towards the work itself. This presents an insurmountable obstacle for definitive, precise descriptions of what musical events took place if one relies only on score-based analysis. One solution lies in changing the discursive object from interpretable and therefore fluid entities into absolute ones. In other words, a more precise understanding comes by reducing the performance into its sonic phenomena and analyzing the absolute data of a performance—tempo (bpm), dynamics (decibels), pitch level (Herz), and so on—emancipated from the ambiguity inherent in scores and language.

64 See the discussion about the relationship between scores, performances, and works in light of Martin Heidegger’s theory of the relationship between technē and poiēsis at the beginning of Chapter 3.

41

Because of its nature as a chronologic art, just as painting is a spatial art, music’s primary ordering is temporal.65 In its realization in performance, music is at once ordered by and provides a form of order for humans. Performers order the temporal progression of realized music according to a number of factors, but they generally do so with respect to the musical ideas embedded in a score.66 This makes time a natural and universal metric for studying how human behaviors – Stiegler’s “tendencies” – change, both within individual performances of works, and performances of works across different chronological and ontological times.67 As the times and aesthetic atmospheres in which people perform music change and, due to the inability of humans to mechanically reproduce exactly the same performance on any two occasions, the traditions performers exhibit also change. As a numerically ordered art form in terms of its physical properties—tempo, Hertz, decibels—music is rich in data, and performed music’s atomic-level parts can be expressed by their numerical properties in relationship to time. Digital tools for audio signal processing are the most appropriate for the

65 Stravinsky, Poetics of Music, 28.

66 One useful analogy here is that of a recipe. For example, there are a number of ways to make a macaroni and cheese dish—various kinds of cheese, milk or roux based sauce, bread crumb topping, and so on; —this is the ordering by humans. Even so, without the basic elements of macaroni or cheese, a person cannot make such a dish; this is how this enterprise is ordering for humans.

67 While illuminating to consider with respect to musical tradition, the differing kinds of time – chronological, psychological, ontological – remain outside the scope of the current project. Theorists have categorized time in a wide variety of ways. These include: messianic time, dead time, virtual time, the time of the event, and time out of joint. Other ontological designations include: creative time, creature-less time, untimely time, demented time, other time, the time of the Other, absolute time, the time zone of the inhuman, generative time, empty time, and quasi-time. Similarly, chronological time has a number of equivalents – linear time, phenomenological lived time, human time, ordinary time, clock time, calendar time, actual time, historical time, generated time, and objective time. For a fuller discussion of ontological time, see Serge F. Hein, “Thinking and Writing with Ontological Time in Qualitative Inquiry,” Qualitative Inquiry XX (X) 1–9 (Thousand Oaks, CA: Sage Publications, 2003), .

42

extraction of this data from recorded music. Further mathematical expression of musical properties is possible and is a primary aim of my future research.

This document offers a new possibility for addressing the limits of how scholars currently engage the concepts of musical performances in relationship to musical works and the tradition their existence creates. In so doing, it adds digital techniques to extant analog methodologies, which proffers more comprehensive analysis and understanding of musical performances and the works of which they are iterations. The new objects this document creates expand the discussion of performance and tradition into the realm of images, thereby broadening and deepening the knowledge scholars, performers, and the public alike can have about an integral part of the creative process.

These images are documents in their own rights, capable of communicating in a way different from that of either text or music (in score or sounded form) can. This multivalent expansion is a move toward the three-fold model for which Roland Barthes advocates and which emerges in Image-Music-Text.68 Such a shift requires the integration of quantitative data about sounded music’s physical properties into the longstanding conceptual and qualitative discussion about the meaning of performance practice. This inclusion enables the employment of all levels of statistical analysis about individual performances, their traditions, and how they interrelate. Digital tools and techniques make all of the new analysis in this document possible, further demonstrating their benefit to musicological enquiry. The primary objective of this research is to enhance the study of performance, treating each instantiation of a musical work as a new work of art connected by common text and historical precedent.

68 Roland Barthes, Image-Music-Text, trans. Stephen Heath (London: Fontana Press, 1977).

43

This approach to the study of tradition responds to Adorno’s challenge of “[remaking] the meaning of past and present alike.”69

69 Adorno, 81.

44

CHAPTER 2 CASE STUDY PARAMETERS AND METHODS

Traditions inhabit a foundational role in operatic composition, production, and performance. The precise analysis of performance tradition benefits substantially from documentation of performances via recording technologies. The works of well-known composers who were active as recording technology developed and as the recording industry emerged thus make ideal case studies.1 Of these composers active at the beginning of the twentieth century (including the giovane scuola),2 Puccini is uniquely suited to a study of tradition through technological means: he has perhaps the best- documented relationship to advances in technology and the resulting shift in entertainment aesthetics of any composer of this era.3 Of his twelve operas, the final one, Turandot, is the most appropriate for this kind of inquiry because of its position at the crossroads of technical practices and social aesthetics stemming from technological conditions in multimedia entertainment in the 1920s. This chapter begins by establishing

Turandot as the ideal operatic case study, sets forth logical parameters for investigation, and concludes with an explanation of the methodology.

1 Four operatic composers emerge as strong possibilities for their individual renown and respective careers having overlapped the emergence of the recording industry: Erich Wolfgang Korngold, Pietro Mascagni, Richard Strauss, and Giacomo Puccini.

2 The giovane scuola consisted of Italian composers whose careers straddled the nineteenth and twentieth centuries, and whose operas demonstrate strong influence from literature’s verismo movement. Prominent composers of this group include Ruggero Leoncavallo (Pagliacci), Pietro Mascagni (Cavalleria rusticana, Iris, L’Amico Fritz), Umberto Giordano (Fedora, Andrea Chenier), and Giacomo Puccini.

3 Budden, 476.

45

Case Study Parameters

Turandot: A Brief Overview

A brief synopsis of Turandot, which premiered in 1926 and is the last Italian opera to enter the international repertoire on a permanent basis, provides dramatic context for each of the six excerpts included in this study. This context is particularly valuable for consideration of the creative process, of which Chapter 3 offers some examples, and of how performer perspective on these dramatic moments further shapes their respective performance traditions. Act 1 opens with a Pekinese court official reading a decree that because the Prince of Persia failed to correctly answer

Turandot’s riddles, he must surrender his head to the swordsman. As the primary protagonists—Timur (blinded and deposed king of Tartary), Liù (slave and attendant to

Timur), and Calaf (Timur’s son, though his name is unknown to all but Timur and Liù)— enter the action, they reunite in exile and bemoan the horror unfolding before them.

Calaf is particularly vocal in denouncing Turandot’s brutality until he catches a glimpse of her, at which point he swoons and declares his intention to attempt her riddles in order to win her hand in marriage. Liù pleads with him not to take such a risk in her aria

“Signore, ascolta!” Calaf will not acquiesce, instead pleading with her to remain faithful to Timur should his attempt at the riddles fail in his aria, “Non piangere, Liù.” The act closes with Calaf striking the gong and announcing his desire to try the riddles.

Act 2 opens with three ministers commenting in their chambers to each other on the awful state of affairs in Peking. The scene then changes to the throne room, where

Turandot offers an explanation for her riddles and the severe consequence of failing to answer them in her aria, “In questa Reggia.” Immediately following this aria comes

“Straniero, ascolta!” (also known as the riddle scene), in which Turandot issues and

46

Calaf accurately responds to each of three riddles. Following Calaf’s success, Turandot pleads with her father, Emperor Altoum, to be freed from her vow to marry the man who triumphs over her intellectually. Calaf in turn offers his own riddle: if Turandot can learn his name by dawn, he will surrender to the swordsman, freeing her from her obligation.

Act 3 begins with heralds relaying Turandot’s commandment that none shall sleep in Peking until she knows Calaf’s name. Calaf reflects on this command, with the famous romanza “Nessun dorma!” as his monologue. When the crowd realizes that he had been speaking with Timur and Liù earlier, they attempt to extract the name from them via torture. Liù insists that Timur does not know the name, that only she does, and that she will not reveal it under any circumstance. The crowd summons Turandot, who orders more torture, bringing Liù to her breaking point. Liù tells the princess that love gives her the fortitude to remain quiet. Realizing that she can endure no more, Liù makes a final address to Turandot, “Tu che di gel sei cinta,” before grabbing a guard’s dagger and stabbing herself. Calaf and Turandot then have a final confrontation in which Calaf kisses the princess, melting the last vestiges of ice around her heart, and then reveals his name to her. When she announces that she has learned his name, she declares that it is love, and the opera closes with a reprise of the main theme from

“Nessun dorma!”

Turandot’s Technoshpere

An overview of the technosphere—the technological atmosphere and behaviors that mark a given era—that multimedia entertainment occupied in the early twentieth century will clarify Turandot’s appropriateness as a subject for technologically driven analysis of tradition. Examining the history of opera in conjunction with technology reveals that many prominent technological advances—acoustic ducting, binoculars,

47

electronic sound transmission, and motion pictures—came about, at least in part, because of opera.4 A brief overview of these developments appears in Table 2-1.

Figure 2-1. Emile Berliner’s direct, hand powered device for making sound recordings.5

Closer chronologically to Turandot, nineteenth-century inventors developed various technologies to capture and transcribe sound waves into multiple storage media—first as etchings in smoke-plated glass, then in tin-foil, wax, shellac, magnetic strip, and, finally, electronic coding. In addition to these advances, the machines used for transcribing have undergone parallel advances. The earliest phonographs and gramophones required hand cranks to provide the necessary energy for sound transcription to occur.6 This led to a variance in recording and playback speeds, as

“even the most practiced operator could not keep the turntable speed constant.”7 Figure

4 Mark Schubin, “The Fandom of the Opera: How Opera Helped Create the Modern Media World,” unpublished presentation at the Library of Congress, Washington, DC, 06 October 2011.

5 Nicholas Cook, The Cambridge Companion to Recorded Music (Cambridge, UK: Cambridge University Press, 2009), 153.

6 In the earliest days of the phonograph and gramophone, the hand crank directly turned the disc for both recording and playback. Later, the hand crank stored kinetic energy in the mechanical workings of a device that in turn powered the rotation of the disc or cylinder, akin to the process of music boxes storing and then exhausting energy for musical playback.

7 Roland Gelatt, The Fabulous Phongraph: 1877-1977 (New York, NY: MacMillan Publishing Co., Inc., 1977), 83.

48

2-1 illustrates Emile Berliner’s device for two-step record production with the hand- driven mechanical apparatus clearly visible.

Table 2-1. Historical Overview of Opera and Technology8 Date Development 1581 Vincenzo Galilei publishes a book containing principles and acoustics of opera as discussed by Florentine Camerata 1619 Drawing upon Galilei, Johannes Kepler posits and publishes 3rd law of orbital motion, which is evident in modern satellites 1637 Opera houses begin to use acoustic ducts 1673 Athanasius Kircher proposes ducting music outside the opera house 1726 Thomas Lediard uses moving-image projection at Hamburg opera 1730 Collapsible telescope (monocular) in use in London 1823 Voigtländer company sells binocular opera glasses 1849 Antonio Meucci experiments with electronic sound transmission at the Gran Teatro Tacón opera house in Havana 1849 Jules Dubosq projects an electric-light sunrise effect at the Paris Opéra 1880 Edward P. Fry listens to opera via telephone 1881 Clément Ader transmits stereo sound from Paris Opéra 1881 British patent filed for gas-jet-illuminated libretto display in opera house 1882 Albert Robida’s Le Vingtième Siècle predicts (& illustrates) opera delivered to homes via “téléphonoscope” 1885 Based on 1884 telephone transmission, opera pay-cable service opens in Lisbon 1888 Edison’s motion-picture patent caveat declares it is for opera 1893 Tivadar Puskás creates electronic newscast to increase utilization of opera-by- phone lines 1894 William Dickson plays music in first synchronized-sound movie 1895 Opera via headphones (available in Lisbon since 1888) offered by Electrophone Electrophone 1896 First operetta films shown in Germany; Georges Hatot's Faust, released by Lumieres in 1897, based on Gounod's opera 1898 La fille du regiment (Donizetti) is the 1st filmed opera (two minutes) to be shown; Martha (Flotow) is shot on film 1899 Martha is shown at the Eden Musée with live performers singing behind the screen; technique popular though 1922 cinema opera Jenseits des Stromes (Hummel) (with projected score at bottom of screen; other systems for cueing musicians in opera movies used small conductor image or signal lights built into the set) 1900 Horace Short broadcasts opera from the top of the Eiffel Tower via a compressed air amplifier; Lionel Mapleson records from a non-interfering location (the wings of the Metropolitan Opera), creating some of the earliest live performance recordings 1903 Ernani (Verdi) becomes the first “full-length” (abridged) recording of a complete opera, consisting in forty discs (eighty sides)

8 Schubin. Transcription of chart included in handout for this lecture, used with permission.

49

Table 2-1. Continued Date Development

1907 Radio transmission of opera singer leads to U.S. Navy use of radio as communication technique; “complete” Faust (22 arias) film using Chronophone; Caruso records “Vesti la giubba” (Pagliacci) a 3rd time and it becomes the first million copy record (Platinum level according to RIAA) 1913 First complete opera shot in U.S. with synchronized sound, Pagliacci (Leoncavallo) 1915 Geraldine Farrar becomes silent-movie star based on success onstage as Carmen 1919 Hugo Gernsback proposes live sound, distributed by radio, to cinemas projecting silent opera movies; Opera radio broadcasts in Chicago and New Brunswick, NJ via military transmitters, latter heard live 2000 miles out to sea 1925 Electronic recording and playback technologies emerge, standardizing techniques for both practices 1928 Washington Post reports opera singers on television 1931 Metropolitan Opera broadcasts via radio for the first time, Englebert Humperdinck’s Hansel und Gretel 1952 Carmen (Bizet) transmitted live to cinemas in 27 cities 1976 First commercial digital recording is opera 1977 Metropolitan Opera broadcasts via television for the first time, Giuseppe Verdi’s Rigoletto 1986 Opera Basel first plazacast of Gaetano Donizetti’s Lucia di Lammermoor 1995 Houston Grand Opera first plazacast 2006 Met HD plazacast to Times Square, international and HD cinema transmission 2009 Opera de Rennes broadcasts Wolfgang Amadeus Mozart’s Don Giovanni live to cinemas in 3-D 2009 International, digital, high-definition, satellite, multi-language-subtitled, surround- sound, live cinema versions of opera

Spring motor gramophones appeared on the market in the fall of 1896 and sales quickly outpaced those of the twice-as-expensive spring motor Edison phonographs.1

As Figure 2-2 illustrates, early (pre-1920s) equipment for recordings and playing music had purely mechanical means of storing and using energy to drive the turnstile. While more reliable and regulated than a hand-cranked device, as the springs approach the end of their stored tension, rotation speeds dropped and required re-tensioning via a winding key. Frequently recharging the potential energy in the springs resulted in some

1 Gelatt, 86.

50

inconsistencies of rotation speed, whether recordings or playing, making matching recording and playback speed difficult. Many of these discs, which Caruso helped popularize, were recorded in the 68-70rpm range.2

Figure 2-2. Three-spring (two in the larger spring barrel, one in the smaller) drive mechanism for powering the rotation of a disc or cylinder. This required the turning of a winding key to store energy in the springs.

Apart from the invention of sound transcription, perhaps the most important development is that of electronic recording and playback technologies in 1925.

Electrical recording via telephone technology for the transformation of acoustical sound waves into electrical impulses and back again came first.3 Electric motors for turning the

2 Roger Beardsley, “Speeds and Pitching of 78rpm Gramophone Records,” AHRC Research Centre for the History and Analysis of Recorded Music, accessed 24 August 2014, http://www.charm.rhul.ac.uk/history/p20_4_3.html.

3 J.P. Maxfield and H.C. Harrison, High Quality Recording and Reproducing of Music and Speech: Methods of High Quality Recording and Reproducing of Music and Speech Based on Telephone Research (New York: Bell Telephone Laboratories, Inc., 1926). This text outlines the entirety of the proposed method, including calculations of frequency requirements, electrical energy requirements and translation characteristics, and circuitry and acoustical diagrams. While this publication dates to 1926, it is a descriptive account of technologies and practices already in existence and use, not of theoretical future possibilities.

51

disc came with the Brunswick Panatrope, built by the Brunswick Company, and provided constant energy for rotation.4 These developments led to increased standardization of machine speeds to the point of the recording industry claiming a standard of 78rpm. There was, however, some variance resulting from the setting of speeds for both functions by human operators, and the industry seemed to be aware of it as machines included speed adjustment dials.

The introduction of electrical recording in 1925 should have simplified matters. The universal recording system was the Western Electric process. . .All the recording and replay characteristics were carefully calculated and controlled but not the speed, even though much of the recording world recognized 78rpm as the standard. Many early electrical Victors and their HMV equivalents were recorded well below 78. . .Not all records continue and finish at the speed at which they started. The recording machine could vary its speed considerably during a side, sometimes by 5rpm.5

Despite the occasional problematic recording, the merging of electrical and recording technologies enabled greater and more nuanced control over the recording and playback process than was possible in the acoustic era.6 Sound quality and mechanical limitation of pre-1925 recording and playback devices posed significant challenges to creating accurate or even aesthetically pleasing records. The sound quality improved dramatically as Roger Beardsley has noted: “the whole ambience and feel of a performance and its surroundings was reproduced. The nearest analogy is that

4 Pekka Gronow and Ilpo Saunio, An International History of the Recording Industry (London: Cassell, 1998), 55.

5 Beardsley, “Speeds and Pitching.”

6 Some early Pathé recordings were recorded at c. 100rpm. Even discounting these examples, the variance on early recordings, recorded anywhere between 60 and 80rpm, could be as much as 10 rpm above or below the quoted recording speed. At most, this variance could be anywhere from 12.5% (for a disc recorded at 80rpm) – 16.667% (for a disc recorded at 60rpm) out of sync. In comparison, electrical recordings might vary up to 6.4% for a disc recorded at the proclaimed standard of 78rpm.

52

of hearing a performance of say [sic] an orchestra though a closed door and down a corridor, and then being brought into a box at the venue.”7 Beardsley also notes that the difference was the result of subjecting “recording (and reproduction). . .to a proper system of scientific research, as against the largely empirical developments of the mechanical recording system.8

The new technology found a perhaps unexpected proponent in John Philip

Sousa, who had once derided the phonograph as a failed “substitute for human skill, intelligence, and soul.”9 Sousa once prophesied “a marked deterioration in the musical development of the country, and a host of other injuries to music in its artistic manifestations, by virtue—or rather by vice—of the. . .reproducing machines.”10 This view seemed to change when Sousa heard the electronically-recorded-and-played-back recording of the Soldier’s Chorus from Charles Gounod’s Faust in the fall of 1925.

“Gentleman (sic), that is a band. That is the first time I have ever heard music with any soul to it produced by a mechanical talking machine.”11 Despite these advances and the seeming approval of one of recording’s most prominent opponents, several years passed before Puccini’s final work materialized. For reasons yet unclear, Turandot’s first excerpted recordings, which happened to be of live performance live, did not emerge

7 Roger Beardsley, Introduction to “High Quality Recording and Reproducing of Music and Speech,” AHRC Research Centre for the History and Analysis of Recorded Music, accessed 24 August 2014, http://www.charm.rhul.ac.uk/history/p20_4_1_3.html.

8 Ibid.

9 Gelatt, 146.

10 Ibid., 146-7.

11 “New Music Machine Thrills All Hearers at First Test Here,” New York Times, 7 October 1925, accessed 31 August 2014, http://search.proquest.com/docview/103526509/6731A389EBD44538PQ/1?accountid=10920.

53

until 1934. The first full studio recording followed in 1938, and the first full-length live recording came in 1953.12

Roughly synchronous to advances in sound capture and reproduction experimenters developed the means by which to mechanically capture and preserve light waves, first in single frames, then frames in succession. This led to successive image capture and playback (motion pictures). Initially conceived of as a means for preserving and disseminating opera, motion pictures first transcribed sound and image in separate media, which in turn required resynchronization for playback.13 Despite earlier experiments of fusing sound and image into a single, permanent entity, the first sound film was Alan Crosland’s The Jazz Singer from 1927. As mentioned above,

Puccini’s Turandot premiered in 1926 and became the last opera of the Italian tradition to permanently enter the international repertoire. These two significant events occurring just one year apart are more than a convenient coincidence. Before Turandot and The

Jazz Singer, Italian film relied significantly on operatic imagery and style as its practitioners sought to develop idiomatic techniques for the emerging medium. In addition, early Italian film directors turned to prominent opera composers to supply the soundscapes for their films.14 Allan Mallach stipulates that some composers, including

Gabriele D’Annunzio, “embraced the cinema, an engagement that began with his

12 Roger Flury, Giacomo Puccini: A Discography (Lanham, MD: Scarecrow Press, 2012), 727, 709, 710. While the reasons for such a delay remain unclear for the present, waiting to see if Turandot would have any staying power before making a recording, and the tumultuous socio-political climate of the late 1920s and 1930s seem viable possibilities.

13 Schubin, “The Fandom of the Opera.”

14 Alan Mallach, The Autumn of Italian Opera: From Verismo to Modernism, 1890-1915 (Boston, MA: Northeastern University Press, 2007), 360. Puccini’s roommate from the Milan conservatory, Pietro Mascagni, supplied a score for the 1915 film Rapsodia satanica.

54

approving the cinematic adaption of his works, and culminated in his taking part in the making of Cabiria in 1914.”15 As cinemas began to populate Italy, they became the preferred gathering places of society’s elite, especially in Milan.16 Mallach also notes that eventually, and perhaps expectedly, opera in Italy was “on a path toward a repertory dominated by works of the past.” 17 He draws a clear contrast, asserting that

“Italian film studios were producing hundreds of new works every year,” and that these films “appealed to every social class or demarcation of taste.”18

A demonstrable shift away from opera and toward film occurred in the 1910s. In light of this fact, one might expect that Puccini’s works predating this decade exhibited tendencies reflective of both film and opera. Gianni Schicchi contains musical parallels to cinematographic techniques: direct cuts, cut-ins, and matting. Direct cuts are instantaneous transitions between one scene and the next. Cut-ins are the insertions into a film that interrupt the action before returning to it. Matting is a special effect which combines two or more images into a single image. Respective parallels in opera include: rapid alternations of musical style and stage action in the opera (direct cut); the emergence of the arias from the ensemble texture before returning to the action (cut-in); and Schicchi singing on top of, and independently from, Lauretta and Rinnuccio

(matting).19 Turandot does likewise, mirroring the cinematic dissolve in the transition

15 Ibid., 358.

16 Ibid., 359.

17 Ibid., 362.

18 Ibid.

19 Andrew Davis, Il Trittico, Turandot, and Puccini’s Late Style (Bloomington, IN: Indiana University Press, 2010), 156-63.

55

from the ministers’ trio to the riddle scene in act 2.20 Moreover, Puccini’s mosaic-like stitching together of scenes from many diverse pieces in La bohème, Madama Butterfly, and La fanciulla del West generated considerable criticism for what critics perceived as more appropriate to cinema than opera.21 Such similarity is perhaps expected because, as Alexandra Wilson indicates, “Puccini appealed to all classes, the workers as well as the bourgeoisie with whom he shared a special affinity.”22

Alongside the burgeoning overlap of opera and film, Puccini’s own fascination with technological gadgetry is well known. Arman Schwarz has shown that Puccini’s home in Viareggio contained a “number of new-fangled gadgets: the aerial for radio reception, the automatic sprinklers that operated from the trees. . .[and] the entrance door opened by remote control.”23 Alexandra Wilson’s portrayal of Turandot as a mechanical character in what Puccini described as “the most. . .human of Gozzi’s works” further distinguishes the overlapping of technology and humanity in Puccini’s life and work.24 Schwarz agrees, maintaining that “Turandot, after all, is herself a sort of

‘mechanical idol’. . .[even] the whole opera’s. . .relationship is embodied in. . .her extremes of mechanism and enchantment.”25 For Schwartz, Puccini, Turandot, and

20Ibid., 209.

21 Alexandra Wilson, The Puccini Problem (Cambridge: Cambridge University Press, 2007), 225.

22 Ibid., 186. Critics and audiences have long held that film’s more democratic essence makes it the most appealing multimedia art form of bourgeois audiences. See Mallach, Ch. 15 for a full discussion of the intellectual and social circumstances that led to the general populace embracing cinema and abandoning opera.

23 Arman Schwartz, “Mechanism and Tradition in Puccini’s Turandot,” The Opera Quarterly 25, no. 1-2 (2009), 29.

24 Giacomo Puccini and Eugenio Gara, Carteggi Pucciniani (Milan; G. Ricordi, 1958), 490.

25 Schwartz, “Mechanism and Tradition,” 45.

56

Turandot together embody a stark dichotomy between technological advances and aesthetic practice. This difference seems natural to Wilson, who asserts that “every technical renewal not accompanied by a corresponding evolution in feeling has the sole result of accentuating the contrast between the new form and the substance that no longer corresponds to it.”26 This juxtaposition of emerging technology and long-standing musical practice highlights a critical question about the fundamental differences between Puccini’s operas and film. Perhaps the most obvious is the liveness factor.

These differences inevitably raise questions surrounding the interactions between performance as musical practice, and technological practices within the technosphere and the social sphere.

The technosphere surrounding Turandot renders it the only opera appearing regularly on operatic stages whose premiere post-dated the advent of electrical recording and playback devices. Roger Flury’s recent compilation of the most comprehensive discography of Puccini to date focuses attention on recordings as historical documents of performances, and demonstrates the democratization possible through mechanical reproduction.27 Thus, Turandot is the most prominent opera whose performance history exists entirely in the era of standardized equipment. As a result, tracing Turandot’s performance history is easily attainable. Furthermore, the technosphere, which helped facilitate the rise of the bourgeoisie in the early- to mid- twentieth century enabled film to overtake opera in the social imagination. These events

26 Wilson, The Puccini Problem, 214. See also the broader discussion of Turandot as a mechanical woman, pp. 201-210.

27 This is the most comprehensive discography in existence of any composer, highlighting Puccini’s relationship to technologies still extant today.

57

solidify Turandot’s place as the last work in the 300+ year history of opera as the primary multimedia entertainment form. Thus, technology allows scholars to investigate the opera more closely as a work at the center of this dramatic shift in cultural aesthetics.

Place and Time

The performance setting for the study and musical phenomena within the opera are two further parameters necessary for consideration. Of the world’s major opera houses, New York’s Metropolitan Opera has perhaps the most comprehensively documented and holistically accessible performance history. This reality—one in which the entire current accessible corpus undergoes analysis—makes future comparative analyses of performance history with regard to place or a house style more efficient.28

The launch of “The Metropolitan Opera Archives” online with open access in 2005 resulted in the most accessible source of information for operatic performing history in any opera house in the world.29 By making this information publicly and readily available, the Met made significant strides in democratizing opera in a way similar to what recordings had done nearly a century earlier. With this new tool, fans now had the ability to deepen their learning and, by extension, their enjoyment of opera at the Met.30

Thus, by extending the activity of learning about opera audiences to the digital realm, the Met thrust itself into the forefront of the larger operatic social sphere.

28 Another possibility would be to consider all accessible performances of Turandot in a given year or frame of years. While this approach will certainly provide a more global perspective, and is an area for future analysis, it remains outside the scope of the current project.

29 Anthony Tommasini, “How to View 26,000 Operas at Once,” New York Times, 31 July 2005, accessed 02 September 2014, http://www.nytimes.com/2005/07/31/arts/music/31tomm.html?_r=0.

30 Claudio E. Benzecry, “Becoming a Fan: On the Seductions of Opera Qualitative Sociology, 32, no. 2 (2009), 134.

58

Table 2-2. Operas broadcast from The Metropolitan Opera more than 50 times Opera (Composer) Number of First Broadcast Latest Broadcast Broadcasts La bohème (Puccini) 85 01 Jan. 1932 18 April 2014 La traviata (Verdi) 82 20 Feb. 1932 03 April 2013 Der Ring des Nibelungen 72 26 Feb. 1932 11 May 2013 (Wagner)* Aida (Verdi) 70 02 April 1932 15 Dec. 2012 Rigoletto (Verdi) 70 18 Feb. 1933 07 Dec. 2013 Carmen (Bizet) 67 01 Feb. 1936 26 Feb. 2013 Tosca (Puccini) 63 12 Jan. 1910 28 Dec. 2013 Il barbiere di Siviglia (Rossini) 59 23 Jan. 1932 03 Jan. 2013 Madama Butterfly (Puccini) 58 25 Jan. 1941 05 May 2014 Don Giovanni (Mozart) 55 17 Dec. 1932 20 Dec. 2012 Die Zauberflöte (Mozart) 51 10 Jan. 1942 01 April 2014 Il trovatore (Verdi) 51 16 Jan. 1932 24 Jan. 2013 Lucia di Lammermoor 50 26 Nov. 1932 19 March 2011 (Donizetti) * This number accounts for the times portions of the Ring Cycle have been broadcast. Though it might initially seem to reflect 18 broadcasts of the complete Ring Cycle, this is not necessarily the case.

Prior to the archives becoming publicly available online, information on performance histories at the Met in published form existed primarily in Paul Jackson’s tripartite account of The Metropolitan Opera Broadcasts prior to 1976 and in a book with limited circulation that accompanied an exhibit at the Museum of Broadcasting.31

Perhaps sensing the interest in this history, and perhaps acknowledging scores of home recordings, the Met established its online database, which demarcates every broadcast

31 Paul Jackson, Saturday Afternoons at the Old Met: The Metropolitan Opera Broadcasts, 1931-1950 (Portland, OR: Amadeus Press, 1992), Paul Jackson Sign-Off for the Old Met: The Metropolitan Opera Broadcasts, 1950-1966 (Portland, OR: Amadeus Press, 1997), and Paul Jackson, Start-Up at the New Met: The Metropolitan Opera Broadcasts, 1966-1976 (Portland, OR: Amadeus Press, 2006). The Metropolitan Opera: The Radio and Television Legacy, September 19 – November 22, 1986 ([S.l.]: Museum of Broadcasting, 1986) is the accompanying book to the exhibit at the Museum of Broadcasting.

59

and performance and includes selected reviews.32 Of the 2,482 broadcasts, the operas transmitted most frequently appear in table 2-2.

The Metropolitan Opera aired two operas in 1910 – Tosca on January 12 and a double bill on January 13 of Cavalleria rusticana and Pagliacci – as part of early radio experiments. Because of competing transmissions that interrupted the second broadcast, the audience’s “ecstasy” resulting from the “divine singing” of Enrico Caruso and Emmy Destinn inside the opera house only reached remote listeners sporadically.33

Regular broadcast programming did not commence until Christmas Day, 1931, with

Englebert Humperdinck’s Hänsel und Gretel. Many recordings of these transmissions are available online through The Metropolitan Opera On Demand, while the Rodgers and Hammerstein Archive of Recorded Sound at the New York Public Library serves as the official repository for hard and digital copy recordings of the broadcasts.34

Of the other preeminent opera houses in the world, London’s Royal Opera House has hosted at least 192 performances of Turandot (this number includes productions by the resident companies . . . The Royal Italian Opera, The Covent Garden Opera

Company, and The Royal Opera) While the demographic information in the Royal

Opera House Collections Online Performance Database is similar to that in the Met’s

32 One strong example of the home recording culture resides at the University of Texas in the form of the Dr. Adams Collection. A “collecting goal of Dr. Adams was to acquire every broadcast performance of the Metropolitan Opera. Dr. Adams also made numerous in-house recordings and thus the collection contains many unique items, especially of vocal recitals by the most significant twentieth-century singers.” University of Texas at Austin Libraries, Historical Music Recordings Collection, accessed 26 August, 2014, http://www.lib.utexas.edu/fal/historical-music-recordings-collection/named-collections.

33 “Wireless Melody Jarred,” New York Times, 14 January 1910, accessed on 6 September 2014, http://archives.metoperafamily.org/archives/frame.htm. This review appears on the page with the information regarding the double-bill performance of Cavalleria rusticana and Pagliacci.

34 The websites for each of these are, respectively, http://www.metopera.org/ondemand/, and http://www.nypl.org/locations/tid/55/node/35925.

60

database, there is no online repository for performances at Covent Garden before

1946.35 In addition to this repository’s incomplete status—there is no accounting for any of the performances during the 2013/2014 Season—the Royal Opera House Collection

Online does not indicate which performances were broadcast. Milan’s Teatro alla Scala hosted the premiere performance of Turandot, but its online archive offers little indication of broadcast history, and any availability of archived materials is unclear.

Thus, given the availability of resources at the Metropolitan Opera, and the continuity of its compiled performance history, the Met is the most appropriate major opera house for studying performance traditions as they are elucidated by recording technology.

Turandot’s performance history at the Metropolitan Opera extends backwards in time to November 16, 1926, when it had it United States premiere. On this night, Maria

Jeritza sang the title role, Giacomo Lauri-Volpi the unknown prince, and Martha Attwood portrayed the slave girl, Liù, with Tullio Serafin on the podium.36 Turandot’s appearance on the Met stage at this time was part of the opera world’s homage to Puccini, following its world premiere on 25 April 1926 at Milan’s Teatro alla Scala under Toscanini’s baton.

As Alexandra Wilson demonstrates in her reception history of Turandot, each performance of Turandot around the world from 1926 to 1928 offered audiences and

35 http://www.rohcollections.org.uk/Default.aspx. The best source of information for performances before the founding of the Covent Garden Opera Company (which became the Royal Opera in 1968) is Harold Rosenthal, Two Centuries of Italian Opera at Covent Garden (London, UK: Putnam, 1958).

36 At the Milan premiere, varying accounts of Toscanini’s words at the conclusion of Liù’s funeral cortege exist. Eugenio Gara, present at the premiere, indicates the conductor’s words were “Qui finisce l'opera, perché a questo punto il maestro è morto.” (Here the opera finishes, because at this point the maestro has died). Ashbrook and Powers, 126-132. Conversely, The American premiere of Turandot included the Franco Alfano ending.

61

performers the opportunity to mark Puccini’s passing.37 The Met performed Turandot twenty-seven times, both in Manhattan and as part of tours to Brooklyn, Philadelphia,

Baltimore, Atlanta, and Cleveland from its American premiere until January 1930.38 For reasons yet unclear, it then entered a thirty-one year dormancy.

Turandot re-emerged in the Met’s repertory to much fanfare in the spring of 1961.

With a star-studded cast of Birgit Nilsson in the title role, Franco Corelli as Calaf, Anna

Moffo as Liù, and Leopold Stokowski on the podium, expectations were high. New York

Times reviewer Harold C. Schoenberg declared that Nilsson “left no question about her mastery of the role. . .She was triumphant. Her voice soared full (sic) and solidly, over orchestra, chorus and soloists, up to a tremendous high C that sounded as if there were still plenty in reserve.”39 Schoenberg was equally effusive about Corelli: “His voice, solidly anchored, was easily produced, and he matched Miss Nilsson note for note. . .it is a tenore di forza with a suave vocal quality. . .He was able to attain great volume without straining his vocal cords, and by doing so, he preserved the stylistic integrity of the role. In addition, he made a handsome prince, and acted with a good deal of sympathy.”40 Schoenberg’s tempered slightly his praise for Anna Moffo as Liù, who

“sounded a little uncomfortable in her first-act "Signore, ascolta," but she more than

37 See the preceding discussion of Turandot’s place as the end of the tradition, particularly Alexandra Wilson’s account of the atmosphere of the Milan premiere.

38 MetOpera Database, accessed 12 September 2014, http://archives.metoperafamily.org/archives/frame.htm, search terms “Turandot” and “Met Performance.”

39 Harold C. Schonberg, “Opera: ‘Turandot’ Back,” New York Times, 25 February 1961, accessed on 14 September 2014, http://search.proquest.com/docview/115347153/1D2B9BB753B84EDDPQ/2?accountid=10920.

40 Ibid.

62

redeemed herself in the third-act death scene, sung with pathos and conviction.”41 Thus, the reappearance of Turandot at the Met seems to have been a success, despite any misgivings regarding the staging.

The second performance of this production, on 4 March 1961, was the first broadcast of Turandot from the Metropolitan Opera. The broadcasts are the means by which the Metropolitan Opera reached a broad audience. Recordings are extant, and they form my primary set of original sources. Figure 2-3 summarizes the pertinent performance demography for the 36 broadcasts of Turandot at the Metropolitan

Opera.42 Broadcasts with located recordings appear in boldface; an asterisk following a name indicates a singer’s house debut.

Date Turandot Calaf Liù Conductor House Manager 4 March Birgit Franco Anna Moffo Leopold Rudolf 1961 Nilsson Corelli Stokowski Bing 24 February Birgit Franco Licia Albanese Kurt Adler Rudolf 1962 Nilsson Corelli Bing 16 January Birgit Jess Lucine Amara Fausto Rudolf 1965 Nilsson Thomas Cleva Bing 3 December Birgit Franco Mirella Freni Zubin Mehta Rudolf 1966 Nilsson Corelli Bing 22 March Marion James Martina Zubin Mehta Rudolf 1969 Lippert McCracken Arroyo Bing 21 February Birgit Plácido Lucine Amara Kurt Adler Rudolf 1970 Nilsson Domingo Bing 27 April Elinor Ross Franco Edda Moser Gabor Ötvös Schuyler 1974 Corelli Chapin 2 July 1974 Elinor Ross Richard Betsy Norden Martin Rich Schuyler Tucker Chapin 28 Ingrid Franco Adriana Alberto Schuyler December Bjoner Corelli Maliponte Erede Chapin 1974 28 March Eva Marton Plácido Leona Mitchell James Bruce 1987 Domingo Levine Crawford

41 Ibid.

42 As of 15 September 2014.

63

4 April 1987 Eva Marton Plácido Leona Mitchell James Bruce Domingo Levine Crawford 13 February Ghena Nicola Leona Mitchell Nello Santi Bruce 1988 Dimitrova Martinucci Crawford 13 January Gwyneth Vladimir Aprile Millo Nello Santi Hugh 1990 Jones Popov Southern 1 February Gwyneth Vladimir Teresa Stratas Nello Santi Joseph 1992 Jones Popov Volpe 11 February Gwyneth Lando Maria Nello Santi Joseph 1995 Jones Bartolini Spacagna Volpe 17 February Ghena Lando Angela Nello Santi Joseph 1996 Dimitrova Bartolini Gheorghiu Volpe 13 Sharon Richard Hei-Kyung Nello Santi Joseph December Sweet Margison Hong Volpe 1997 1 March Adrienne Richard Norah Marco Joseph 2003 Dugger Margison Amsellem Armiliato Volpe 29 January Andréa Johan Botha Krassimiri Bertrand de Joseph 2005 Gruber Stoyanova Billy Volpe 30 March Andréa Richard Hei-Kyung Fabio Luisi Peter Gelb 2007 Gruber Margison Hong 05 April Erika Richard Hei-Kyung Fabio Luisi Peter Gelb 2007 Sunnegardh Margison Hong 14 April Andréa Richard Hei-Kyung Marco Peter Gelb 2007 Gruber Margison Hong Armiliato 16 April Andréa Richard Hei-Kyung Marco Peter Gelb 2007 Gruber Margison Hong Armiliato 25 April Erika Richard Liping Zhang Marco Peter Gelb 2007 Sunnegardh Margison Armiliato 03 May 2007 Erika Richard Liping Zhang Marco Peter Gelb Sunnegardh Margison Armiliato 08 May 2007 Erika Richard Liping Zhang Marco Peter Gelb Sunnegardh Margison Armiliato 28 October Lise Marcello Marina Andris Peter Gelb 2009 Lindstrom* Giordani Poplavskaya Nelsons* 07 Maria Marcello Marina Andris Peter Gelb November Guleghina Giordani Poplavskaya Nelsons 2009 10 Lise Marcello Marina Andris Peter Gelb November Lindstrom Giordani Poplavskaya Nelsons 2009 07 January Maria Philip Webb Maija Andris Peter Gelb 2010 Guleghina Kovalevska Nelsons 20 January Maria Salvatore Maija Andris Peter Gelb 2010 Guleghina Licitra Kovalevska Nelsons 28 January Lise Frank Grazia Julien Peter Gelb 2010 Lindstrom Porretta Doronzio Salemhour* 26 Maria Marco Berti Hibla Dan Ettinger Peter Gelb September Guleghina Gerzmava 2012

64

3 October Maria Marco Berti Hibla Dan Ettinger Peter Gelb 2012 Guleghina Gerzmava 02 Iréne Theorin Marcello Janai Brugger* Dan Ettinger Peter Gelb November Giordani 2012 7 January Iréne Theorin Walter Hibla Dan Ettinger Peter Gelb 2013 Fraccaro Gerzmava 23 Christine Marcelo Hibla Paolo Peter Gelb September Goerke Álvarez Gerzmava Carignani 2015 30 Christine Marcelo Hibla Paolo Peter Gelb September Goerke Álvarez Gerzmava Carignani 2015 26 October Lise Marcelo Leah Crocetto Paolo Peter Gelb 2015 Lindstrom Álvarez Carignani 26 January Nina Stemme Marco Berti Anita Hartig Paolo Peter Gelb 2016 Carignani 30 January Nina Stemme Marco Berti Anita Hartig Paolo Peter Gelb 2016 Carignani Figure 2-3. Broadcast history of Turandot at the Met

Musical Moments for Investigation

Within this opera, six moments are most appropriate for examination for both practical and dramaturgical reasons. Arias in Puccini’s operas offer dramatic pauses for characters; they are moments of emotional exposition, or the result of reflecting on surrounding events. Both of these types of arias occur in Turandot. Liù’s first aria,

“Signore, ascolta!”, exhibits both functions, as she reacts emotionally to Calaf’s stated intent to answer Turandot’s riddles. Calaf’s response, “Non piangere, Liù,” gives him the opportunity to respond tenderly to Liù, and show concern for his father, Timur, by requesting she remain if the prince surrenders his life. Turandot’s entrance aria (in act

2), “In questa reggia,” recounts the reasons for her riddles and brutality – her desire for vengeance on behalf of her ancestress Lou-o-Ling. “Nessun dorma!” is Calaf’s reflection on Turandot’s commandment that no one in the realm shall sleep until she knows his name and shows his resolve in conquering the princess. Liù’s suicide aria, “Tu che di gel sei cinta,” reveals both her crumbling determination under the duress of torture and

65

her desire for Calaf to win the hand of Turandot. These moments are of thinner musical texture as they have fewer performers producing sound and therefore have fewer note onsets to detect, which simplifies beat placement. They are thus logistically more accessible as a starting point for examining performing traditions. Texts and translations for each of these arias appear in Figure 2-4.43

In addition to these arias, one other moment of musical practicality and dramaturgical significance merits consideration for its performance traditions. In act 2’s riddle scene (“Straniero, ascolta!”), where Turandot poses each of her three riddles and

Calaf successfully answers them, fermatas precede the statement of each riddle and

Calaf’s response to the first and third. How performers observe these fermatas in relationship to the surrounding musical fabric can indicate varying degrees of gravity, levity, or expediency as the opera’s second dramatic conflict emerges from the resolution of its first.

Liù (avvicinandosi al principe, supplicante, Liù (approaching the prince, imploring, piangente) weeping) Signore, ascolta! Deh!, signore, ascolta! Lord, listen! Ah! Lord, listen! Liù non regge più! Liù can stand no more! Si pezza il cuore! Her heart is breaking! Ahimè, ahimè, quanto cammino Alas, alas what a long way col tuo nome nell'anima With your name in my soul, col nome tuo nell'labbra With your name on my lips Ma se il tuo destino, doman, sarà deciso, But if your fate is decided tomorrow, noi morrem sulla strada dell'esilio. We’ll die on the road of exile! Ei perderà suo figlio... He will lose his son. . . io l'ombra d'un sorriso! I, the shadow of a smile Liù non regge più! ha pietà! Liù can stand no more! Ah, pity! Il Principe Ignoto (avvicinandosele, con The Unknown Prince (approaching her, with commozione) emotion) Non piangere, Liù! Don’t cry, Liù! se in un lontano giorno If on one far-off day I smiled at you, For that smile, my sweet girl,

43 The translations are from: William Weaver, Seven Puccini Librettos (New York: W.W. Norton & Company, Inc., 1981).

66

io t'ho sorriso Listen to me: your Lord per quel sorriso, dolce mia fanciulla will be, tomorrow, perhaps alone in the world. . . m'ascolta: il tuo Signore Do not leave him, take him away with you! sarà, domani, forse, solo al mondo Liù Non lo lasciare, portalo via con te! We’ll die on the road of exile! Liù Timur Noi morrem sulla strada dell’esilio! We’ll die! Timur The Unknown Prince Noi morrem! Make gentle the roads of exile for him Il Principe Ignoto This. . .this. . .Oh, my poor Liù, Dell'esilio addolcisci a lui le strade Of your little heart that doesn’t fail Questo...questo, o mia povera Liù, He asks who smiles no more. . . al tuo piccolo cuore che non cade Who smiles no more! chiede colui che non sorride più. . . Che non sorride più! Turandot Turandot In questa reggia, or son mill'anni e mille, In this palace, a thousand, thousand years ago, un grido disperato risonò. A desperate cry resounded. E quel grido, traverso stirpe e And that cry, through descendant and stirpe descendant, qui nell'anima mia si rifugiò! Took refuge in my soul! Principessa Lou-Ling, Princess Lou-Ling, ava dolce e serena che regnavi Sweet and serene ancestress, who reigned nel tuo cupo silenzio in gioia pura, In your dark silence, in pure joy, e sfidasti inflessibile e sicura And who defied, inflexible and sure, l'aspro dominio, oggi rivivi in me! Bitter domination, you relive in me today! La Folla (sommessamente) The Crowd (quietly) Fu quando il Re dei Tartari It was when the king of the Tartars le sette sue bandiere dispiegò. Unfurled his seven flags! Turandot (come cosa lontana) Turandot (like something remote) Pure nel tempo che ciascun ricorda, Still, in the time that everyone remembers, fu sgomento e terrore e rombo d'armi. There was alarm, and terror in the rumble of arms! Il regno vinto! Il regno vinto! The kingdom defeated! The kingdom defeated! E Lou-Ling, la mia ava, trascinata And Lou-o-Ling, my ancestress, dragged away da un uomo come te, come te straniero, By a man like you, stranger, like you, foreigner, là nella notte atroce There in the atrocious night, dove si spense la sua fresca voce! Where her fresh voice was extinguished! La Folla (momora reverente:) The Crowd (murumrs reverently:) Da secoli ella dorme For centuries she has slept nella sua tomba enorme! In her enormous tomb! Turandot Turandot O Principi, che a lunghe carovane O Princes who in long caravans d'ogni parte del mondo From every part of the world qui venite a gettar la vostra sorte, Come here to try your fate, io vendico su voi, su voi quella purezza, I avenge upon you, upon you that purity, quel grido e quella morte! That cry and that death! Quel grido e quella morte! That cry and that death! Mai Nessun m’avra! Mai nessun, No one will ever possess me! No one ever, nessun m'avrà! No one will possess me! L'orror di chi l'uccise The horror of him who killed her

67

vivi nel cuor mi sta. Is alive in my heart! No, no! Mai nessun m'avrà! No, no! No one will ever possess me! Ah, rinasce in me l'orgoglio Ah, in me is reborn the pride di tanta purità! Of such purity! (e minacciosa, al Principe:) (and, threateningly to the Prince:) Straniero! Non tentar la fortuna! Never anyone, anyone shall have me! “Gli enigmi sono tre, la morte una!” “The riddles are three, the death is one!” Il Principe Ignoto The Unknown Prince No, no! Gli enigmi sono tre, No, no! The riddles are three, una è la vita! one the life! Turandot Turandot No, no! No, no! [Gli enigmi sono tre, la morte è una! [The riddles are three, the death is one! Il Principe Ignoto The Unknown Prince Gli enigmi sono tre, una è la vita!] The riddles are three, one the life!]

Turandot Turandot Straniero, ascolta: "Nella cupa notte Stranger, listen! “In the gloomy night vola un fantasma iridescente. Sale an iridescent ghost flies. It rises e spiega l'ale and spreads its wings sulla nera infinita umanità. over the black, infinite mankind. Tutto il mondo l'invoca All the world invokes it, e tutto il mondo l'implora! and all the world implores it! Ma il fantasma sparisce coll'aurora Bu the ghost vanishes with the dawn per rinascere nel cuore! to be reborn in the heart! Ed ogni notte nasce And every night it is born ed ogni giorno muore. . ." and every day it dies. . .” Il Principe Ignoto (con improvvisa sicurezza) The Unknown Prince (with sudden confidence) Sì! Rinasce! Rinasce! E in esultanza mi porta via Yes! It is reborn! It is reborn! And in triumph con sé, Turandot: It carries me away with itself, Turandot, La Speranza! “Hope.” I Sapienti (si alzano e ritimicamente aprono The Sages (stand up and rhythmically open insieme il primo rotolo) together the first scroll) La Speranza! La Speranza! La Speranza! Hope! Hope! Hope! Turandot (gira gli occhi fierissimi. Ha un Turandot (turns her very proud eyes. She freddo riso. La sua altera superiorità la laughs a cold laugh. Her haughty superiority ripiende) grips her again) Sì! la speranza che delude sempre! Yes! Hope which always disappoints! "Guizza al pari di fiamma, e non è fiamma! “It darts like a flame, and is not a flame! È talvolta delirio. È febbre At times it is delirium! It’s a fever d'impeto e ardore! of impulse and ardor! L'inerzia lo tramuta in un languore! Inertia transforms it into languor! Se ti perdi o trapassi, si rafredda! If you are lost or die, it grows cold! Se sogni la conquista, avvampa, avvampa! If you dream of conquest, it flames, it flames! Ha una voce che trepido tu ascolti, It has a voice that you listen to in fear e del tramonto il vivido baglior. . ." and the vivid glow of the sunset. . .” L'Imperatore The Emperor Non perderti, straniero! Don’t destroy yourself, foreigner! La Folla The Crowd È per la vita! It is for your life! È per la vita! Parla! It is for your life! Speak!

68

Non perderti, straniero! Don’t destroy yourself, foreigner! Parla! Parla! Speak! Speak! Liù Liù È per l'amore! It’s for love! Il Principe Ignoto (perde ad un tratto la The Unknown Prince (loses suddenly the dolorosa atonìa del viso. E grida a Turandot:) painful, lost expression. And he cries to Turandot:) Sì, Principessa! Avvampa e insieme langue, Yes, Princess! It flames and also languishes, se tu mi guardi, nelle vene: Il Sangue! In my veins, if you look at me. “Blood!” I Sapienti (aprendo il secondo rotolo) The Sages (opening the second scroll) Il Sangue! Il Sangue! Il Sangue! Blood! Blood! Blood! La Folla (prompendo gioiosamente) The Crowd (bursting out, jouyously) Coraggio, scioglitore degli enigmi! Courage, solver of riddles! Turandot (raddrizzandosi, come colpita da Turandot (stiffening, as if struck by a lash, una frustrate, urla alle guardie:) shouts to the guards:) Percuotete quei vili! Strike those wretches! "Gelo che ti dà foco e dal tuo foco “Frost that sets you afir! And from your fire più gelo prende! Candida ed oscura! gains more frost! Candid and dark! Se libero ti vuol ti fa più servo! If it wants you free, it makes you more enslaved! Se per servo t'accetta, ti fa Re!" If it accepts you for slave, it makes you king!” Su, straniero! Ti sbianca la paura! Come, foreigner! Fear makes you blanch! E ti senti perduto! Su, straniero, And you feel yourself lost! Come, foreigner, il gelo che dà foco, che cos'è?" the frost that gives fire – what is it? Il Principe Ignoto (desolato ha piegato la The Unknown Prince (desolate, he has bowed testa fra le mani. Ma è un attimo. Un lampo di his head in his hands. But it is only for a gioia lo illumine. Balza in piedi, magnifico moment. A flash of joy illuminates him. He d’alterigia e di forza. Esclama:) springs to his feet, magnificent in pride and strength. He exclaims:) La mia vittoria ormai t'ha data a me! My victory now has given you to me! Il mio fuoco ti sgela: Turandot! My fire thaws you: “Turandot!” I Sapienti (che hanno svolto il terzo rotolo, The Sages (who have unfurled the third esclamano:) scroll, exclaim:) Turandot! Turandot! Turandot! Turandot! Turandot! Turandot! Il Principe Ignoto The Unknown Prince Nessun dorma! Nessun dorma. . .! Let no one sleep! Let no one sleep! Tu pure, o Principessa, You too, O Princess, nella tua fredda stanza, In your cold room, guardi le stelle are looking at the stars che tremano d'amore e di speranza! That tremble with love and with hope! Ma il mio mistero è chiuso in me, But my mystery is locked in me, il nome mio nessun saprà! No one will know my name! No, no, sulla tua bocca lo dirò No, no, upon your mouth I’ll say it quando la luce splenderà! when the light shines. Ed il mio bacio scioglierà il silenzio And my kiss will break the silence che ti fa mia. . .! that makes you mine. . .! Voci di Donne (misteriose e lontane) Women’s Voices (mysterious and remote) Il nome suo nessun saprà!... No one shall know his name! e noi dovrem, ahime, morir! And we must, alas, die! Il Principe Ignoto The Unknown Prince Dissolve, o night. . .! Set, stars! Set, stars!

69

Dilegua, o notte. . .! Tramontate, stelle! At dawn I’ll win! I’ll win! I’ll win! Tramontate, stelle! All'alba vincerò! Vincerò! Vincerò!

Liù Liù Tu che di gel sei cinta, You who are girded with frost, Da tanta fiamma vinta, overcome by such flame, l'amerai anche tu! you will love him also! l'amerai anche tu! you will love him also! Prima di questa aurora, Before this dawn Io chiudo stanca gli occhi, I close, weary, my eyes, Perchè Egli vinca ancora. . .ei vinca ancora. . . that he may win again. . .that he may win again. . Per non. . .per non vederlo più! . Prima di questa aurora, di questa aurora To not. . .to not see him any more! io chiudo stanca gli occhi, Before this dawn, this dawn Per non vederlo piu! I close, weary, my eyes to not see him any more! Figure 2-4. Text and translations of arias in Turandot

Musical Phenomena for Investigation

Sounded music contains measurable (quantifiable) elements based on the physical properties of sound: rhythm, meter, tempo, pitch, melody, harmony, key, form, timbre, phrasing, articulation, dynamics, and (for vocal music) diction. Pitch, meter, rhythm, key, dynamics, articulation, and phrasing have direct parallel notational encodings; the other elements are indirectly discernable from notation.

Tempo is the most readily analyzable musical element, owing to its limited parameters of marking beats in relationship to the passage of time. It is also foundational to understanding how other musical phenomena unfold in a given performance. José Bowen has noted that earlier examinations of performance practices have considered tempo because performers—especially conductors—have, taking after

Beethoven, considered it is the key interpretive element. Richard Wagner also exulted tempo as the most important interpretive component of performance.

The whole duty of a conductor is comprised in his ability always to indicate the right tempo. His choice of tempi will show whether he understands the

70

piece or not. With good players again the true tempo induces correct phrasing and expression, and conversely, with a conductor, the idea of appropriate phrasing and expression will induce the conception of the true tempo.44

Moreover, Puccini was notoriously fastidious with regard to interpretations of his works, particularly as performer’s choices regarding tempo – and thereby dramatic pacing – reflected them. A collection of notes by vocal coach and accompanist Luigi

Ricci, who worked with Puccini in the 1910s, reveals some of the composer’s most profound thoughts about interpretation and tempo. Even the collection and publication of the notes occurred “so that the superficial and hurried interpreters (and those who are satisfied with the “classical half-attempt”) (sic) can at least be shown that which the exacting Puccini wanted.”45 Ricci prefaces his notes on each of Puccini’s opera through those comprising Il Trittico with a summary of Puccini’s interpretive guidelines. These address aspects of performance, including tempo, colors of expression, colors of sonorities, fermatas, portamentos, artistic dedication, staging and dramatic atmosphere, backstage music, and use of bells.46 Understanding the tempo relationships within individual performances is thus the necessary first step to being able to compare how performers interpret a work.

44 Richard Wagner, Wagner on Conducting, trans. Edward Dannreuther (New York, NY: Dover Publications, Inc., 1989), 20.

45 Luigi Ricci, Puccini Interprete di se stesso (Milan: Casa Ricordi, 1954), prefatory material (unpaginated). Harry Dunstan, in “Performance Practices in the Music of Giacomo Puccini as Observed by Luigi Ricci” (PhD diss., Catholic University of America, 1989) provides an English translation. Ricci and Puccini had stopped working together by the time of Turandot.

46 Ibid. The discourse regarding these precepts occupies pages 11-14.

71

Ricci situates tempo as Puccini’s significant concern for interpretation, almost to the level of neuroticism:47 “Puccini aveva una sensibilità acuttissima nei riguardi di tutta questa varietà di movimenti ritmici:. . .In generale quando si trattava di tempi lenti, il suo fiuto era d’una finezza straordinaria. Diceva che i tempi ‘troppo lenti’ fanno morire l’azione, la narcotizzano, la rendono acciddiosa e pesante, come tutte le cose morte. ”48

Despite what Ricci termed Puccini’s phobia of tempi that are too slow draining the dramatic life from his works, Puccini appears not to have given strict instructions to performers on this point.49 Rather, he allowed for the individuality of interpretation as part of “lo svolgimento della nastro melodico,” which he stipulated comes from the metronome of the heart, what he termed “il Maelzel che sta dentro di noi.”50 Ricci’s accounts of Puccini’s remonstrations about tempo choices not meeting his aesthetic are many. Ricci offers anecdotal evidence to suggest that each performance must be at a tempo perhaps organically appropriate to its own needs. Furthermore, Ricci’s anecdotes imply that too slow tempi were perhaps the worst crime a performer could commit against Puccini’s aesthetic sensibility.

Maestro! Se s’addormenta lei, ci addormentiamo tutti!. . .Vita, vita maestro! Non rallenti troppo. Non sente che questo brano casca a brandelli, che questo frammento si sbricola, quest’altro s’impadula?. . .La

47 Of the pages devoted to Puccini’s “rules” (in translation), the discussion regarding tempi occupies two full pages, precisely the length devoted to the discussion of all of the other concerns.

48 Ricci, 11. “Puccini had a most acute sensibility in regards of the variety of rhythmic movements. . .In general, when he used slow tempi, his intuition was of extraordinary finesse. He said that tempi that are ‘too slow’ make the action die, they drug it, they render it slothful and heavy, like all dead things.” Translations from the Italian are my own except where noted.

49 Ibid., Ricci’s phrase it “fobia dei tempi troppo lenti,” which, while seemingly an exaggeration, certainly makes a point about Puccini’s sense of dramatic pacing and the importance of forward momentum, both musically and dramatically.

50 Ibid., 11,12. “The development of the melodic ribbon,” and “The Maelzel inside of us.” The latter is a reference to the metronome Johann Nepomuk Maelzel patented in 1815 and indicates at least a conceptual incorporation of technology into aesthetic practice.

72

stasi è la negazione della musica, specie della musica teatrale. Anche un passo grave deve accusar la vita. La gazzella e l’elefante si muovono con la deambulazione che è loro propria: ma guai se piegano le gambe e stramazzano a terra. . .51

As fermatas appear throughout the riddle scene, and again at the revelation of

Calaf’s name, Puccini’s recorded thoughts about them are pertinent. Ricci’s notes, however, do not deal with notated fermatas, only moments in a musical line where a performer might be tempted to elongate a note’s rhythmic value. They only say that ‘le note coronate’ (the crowned notes – i.e. those at the apex of a line) should be

“esattamente il doppio del loro valore morfologico. La qual cosa rientrava nel suo modo di intendere l’arte: mai effetti esagerati. Niente acuti molto lunghi e nemmeno esageratamente forti, violenti, aggressivi. Niente note dalla lunga agonia: note che a smorzarle, ci vorebbe un collasso.”52

With regard to tempi, three main categories are viable options for analysis, and each reveals something different about a performance. First, and perhaps most general is global tempo. This is the average tempo of an entire performance of a work, or

# 표푓 푏푒푎푡푠 푖푛 푤표푟푘 60 푠푒푐. section thereof ( × ). Global tempo considerations are 푡푖푚푒 (푠푒푐.)표푓 푝푒푟푓표푟푚푎푛푐푒 1 푚푖푛푢푡푒 useful in that they provide a broad sense of the overall tempo taken in a performance

51 Ibid., 11-12. “Maestro! If you fall asleep, we all fall asleep! . . . Life, life, maestro! Do not slow too much. Do you not sense this part falls to pieces, that this fragment crumbles, this other becomes a swamp? . . . The stasis is the negation of the music, especially of the theater music. Even the serious step must acknowledge life. The gazelle and the elephant move each with their own steps: but heaven help them if their legs fold and they fall to the ground.”

52 Ibid., 13. “exactly double their morphological value. This goes back to his mode of understanding art: effects should never be exaggerated. Nothing sharp, very loud, nor exaggeratedly loud, violent, or aggressive. No notes of long agony; notes that grimace, or that would collapse.” Will Crutchfield’s highlighting of the seeming offences of this rule regarding the final b-natural at the end of “Nessun dorma!” (discussed in Chapter 1) indicates perhaps a changing cultural aesthetic in this regard, or perhaps a complete disregard for Puccini’s wishes. In either case, the reality Crutchfield acknowledges raises numerous questions about authenticity, intention, and the evolution of aesthetics.

73

and are relatively easy to calculate. The limitation with global tempo calculations is that there is no way to account for the tempo between beats or groups of beats. Even so, global tempo calculations are useful for establishing overall trends in musical speed, such as slowing or speeding in different recording contexts or in relationship to circumstances of a cultural milieu (such as the premiere of Turandot as a memorial to

Puccini). Local tempo is the most specific, accounting for the time passing between

1 (푏푒푎푡) 60 푠푒푐. each discrete pair of beats in a given performance ( × ). The # 표푓 푠푒푐표푛푑푠 푏푒푡푤푒푒푛 1 푚푖푛푢푡푒 limitation here is the inverse of global tempo; there is no straightforward way to account for tempo relationships between groups of beats or of the whole. If global tempo is the broadest view of a performance’s tempo, and local tempo is the narrowest, varying degrees of phrasal tempo occupy the spectrum between these poles. Phrasal tempo can reveal where performers – consciously, intentionally, or not – emphasize certain moments in a work of music.53 Finally, performer decisions regarding musical and dramatic pacing in the outlined moments are vital to understanding dramatic aspects of performing traditions in Turandot. This study considers these three kinds of tempo as they relate to a performance’s surface level (local – what one hears in listening multiple times), its mid-level structure (phrasal – where connections between text and music sections emerge), and its foundational or background level (global tempo). The current study applies this paradigm to each excerpt as a distinct unit; future research will

53 In the sense that I have described three levels of tempo hierarchically (global, phrasal, and local), these structures and relationships provide a very close parallel to the levels of a Schenker graph. Global tempo is akin to the background, or Ursatz; local tempo is akin to the surface level events; phrasal tempo is the wide span of the Schenkerian middle-ground. Further analysis could likely establish these tempo levels and relationships as integral to a performance’s structure and, by extension, a ‘canon’ of interpretation which evolves, like harmonic language, over time, reflecting shifting aesthetics.

74

consider each excerpt in the context of the entire musical and dramatic fabric of the opera.

Case Study Methods

The methodology used in this study contrasts markedly with traditional approaches. As mentioned earlier, scholars have generally engaged musical works and their attending traditions largely with qualitative instead of quantitative means. While qualitative definitions of tradition populate the history of aesthetic analysis, scholars have historically paid little attention to quantitative methods of identifying and defining it.

Robert Philip and Mark Katz have each discussed at length how certain performance behaviors have changed over time in relationship to preservation and dissemination technology.54 Studies devoted to the psychology of performance, the semantics of performance, and the relationships between performance and analysis relate to this body of literature.55 All of these studies used available methods for the analysis of recorded music as a means to a greater end, but none focused solely on analyzing recorded music as an end.

The establishment of the Centre for the History and Analysis of Recorded Music

(CHARM) was one of the first large-scale undertakings concerned primarily with recordings as documents of performance. This five-year, AHRC funded project sought

54 Robert Philip, Performing Music in the Age of Recording (New Haven: Yale University Press, 2004) and Mark Katz, Capturing Sound: How Technology Has Changed Music (Berkeley: University of California Press, 2010).

55 John Rink, ed. The Practice of Performance: Studies in Musical Interpretation (Cambridge, UK: Cambridge University Press, 1995). I have written about performance practice and performers’ differences in mentality apparent in various circumstances of recording [Joshua O. Neumann, “Performance practices in four Puccini arias: tempo choices and choosers” (MM thesis, University of Nebraska—Lincoln, 2008)].

75

to broaden the perspectives of musical works by developing methods for engaging performed music as this experience becomes the increasingly dominant means of musical activity. In short, “CHARM was established to promote a musicology that better reflects the nature of music as experienced in the twentieth century and beyond.”56 One of the primary aims of this project was to develop methodologies for analysis recorded music. As part of this aim, scholars developed computational analysis of individual performances and highly detailed and increasingly complex statistical analyses to large groups of musical performances.

Nicholas Cook, Andrew Earis, and Craig Sapp collected over 3,000 recorded performances for 49 of Frederic Chopin’s mazurkas—on average over 80 performances for each mazurka—as part of CHARM’s Mazurka Project.57 The scope of this work made keeping track of differences and similarities between numerous performances, of even a single mazurka, prohibitively difficult, especially in comparing recordings heard weeks, months or even years apart. Thus one of the primary obstacles to a corpus study of this scale was that remembering the distinguishing features of 80 individual performances of a composition would be taxing on anyone’s memory. They discovered that “Often the surface acoustics of a performance (such as reverb, microphone placement, piano model, recording/playback noise) are more noticeable and memorable than the actual performance, so identifying related performances solely by ear can

56 About CHARM, AHRC Research Centre for the History and Analysis of Recorded Music, accessed 15 March 2015, http://charm.rhul.ac.uk/about/about.html.

57 http://mazurka.org.uk.

76

sometimes be difficult.”58 Aware of the various kinds of sonic data in any performance, the Mazurka Project’s researchers’ solution was to develop means to deconstruct audio files into their constituent phenomena for analysis and then to analyze it—work best suited for digital tools.

An overview of pertinent terms will help clarify CHARM’s methodologies and my uses of them. I have organized this vocabulary here according to relevant conceptual areas: software tools, data representation and evaluation of individual performances, and data evaluation and representation between performances. Sonic Visualiser allows for identifying several aspects of musical performance through multiple analytical layers: note onset locations – related to tempo, pitch relationships across time, and dynamic levels of recorded sounds.59 The Time Instants Layer enables Sonic Visualiser to generate data sets comprised of both time and tempo relationships resulting from tempo markers tapped while listening to an audio track in real time. It can then calculate global tempo by averaging the time between all supplied tempo markers. The availability of multiple plug-ins and analytical layers allows for further nuancing of the kinds of analysis possible with Sonic Visualiser.60 Three particularly useful plugins for enhancing beat placement accuracy in Sonic Visualiser are the Spectral Reflux, Harmonic Spectrogram, and Power Curve tools. Figure 2-3 illustrates the analysis of beat placements in Sonic

58 Craig Stuart Sapp, “Computational Methods for the Analysis of Musical Structure” (Ph.D. diss., Stanford University, 2011), 111.

59 Chris Cannam, Christian Landone, and Mark Sandler, Sonic Visualiser: An Open Source Application for Viewing, Analysing, and Annotating Music Audio Files, in Proceedings of the ACM Multimedia 2010 International Conference.

60 For a list and description of these plug-ins, see: http://www.mazurka.org.uk/software/sv/plugin/.

77

Visualiser, the Harmonic Spectrogram, and how these tools relate to notated musical phenomena in the score of Chopin’s Mazurka in A minor, Op. 17, No. 4, measure 29.

Spectral Reflux. . .identifies note onsets in percussive timbres (such as piano [or string pizzicato or staccato attacks). It [helps] remove spurious peaks following a note onset to improve efficiency with visual correction of the data and increased sensitivity at low sound levels.

Harmonic Spectrogram. . .displays the raw output of the harmonic product spectrum as a spectrogram. [5] This process was originally used to identify mono- phonic pitch in vocal timbres, but it functions well to highlight the melodic line. . .by removing redundant information generated by harmonics...

Power Curve61. . .is also used to help localize note onsets by examining the slope of the smoothed power values extracted from the audio data. The smoothing process hides clicks and pops (very short bursts of noise). And by smoothing the raw power measurements in both the forward and reverse directions, the slope of the smoothed power curve will contain peaks centred at note onsets.62

61 http://sv.mazurka.org/MzPowerCurve.

62 Sapp, “Computational Methods,” 22-23.

78

Figure 2-5. Screen capture of time instants layer with adjusted beat placements (middle), and Harmonic Spectrogram (bottom) of Chopin Op. 17, No. 4, m. 29.63

Points in time representing beats are marked with dashed lines above the orange-coloured audio waveform linking to the score. Sonic Visualiser allows for analytic data to be superimposed on this audio waveform display. The black curve underneath the waveform is a continuous onset detection function (sampled every 10 milliseconds) which is generated by the Spectral Reflux plugin, and which is very useful for accurately finding note onsets in piano music. The vertical orange lines rising through the centre of each peak in the continuous onset function are the identified note onset times as defined by these peaks. The purple vertical lines marking the beat locations were initially tapped in real time while listening to the performance, and then subsequently adjusted to coincide with the correct note onset. . . corrected tap times are [later] loaded back into Sonic Visualiser and proof-audited to ensure that the tap events are attached to the correct note onsets. Tempo values are used instead of

63 Ibid., 23.

79

delta times for performance comparisons in order to minimize the outlier behaviour of long pauses in the music when calculating correlation values, although this is a small effect. These final beat-tempo values are the basis for numerical measurements made for comparing multiple performances with each other.64

For all recordings under consideration in this study, I extract tempo data in the way that Craig Sapp delineates in the foregoing excerpt. For recordings that exist in digital format, this is a straightforward process. For recordings that do not exist in digital format, and of which I cannot procure digital versions, an added step is necessary to generate tempo data. Regulations at the New York Public Library’s Rodgers and

Hammerstein Archive of Recorded Sound prohibit the handling of sound recordings by listeners.65 Employment of Sapp’s method for analyzing individual recordings requires anyone accessing recordings at this institution (hereafter the NYPL) to generate tempo- tap tracks.66 Tempo-tap tracks are the result of listening to a recording (remotely in this instance) and tapping the tempo into a microphone attached to a digital recorder by hand.67 As variances in this sort of data entry are inevitable, creating several tempo-tap

64 Ibid., 23-4.

65 “Users never directly handle sound recordings. . .Upon receipt of a call slip, the reference librarian communicates by computer with a playback technician located in the basement storage area. Selections are piped into specially designed listening and viewing booths; subsequent communications take place directly between the listener and audio technician through computer terminals stationed in each booth.” Taken from: “Using the Archives of Recorded Sounds: Avenues of Access,” New York Public Library, Rodgers and Hammerstein Archive of Recorded Sound, accessed 17 September 2014, http://www.nypl.org/locations/tid/55/node/7454.

66 The same method is usable by anyone whose archive access to recordings is solely via remote means, i.e. – where digital copies of recordings are not available.

67 Any cellphone with the capacity to record voice notes will suffice. My process was multi-step. I first listened to the recording and annotated a score with indications about tempo modifications (faster, slower, extra time at fermatas, and so on). Next, I listened again and tap into the voice-notes recorder on my cell- phone. Then I synchronized the tap track on my phone to the archival recording and listened, making further annotations about discrepancies in beat placement between the two audio tracks. I then made a second tap track, which I subsequently compared against the recording. I repeated this process of annotating, tapping, and comparing until the tap track lined up as closely as possible with the archival recording without the advantage of digital refinements such as Sonic Visualiser’s plugins.

80

tracks, finding averages, and double-checking the results against the original sound recording all become necessary for precision.

This data alone is not enough to cultivate a readily accessible understanding of tempo relationships within a performance. Presenting this data in a scape plot makes the tempo data more accessible and facilitates easier analysis of individual performances by representing tempo relationships between beats in a performance.

According to Sapp, “Scape plots take their name from the word landscape since they show small-scale features analogous to the foreground in a picture, as well as large- scale features similar to the background. And like a painting, the interesting parts of the scape plot usually lie somewhere in the middle-ground.”68 Scape plots reflecting individual performances are Time Scapes, which visually represent the hierarchy of tempo relationships within a single performance. Time Scapes “were intended to examine similarities between performances by the same [performers], between students and teachers, and to enable studies of performance style changes over time.”69 A simple example illustrates how a Time Scape illustrates tempo hierarchy.

A musical performance consists of six beats, which are labelled: A, B, C, D, E, and F. These six beats can be chopped up into 21 unique sub- sequences (n-grams). Firstly, the elements can be considered in isolation. Next they can be grouped by sequential pairs: AB, BC, CD, DE, EF. Then by threes: ABC, BCD, CDE, DEF; by fours: ABCD, BCDE, CDEF; by fives: ABCDE, BCDEF; and finally one sequence covering the entire performance: ABCDEF. All of these possible sub- sequences of the basic six-beat performance, can be arranged on top of each other to form the arrangement shown in Figure 2-6.70

68 Sapp, “Computational Methods,” 115.

69 Ibid., 33.

70 Ibid., 115.

81

Figure 2-6. Scape plotting domain structure (left) and example using mathematical average (right).

The pyramid on the left demonstrates the relationships between data points

(tempo between beats) individually on the bottom, in varying combinations in the middle, and the whole on top. The right pyramid is “an example application” showing hierarchy “where the original data sequence is (7, 6, 2, 5, 8, 4).”71 Time Scapes demonstrate the hierarchical tempo relationships in a recording in a similar fashion to

Schenker graphs with tonal relationships. In these, surface, mid, and background relationships are easily visible. For each Time Scape, Color-coding reflects tempi in relationship to the global tempo of that individual performance. Based on the structure of the color spectrum acronym ROYGBIV, Green represents the global tempo while

Yellow, Orange, and Red reflect increasingly faster tempi than average (Red is the fastest), and Blue, Indigo, and Violet indicate tempi increasingly slower than average

(Violet is the slowest). In Chapter 3, Time Scapes (one for each musical moment outlined above) illustrate the hierarchical relationships of phrasing and tempo in for each recording. Using these graphs, understanding how performers have realized this music

71 Ibid.

82

with respect to tempo and rhythmic phrasing, both locally and globally, is straightforward and readily accomplished without having to recall the performances they represent.

As useful as Time Scapes are for understanding individual performance, comparison between any two performances using only their respective Time Scapes provides only a general perspective. This approach would show approximate musical locations of similarity and difference, but no detail about underlying similarities or differences between the performances they represent. Correlation analysis can provide insight into the tempo/rhythmic structure of individual performances and across multiple sets thereof. Mathematical averages and correlation coefficients embody a tension between retaining old and embracing new data, and they change as new numbers join the data pool. Tradition connects what has gone before to what is current. In this sense, then, the mathematical mean represents the quantitative aspects of tradition, providing the necessary means for augmenting current, qualitative, understanding of performing tradition. Chapters 4 and 5 contain chronologically-ordered Time Scapes based on a growing tempo average aggregate with respect to chronology alone (Chapter 4) and to production, conductors, and singers (Chapter 5).

Arranging the data table and plots chronologically affords a view of Turandot’s evolving performance tradition at The Metropolitan Opera. To do so, the data set must account for the average tempo values as each successive recording enters the data pool and changes the average, as table 2-3 illustrates for the first four performances.

83

Table 2-3. Illustrative method for calculating correlation with respect to performance chronology and to resulting evolving arithmetical mean Perf. 1 Perf. 2 Avg.1: 1,2 Perf. 3 Avg.2: Perf. 4 Avg.3: 1,2,3 1,2,3,4 Tempo, Tempo, Tempo, Tempo, Tempo, Tempo, Tempo, Beat 1-2 Beat 1-2 Beat 1-2 Beat 1-2 Beat 1-2 Beat 1-2 Beat 1-2 Tempo, Tempo, Tempo, Tempo, Tempo, Tempo, Tempo, Beat 2-3 Beat 2-3 Beat 2-3 Beat 2-3 Beat 2-3 Beat 2-3 Beat 2-3 Tempo, Tempo, Tempo, Tempo, Tempo, Tempo, Tempo, Beat 3-4 Beat 3-4 Beat 3-4 Beat 3-4 Beat 3-4 Beat 3-4 Beat 3-4

As each new set of tempo data from a new recording enters the data pool, the mathematical mean reflecting its inclusion supersedes the previous mean in order to reflect only the performance data and the new average. For analysis of performances 1 and 2 in table 2.5, three data sets comprise the sequences for polycorrelation: performance 1, performance 2, and average 1. For analysis including the first 3 performances, four data sets comprise the sequences: performance 1, performance 2, performance 3, and average 2. Including averages in this way makes the analysis of a later performance account only for its relationship to preceding performances, and the average to which it contributed, not of the previous average values (i.e. – the analysis of performance 10 in relationship to the whole body of data includes sequences for performances 1-10, and only average 9).72 Thus, the data set for this approach when considering the entire body of recordings (19) in the given time frame (1961 – 2014) contains 20 sequences (one per recording, plus one additional for the average). This

72 Another option here for later study is the analysis of all performance data sequences and all averages preceding it chronologically. This TW? would enable one to relate a performance in 1996 to the average of the first two performances in the study, and would result in a data set containing 33 sequences (17 recordings, 16 averages). A third option includes averages between all sets of performances (each performance in combination with 1 other performance, 2 other performances, 3 others, and so on.

84

approach exists only to reveal a chronological evolution of a tradition. Determining influence from the kinds of quantitative data and analysis in this project is not possible.

Furthermore, notions of influence are strikingly similar to those of authenticity, which the current project has no aspirations of establishing or perpetuating. This project thus ultimately differentiates between convention, which implies some sense of an active expectation or authenticity, and tradition, which emerges as a passive and resultant rather than governing component of a musical work.

Chapter 6 uses network mapping and analysis to highlight relationships both among the performances of each of the aforementioned six musical moments, and to the average, or quantifiable tradition of tempo rubato since 1961. Pearson’s product moment correlation allows the comparison of averages of individual performances

(global tempo) and restricts the resulting correlation values (coefficients called r values) to the range of -1  +1. An r-value of -1.0 represents an inversely correlated relationship (e.g. – one sequence increasing by a value of .5 at each point and the other decreasing by .5). An r-value of 1.0 reflects an exact math (e.g. – both sequences increasing by a value of .5 from point to point). An r-value of 0.0 indicates no detectable relationship between compared sequences. The following equation defines Pearson correlation. X and y are number sequences of beat-to-beat tempo data of equal length – the length of each excerpt in beats. 푋̅ and 푦̅ are the average values of the numbered sequences (global tempi).

∑ (푥 − 푥̅)(푦 − 푦̅) (2-1) 푟(x, y) = 푛 푛 푛 2 2 √∑푛(푥푛 − 푥̅) ∑푛(푦푛 − 푦̅)

R-values alone can be difficult to understand in relationship to musical performance. As such, correlation plots are useful for illustrating them. These show

85

graphically to what extent two recordings exhibit similar patterns.73 Individual correlation plots only account for similarity and difference between two sequences of data. They are thus not amenable to representing large sets of data. Polycorrelation plots, however, represent the comparison of tempo data of multiple recordings against each other, using color-coding to distinguish the various recordings from each other. While these compare recordings, they also reflect how each recording relates to the mathematical mean of the data sets (in the standard color-coding, black represents the average).

Polycorrelation plots reveal more information about similarity and difference between performances than correlation plots or Time Scapes. But, like these,

Polycorrelation plots cannot display a complete picture of how various performances relate to each other, or the average. Given the connective underpinning of performances of a common work and the traditions they create, network analysis is a useful tool for understanding the structure and nature of their relationships. Correlation network diagrams graphically and spatially represent in a single plot how recordings relate to each other, based on the analyzed data sets. For the final step in this study, a tempo similarity network will illustrate how each performance relates to others and to the average, which is a quantitative parallel to qualitatively defined tradition. A prototype network reflecting performances of Chopin’s Mazurka in c# minor, Op. 63, no. 3 appears as figure 2.7.

73 The Scape Plot Generator developed by CHARM researchers can create Time Scapes and single Correlation Plots. http://www.mazurka.org.uk/software/online/scape/.

86

Here, spatial arrangement is largely arbitrary, but Sapp grouped stronger connections with shorter lines. Line color and thickness between points (which represent individual performances) indicate strength of similarity. Sapp indicates “when a performance has no particular similarity to any other performance, a single dashed grey line connects it to its most (weakly) similar performance.”74 The numbers in the key are Pearson correlation r-values. With Sapp’s prototype (figure 2.7), a further layer of analysis of performance clustering is applicable. Sapp considered performers’ nationalities and prominent individuals and discovered distinct groups for: older

Russian; modern Russian and Polish; Hungarian French, English, American; southern

European disciples of Paderewski; and those closest to Rubinstein.75

74 Sapp, “Computational Methods,” 36.

75 Ibid., 37.

87

Figure 2-7. Prototypical tempo similarity network.76

Conclusion

While such considerations are relatively straightforward for solo piano, understanding them in the context of opera is inevitably more complex. Since the art world of opera production has so many moving parts, the first step is to try to ascertain which members either maintain or deviate from tradition (the mathematical average).

Any number of persons could be responsible for shifting or standing firm: singer,

76 Ibid., 36.

88

conductor, chorus master, production designer, stage director, or general manager. The final chapter will consider how any of these persons were involved in performances correlated to tradition, both those with strong and weak statistical correlation to tradition.

This process is a first step in understanding who among these members of a production’s artistic personnel may have acted as purveyors or dissidents of tradition.77

In so doing, this document will establish a means to come closer to defining performance tradition in Puccini’s Turandot at the Metropolitan Opera and understanding who the purveyors of that tradition have been. The analysis of individual performances at the level of their basic sonic phenomena (tempo) offers unprecedentedly detailed information about these performances’ respective iterations in and across time. Analyzing this data for a chronologically progressive arithmetical mean provides a similar glimpse of performance tradition as it evolves, and in relationship to various elements of the performance history. Finally, network analysis better equips scholars to understand the dynamic natures of performance and tradition, and how they interrelate.

77 As statistical correlation shows only similarity or difference, more work is necessary for making such a determination. Such an investigation would require source material (such as interviews, memoirs and similar sources) from the persons involved in order to understand their respective performance psychologies and motivations for their stage actions.

89

CHAPTER 3 PROFILING PERFORMANCE VIA TEMPO HIERARCHY ANALYSIS

Rather than trying to change directly the language on music, it would be better to change the musical object itself, as it presents itself to discourse. --Roland Barthes Image-Music-Text1

Rethinking Musical Objects

Musical activities, especially operatic ones, are inherently social, with performance being the primary interface between artistic producers and consumers.2 All practice involves experience in one way or another; all products resulting from operatic endeavor are therefore consumable phenomena. The score and performance are the two most visible components of a musical work, and yet neither alone is the entirety of the work itself. As such, performance analysis is an essential component of understanding any operatic work. This view is echoed by Sergio Durante, who stipulates that considering singers and their impact on opera “is such a natural part of any history

[of it] that it needs no justification.”3 This reality highlights the importance of studying singers and what they do on stage for a more comprehensive understanding of opera.

Until recently, notational encoding has formed the basis for scholars’ focus on performance, resulting in text-based, rather than sound-based analysis. Such an approach examines components of operatic (and musical) analysis at the cost of examining how performers realize the information of a score.

1 Barthes, 180.

2 Here, I use producers and consumers in the same sense espoused in: Howard Becker, Art Worlds (Berkeley: University of California Press, 1982), 1-4. Producers are those who are responsible for bringing any work of art to its audience and critics—the consumers.

3 Sergio Durante, “The Opera Singer,” in Opera Production and Its Resources, ed. Lorenzo Bianconi and Giorgio Pestelli (Chicago: The University of Chicago Press, 1998), 345.

90

Once scholars establish the means to understand performance practices more accurately, establishing the unique identity of each performance is possible, and discerning performance tradition is feasible. Opera performances are simultaneously products of experiences and accumulations of musical phenomena, such as tempo, phrasing, pitch relationships, breathing, rhythm, dynamics, articulation, and diction.

Phenomenology asserts that reality results from the perceptions of experiences in human consciousness and not from anything independent of that consciousness. Thus, phenomenology, conceived as relating empirical observations of phenomena to each other, offers a natural origin point for opera, which affects both its practitioners and consumer experience through perception and even consciousness. It opens an empirical space to understand those practices and behaviors that constitute a performance and, over time, construct a tradition. Responding to Roland Barthes’s exhortation to fashion a new musical object for discourse, presented in the epigraph above, this chapter begins with an analysis of the relationship between scores, performances, and a musical work and concludes by exhibiting a new method of documenting performance in order to enable new discourse.

Scores, Performances, and Works

This document seeks to forge new connections between quantitative and qualitative analyses. Considering these connections in the context of Edmund Husserl’s foundational work in phenomenology can help illuminate their significance. Husserl developed a phenomenological reduction concerning a subject’s wesen (its “essence”).

Perceptions of an event are the result of that event’s discrete phenomena, which, occurring in a chronological sequence, are measureable against the time in which they transpire. Husserl’s work laid the foundation for one of the foremost philosophers of the

91

twentieth century, Martin Heidegger. Heidegger’s Sein un Zeit (Being and Time) and

“The Question Concerning Technology” examine the emergent technological world of the mid-twentieth century.4 For him, “technology is a way of revealing” because it is “that which belongs to technē,” which consists of the skills and practices of a craftsman—i.e. the work of hands—along with “the arts of the mind and the fine arts. Thus, technē belongs to bringing-forth, to poiēsis; it is something poietic.”5 Written language is perhaps the most basic form of technology.

Writing is a ‘machine’ to supplement both the fallible and limited nature of our memory (it stores information over time) and our bodies over space (it carries information over distances). So it’s not so much that we humans made technology: technology also made us. As we write, so writing makes us. It is technology that allows us history, as a recorded past and so a present, and so, perhaps a future. So to think about technology, and changes in technology, is to think about the very core of what we, as a species, are and about how we are changing. As we change technology, we change ourselves.6

In addition, writing reflects the thoughts and creativity of an author. This principle remains valid regardless of the kind or medium of writing in question. Thus, written music represents not only a composer’s creativity, but rather the result of a technological practice that encodes ideas in a transmissible form. Musicologists have by now accepted that viewing the score as the sole representative of a musical work substantially limits understanding of a work. In fact, writing music conceals the actual musical entity: the ideas that lead to the sonic phenomena. This concealment occurs

4 “Wesen” is notoriously problematic in translation. While acknowledging the normal translation, I have elected to leave this word untranslated as its meaning is far fuller than “essence.” See Martin Heidegger, The Question Concerning Technology and Other Essays, Translated by William Lovitt, (New York, NY: Harper Perennial, 1977), 36.

5 Heidegger, The Question Concerning Technology, 13.

6 Robert Eaglestone, Contemporary Fiction: A Very Short Introduction (Oxford: Oxford University Press, 2013), 87.

92

because of philosopher Jacques Derrida’s stipulation that “Writing reduces the dimensions of presence in its sign.”7 Thus, a written score is Heideggerian “Enframing” because it “conceals that revealing, which, in the sense of poiēsis, lets what presences come forth into appearance.”8 The score enables the sounds of a musical entity (what

Heidegger characterizes as that which “presences”) to come into (aural) appearance repeatedly, but is not itself the musical entity.9

Lingering adherence to the conception that the score is the only authoritative musical entity by some reflects what Derrida considers “the epoch of logocentrism,” which “is a moment of the global effacement of the signifier [the written word].”10 This effacement creates a mode of thinking in which “one then believes one is protecting and exalting speech [by adherence to the written text], [but] one is only fascinated by a figure of the technē.”11 This fascination stems from Derrida’s steadfast belief in Jean-

Jacques Rousseau’s assertion of two hundred years earlier: “Writing is nothing but the representation of speech; it is bizarre that one gives more care to the determining of the image than to the object.”12 In other words, written representations, because they are representations, are not the ideas that they represent, and yet some promulgate the

7 Jacques Derrida, Of Grammatology, trans. Gayatri Chakravorty Spivak (Baltimore, MD: The Johns Hopkins University Press, 1974), 281.

8 Heidegger, The Question Concerning Technology, 27.

9 Heidegger characterizes “coming into being” by using “presences” in a verb form to reflect an internal impetus for this process rather than an external one. Alternatively, this can also reflect any internal aspects of any process by which what Heidegger terms “revealing” or “unconcealing” comes to pass.

10 Derrida, Of Grammatology, 285-6.

11 Ibid., 286.

12 Ibid., 27.

93

idea that the textual representation of an idea is more important to the idea itself.

Similarly, speaking produces auditory phenomena representing the idea, rather than the idea itself. Derrida’s analysis of written language thus concurs with Rousseau’s stipulation: “The art of writing does not depend at all upon that of speaking.”13 Hans

Georg Gadamer echoed these conclusions, arguing that the written component of an aesthetic entity lies in its ability to connect the past and present.

. . .the word [written work] is called upon to mediate between past and present. . .Neither the being that the creating artist is for himself – call it his biography – nor that of whoever is performing the work, nor that of the spectator watching the play, has any legitimacy of its own in the face of the being [or the coming into being] of the artwork itself.14

Gadamer’s stipulation about written works also applies to musical practice. It implies that there is no inherent codependency or inherent importance differentiation possible between a written score and the sounded notes. “The fact that aesthetic being depends on being presented, then, does not imply some sort of deficiency, some lack of autonomous meaning. Rather, it [aesthetic being] belongs to its very essence. The spectator is an essential element in [what] we call aesthetic.”15 Gadamer goes so far as to claim that it is in performance, and not in the written score that one encounters the musical work, advocating an embrace of opsicentrism. This idea suggests that what is more important is the action of bringing a work into being, what I term “opsicentrism,”

13 Ibid., 294.

14 Gadamer, Truth and Method, 129.

15 Gadamer, Truth and Method, 130.

94

rather than the text of a work itself, unlike “logocentrism,” which preferences the written word as the primary embodiment of a musical work.16

The performance of a play, like that of a ritual, cannot simply be detached from the play itself, as if it were something that is not part of its essential being, but is as subjective and fluid as the aesthetic experiences in which it is experienced. Rather, it is in the performance and only in it – as we see most clearly in the case of music – that we encounter the work itself. . .A drama only really exists when it is played, and ultimately, music must resound.17

Of course, the challenge with music is its evanescent temporality—its sounded and intangible existence ceases almost as soon as it comes into being. The ephemerality of a performed music entity requires something to continue in order to allow its ongoing realization. Each realization (performance) of a musical work is the revealing that brings it forth: poiēsis.18 “[A work’s] own original essence is always to be something different (even when [performed] in exactly the same way. It has its being only in becoming and return.”19 Furthermore, “a written score contains only the most basic of expressive instructions. The composer relies on the performer to interpret the work according to implicit rules as well as the written instructions.”20 Put another way, as

16 I use opsicentrism to describe the view, advocated by Gadamer, that the performance is closer to the Heideggerian ‘wesen’ of the musical work than the score can ever be. While Gadamer’s stance might at first seem a reactionary position to adopt, it bears consideration. With regard to musical meaning, and what actually the musical entity ‘is,’ the understanding of the meaning between what appears on paper and what one hears is of utmost pertinence.

17 Gadamer, Truth and Method, 120.

18 Heidegger, The Question Concerning Technology, 29.

19 Gadamer, Truth and Method, 126.

20 Craig Stuart Sapp, “Computational Methods for Analyzing Musical Structure” (Ph.D. Diss – Stanford University, 2011), 111.

95

Nicholas Cook avers, “Music’s communicative potential—its ability to enact openness to the other—. . .is built into practically all musical performance and listening.”21

Musical works, which, by their nature intersect again and again with audiences only in their coming into being or presencing (Anwesen) via the performers bringing- forth (Her-vor-bringen), are convincing testaments to humans’ inability for exact mechanical reproduction.22 Gadamer expands on music’s Hervorbringen as evidenced in its occasionality. “The work of art. . . experiences a continued determination of its meaning from the ‘occasion’ of its coming-to-presentation. This is seen most clearly in the performing arts, especially theater and music, which wait for the occasion in order to exist and define themselves. . .”23 Thus, acknowledging the viability of opsicentrism is an integral step in coming closer to encountering the musical work.

Rather than blindly accepting either logocentrism or opsicentrism and abandoning the other, however, I suggest a reinvestment in Heideggerian ideals.

Neither the score nor the performance contains the ‘wesen’ of the musical entity, thus neither one alone constitutes that entity completely. Instead, both point to it, and the aesthetic being that is music is perhaps like a performance of music itself: intangible.

Because the score endures, and is therefore an embodiment of the Heideggerian

Enframing of music, the score is the essencing of technology, the technē of the musical work. Similarly, “essential to dramatic or musical works, then, is that their performance

21 Nicholas Cook, Beyond the Score: Music as Performance (Oxford: Oxford University Press, 2013), 412.

22 Heidegger, The Question Concerning Technology, 10-11.

23 Gadamer, Truth and Method, 147.

96

at different times and on different occasions is, and must be, different. . .specifically occasional art forms determine [themselves] anew from occasion to occasion.”24

In accordance with Heidegger’s foregoing characterization of the relationship between technē and poiēsis, both a musical score and its realization point to the revealing of the musical work. Heidegger reminds his audience that “There was a time when it was not technology alone that bore the name technē. Once that revealing that brings forth truth into the splendor of radiant appearing also was called technē. Once there was a time when the bringing-forth of the true into the beautiful was called technē.

And the poiēsis of the fine arts was also called technē.”25 The resolution of the

Heideggerian danger of viewing scores as musical works in and of themselves lies not in disparaging this view. Rather, the danger lies primarily in limiting an understanding of technē to only one-dimension. For Heidegger, the antidote to this danger, its saving power, lies in the truth that technē multi-modal nature. For in understanding that both performances (“oral” language) and scores (“written language”) enable a revealing, both are forms of technē, and each therefore belongs to poiēsis. Those who endeavor in either, in the pursuit of revealing, demonstrate how “poetically dwells man upon this earth.”26

These theoretical concepts emerged in a very concrete way decades before

Heidegger ever expressed them. The dual modality of encountering a musical work did not escape Italian publishing magnate Giulio Ricordi, manager of the publishing house

24 Ibid., Truth and Method, 147-8.

25 Heidegger, The Question Concerning Technology, 34.

26 Heidegger, The Question Concerning Technology, 34.

97

Casa Ricordi from 1888 until his death in 1912. In a 1906 Casa Ricordi board meeting, the recent emergence of motion pictures and its possible effects on the firm’s business interests was a topic of conversation. With insight that predated the emergence of Alan

Crosland’s The Jazz Singer (1927) by more than 20 years, Ricordi anticipated the effect that film would eventually have on opera, though perhaps not to the same extent that he feared.

A Lodi, a Crema, a Monza, potrebbero riprodurre La Traviata come in rappresenta ora alla Scala e di comprende [sic] come il pubblico di quei piccoli teatri andrebbe in folla ad udire, sia pure più o meno bene riprodotte, le abilità canore di artisti che mai avrebbero sognato di avere nelle loro città. E così si rinuncerebbe ad organizzare spettacoli relativamente costosi e che necessitano un certo prezzo d’ingresso, mentre col mezzo del cinematografo, il pubblico si divertirebbe spendendo solo venti e trenta centesimi.27

Ricordi’s objection to the reproduction of the sounds of operas for which his firm held copyrights was strong enough to motivate him to address what he considered clear violations of his firm’s rights. On one occasion no later than early 1906, he discovered that someone associated with the copyright collection agency Società Italiana degli

Autori had not consulted him before showing a moving picture of portions of Verdi’s

Otello (the rights of which were held by Ricordi) in Rome. Because this was still in the

“Silent Film” era (before 1927’s The Jazz Singer), a recording of Francesco Tamagno singing music from those scenes accompanied the showing. This occurrence understandably incensed Ricordi, who wrote to the Società, demanding that they fulfill

27 Umberto Campanari, Libro delle sedute del Consiglio di Vigilanza 1900-1940, minutes from 27 February 1906, from the Archivio Storico Ricordi. “In Lodi, Crema, and Monza, they could reproduce La traviata as now staged at La Scala, and understand how the public of those small theaters would go in droves to hear, albeit more or less well reproduced, the singing skills of artists that they would never have dreamed of having in their cities. And in this way one would forgo organizing relatively expensive performances that require a certain entrance fee, while the public would be entertained by means of the cinema, spending only twenty or thirty centesimi.”

98

one of their aims: protecting performances of written works, and by extension, reproductions of performances. “Di conseguenza se la vostra Società crede di occuparsi di tutelare la esecuzione di pezzi di opere con fonografi che si possano dare in pubblico, lo faccia.”28

This episode led the Ricordi company to invoke the dual Italian legal protections applicable to artistic works: diritto d’edizione to protect publishing, reproducing, and selling of printed works, and diritto di rappresentazione, to protect the right to authorize performances of a work.29 The Italian legal code that Casa Ricordi invoked managed to reserve the right to grant or deny permission for any performance of an opera within the first forty years of its composition. Ricordi’s application of this code included Verdi’s works from Aida onward, and all of Puccini’s oeuvre.30 Ricordi’s legal maneuvering in response to cinema and the reproduction of operatic scores and performances reveals his conception of the nature of an operatic work: texts constitute only one avenue by which interested parties can access an opera, because opera inheres in performances just as greatly as it does in texts themselves. This example, paired with Puccini’s own evident conceptions of the dynamic, performance-based nature of his works, reinforces realizations as technē, just as scores also serve to reveal aspects of a work’s poiētic nature.

28 Giulio Ricordi, letter to the Società Italiana degli Autori 23 March 1906. This letter appears in Coppialettere 1905-1906, vol. 15 #59. The volumes of Coppialettere hold archival copies of all out-going correspondence form the Ricordi main offices in Milan. “So if your company believes that it should protect the performance of pieces of works with phonographs that can be given in public, do it."

29 Christy Thomas, “When Opera Met Film: Casa Ricordi and the Emergence of Cinema (paper presented at the first Transnational Opera Studies Conference, Bologna, Emilia-Romagna, Italy, 30 June- 2July, 2015).

30 Ibid.

99

New Musical Objects for Discourse

The remainder of this chapter demonstrates a manner of profiling the tempi recorded in available recorded broadcasts of Turandot at the Metropolitan Opera for the period 4 March 1961 – 7 November 2009. This method of profiling a performance’s tempo properties creates new musical objects for discourse, to recall Barthes’s notion presented above—objects that exhibit and therefore render analyzable the unique surface and structural features of any performance. Here, I analyze each of the segments of the opera detailed in Chapter 1 for both the first and most recent available recordings in the corpus. In addition, this discussion offers detailed analysis of one recording of each extract that contains significantly different features in the tempo hierarchy, demonstrating some of the capabilities of the analytical method in use can facilitate in future investigation. Tempo is a logical first step in the analysis of musical performance, as it compares only the passage of beats against time—a process that is well within the capabilities of current technology, and one that, as Chapter 2 emphasized, is central to understanding Puccini’s works. Future analysis of other physical properties of sounded music—dynamics, pitch, timbre, articulation, phrasing, and diction—will only enhance understanding of what a performance was or is.

While the following graphs are each unique, owing to their respective representations of individual performances, some general principles for understanding and interpreting them are useful. As Chapter 2 elucidates, each graph’s color-coding reflects the performance’s tempo relationships compared against each its global tempo.

As most of the tempo modulations are ritardandos, being aware that their musical causes (ritardandos) at the level of one or two beats typically include notated and non- notated fermatas, tenutos, and other note-specific articulation marks is imperative. Such

100

instances highlight individual words or portions of words, which, when their extensions’ colors extend into the middle and higher structural levels, can help form hypotheses about a performer’s interpretation. Extensions radiating from sections imply temporal emphasis on the text and music within them imply temporal emphasis on some aspect of these contents. These moments are useful in constructing a more comprehensive projection of what a performer’s interpretation of an aria’s musical and dramatic content might be on a given performance. As these extensions cross at higher levels of a performance’s tempo hierarchy, their importance in the overall structure of that performance’s identity increases. As the following graphs show, most of the temporal- textual emphases that appear result from the varying applications of tempo rubato

(including ritardando and stringendo), and not simply from the beat structure of Puccini’s settings.

The following analyses arrange the Time Scape and analysis of each musical segment as it appears in the score (“Signore, ascolta!”, “Non piangere, Liù,” “In questa

Reggia,” “Straniero, ascolta!” “Nessun dorma!”, “and “Tu che di gel sei cinta”). Each aria’s analysis is in two parts, first an examination of surface level events, and then of the entire tempo hierarchy. The discussion of highlighted arias will proceed in the same order, and with the same demographic information. For demographic information and

Time Scapes of all recordings included in this corpus, see Appendix A – Demographics and Documents of Performance.

Reading an Operatic Time Scape

Chapter 2 outlined the quantitative structure of building a Time Scape. Since

Time Scapes form the basis of the following three chapters, an overview of reading them with the particular overlay in use here is in order. I used CHARM’s web-based

101

Scape Plot Generator to create each of the Time Scapes in this document.31 Four elements accompanying each Time Scape given here nuance the understanding one can draw from them. First, each Time Scape has a structural overlay indicating approximate divisions of surface (beat-to-beat tempo relationships), mid-ground

(phrasal connections) and background (global tempo). The second element is the color- coding key reflecting tempo relationships to each Time Scape’s global tempo. Below each Time Scape is a textual underlay, which serves both to help listeners locate themselves in a recorded performance, and to reveal temporal-textual emphases.

Finally, a tempo contour graph indicates precise tempo values of the surface level, while a linear trend line reflecting the overall pacing contour of the performance.

The bottom third of each Time Scape reflects the structural surface of a performance. The color-coded tempo relationships here—yellow, orange, and red are progressively faster than average, which is green, and blue indigo and violet are progressively slower than average—are discernable in listening to a performance

(recorded or live). The tempo contour graph is another representation of the same information, but with the specific tempo values discernible. Tempo contour graphs cannot, however, reveal definitive connections between non-neighboring parts of the text, as they reflect only the surface and not the mid-ground or background of the time and tempo structure or a performance. The middle third of each Time Scape reflects the structural mid-ground of a performance, and in this space connections between non- neighboring parts of the text can emerge. While it may be possible to discern such connections in listening alone, they generally become more difficult to discern as one

31 http://www.mazurka.org.uk/software/online/scape/

102

moves higher in the structure. The top third of each Time Scape is predominantly green, unless there are connections between parts of the text at the beginning and end of an excerpt.

On the surface level, deviations from the global tempo have the effect of emphasizing or de-emphasizing different elements of the text; sometimes these result from indications in the score such as a fermata or ritardando, while at others, they result from performer decisions. In both instances, performances have unique temporal- textual emphases, which are places where a performer’s use of time emphasizes or deemphasizes particular words or musical phrases. If one spends more time delivering the word “amore” than the word “colore,” then the word “amore” receives a greater temporal-textual emphasis, suggesting the speaker’s designation that it is the more important word. Of course with texted music, the composer’s setting of text influences this as well, though sometimes performers re-interpret the composer’s notation to fit their own understanding. On their own, each of these surface-level emphases highlights individual words, but when their emphasis crosses another one in a higher (or deeper) structural level, a structural crossing links the two words. These crossings, which occur in the mid-ground and background of a performance’s tempo hierarchy, are evident by the intersections of color extensions from the graph’s surface level. Analysis of structural crossings can reveal the unique timing structure each performer imposes on the text and what it communicates, thus providing fresh and nuanced insight into a performer’s interpretation of a text. While Time Scape analysis of different performances of the same work reveals unique musical behaviors, it can also point toward the unique interpretations or other mitigating factors that influence practice.

103

Analysis of Tempo Hierarchies from 4 March, 1961

The performance of Turandot on 4 March 1961 is a natural origin point in this analysis as it was this opera’s first broadcast, its second performance in the 1960-61 season, and, most important, the first recording of it at the Metropolitan Opera. Turandot received sixteen performances this season, and this performance was the 29th in this opera’s performance history at the Met. Birgit Nilsson sang Turandot, Franco Corelli was Calàf, Anna Moffo was Liù, and Bonaldo Giaiotti was Timur. Leopold Stokowski conducted.

The tempo data generated by analyzing Anna Moffo’s rendition of “Signore, ascolta!” on 4 March, 1961 produces the Time Scape shown in Figure 3-1. This figure’s accompanying tempo contour graph reflects the precise tempo values for Moffo’s performance, and the linear trend line indicates that the overall arc of her tempi slows from the upper to the lower 30s. While tempo contour graphs are useful for this kind of analysis (Moffo slows throughout her performance), they do not reveal structural crossings of temporal-textual emphases. Heeding Timur’s plea for Liù to address Calaf,

Moffo begins the aria significantly faster than what will become the global tempo as she addresses Calaf directly. Moffo’s performance first slows as Liù’s words turn to describe her own heartbreak at having carried Calaf’s name in her soul and on her lips for so long.32 Turning Liù’s attention back to Calaf, Moffo and Stokowski return first to the global tempo, and then to a tempo slightly faster than average as she exclaims to Calaf that his actions will result in her death, and the death of Timur. The tempo slows again

32 “Si spezza il cuor! Ahimè, ahimè, quanto cammino col tuo nome sull’anima, col nome tup sulle labra!”

104

as Liù indicates the location of their death, “sulla strada dell’esilio” (on the road of exile), and explains that Timur will lose his son, and she will lose the shadow of a smile (from

Calaf).33 Moffo’s performance speeds up as she delivers the closing plea for Calaf to have pity.

The opening stringendo is the most significant tempo modulation of this kind, given its height (no other pushing of the tempo reaches higher into the hierarchy) and its color (red is fastest in relationship to the global tempo). The two significant ritardandos occur at Liù’s two descriptions of herself in the case that Calaf fails to answer

Turandot’s riddles (b. 17-30, 46-56). Their structural conflation throughout the entire mid-range of the tempo hierarchy indicates that, from a temporal perspective, the emphasis on these moments is integral to the identity of Moffo’s performance. Given the dominance of this feature in the structural hierarchy, postulating that for Moffo and

Stokowski, Liù’s anguish is the most important aspect of the drama to communicate in this aria’s performance is entirely reasonable. This is in line with what one might expect of a conventional interpretation of this aria and this role. Other examples (Figure 3-7 and Figure 3-13, e.g.) reveal alternative interpretations, even if the alterations are nuanced.

33 “Ei perderà suo figlio, io l’ombra d’un sorriso!”

105

Figure 3-1. Time Scape, text, and tempo contour graph (with linear trend line) for Anna Moffo 04 March 1961 "Signore, ascolta!"

106

Figure 3-2 reveals surface level events and textual emphases readily apparent from listening to Corelli’s rendition of “Non piangere, Liù” on 4 March 1961 a few times.

On the hierarchical surface, slight ritardandos appear in varying shades of blue or purple as Calaf begins to address Liù, and again as he characterizes her as “dolce mia fanciulla.” The next section of the aria, in which Calaf explains why he implores Liù to listen to him—his possible death and thus abandoned father—the tempo increases slightly (yellow) above the global tempo of the performance, only to slow again when implores Liù, “Non lo lasciare, portalo via con te.” Both Moffo and Giaiotti push the tempo at the beginning of Liù’s and Timur’s respective interjections. Corelli remains near the average tempo before slowing significantly to set up the return to the aria’s primary theme with “questo, questo.” Corelli then surges through the melodic return as

Calaf’s text focuses on concern for Timur prior to observing the aria’s lone marked fermata on the melodic climax, which Puccini linked to the negation of Timur’s smiling by aligning it with “non” in “non sorride più.”

107

Figure 3-2. Time Scape, text, and tempo contour graph (with linear trend line) for Franco Corelli 04 March 1961 "Non piangere, Liù."

108

In higher levels of the tempo hierarchy, several swatches of color extend from groups of the surface level events. The height of these swatches indicates the structural significance of the collections of surface level events. Thus, the first stringendo is significant in the tempo hierarchy of the entire performance. Calaf’s subsequent two ritardandos are less significant independently, but the crossing of beams indicates that these two events combine to produce a similar level of hierarchical significance to the preceding stringendo. The faint crossing about half way up the vertical axis indicates the greatest tempo-structural significance of the performance and links “Non lo lasciare, portalo via con te” with “che non sorride più.” Such a crossing might provide insight into

Corelli and/or Stokowski’s understanding of the drama at this point on this occasion. A reasonable inference is that the important dramatic emphasis in this aria is: “Do not leave him (38-44), oh my poor Liù (66-71), (him) who does not smile (84-88).”

Figure 3-3 reflects Birgit Nilsson’s performance of “In questa Reggia” on 4 March

1961. Here, tempo modifications reflect the musical structure of the aria: twenty measures of 2/4 marked molto lento (♩ = 46) (beats 1-40), sixty-eight measure of 4/8 interspersed with three measures of 2/8 marked with lento (♪= 66) (beats 41-111), and twenty-two measures of 4/4 marked largamente (♩ = 56) (beats 113-199). In the first section of the aria, as Turandot explains her historical connection to her ancestor, Lo-u-

Ling, and her ancestor’s scream, Nilsson hovers just above what will become the global tempo of the aria in this performance (38.7575 ♩/minute). For the second section’s explanation of the atrocity of the political conquest that eventually claimed the purity and life of Lo-u-Ling, the tempo drops significantly and in accordance with Puccini’s indication metronome marking.

109

Figure 3-3. Time Scape, text, and tempo contour graph (with linear trend line) for Birgit Nilsson 04 March 1961 “In questa Reggia”

110

Nilsson and Stokowski undertook the third section of the aria faster than the first or second, but not to the degree indicated in the score. Here, however, the range of tempo modifications is greatest, as the alternations of red and purple indicate. Two striking uses of ritardando occur as Turandot ascends to a b-natural, first with a fermata on the first word of her exclamation, “Ah, rinasce in me l’orgoglio di tanta purità!” (Ah, pride of such purity is reborn in me!), and the second (without fermata) on the “–glio” of l’orgoglio. Nilsson then accelerates through her admonition to Calaf not to tempt fate before slowing on the articulation-and-textual-instruction-unmarked crowning note (g- natural) of her final warning, “gli enigmi sono tre, la morte è una!” (The riddles are three, the death is one!). Corelli responded in musical kind, emphasizing the first syllable of his utterance of “sono” on an a-natural, though this appears with a tenuto line and “ten.” directly above the note.34 As Calaf and Turandot utter their challenges to each other simultaneously, they ascend in octaves to the high-c, which Puccini marked with a fermata. These practices account for the purple sections occurring between beats 170-

190.

While there are structural crossings for the two sections of stringendo linking the events underneath them, their hierarchical significances pale in comparison to the importance of the aria’s second section. This section dominates the tempo hierarchy and faintly crosses an upward extension from the ritardandos occurring at the end of the aria. This provides a structural link between them and the temporal emphasis on the

34 Puccini, Turandot, 262. While this marking appears in every version of the score I have seen, I have not seen the autograph due to its restricted access in the Ricordi archive, and my still-pending access request. Thus, this marking’s provenance is not absolute. Nevertheless, as this document is about the performance of music, rather than its encoding in notation, such information could be informative about articulation interpretation. This avenue of inquiry, however, remains a project for future work.

111

retelling of the crime against Lo-u-Ling, perhaps indicating an interpretive connection that reinforces the textual material of the aria: Lo-u-Ling suffered horribly, and she, now reborn in Turandot, demands the death of any man who cannot outsmart her.

Act II’s dramatic crux, the riddle scene, is the longest and most complex musical material in the current study. Here, Figure 3-4 exhibits the varying degrees of ritardando that demarcate the delivery of each riddle. Strikingly, each of Calaf’s answers comes forth so near the global that only slight deviations are perceptible, perhaps reflecting stoicism in his character in this performance. The tempo pushes ahead at various degrees with each answer’s confirmation by the wise-men (I Sapienti), along with interjections by the crowd, emperor, and Liù. The most intense tempo increase occurs at Altoum’s interjection, which the crowd then echoes with rising fervor, “Non perderti, straniero. È per la vita! Parla!” (Don’t lose yourself, stranger. It’s for your life! Speak!, beats 273-305). Liù’s exclamation “È per l’amore!” (It’s for love!) is the only slowing of the tempo throughout this section (beats 306-309). Apart from the ritardando at the first riddle, the most significant occurrence of this kind of tempo modification occurs in the midst of the third riddle, as Turandot taunts Calaf that she has triumphed and incites him to admit defeat. The strongest ritardando here occurs as Turandot issues her ultimate challenge to Calaf, “Su, straniero! Ti sbianca la paura! E ti senti perduto!”

(Come, stranger! You are white with fear! You feel that you are lost!”). This is entirely reasonable, as stage directions in the score indicate that Calaf is barely breathing, and

Turandot is to be “curved as over her prey, and grinning.”

112

Figure 3-4. Time Scape, text, and tempo contour graph (with linear trend line) for 04 March 1961 riddle scene

113

Moderately significant structural crossings in the mid-ground clearly demarcate the second riddle and the stage and musical action between it and the third. In addition, given the metric scope of the third riddle, including Turandot’s taunting, the height of the crossing between the beginning of the riddle and the taunt elevates this segment’s temporal emphasis to a level just under that of the first and second riddle. Here, it might be as if Turandot’s taunt constitutes a feeble, fourth riddle. The alternation between faster-than-average and slower-than-average tempi here highlights the distinct roles of each of the cast members musically active in this scene.

Figure 3-5 represents Corelli’s rendition of “Nessun dorma!” from 4 March, 1961.

In this performance, Corelli slows significantly from the global tempo to address the

Princess reminding her that even the stars in the night she has decreed shall be a sleepless should one tremble with love and hope. Corelli slows most emphatically on “e di speranza,” perhaps asserting his confidence that she will not be able to answer his riddle before dawn, or perhaps even mocking the princess with his response to her first riddle in Act II. Corelli speeds up significantly at perhaps the most famous motive of the entire opera, only to slow to emphasize that his secret, the answer to his riddle to the princess, remains closed in himself. Corelli’s next significant rubato occurs at the tag of the first strophe of the romanza, as he slows to the fermata notated on the first syllable of “splenderà,” where he employs his highly distinctive and celebrated decrescendo.

The structural linking of these moments suggests Corelli’s sense that the hope-and- love-trembling stars yearn, like him to bask in the glory of the morning light.

114

Figure 3-5. Time Scape, text, and tempo contour graph (with linear trend line) for Franco Corelli 04 March 1961 "Nessun dorma!"

115

As Figure 3-5 illustrates, Corelli spends the remainder of this performance close to the global tempo, mixing rallentando with stringendo. He surges urgently through his command to the stars to depart before slowing in accordance with an indicated “poco ritard” to emphasize his moment of triumph, “all’alba vincerò.” He speeds through his second iteration of “vincerò,” and drastically pulls back at the final iteration, observing a fermata on the high B that is the last sixteenth note of the measure. Conductor Leopold

Stokowski stopped the orchestra in order to allow applause before continuing with what has become the standard ‘play-out’ for this aria. This practice appears in no other recorded performance constituting the current corpus.

The high-level structural crossings of Corelli’s many uses of rubato reflect links between Corelli’s tempo emphases on “splenderà” following his declaration that he will tell Turandot his name with a kiss then and “all’alba vincerò” and at this higher level his address to Turandot and conviction that he will win her at dawn. One possible conclusion from these structural features is that, for Corelli in this performance, this aria is less about Calaf’s ruminating in a monologue about the Princess’s decree and his determination and more about fashioning an address directly to Turandot.

116

Figure 3-6. Time Scape, text, and tempo contour graph (with linear trend line) for Anna Moffo 04 March 1961 "Tu che di gel sei cinta"

117

Figure 3-6 represents Moffo’s rendition of Liù’s suicide aria, “Tu che di gel sei cinta,” on 4 March 1961. On the surface of her performance, Moffo pushes the tempo as she describes Turandot as icy, but to-be-won by fiery passion (b. 1-17), slowing slightly to deliver her ultimate assessment of the situation, and again more significantly to repeat it: Turandot will also love Calaf (b. 18-28).35 She next employed ritardandos as

Liù declares her intention to close her tired eyes (b. 35-38), and then reveals (b. 40-44) and repeats (b. 47-50) Calaf’s pending victory as her motivation for her actions.36 The most significant slowing occurs as Liù divulges the extent to which she is going for

Calaf’s victory, an extent that will claim her life (b. 52-57).37 Liù then repeats herself textually and musically, but Moffo’s reiteration redistributes the textual and musical emphases. In the repeat, she slows to emphasize “questa aurora” (b. 65-69), the accelerates through reminding those present that she will close her eyes (b. 70-76) and again slowing, aided by three subsequent fermatas, on insisting that never seeing Calaf again assures his victory (b. 77-80). After Liù stabs herself (sometime between beats

80-82), the crowd anxiously and agitatedly pleads that she reveals Calaf’s name with her dying breath. The end of this aria is unique, both within the context of Turandot, and also within the wider context of all of Puccini’s arias, because of the amount of text

Puccini elected to repeat.38The rather large swatch of blue extending upward in the

35 “Tu che di gel sei cinta, da tanta fiamma vinta, l’amerai anche tu!”

36 “Io chiudo stanca gli occhi, perché vinca ancora. Ei vinca ancora.”

37 “Per non vederlo più!”

38 In all, Puccini set four lines of the text twice, accounting for 27 or 28 syllables, depending on one’s interpretation of rules of Italian poetry scansion (Prima di questa aurora (7), Io chiudo stanca gli occhi (7), Perchè Egli vinca ancora (7), Per non vederlo più! (6 or 7)). The only other aria in Puccini’s oeuvre which repeats more than 20 syllables of text is Tosca’s “Vissi d’arte!”, which reprises twenty-five. (Nell’ora del dolore (7), perché, perché Signore (7), perché me ne rimuneri così? (11)). Investigation into

118

horizontal-middle of the Time Scape indicates that Moffo’s presentation of Liù’s stated motivation and course of action combine to form a significant part of the dramatic thrust of this aria. Despite its small size as a surface level event, the emphasis on “l’amerai anche tu!” serves as the primary relational point of textual emphasis. That its upward extension crosses the upward extensions of all subsequent events indicates, temporally, that it plays a subtle, yet critical role in the tempo hierarchy. Interpreting the crossing resulting from this point and Liù’s final words, one might consider that a fundamental component of Moffo’s expression of this moment is the embodiment of utmost romanticism – dying with unrequited love for another love to flourish.

Analysis of Tempo Hierarchies from 7 November 2009

This instantiation of Turandot was the fourth broadcast of the 2009-2010 season, the second telecast (the first HD transmission), and the 274th performance of Turandot at the Metropolitan Opera. Maria Guleghina sang the title role, while Marcello Giordani sang Calaf, Marina Poplavskaya sang Liù, and Samuel Ramey was Timur. Andris

Nelsons conducted.

As Figure 3-7 illustrates, Marina Poplavskaya and Andris Nelsons employed mild and moderate (primarily yellow and orange) stringendo throughout their performance of

“Signore, ascolta!” on 07 November 2009. Given the comparatively low heights of these three swatches’ extensions, and that they indicate no structural or projected crossings, their effect on the overall tempo hierarchy is limited. They do however demarcate the beginning of each of the three major textual sections of the aria. First is Liù’s initial plea

the musical and dramaturgical implications of such comparatively anomalous text-settings is one area for future endeavors.

119

to Calaf about the result his assumed impending failure will have on her emotional state

(heartbreak)— “Signore, ascolta! . . .col nome tuo sulle labbra” (Master, listen! . . .with your name on my lips). Her concern about the effect on hers and Timur’s physical state

(abandonment on the road of exile) follows immediately— “Ma se il tuo destino. . .io l’ombra d’un sorriso!” (“But if your destiny. . .I, the shadow of a smile). Her plea for Calaf to consider her needs because she can bear no more is the aria’s coda in text, music, and performance as it displays a distinct tempo hierarchy feature, largely independent of what preceded it— “Liù non regge più! Ah, pietà!” (Liù cannot bear more! Ah, have pity!).

The structural crossing between the two larger sections of ritardando reveals a mild to moderate link between the cause and effect of Calaf’s intended pursuit, albeit in reverse order. The height of this swatch of blue and indigo establishes that this link might be the most significant aspect of this aria that Poplavskaya and Nelsons decided to communicate, though to a lesser degree than is evident in Figure 3-1. Two other features bear on the identity of this particular performance—the upward extension from the aforementioned ritardandos against the right edge of the graph approximately 1/3 and 2/3 up the vertical axis, respectively. These might imply that in this performance,

Calaf’s potential physical absence contributes to Liù’s heartbreak, but that the real tragedy for Liù is the emotional burden that comes from knowing she might never see

Calaf smile at her again. Nevertheless, Poplavskaya’s Liù on this occasion seems more concerned with Calaf and Timur than with herself, given that these latest structural features are decidedly faint and appear only in small areas.

120

Figure 3-7. Time Scape, text, and tempo contour graph (with linear trend line) for Marina Poplavskaya 07 November 2009 “Signore, ascolta!”

121

Figure 3-8. Time Scape, text, and tempo contour graph (with linear trend line) for Marcello Giordani 07 November 2009 “Non piangere, Liù”

122

Figure 3-8 is a visualization of tempo in Marcello Giordani’s interpretation of “Non piangere, Liù” on 07 November 2009. On the surface, Giordani delivers the opening words and “dolce mia fanciulla” (emphasis added) slower than the global tempo. He provides Calaf’s explanation that Timur may be alone in the world at dawn at a mildly- to-moderately quicker than average tempo before slowing significantly to exhort Liù to

“portalo via con te.” Giordani focuses Liù, following her slowed interjection that she and

Timur will die on the road of exile, with a significant slowing at “questo, questo.” The greatest temporal emphasis in Giordani’s performance builds on this, as Calaf re- addresses and re-characterizes the slave girl as “mia povera Liù” with the “piccolo cuore.” This is perhaps unsurprising, as Puccini indicated an espressivo here. Giordani then contrasted this by reminding Liù that her heart does not falter as he surged towards the melodic climax, placing Liù’s attention on Timur, before observing the fermata on the repetition of “che non sorride più.”

Several features of the tempo hierarchy in this performance are noteworthy. First, and most striking, are the heights to which extensions from the most significant stringendo and ritardando extend. More interestingly, they do not actually cross, but offer only a projected or implied intersection almost in the middle of the triangle. That they do not cross establishes the structurally significant independence of two specific points as the primary components of this performance’s identity. The first results as

Giordani pushed through his initial explanation to Liù (which is his concern for Timur) and the second from his slowing at Calaf’s re-focusing on Liù identity. Giordani has indicated that his interpretation in this moment generally tends to be Calaf’s focus on everyone as individuals, and that this universal concern offers at least a shadow of

123

connection amongst the characters.39 Furthermore, the blue and indigo shading along the right edge of the graph are increasingly higher structural projections from “O mia povera Liù,” Liù’s “sulla strada dell’esilio,” and “portalo via con te,” respectively. All of these cross with the subtle upward projection from the fermata emphasizing Timur’s inability to smile any more, and that the highest structural significance comes from the crossing with “portalo via con te” could mean an underlying or subconscious focus in

Giordani’s performance was a greater concern for Timur than for Liù.

The tempo relationships in Maria Guleghina’s 07 November 2009 performance of

“In questa Reggia” align neatly to each of the aria’s three meter changes, which Puccini used to establish a tripartite musical form. The music’s three primary formal divisions also mirror the three sections of the text Turandot delivers in her first face-to-face encounter with Calaf. In the introduction, Guleghina employed mostly only localized fluctuations in tempo as she invoked the millennium resounding scream of her ancestress, as Figure 3-9 indicates. A very low level connection between ritardandos on

“e quel grido” and “nell’anima” appears in blue near beats 20-30. A slightly higher level connection occurs between Turandot’s first words and the silence between the princess’s claim that her ancestress’s soul has taken refuge in her and the first utterance of “Lou-o-Ling,” as the crossing faint yellow extensions from near beats 5-17 and 33-37 indicate.

39 Marcello Giordani, interview with the author, 11 July 2013.

124

Figure 3-9. Time Scape, text, and tempo contour graph (with linear trend line) for Maria Guleghina 07 November 2009 “In questa Reggia”

125

The second section of the aria is significantly slower than the rest, which is unsurprising, given Puccini’s score indications.40 Unlike Nilsson’s 1961 performance

(see Figure 3-2), two downward extensions of green from the central mass of purple interrupt the stringendos in the first (c. beats 16-19) and third (c. beats 145-50) sections of the aria, respectively. The first of these links “un grido disperato risono” (b. 13- 18) and Turandot’s final exclamation of the second section, “Quel grido e quella morte (b.

110-113). One possible reading of this is that, on some deep level, Guleghina’s

Turandot on this occasion communicated an obsession with Lo-u-Ling’s “grido disperato,” and that the horror of this was in part Turandot’s motivation to behead failed suitors. Higher in the tempo structure, this extension also links “un grido disperato risono” to “di tanta purità” (b. 153-156). At this more fundamental level, a stronger possible motivation emerges: the millennia-travelling scream becomes a semiotic index of Lo-u-Ling’s purity, which Turandot is certain has re-emerged in herself.41 Guleghina’s tempi in this section dominate the structure, thus indicating that this link was the foundation of her performance’s identity.

The outline of a triangle indicating mild stringendo dominates the aria’s final section, its apex extending approximately halfway up into the tempo structure. This reveals that, as expected due to Puccini’s score indications, this section is, overall, mildly faster than either of the first two sections is. The increase in tempo, both in performance and in the score, might indicate an increase in warmth and emotional energy, given Puccini’s instruction con energia, and that the music here is the lushest of

40 See discussion following Figure 3-3.

41 “Ah, rinasce in me l’orgoglio’ di tanta purità!”

126

the aria. The fastest section in Guleghina’s performance occurs as Turandot and Calaf

(Marcello Giordani) throwing down the proverbial gauntlet, first separately, then in unison at the octave: “No! No! Gli enigma sono tre, la morte è una!” (Turandot), and

“No! No! Gli enigmi sono tre, una è la vita!” Overall, Guleghina’s performance, from a temporal-textual emphasis perspective most closely reflects the aria’s textual and musical structure at all levels of its tempo hierarchy.

Figure 3-10 reflects the tempo hierarchy of the riddle scene (“Straniero, ascolta!”) for 07 November 2009. As the text underlay exhibits, most of the tempo features here reflect the second-level of segmentation in the scene’s formal structure.42 Perhaps unsurprisingly, the tempo slows significantly at each riddle’s delivery, perhaps underscoring a deliberateness of communication Guleghina and Nelsons deemed appropriate for the gravity of this moment in the opera’s dramatic trajectory. Also, given the rising tension for Calaf and the “Popoli di Pekino” preceding each of his responses and the simultaneous tension for Turandot and anxious elation for the crowd immediately following each successful response, is understandable within what one might consider a conventional interpretation of this scene.

Unsurprisingly, links, albeit faint ones, between the two successive pairs of riddles (1&2, 2&3) appear in the mid-ground of this Time Scape. A strong link between the two response phases of these riddles’ respective postulations is also apparent in the mid-ground. The other large-scale structural feature in Guleghina’s tempo hierarchy is the pale link between the first and third riddle. This might appear to be straightforwardly

42 Here, I use segmentation to denote the levels of formal structure within the scene. Thus, the first level of segmentation is the entirety of each of the three riddle complexes, and the second is the subdivision of the events combining to form each riddle complex.

127

a result of these riddles’ respective locations in the aria, though recalling that higher structural levels reflect everything underneath them is imperative. With this in mind, what emerges is that the surface-level intensity of the slower-than-average tempi for each riddle combine to dominate the overall structure of Guleghina’s performance. The identity of this aria in this instance relies almost entirely upon the delivery of the riddles, and the answers, while necessary for propelling the dramatic action forward, are entirely subservient to the combined delivery of the riddles.

Though such a reading might seem less varied or interesting than others in this study, its immediate dramatic effect can heighten the pathos of Turandot’s plea to her father, “Figlio del cielo.” Projecting this effect into the third act produces two strong possibilities. First, Liù’s (Puccini’s ultimate ingénue) death resonates more strongly with

Turandot, perhaps making her process toward transformation more dramatically believable. Second, it sets up the moment of Turandot’s final transformation, the entendre-laden kiss from Calaf, as the logical outcome of Calaf’s proclamation at the end of Act II, “No, no principessa al terra, ti voglio (tutto) ardente d’amor!” and Liù’s dying prophecy, “Tu che di gel sei cinta. . .l’amerai anche tu!”43

43 Though this analysis might seem to be justification to advocate for performing in a stricter spirit of “come scritto,” it is only to demonstrate how the tempo hierarchy of one section of the opera can affect later events and developments in the plot. In this way, performances which appear to adhere to the textual or musical structure can be useful in better understanding those structural features, which in turn enriches understanding of a work and its identity. The (tutto) in Calaf’s proclamation indicates that this word appears in one version inscribed in the score, but is non-existent in the ossia with requires Calaf’s ascent to the tenor high-c, c’.

128

Figure 3-10. Time Scape, text, and tempo contour graph (with linear trend line) for 07 November 2009 riddle scene

129

Figure 3-11. Time Scape, text, and tempo contour graph (with linear trend line) for Marcello Giordani 07 November 2009 “Nessun dorma”

130

The Time Scape for Marcello Giordani’s performance of “Nessun dorma!” appears as Figure 3-11. In it, several various nested links of temporal-textual emphasis are clear. First, the extension from the slowing throughout the initial address to the

Princess, which is very similar to Franco Corelli’s 1961 version (see Figure 3-5), forms a link with the extension resulting from the slowing at the first syllable of “splenderà!” This latter extension also forms an implied link with the stringendo Giordani and Nelsons employed between the first and second phrase of Turandot’s most famous motive (b.

43-48). There is also a link between the ritardando on “splenderà!” and the ritardandos of Calaf’s proclamation that Turandot’s silence will make her his (“che ti fa mia!” b. 78-

82) and the chorus’s interjection (b. 83 – 96).

Extensions from each of these moments form links (of varying intensities) with one or more of the threefold climactic iterations of “Vincerò!” (b. 106-120). The distinct nesting of temporal-textual emphases in Giordani’s performance suggests that these moments, if considered in increasing hierarchical order, share a commonality of effect.44

The commonality of their effect, that Calaf will win Turandot’s heart, and hand in marriage, suggests the possibility of a commonality of motivation. In other words,

Giordani’s performance evokes a structure in which Calaf addresses everyone’s concerns through his love for all. This embodies Giordani’s interpretation of Calaf’s state of being at the aria’s point location in the opera. For him, “Lui ama tutti, perché a questo punto, lui è amor.”45 Love in its truest, self-sacrificial sense is the motivation Giordani attempts to project in his performances as Calaf. This is also what makes Liù’s death all

44 Understandably, this hierarchical structure depends largely on the imposed ordering of text and music by Puccini and his librettists, Adami and Simoni.

45 Marcello Giordani, interview, 2013.

131

the more moving for him as a singer who then embodies a character, and as that character. In Giordani’s view, this also makes Liù’s death a genuine turning point for

Turandot, rather than an awkward place in the opera’s composition for Puccini to have died, as the princess sees someone die for no personal benefit.46

The Time Scape for Marina Poplavskaya’s performance of “Tu che di gel sei cinta” reveals that the most significant structural temporal-textual emphasis accentuates the second half of the aria. The slight stringendo over the beginning of the aria (b. 1-23) and the lower-level average tempo (green, b. 26-56) features form the two primary building blocks for the tempo structure of the first half of Poplavskaya’s performance. In the first of these, the temporal-textual emphasis links Liù’s direct address of Turandot

(“tu”) to her insistence that the princess will also love Calaf before dawn (“l’amerai”), specifically her emphasis of the future tense, second person (Turandot) and its grammatical object (Calaf). While the second structural block reflects tempi closer to average, its outline is distinctly blue, with even stronger connections between surface- level ritardandos in Poplavskaya’s performance.

The strongest of these connections exists between the emphases on “io chiudo stanca gli occhi” (b. 34-8) and “Egli vinca ancora” (b. 42-4). Extensions from these moments intersect with extensions from “l’amerai anche tu!” and “Per non vederlo più!”

(b. 55-60). The crossing of the extensions from these two points forms the second building block of the aria’s first-half tempo hierarchy for this performance. One strong

46 Ibid. In the interview, Giordani frequently characterized his view of Calaf’s embodiment of love as “simile a Dio,” which challenges longstanding assertions that Calaf at this point is a sexually motivated, autobiographical projection of Puccini’s younger self. Analysis of the spectrum of dramatic types of characters as they result from the temporal-textual emphases of different performances will be a future project.

132

possibility is that the relative rushed tempi reflect Liù’s anxiety over Calaf’s yet-unknown fate, or over Turandot’s reticence to submit to love. Counter-balancing this are the strikingly slow tempi linking Liù’s knowledge that only by her death in and for love will

Turandot transform, which might indicate a desire to reflect on her impending action, to accept the gravity of it, to resist it, or any combination of these. Thus, the alternation of stringendo and ritardando might reflect an interpretation in which Poplavskaya views Liù as understanding that the way forward is for Calaf and Turandot in love, and that it is only achievable with her death.

The intensity of the blue and purple swatches that form an inverse triangle on

Liu’s first utterance of “Per non vederlo più!” indicate that this moment bears the greatest non-musically altered temporal-textual emphasis of Poplavskaya’s performance.47 The fact that extensions from this moment project to both edges of the

Time Scape demonstrates that this surface-level event is this performance’s most structurally significant in the tempo hierarchy. The rays of blue, indigo, and violet protruding upwards from “aurora” (b. 66-9), and the repetition of “non vederlo più!” (b.

77-9) emphasize Liù’s dramaturgical need to die (at dawn) in order to allow or inspire

Turandot’s metamorphosis. Thus, the identity of Poplavskaya’s performance seems to rely first on an energetic assertion to Turandot that she will love Calaf, and second, on her awareness of how the ice around the princess’s heart will thaw. In this way, like with the riddle scene on this occasion, Liù’s sacrifice becomes a plausible motivation for the opera’s remaining dramatic trajectory.48

47 Fermatas accompany the slower tempi at Liù’s final words, but not at her first indication that she will die.

48 See discussion of Figure 3-Figure 3-Figure 3-10 and footnote 15.

133

Figure 3-12. Time Scape, text, and tempo contour graph (with linear trend line) for Marina Poplavskaya 07 November 2009 “Tu che di gel sei cinta”

134

Further Examples in Textualizing Performance

“Signore, ascolta!”

Unlike the other performances this chapter examines (see Figures 2.1 and 2.7),

Figure 3-13 demonstrates that Hei-Kyung Hong’s 14 April 2007 performance of

“Signore, ascolta!” emphasizes the ritardando at the aria’s conclusion to a degree such that the emphasis becomes structurally important to the identity of Hong’s realization.

The additional structural layers of blue, indigo, and violet along the right edge of this

Time Scape result from Hong’s employment of tempo at or near the global average between beats 57-68 and her increased temporal emphasis from beats 68-73.

On the surface, and a likely perception for anyone listening to this performance,

Hong does not slow noticeably, owing to the comparative closeness of her ending ritardando to its immediately preceding global tempo. Such a small difference in surface level events, however, can have a substantial effect on the rest of the tempo hierarchy.

The performer approach to the tempo over the aria’s last eighteen beats adds new layers to the hierarchy. These layers diminish the intensity of the central crossing point resulting from ritardandos over beats 17-30 and 46-57, respectively. This lowered intensity results from the crossing of extensions from beats 17-30 and 68-73, which subsumes the aforementioned focal point, which was prominent in both Figure 3-1 and

Figure 3-7, along with the similarly significant crossing resulting from beats 46-57 and beats 68-73.

135

Figure 3-13. Time Scape, text, and tempo contour graph (with linear trend line) for Hei- Kyung Hong 14 April 2007 “Signore, ascolta!”

136

Interpretively, these specific tempo relationships could reflect a significant differentiation from Anna Moffo 46 years earlier and from Marina Poplavskaya two years later. These singers’ respective temporal-textual emphases seem to suggest a Liù with a relatively balanced concern for Calaf (beats 17-30) and Timur (46-57), and perhaps reticent to let too much of her own self-interest show, owing to her social class.

Conversely, Hong’s temporal-textual emphasis perhaps elicits a Liù who is less observant of social mores regarding her servitude. This Liù might thus be willing to allow herself a more expanded moment than Moffo or Poplavskaya to reflect on the personal cost of Calaf’s presumed failure to answer Turandot’s riddles.

“Non Piangere, Liù!”

Figure 3-14 reflects “Non piangere, Liù” from Plácido Domingo’s third appearance as Calaf, on 21 February 1970. Here, there are several lower-level structural links resulting from Domingo’s and Kurt Adler’s uses of stringendo and ritardando. The first of these links does not appear in Figure 3-2, Franco Corelli’s 1961 performance, but does emerge in Figure 3-8, Marcello Giordani’s 2009 performance— the link between “Non piangere, Liù,” and “dolce mia fanciulla” (“My sweet girl,” emphasis added). The next such feature covers beats 21-33, wherein Calaf implores

Liù to listen to him and explains that Timur may be alone in the world one day into their future.49 This section forms a link with the ritardando Domingo employed to emphasize

Calaf’s exhortation to Liù, “Non lo lasciare portalo via con te,” (“Do not leave him; carry him with you”) highlighting the “if/then” scenario Calaf describes. The following lower- level stringendo section (beats 45-64) accounts for Liù’s impassioned interruption and

49 “m’ascolta: Il tu Signore, sarà domain, forse solo al mondo.”

137

Timur’s seeming acceptance of his and Liù’s fate (“noi morrem”). Interestingly, Domingo iterates “questo, questo” just before the return of the primary theme above the global tempo, something neither Corelli nor Giordani did in 1961 or 2009, respectively. The final lower-level feature connects “O mia povera Liù” to “chiede colui che non sorride più,” intimating another “if/then” argument. Here, the “if” is the realized amalgamation of all the aria’s aforementioned un-pleasantries and the “then” is what Liù can do for Timur under such circumstances.

The Time Scape for this performance reveals three primary higher-level crossings. First, there is a link between the acceleration following “dolce mia fanciulla” and Calaf’s insistence that Liù “Dell’esilio addolcisci lui le strade!” This link appears in the vertical mid-range of the graph, just left of the triangle’s centerline. Rather than forming an “if/then” argument, this connection highlights the core of this aria as an exhortation—not for the sake of assuaging Liù’s own fear, but to remind her of her duty to accompany Timur regardless of any circumstances. The second connection here, between Liù’s assertion that she and Timur will die on the road of exile “sulla strada dell’esilio” and Calaf’s reiteration of “che non sorride più” appears in green about half way up the right edge of the graph. Interpretive ramifications here are numerous, owing to the linkage of temporal-textural emphases of two performers. Even so, this link highlights the contrast between Liù, whom Calaf expects take an active role should he die, and Timur, who is elderly, disabled, in exile, and therefore at least sad, if not despondent at the prospect of losing Calaf.

138

Figure 3-14. Time Scape, text, and tempo contour graph (with linear trend line) Plácido Domingo 21 February 1970 “Non piangere, Liù”

139

The final higher-level structural crossing in Domingo’s performance links the stringendo at the moment of Calaf’s first characterization of Timur as “Il tuo Signore” with his last of him as “che non sorride più.” Thus, the temporal-textual emphasis of

Domingo’s performance in 1970 reflects a primary focus on text devoted to Timur, rather than to Liù. This particular emphasis could mean that, on this occasion, Domingo and Adler perceived this aria as an address to Liù to exhort her to care for Timur, but only inasmuch as this gives her agency; if Calaf dies, then Liù remains Timur’s primary caretaker.

“In questa reggia”

The Time Scape representing Andréa Gruber’s performance as Turandot on 29

January 2005 appears in Figure 3-15. Much of the tempo hierarchy in Gruber’s performance is similar to the other eighteen performances of this aria in the current study (see Appendix C). Unlike for Birgit Nilsson in 1961 (Figure 3-3) or Maria

Guleghina in 2009 (Figure 3-9), the higher-level structure of Gruber’s large-scale slower-than-global-tempo section (b. 41-113) does not extend to the left edge of the graph in a significant manner. This feature is the result of Gruber’s comparatively quicker tempi through the aria’s introductory text (b. 1-41), with respect to her performance’s global tempo, as the lack of surface-level green in this section indicates.

This subtle difference provides a clear structural feature absent in many of the Time

Scapes documenting performances of this aria. As is evident in Appendix 2.1, this feature does appear in the tempo hierarchies for each of Birgit Nilsson’s, along with those of Ingrid Bjoner, Elinor Ross, Eva Marton, and Adrienne Duggar.

The most striking higher level features in this performance’s tempo hierarchy are the structural crossings resulting from the extension of Turandot naming Lo-u-Ling

140

crossing two others, both involving Calaf. The first extension this one crosses emanates from the prince’s first interjection in this aria, “No! No! Gli enigmi sono tre, una è la vita!”

(b. 173-180). The second extension originates at Calaf’s and Turandot’s dual ascent challenging each other (b. 181-188). These are the highest significant structural crossings in this tempo hierarchy, and the tempo-textual emphasis they highlight points to the crux of Gruber’s interpretation of this role.50

While the libretto indicates that Turandot seeks revenge on all men for the travesty inflicted on Lo-u-Ling, this does not reflect the complexity necessary for

Gruber’s preferred characterization. In her view, Turandot is a coming-of-age thirteen- year-old who is a pawn in her father’s game of suitors, riddles, and beheadings.

Believing Lo-u-Ling is reborn in her or that she is Lo-u-Ling provides a two-fold escape for Turandot—one from her father’s domination of her will and actions (which explains part of Turandot’s identifying with her ancestress), and the other is from her burgeoning sexuality, which she has no way to control. Also for Gruber, Turandot knows from the moment she sees Calaf that he is different from all previous suitors. This cognizance eventually leads to Turandot’s Act III confession that she knew from the moment she saw Calaf, he was different than all others who had died trying to win her.51

50 Andréa Gruber, interview with the author, 29 January 2015.

51 Puccini, Turandot, 429-36. “Del primo pianto, sì, straniero, quando sei giunto, con angoscia ho sentito il brivido fatale di questo mal supremo...ma ho temuto te! C’era negli occhi tuoi la luce degli eroi!” (My first tears, yes stranger. When you arrived, I felt with anguish the fateful thrill of destiny, but I trembled for you! There was in your eyes the light of heroes!”

141

Figure 3-15. Time Scape, text, and tempo contour graph (with linear trend line) for Andréa Gruber 29 January 2005 “In questa Reggia”

142

Under these circumstances, the lower link, which relies upon the extension from

Calaf’s interjection, indicates his awareness of Turandot’s situation and his confidence that he will free her from her present position. Singers who approach Calaf in this way are the most dramatically convincing for Gruber.52 In the subsequent parallel octaves,

Calaf senses Turandot’s trepidation about a life different from the one she knows, even as masks it in her threat of death. Gruber summarizes this moment: “Calaf knows he can break the spell of Turandot’s release-by-obsession by outsmarting her father and arousing her erotic love.”53

“Straniero, ascolta!”

Figure 3-16 exhibits the tempo hierarchy for the performance of the riddle scene

(“Straniero, ascolta!”) on 01 February 1992, featuring Gwyneth Jones as Turandot and

Vladimir Popov as Calaf, with Nello Santi conducting. Here, segmentation of the various parts of the riddle sequence appears clearly in lower to lower-mid levels. The most striking moments of this performance’s surface come at the conclusion of each of

Calaf’s responses (b. 143, b. 342, b. 553), which the black brackets identify. Here, Nello

Santi emphatically observed the instruction in the score that the ensuing music should commence “a tempo affrettando (dopo la parola),” after Calaf exclaims the answer.54

Even though this instruction appears only at the response to the first riddle, Santi clearly

52 Gruber, interview.

53 Ibid. Such a detail emerging in the tempo hierarchy and temporal-textual emphasis of Gruber’s performance demonstrates, as it did with Marcello Giordani’s performances, the applicability of this methodology to better understanding the creative process. Being able to connect singers’ dramatic interpretations to their onstage musical behaviors in this manner represents perhaps the most significant contribution to this vein of inquiry.

54 Puccini, Turandot, 269.

143

appended the parenthetical instruction to the second and third responses, each of which carriers the “a tempo affrettando.”55 In addition to clearly separating each riddle, this also heightens the immediate drama by breaking the flow of Turandot’s delivery with each answer.

Furthermore, Santi observes the fermata assigned to the measure preceding the start of Calaf’s first response in the silence separating the conclusion of the sustained

nd orchestral G4 and the 32 notes serving as a pickup to the answer (b. 113-4). The tempo for this beat is approximately half (36.8 bpm) of the average (72.8 bpm) of the preceding three beats. Santi approached the onset of the third response similarly (b.

521-4).56 Here, the average tempo of the preceding beat is 53.88 bpm, and Santi’s observation of the fermata, which occupies a single beat, is 23.28 bpm.

In the mid-ground appear three faint triangles—first in blue, then in yellow, and then again in blue. Each of these overlaps the following triangle, which contrasts with the clear sectional divisions apparent at the surface. Perhaps expectedly, the first two of these triangles with mid-level significance reveals clear temporal-textual links between the first two riddles and the activities of everyone other than Turandot and Calaf in this performance. The third of these triangles links Calaf’s second answer—“il sangue!”—to the third riddle and its answer.57

55 Ibid., 280, 289.

56 There is no fermata immediately preceding the second response.

57 “Blood!”

144

Figure 3-16. Time Scape, text, and tempo contour graph (with linear trend line) for 01 February 1992 riddle scene.

145

The overlaps of these sections illustrate the interchange of exchanges that occur between Turandot, Calaf, and the rest of the ensemble. Furthermore, they might indicate an increase in the importance of the crowd’s (including I Sapienti) agency throughout this scene. This could equally easily indicate two different scenarios for the crowd’s involvement. One options follows from the Act I demonstration of bloodlust in

“Gira la cote!” and eight cries for the executioner “Pu-Tin-Pao!”58 Here, the numerous repetitions of “dove regna Turandot” throughout the scene would be exclamations of ecstasy in the conventions of their culture. The crowd, with its unquenchable thirst for bloodshed via the spectacle of public decapitations and severed heads on pikes, views all suitors first as threats, then as necessary victims. Alternatively, the crowd could harbor the semi-private thoughts of Ping, Pang, and Pong that emerge in Act II’s first scene: yearning for peace and a return to life away from the charade of riddles. In either of these cases (or in many others), this performance of the riddle scene parallels many others in the current study. It, like them, evinces a strong interweaving of the characters and character classes involved.59

“Nessun dorma!”

Figure 3-17 reflects the tempo hierarchy for Nicola Martinucci’s performance of

“Nessun dorma!” on 13 February 1988 with Nello Santi conducting. This tempo hierarchy contains the highest density of nesting and interconnection, thus creating a unique identity for this performance amidst the other performances of this aria in the

58 Puccini, Turandot, 25-47, 61. “Turn the stone!” “Pu-Tin-Pao” is the name of the executioner.

59 The other two examples of the riddle scene discussed above do not share the intensity of this structural interlocking of surface level segmentation. However, as Appendix 2.1 exhibits, this feature exists in most of

146

current study. In this performance’s tempo hierarchy, thirty crossings exhibiting temporal-textual linkages at least one phrase apart (more than eight beats) appear. The black lines in Figure 3-17 demonstrate where these occur, and what they link.60 Some of the highlighted extensions emanate from individual moments (b. 33, 41, 45, 49, and 69), while others demarcate the boundaries of slower tempi over groups of beats (b. 17/37,

53/65, 77/85, 106/117). While the graph’s structure, which Chapter 1 describes, influences the crisscrossing evident here, the degree to which it appears in color results solely from the structural linkages of surface level events in Martinucci’s performance.

Like in Corelli’s 1961 performance (Figure 3-5) and Giordani’s 2009 performance

(Figure 3-11), the lower level ritardando covering the entire opening address to

Turandot (b. 13-36) connects first with Calaf’s assertion, “Quando la luce splenderà!” (b.

57-64). Furthermore, this latter section’s extension crosses that of its subsequent ritardando, which covers Calaf’s strongest-yet assertion that Turandot’s silence will make her his (b. 77-84). Also like Corelli’s and Giordani’s tempo hierarchies, extensions from this moment cross extensions from each of Calaf’s first and last exclamations of

“Vincerò!” (b. 106-111, 116-120).

The density of color extensions and the volume of resulting crossings here can, depending on one’s interpretation, afford either great clarity or great difficulty in attempting to ascertain a possible interpretive scheme for Martinucci on this occasion.

One straightforward possibility here is that the overarching dramatic impulse at this

60 There are thirty-six total crossings resulting from the overlay of the black lines on Figure 3-17. However, the six lowest of these all occur over the span of one phrase (about eight beats) or less. As the music naturally connects the text occurring within these phrases, their structural significance is comparatively, and understandably minimal.

147

moment is Calaf assuring himself, the crowd, Timur, Liù, and/or Turandot that he will win her game and claim her as his wife. A complex interpretation of these crossings seeks to create a hierarchy of emphasized words or phrases occurring on similar or equal structural levels. Crossings in this Time Scape’s lower quarter demarcate the aria’s larger phrase structures. The first significant set of links begins in the mid-ground, near the left axis, connecting “Tu pure, o principessa” (“Even you, o Princess!” b. 14-18) with “qunado la luce splenderà! Ed il mio bacio” (“when the light shines!”, b. 57-71). This set then extends to higher levels of the temporal-textual emphasis structure at the intersection of the extension from “che ti fa mia!” (b. 77-81). This initial extension also intersects with both rays emanating from Calaf’s concluding three-fold exclamation

(106-120).

One conclusion from this level of analysis is that Martinucci and Santi felt the most important information to communicate in this performance might be “Princess, when the light shines, my kiss makes you mine. I will win.” The connections between each of these extensions form the prominent intersections at the highest levels of the tempo hierarchy. Such a determination may seem straightforward, given the opera’s dramatic trajectory. The other vocally expressive components of Martinucci’s (or anyone’s) rendition—articulation, dynamics, timbre, diction—complete the profile of

Martinucci’s interpretation in this performance.61

61 The analysis of the other components of a performance’s profile, especially with respect to opera, has remained wholly unexamined with digital means. This is an area for research in the immediate future.

148

Figure 3-17. Time Scape, text, and tempo contour graph (with linear trend line) for Nicola Martinucci 13 February 1988 “Nessun dorma”

149

“Tu che di gel sei cinta”

Figure 3-18. Time Scape, text, and tempo contour graph (with linear trend line) for Licia Albanese 24 February 1962 “Tu che di gel sei cinta”

150

Figure 3-18 exhibits the tempo hierarchy for Licia Albanese’s performance of “Tu che di gel sei cinta” on 24 February 1962. Two groups of tempo modulation affecting the rest of the hierarchy are immediately apparent—the significantly faster than global tempo at the aria’s beginning (b. 1-13) and significantly slower than global tempo at its end (b. 79-83). The intensity of the extension at the aria’s conclusion results from fermatas over each of the three eighth notes and syllables of Liù’s penultimate word,

“vederlo.”62 These features’ intersection near the top of the Time Scape outlines a distinct set of nested triangles, many of which have apexes near the horizontal middle of the graph (above beats 46-51). Two of these figure prominently in the performance’s tempo hierarchy. First, extensions originating at beats 36-39 and 59-61 link Liù’s first indication of her plan, “io chiudo stanca gli occhi” to the explanation of the gravity of her meaning, “non vederlo più” forming the lower mid-ground’s most significant textual relationship.63

The extension emanating from beats 36-39 also intersects prominently with the one from Liù’s final words slightly higher in the graph. The other prominent intersection at this level of the hierarchy connects the repetition of “l’amerai anche tu” (b. 22-26) and

“questa aurora. . ., Io chiudo” (b. 65-71).64 The intersection of extensions from “l’amerai anche tu” (b. 22-26) and of Liù’s final words (b. 79-83) appears immediately above this.

The combination of these conspicuous intersections seems to indicate that a prominent component of Albanese’s interpretation is Liù’s obsession with what she must do to

62 Puccini, Turandot, 389. “See him.”

63 “I close my tired eyes” and “I will see him no more.”

64 “You will love him too” and “this dawn, I close them.”

151

ensure that Turandot will love Calaf: closing her tired eyes so as never to see Calaf again.

The combination of these links, together with the repetitions of so many textual phrases within Puccini’s setting perhaps demonstrates an interpretation in which Liù uses these words as a mantra to help her accomplish her task. Considering that the overarching intersection links Turandot’s dichotomous existence at this point in the opera, “Tu che di gel sei cinta, da tanta fiamma vinta,” to the final “non vederlo più,” one might surmise that Liu possesses a unique level of awareness. Perhaps Albanese’s Liù understands the role she has to play in Calaf’s and Turandot’s lives. The textual emphases, and the hierarchical levels at which they occur could indicate that, more than being the only one who knows Calaf’s secret, Liù knows Turandot will only accept love when she witnesses self-sacrificial love.

Conclusions

Establishing and studying the musical events in performances of the past--both recent and more distant--facilitates better understanding of those events and the effect interpreters can have on the formation a performance tradition. Such activity sheds new light on a work itself. One might compare tradition to a bookshelf with each book representing a single performance of a work. As traditions surrounding each musical work develop over time, many things will remain the same – most of the drama and music – while others, such as productions, staging, and costuming will change either gradually or from one performance to another. The bookends of a tradition are the events at either end of its performance chronology. Analysis of the first (04 March 1961) and most recent (07 November 2009) available recorded performances focus on the bookends of Turandot’s performance history at the United States’ principal opera house.

152

The similarities and differences in the recordings of these performances demonstrate how they frame a tradition, and, in turn, how they relate to the other participants within it. Examining them can provide valuable insight into the general trajectory of a tradition by highlighting and comparing factors relating to its origin and destination.

Admittedly, analyzing performances at the opposite ends of a performance history provide only a partial perspective on its development and evolution. A more complete one is possible only by examining each realization--all the books on the shelf, together, as it were. Given the usefulness of Scape Plots as new musical objects for discourse, further analysis using them is a natural recourse for establishing the history of how performers realize a work. The third section of foregoing analyses provides and discusses a Time Scape for a different performance for each of the six segments of

Turandot. These analyses reveal more of the sonic events in this opera’s performance history and, in so doing, provide further examples in the application of textual-temporal emphasis analysis to the operatic creative process. For instance, the linear trend lines accompanying each Time Scape indicate that in these examples, performers tend to raise their global tempo throughout the riddle scene and “Nessun dorma” and decrease it throughout “Signore, ascolta!” and “Non piangere, Liù.” For “In questa Reggia” and

“Tu che di gel sei cinta,” however, the trend lines suggest greater variety in performer decisions and behaviors regarding the overall tempo and resulting pacing. Time Scapes for all analyzed excerpts from this opera are viewable in the six appendices of this document.65

65 Appendix A contains Time Scapes for “Signore, ascolta!”; Appendix B contains Time Scapes for “Non piangere, Liù,”; Appendix C contains Tie Scapes for “In questa Reggia,”; Appendix D contains Time

153

Tempo is, of course, just one of a performance’s many elements. Furthermore, considering how tempo influences temporal-textual emphases is merely one approach to that element’s analysis.66 This type of analysis can be applied to other components of a performance. Future research will develop the means to analyze dynamics, topical affectations (including vibrato and portamento), articulation, phrasing and breathing, timbre, and vocal diction. By documenting these components together, researchers will be able to construct a profile of a performance with respect to the relationship between the expressive palette and the communication of the text and drama of an opera.

Each of the foregoing examples documents its respective performance, thereby exhibiting the identity of each with regard to tempo modulation, tempo hierarchy, and the resulting temporal-textual emphasis. Such a means of capturing performance is useful in that it allows one to simultaneously glimpse both surface and hierarchical relationships of musical phenomena. By visually reflecting on-stage behaviors and their organization in time, Scape Plots can move analysis beyond the limits of language and score. These tools emphasize the importance of the point that opened the foregoing discussion: music is not merely its score; rather, its repeated instantiations through performance reveal its dynamic nature, a nature the fixity of a score cannot support.

Scape Plots emerge as a sort of text in and of themselves, one crucial to understanding performance histories and their traditions.

Scapes for “Straniero, ascolta!”; Appendix E contains Time Scapes for “Nessun dorma,”; and Appendix F contains Time Scapes for “Tu che di gel sei cinta.”

66 The Time Scapes for all other accessible performances of Turandot at the Metropolitan Opera appear in chronological order, excluding those appearing in this chapter, in Appendices A-F.

154

Furthermore, documenting performances in this way—establishing the texts—is necessary as a first step toward quantitatively identifying and then analyzing performance traditions as they evolve, which Chapter 4 will demonstrate. Chapter 5 delineates how calculating quantitative models of performance and traditions allows an analysis of how these components of a musical work’s identity interrelate.

155

CHAPTER 4 QUANTITATIVE AND CHRONOLOGICALLY-BASED MODELS OF TRADITION

The mathematical is a fundamental position towards things. . .with regard to the way in which they are already given to us and must be so given. The mathematical constitutes therefore the fundamental presupposition of the knowledge of things.

--Bernard Stiegler Technics and Time1

Since music performance is a form of technē in that it reveals one perspective of a musical work to an audience, it is therefore also a technical behavior. As such, each instantiation of a work is at once a fact while also consisting in many constituent facts, such as the tempo relationships examined in the preceding chapter. In his work on the relationship between technology’s evolution and humans’ interaction with it, Bernard

Stiegler characterized technology as the result of human behaviors. “The experience and organization of. . .human tendencies,” requires then that “the question then becomes that of distinguishing the. . .tendency within. . .facts. The tendency is realized by the facts, and the examination of the links between facts affords us a view of the conditions for the realization of the tendency.”2 Furthermore, “the tendency does not simply derive from an organizing force, [it] operates, down through time. . .in a relation of the human living being to the matter it organizes and by which it organizes itself. . .”

That Stiegler’s tendencies continue to exist and function across time and in relationship to human endeavor aligns them neatly to Chapter 1’s discussion of tradition. Because of their organizing components, tendencies and traditions are linked to conventions, which represent benchmarks of comparison or models to follow. All conventions are thus parts

1 Stiegler, 206.

2 Ibid, 46-7.

156

of a tradition, but not all traditions become conventions. Understanding this delineation further clarifies Stiegler’s stipulation that a chronological and historical organization of tradition-as-tendencies has mathematics at its core.

So, what does the prefix ēpi stuck to the root of mētheia mean? Ēpi carries the character of the accidentality and artificial facutuality of something happening, arriving, a primordial “passability” [passabilité]. With matheia, or mathēsis, we are dealing with something happening, that is from the first passed on, shared: knowledge is primordially. Epimētheia means heritage. Heritage is always ēpimathēsis.3

With regard to music, Stiegler’s ‘tendency’ is much like tradition – the collection of behaviors that supersedes the temporal and geographical locations where performances occur. Stiegler’s ‘matter’ is akin to the musical score, or the unperformed musical object, which humans organize in the act of composition, and by which humans organize themselves by performing. In other words, mathematics are inseparable from tradition; understanding tradition quantitatively thus begins in mathematics. Statistical means (averages) behave in almost exactly the same way that traditions do as Chapter

1 demonstrates. Averages account for all data currently in a given set; they allow the addition of new data; they reflect the change to the overall data set with each addition.

In establishing quantitative representations of tradition, and in keeping with tradition’s chronologically ordered development, Chapters 4 and 5 organize the tempo data averages from select arias in Turandot together in chronological order. Chapter 4 considers three different levels of chronological organization, while Chapter 5 examines the excerpts in relationship to other aspects of performance. As the first performance of each aria in this corpus represents the re-initiation and continuance of performing

3 Ibid, 206-7.

157

Turandot at the Met, it is the initial analyzable point in the trajectory of the tempo rubato tradition for each aria. Thus, the arrangement of the Time Scapes for each aria is as follows: Performance 1 (4 March 1961), Average Tempo Values Performances 1-2 (A1-

2). Average Tempo Values Performances (A1-19). This arrangement facilitates visual inspection of the tradition’s chronological evolution by scanning the graphs in order.

Here, the appearance order of each set of graphs mirrors the order of their occurrence within the plot of Turandot. Performances of three different musical excerpts resulting in the progressively aggregating averages comprise the primary data for the following discussion. In the same manner that no two performances of the same excerpt in

Chapter 3 are identical, traditions are also dynamic and evolve at different rates. Some of these changes occur from moment to moment while others evolve over longer spans of time. This chapter offers views of three different traditions’ evolutions. It begins with a point-by-point examination of the tradition developing around Liù’s Act I aria, “Signore, ascolta!” prior to considering Calaf’s Act I aria, “Non piangere, Liù” by decade and concluding by considering the entirety of the performance tradition of Liù’s Act III suicide aria “Tu che di gel sei cinta” at once.

“Signore, ascolta!”

1960s

The Met’s measurable tradition of “Signore, ascolta!” begins with Performance 1

(P1) on 04 March 1961, sung by Anna Moffo, and represented here by Figure 4-1.4

Moffo originated this aria’s measurable tradition at the Met. Briefly reviewing the tempo

4 For a detailed discussion of this performance and its resulting Time Scape, see Chapter 3, Figure 3-1.

158

hierarchy illustrates that Moffo included a comparatively significant stringendo at the outset of the aria as she pleaded for Calaf to give his attention to her concerns.

Together with conductor Leopold Stokowski, Moffo’s tempi vacillate (in relationship to the global tempo) as Liù’s attention shifts between Timur, Calaf, and herself.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-1. Time Scape of “Signore, ascolta!” 04 March 1961 – Performance 1 (Anna Moffo as Tradition originator)

The first instance in any performance history that exists solely as a point in its tradition follows the second performance, when the mathematical average between multiple performances is first calculable. Figure 4-2 is the Time Scape derived from the data resulting from averaging the tempo values Performance 1 with those of

Performance 2 (Licia Albanese, 24 February 1962).5 Three noticeable differences appear as a result of adding Albanese’s performance to the performance tradition at this point. All of these are in the second half of the aria. First, the slight stringendo Moffo

5 For the table of averaged tempo data, see appendix B.

159

employed over beats 35-45 and its related shallow hierarchical significance are gone, with only a surface-level ritardando. This absence results from Albanese’s inverse approach to tempo at this point (slowing where Moffo sped up). Next, as the amount of red and orange in the low-level structural spike at beats 57-68 increases between P1 and the average, A1-2 reflects a significant tempo increase, reflecting Albanese’s significantly more intense stringendo and a different trajectory for the tradition from its inception one year earlier. Finally, the increased presence of indigo and violet in the final nine beats of the aria in A1-2’s Time Scape reflects a less dramatic alteration in the tradition’s trajectory than the preceding stringendo.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-2. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-2 (A1-2)

Taken together, these changes reflect a more dramatic range of tempo fluctuations in Albanese’s performance than Moffo’s. No dramatic changes appear in

160

higher levels of the tempo hierarchy, as the top two-thirds of the graph remains largely unchanged from P1 to A1-2, though a slightly less amount of blue appears on the right side of the Time Scape, indicating a minor move away from the reflected ritardandos structural significance.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-3. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-3 (A1-3)

The Time Scape reflecting the tradition of “Signore, ascolta!” following Lucine

Amara’s performance on 16 January 1965 appears as Figure 4-3. Here, the most noticeable changes on or near the surface occur in the middle and at the end of the aria, respectively. In the middle, the reflected tempi are closer to the global tempo than in earlier incarnations of the tradition, indicating that Moffo’s 1961 performance is more independent of the trend developing at this point in the tradition’s trajectory (1965). At the end, the slightly increased amount of red throughout the stringendo section results in a broadening of the shape of this lower structural level feature. Furthermore, the

161

increased density of the purple in the feature at the end of the aria indicates a continued slowing of the tempo below the global average at this moment. That the top right corner of this area touches the right side of the graph above the purple indicates it is more important to the structure of the performance than the ending ritardando, and that the concluding ritardando becomes increasingly isolated from the remainder of the tradition’s tempo hierarchy.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-4. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-4 (A1-4)

Figure 4-4 reflects the tempo tradition following Mirella Freni’s 1966 performance under the baton of Zubin Mehta. The very small amount of change in this graph from

A1-3 indicates that Freni’s and Mehta’s tempo choices on 03 December 1966 were very close to the tempo values of the tradition evident from the preceding three broadcast performances. The changes that do affect the shape and color of this Time Scape include a continued slight quickening of the tempo throughout more of the outset of the

162

aria, the near disappearance of tempi distant from the global tempo in the middle, as the wider swath of red and the disappearance of the single purple dot in the middle indicate.

In addition, the structural significance of the stringendo towards the end of the aria abates slightly, and more of the ending is significantly below tempo, as the increased shift from indigo to purple in this swatch reflects.

Few changes are immediately apparent between Figure 4-4 and Figure 4-5, the latter of which represents the measurable tradition for “Signore, ascolta!” after Martina

Arroyo sang it under Zubin Mehta’s baton on 22 March 1969. This indicates that, from a structural standpoint, Arroyo’s performance is remarkably similar to the aggregate of those preceding it. Amidst some subtle changes in the graph’s coloring, which here reflects the average tempo values, one distinct area of surface level features reflects

Arroyo’s most significant departure from the trajectory set by her predecessors.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-5. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-5 (A1-5)

163

The green section between beats 31-45 in Figure 4-4, (given in black bracket) as

Liù envisions the future without Calaf, contains only two miniscule portions (only 1 and 2 pixels, respectively) of blue. These features indicate the presence of a slower than global tempo, first at beat 34, and then at beats 42 and 43. In the same portion of Figure

4-5, which the black bracket encloses, the faint blue pixels appear at beats 34 and 42, while yellow pixels appear at beats 37-9 and 43-4. For Arroyo’s performance to alter the coloring of the Time Scape for the average in this way, her tempo at these two beats needed to be significantly faster than what had been the burgeoning tradition at the date of her performance (A1-4). Her tempo from beat 38-39 is 49.1538 beats per minute.

When compared against the averaged tempo values for the performances by Anna

Moffo, Licia Albanese, Lucine Amara, and Mirella Freni at this point (40.92775 beats per minute), Arroyo’s tempo here is 20.1% faster. This results in an increase in the tempo average after her performance to 42.57296 beats per minute (at A1-5), which is an increase of 4% due to Arroyo’s comparatively strong deviation from the mean. At beats

43-44, the average tempo value of the first four performances (A1-4) is 36.36758 bpm while Arroyo’s tempo is 60.0927—an increase of 65.237%. Arroyo’s significantly faster tempo increases the average to 41.1126 bpm (for A1-5), an increase of 13%. Such striking uniqueness at these two points does not reflect Arroyo’s being outside of the tradition at this point. Instead, it demonstrates how her dissimilar interpretation, even if only in isolated moments, enriches the tradition’s entirety, and changes the trajectory the tradition might follow.6

6 Audience and performer awareness of performance traditions, and how such awareness informs expectations remains an area for future consideration.

164

1970s

Figure 4-6 reflects the average time values following Lucine Amara’s performance as Liù on 21 February 1970. Like in figures 4-4 and 4-5, the most evident difference in the Time Scape appears between beats 31 and 45. Here, the average tempo from beat 36 to 38 is faster than the global tempo this Time Scape espouses.

This continues to reflect the faster tempo from beats 37-9. In addition, there is a new yellow patch appearing on the surface at beat 36, reflecting a growing increase in the tempo of these beats in relationship to the global tempo. Furthermore, as the region tempo increases, the hierarchical significance of it increases slightly, as the faint yellow triangle based on beats 36-9 indicates. The other noticeable change in this Time Scape is the reappearance of a blue pixel at beat 42.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-6. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-6 (A1-6)

165

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-7. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-7 (A1-7)

Following Edda Moser’s 27 April 1974 performance in Boston, with the Met on tour, most of the upper level of the average tempo hierarchy remains unchanged, as

Figure 4-7 illustrates. However, like the previous several average values, this Time

Scape reveals continuing change in the middle section of the aria. Here, the circular annotation is an enlargement of this area; it reveals that both the slightly fast section at beats 36-9 and the slightly slower section at beats 42-5 begin to extend beyond the surface level temporal events. The increased intensity of this latter moment’s slowing causes one other noticeably, albeit minor alteration in the average tempo hierarchy.

Approximately three-quarters of the way up the right edge of the graph, the area that has alternated between being entirely blue and blue with flecks of indigo again appears with indigo. This is likely the result of the increased hierarchical significance of the expansion of the ritardando at beats 42-5. At least for the case of “Signore, ascolta!,”

166

this trend of minor variations in tempo choices at one or few isolated moments either on the temporal surface or higher in the hierarchy seems itself to be part of the tradition of this aria over the first fifteen years of its recorded performance history at the Met.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-8. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-8 (A1-8)

Apart from the disappearance of the indigo flecks in the upper region of the tempo hierarchy, and the slight lessening of red into orange and yellow at the aria’s outset, Figure 4-8 exhibits little change from Figure 4-7. Figure 4-8 represents the tempo hierarchy average and textual-temporal emphasis tradition following Adriana

Maliponte’s performance on 28 December 1974. Despite the slowing of the initial tempo in relationship to the global tempo here, the ritard [or, the difference] is not significant enough to make the blue/indigo extension from beats 46-56 reach the graph’s left edge.

Another subtle shift from Figure 4-7 to Figure 4-8 is the increased amount of red in the hierarchy above beats 57-67. This change, combined with the lack of any intensification

167

or increase of the violet areas in the middle of the graph results in this dominating structure shrinking, albeit nearly imperceptibly.7 Furthermore, the increase of violet at the aria’s end results from Maliponte’s longer observation of the fermata (in relationship to the global tempo) at this location than the measurable average of those who sang this role over the Met’s previous fifteen seasons.

1980s

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-9. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-9 (A1-9)

Figure 4-9 reflects the tempo tradition following Leona Mitchell’s performance on

28 March 1987. Here, the most significant changes are not in isolated surface or structural places, but rather in degrees of intensity of tempo modulation. The beginning

7 This shift is most noticeable when one views the series of average-tempo-data graphs in quick succession. While such a subtle change might be perceivable if viewing in a large enough print format, digital viewing, such as in image viewing or PDF format facilitates, of these graphs with respect to their placement in their respective chronological series is the best option.

168

stringendo (beats 1-16) shows less red and more yellow and orange. Similarly, the ritardandos covering beats 17-20, 46-56, and 69-76 contain less violet and more indigo and blue. These indicate that Mitchell’s tempi over these sections are less variable than what the tradition preceding her indicates. The one shift that does not follow this trend concerns the stringendo section covering beats 57-67. Figure 4-9 has more red in this area than does 4-8, indicating that Mitchell’s performed tempo modulation here stands out significantly from the rest of her otherwise narrower tempo modulations.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-10. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-10 (A1-10)

Figure 4-10, the measurable tradition following Mitchell’s performance under

James Levine on 4 April, 1987, is very similar to Figure 4-9. It shows a continuation of the trend evident in Figure 4-9, wherein the intensity of the beginning stringendo (beats

1-16) and middle (beats 17-20 and 46-56) and ending (beats 69-76) ritardandos all exhibit minor reductions in their respective intensities. This continuation is not, however,

169

a unilateral one across all tempo relationships or in all levels of the tempo hierarchy.

The main difference between these two sets of tempo averages occurs at the second stringendo (beats 57-67). Here, like in the other tempo modulations in the quantitative representation of the tradition at this point, the intensity here abates rather than escalates. This change in this region from Figure 4-9 to Figure 4-10 indicates that

Mitchell’s tempi here were counter to a localized trend that had emerged, and which she had also followed as recently as 28 March 1987.8

The data supports such an observation, as Mitchell’s average tempi on 4 April

1987 over beats 57-67 is 36.68 beats per minute, which is 15% faster than the global tempo of her performance (31.78 beats per minute). In her performance on 28 March

1987, Mitchell averaged 43.31 beats per minute over this same span, which is 25% faster than her global tempo (43.31 beats per minute) on that occasion. The quantitative model of the tradition as it stood between these two performances (following the 28

March performance, but preceding the 4 April performance) reflects an average over these beats of 45.57 beats per minute, which is 20% faster than its global tempo (39.76 beats per minute). The relationship of these beats’ average tempi to is greater in comparison to their global averages for both Maliponte’s December 1975 performance and Mitchell’s in March 1987. For each of these performances (P8 and P9), these tempi are 21 and 25% faster than their global averages, whereas the tempi for the aggregated averages over these beats are 19% (for A8) and 20% (for A9) than their respective global tempi. For Mitchell’s April 1987 performance, this relationship inverts, with her

8 Investigation into reasons for this departure from established convention are currently hypothetical at best, and remain an area for future consideration. Such inquiry will be most effective should Mitchell or Levine grant an interview.

170

performance over these beats being 15% faster than her global tempo, and the average

(A10) tempo over these beats is 20% faster than its global tempo. Such a significant shift in performance practice away from established trends challenges notions that traditions are fixed entities. This is especially true when the same singer evinces the same shifts in such a short time span between performances. This evidence seems to suggest that traditions result from, rather than govern, performers’ behaviors onstage.9

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-11. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-11 (A1-11)

Figure 4-11 reflects the average tempo values of all available recorded performances preceding and including Leona Mitchell’s 13 February 1988 performance under the baton of Nello Santi (P11). At this point in the progression of averages, few

9 One possible cause for this, and therefore an area for future investigation, is that the performance on 4 April 1987 (P10) was video recorded for commercial purposes. If such knowledge was available to the performers, this could reasonably influence performers’ mindsets regarding this performance.

171

changes appear at the surface level of the Time Scape, and those that do appear are most evident in the large blue features that dominate the middle of the graph. Like occurred in the transition from Figure 4-9 to 4-10, the intensification of tempo hierarchy features in the mid-ground lessens between Figure 4-10 and 4-11.

1990s

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-12. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-12 (A1-12)

Figure 4-12 reflects tempo data averages including Aprile Millo’s performance on13 January 1990, also under Nello Santi’s baton. Here again, the most evident changes are those of intensity, and these occur in the same, mid-ground region of the graph. Unlike the trend from Figure 4-10 to 4-11, the shift from 4-11 to 4-12 reveals a growing intensification of slower-than-global tempo modulation. Including data from

Teresa Stratas’s performance as Liù on 1 February 1992, Figure 4-12 reflects very little change in the central hierarchical feature. Instead, its primary change occurs in the

172

stringendo section covering beats 57-67. Here, the intensity lessens, reflecting a comparatively significant impact of her tempo choices on the aggregated data to this point. In other words, throughout these beats, Stratas’s tempi are considerably divergent from the tradition in this section while her tempi in the other sections are largely in line with it. Her average tempo throughout this section (33.155 bpm) is nearly

30% slower than the tradition before her performance. Meanwhile, her average throughout the other parts of this average is only 20.45 percent slower than the average of all measured performances preceding hers.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-13. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-13 (A1-13)

Figure 4-14 illustrates this increased disparity clearly, as well as the difference in the tempo arches for these examples, by mapping the tempo contours and associated linear trend lines for both P13 and A1-12. Despite some individual moments showing a perhaps greater disparity in tempo relationships between A12 and P13, the spaces

173

between the two lines between beats 57-69 reflect the greatest relational distance between the streams of tempo data. In addition, as one might expect a progressive average to do, the tempo arch (similar to a melodic arch) becomes increasingly standardized. The independence of a performance from a tradition is also evident in the differing lines slopes of the trend lines in Figure 4.14. While both P13 and A1-2 exhibit overall decreases in tempo, the trend line for P13 reveals a sharper decrease in global tempo across its duration than in the tradition immediately preceding Stratas’s performance. Some performers (either conductors or singers) will perform similarly, without much variance in tempo across an aria, and some, like Teresa Stratas on 01

February 1992, exhibit frequent and sometimes wide swings in tempo.

Tempo comparison A1-12, P13

70

60

50

40

30 Beats per Minute Beatsper 20

10

0 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 Beats

A-12 P13 Linear (A-12) Linear (P13)

Figure 4-14. Tempo line comparison Average 1-12 and Performance 13 (Teresa Stratas, 01 Feb 1992)

Noticeable changes in the tempo hierarchy nearest the surface emerge in Figure

4-15. Here, the intensity of both stringendo sections diminishes along with the intensity

174

of the central ritardando structure. The concluding ritardando intensifies, indicating a substantially slower-than-average tempo here by Maria Spacagna on 11 February 1995.

Spacagna’s tempo average over the last nine beats of her performance was 33.778 beats per minute, which is 22.1% slower than the average tempo over these beats for all performances preceding hers. Also in this graph, a small structural feature emerges between beats 35-45. This slightly faster-than-average area results from minute, incremental increases in tempo over these beats by successive performers.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-15. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-14 (A1-14)

Figure 4-16 reveals the measurable tradition following Angela Gheorghiu’s performance on 17 February 1996. This set of averages reflects subsiding intensities in the concluding ritardando and both areas of stringendo. Furthermore, the height of the structural feature of the closing ritardando also contracts. Reversing the general trend of lessening intensity, Gheorghiu’s performance re-intensifies the ritardando structure that

175

dominates the middle of the Time Scapes for “Signore, ascolta!” The first of the lower segments expands upwards by one pixel, while the latter’s upper edge changes from predominantly blue and indigo (Figure 4-15) to mostly indigo (Figure 4-16). Together, these changes affect the higher structural feature, which expands upwards by one pixel on both its left and right hand side. These comparatively microscopic alterations reflect the aforementioned trend of the averaged tempo data becoming more standardized.

This demonstrates graphically the mathematical reality that in order for a singer’s performance and its resulting data to significantly alter the tradition’s trajectory, it must be substantially different from the norm of what others have done previously.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-16. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-15 (A1-15)

176

2000s

The average tempo data following Norah Amsellem’s performance under the baton of Marco Armiliato on 01 March 2003 appears as Figure 4-17. Here, very little change is perceptible, except that the locations of ritardando each diminish in their respective intensities. The latter diminution is a continuation of the overall trend.

Amsellem’s tempo throughout the middle section of the aria shows a retrograde motion of this area’s feature’s intensity, thus further highlighting Gheorghiu’s tempo choices as an outlier from the established norm in 1996. If Gheorghiu’s performance had not occurred, or been available for analysis, the Time Scape of averages following

Amsellem’s performance would look almost identical to that following Spacagna’s 1995 performance.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-17. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-16 (A1-16)

177

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-18. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-17 (A1-17)

In Figure 4-18, which reflects the measurable tradition following Krassimiri

Stoyanova’s performance on 29 January 2005, with Bertrand de Billy conducting, one sees the continuing trend of varying tempo modifications. Here, the respective intensities of the first stringendo and the second ritardando sections each diminish. This indicates that, compared to the average reflected in Figure 4-17, Stoyanova’s tempi throughout these sections were significantly closer to the global tempo of these areas.

Furthermore, Stoyanova’s tempi in the first and concluding ritardandos and the second stringendo increase the intensity of these areas in the Time Scape reflecting the tradition after her performance. In addition, her tempi in the first ritardando, combined with the less intense stringendo at the outset, result in the slight growth and intensification of the higher-level ritardando feature of the average hierarchy.

178

Figure 4-19 reflects the measurable tradition following Hei-Kyung Hong’s performance under Marco Armiliato’s baton on 14 April 2007. Here, there is only one noticeable difference of any structural significance. The second instance of stringendo becomes distinctly less red and orange colored, with more orange in the middle of the feature, and it shrinks in size. This indicates, on one hand, that Hong’s tempi throughout most of the aria were not significantly divergent from the average of all measurable previous performance. On the other hand, it indicates that Hong’s tempi in the later stringendo are her most divergent from the established norm; thus, they are what most strongly make her performance unique amongst the measured tradition resulting from all preceding interpretations.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-19. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-18 (A1-18)

179

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, ascolta. . Si spezza. . . Ma se il tuo Del’esilio. . . Liù non regge più, Ah, . (x2, 1-10) labbra! (17-30) destino. . .strada sorriso! (46- pietà! (57-72) (31-45) 56) Figure 4-20. Average Tempo Values Time Scape for “Signore, ascolta!”, Performances 1-19 (A1-19)

Almost no visual changes are perceptible between Figure 4-19 and Figure 4-20, the latter of which reflects Marina Poplavskaya’s performance as Liù under Andris

Nelsons’s baton on 7 November 2009. At this point—the current end of the measurable tradition of singing this aria at the Metropolitan Opera—Poplavskaya’s tempi are very much in line with the trends that had developed up to the date of her performance.

Every feature of significant tempo modulation from the global tempo lessens in intensity, albeit slightly, indicating that Poplavskaya’s performance is almost exclusively within the center of the tradition as it existed before her performance.

Visualizing all of the parsed moments along the tradition’s evolution for “Signore, ascolta!” offers an image-based, surface-level (what one would hear) review of both the tradition and the performances that created it helps to contextualize the chronological progression. Moreover, it further illustrates how each iteration in a performance history

180

shapes a tradition, thereby establishing the primarily causal nature of performances in relationship to traditions. Figure 4-21 reveals the trend in both performances and, because of it, in the tradition, is one of slowing. Three prominent areas of slowing for

“Signore, ascolta!” and its traditions over fifty seasons, beats 8-15, 34-42, and 59-68.

This slowing is particularly evident at the switch from the Beaton production to the

Zeffirelli, which the dashed line at T1-8 demarcates. Further slowing occurs in the tradition, particularly between beats 22-24 (“col tuo nome sull’anima”) and beats 53-56

(“io l’ombra d’un sorriso!”), as results of performances conducted by Nello Santi (P11-

P15). Because this trend emerges over several performances under Santi’s guidance, each with a different Liù, a likely conclusion is that Santi’s own interpretation of this aria emerges in his pacing, thereby shaping the tradition that followed his contribution to it.

This analysis, covering performances from 1961 to 2009, details the evolving development of a performance tradition at the surface, tracking each change in the whole resulting from each new iteration. Thus, the data reflecting the tradition of tempo for Liù’s aria “Signore, ascolta!” demonstrate the slow pace of change. Furthermore, as the overall average accrues more data (following the natural, additive progression of

A1-2. . .A1-19), earlier performances in this aria’s performance history have a more direct influence in shaping the tradition. For later performances, this reality tends to result in nuanced changes, either at surface or structural levels of the tradition’s hierarchy.

181

Figure 4-22. Tempo contour map for entire recorded and available performance history of “Signore, ascolta!” at the Metropolitan Opera, 1961 – 2009.

“Non piangere, Liù”

Viewing the data of a performance tradition in larger units helps to develop a sense of the overall form and trajectory of that tradition. This parsing is particularly relevant for examining changes by chronological period. Doing so reveals a perhaps

182

surprising instance of a formidable performer having less influence on the shape of a tradition over a decade than one might expect. Such an approach to the performance traditions developing for Calaf’s act I aria follows.10 The following analysis thus considers the tradition of “Non piangere, Liù” in groups according to the decades in which that tradition emerged. The images associated with the tradition’s development over each of the five decades for which it is measureable appear at the end of this section. Arranging them in this way allows for quick visual assessment of the tradition’s overall trajectory and evolution, as well as for the sub-group organization by chronology.

Figure 4-22 represents the tradition of tempo rubato application in Calaf’s act I aria, beginning with Franco Corelli’s performance as Calaf on 04 March 1961, over the five recorded performances of the 1960s. Most notably here, the upward extensions from the first significant stringendo section present in Corelli’s performance (P1) weaken immediately after his performance a year later adds to the tradition. Furthermore, the upward extensions from the minor stringendo at the surface level disappear, which results in the elimination of the link between the earliest, small-scale stringendo and the second, larger-scale one. This structural feature re-emerges after Jess Thomas’s 1965 contribution to the performance tradition (in A1-3), but then disappears and does not appear again in the 1960s. That Corelli did not apply tempo rubato in a similar way over the three recorded performances of the 1960s could indicate its initial presence had more to do with conductor Leopold Stokowski’s temporal structuring than Corelli’s.11

10 The best means of viewing a tradition’s trajectory is, as footnote 4 in this chapter explains, to view the graphs in a digital environment that allows an uninterrupted transition from one graph to the next.

11 This hypothesis is yet uncertain, but the uniqueness of this feature with respect to Corelli’s performances here highlights it as an avenue of inquiry to reveal its provenance and possible motivation.

183

Another weakened hierarchical feature is the connection between the two middle sections of ritardando. In Corelli’s 1961 performance (P1), extensions from these two moments clearly connect in the mid-ground of the Time Scape. As subsequent performances enter this aria’s performance tradition, each emphasizes the stringendo section between these slower sections. This linkage between stringendo sections further weakens what had been a primary tempo feature in Corelli’s initial realization of this aria. By the end of the 1960s, any discernable connection between these ritardando segments is almost absent from the tradition’s tempo hierarchy.

Also apparent here are the changing degrees of tempo modulation in hierarchically un-linked portions. The most notable changes occur in the three most significant stringendo sections. The first area lessens to the point that the average value of the entire data set subsumes the second half, making it visually indistinguishable.

The second area intensifies to the extent that its structural significance increases, reflecting a growing adoption of the practice of singing this section faster than the global tempo. The final stringendo section’s tempo modification vacillates, first being significantly faster at A1-2, and eventually diminishes. The strong similarity in coloring in this area between A1-3 and A1-4 suggests a similar approach to the tempo arc between

Jess Thomas’s 1965 and Franco Corelli’s 1966 performances. The tradition’s most evident change following James McCracken’s 1969 performance is a slight diminishing of the stringendo’s intensity in the final faster-than-global tempo section of the aria.

184

Figure 4-22. Time Scapes for “Non piangere, Liù” tradition origin and evolution 1960s (Top row R-L: P1, A1-2, A1-3; Bottom row R-L: A1-4, A1-5)

185

Figure 4-23. Time Scapes for or “Non piangere, Liù” tradition evolution 1970s (L-R: A1- 6, A1-7, A1-8)

186

Three performances in the 1970s contributed to the growing corpus of singing

“Non piangere, Liu” at the Met. This decade also saw the final performances of the Cecil

Beaton production from1961, in which Turandot had reappeared on the Met’s stage in

1961. The Time Scapes representing the tradition’s evolution in this decade appear as

Figure 4-23. Three changes are apparent, and each at a level deeper than the surface.

First, while the intensity of the opening ritardando lessens slightly, the structural connections between it and the second surface level grows slightly, becoming more evident in the third of the Time Scapes of Figure 4-23 than it is in the first. This sequence appears in the circular zoom lens on the left side of each graph. In addition, the higher structural level ritardando, which appears in the zoom lenses on the right side of each graph, intensifies slightly and grows from more of a pyramidal shape to more of a block shape. The third perceptible transition in this decade is the slight increase of yellow (slightly faster than global tempo) section in the central region of the graph. This emergence primarily showcases a link between the second and third stringendo sections.

Figures 4-22 and 4-23 reflect the performance tradition of this aria over the course of the entire recorded history of the Cecil Beaton production of Turandot.

Perhaps most striking about the evolution of this aria’s tempo tradition over the life of this production is the inversion of the structural linkages between Corelli’s first performance in 1961 and his final performance in 1974. In P1, the most notable structural connections result from the nested surface-level events—the middle and subsequent two ritardandos. The only perceptible connection between stringendo sections occurs at the beginning of the aria, where the first significant area of yellow

187

connects to the earliest, surface-level stringendo. In the Time Scape for A1-8, any visual evidence of these connections is absent. Instead, the emergence of the yellow extensions’ crossing in the center of the graph indicates that, at a structural level, the tempo rubato tradition reveals a marked shift in its trajectory from the time of Corelli’s

1961 performance to the end of the Beaton production’s performances.12 Correlation data for this performance (P1) and its relationship to the progressive averages (A1-2. .

.A1-8) also reflects this divergence. Corelli’s performance shares correlation values of

.9001 (A1-2), .8817 (A1-3), .8481 (A1-4), .8548 (A1-5), .8379 (A1-6), .8257 (A1-7), and

.8221 (A1-8) with them. The next weakest correlation lies between A1-2 and A1-8,

.9099, which is stronger than the closest correlation between P1 and any other performance or point in the entire tradition’s evolution.

12 Further investigation into possible reasons for the divergent trajectory is an area for future consideration, and one that becomes increasingly possible as more data and the means to analyze it becomes available.

188

Figure 4-24. Time Scapes for or “Non piangere, Liù” tradition evolution 1980s (L-R: A1- 9, A1-10, A1-11)

189

The emergence of a new production – Franco Zeffirelli’s lavish and iconic one from the mid 1980s – had relatively little effect on the tradition’s path through this decade. Three noticeable changes appear. First, the growing intensity of the yellow patch in the center of the graphs continues along a similar trajectory as it had in the

1970s. This means that the first two significant surface level stringendo sections continue to be features for performances as early as Jess Thomas’s 1965 iteration, whose contribution to the tradition’s temporal structure becomes evident in A1-3 (see

Figure 4-22). Next, the ritardando section near the middle of the surface level grows slightly in intensity, as the zoom lenses in Figure 4-24 emphasize. More significantly, it begins to extend upward and to the right, forming a link with the extension from the next slower-than-global tempo section. Finally, the final two ritardando sections— emphasized by the zoom lens on the right of each Time Scape—begin to increase in size and eventually form a definite link following in A1-11, which follows Nicola

Martinucci’s performance on 13 February 1988 with Nello Santi conducting.

190

Figure 4-25. Time Scapes for or “Non piangere, Liù” tradition evolution 1990s (Top Row, L-R: A1-12, A1-13, A1-14, Bottom Row: A1-15).

191

The next four Time Scapes, which reflect the tradition of tempo rubato application in “Non piangere, Liù” throughout the 1990s, appear in Figure 4-25. These graphs are nearly identical, meaning that the tradition from 1988 (A1-11) through 1996 (A1-15) remained relatively static. Nello Santi conducted each of these performances; perhaps the stasis in tradition occurred because the conductor remained the same. Few noticeable changes appear in this series. First, the ongoing increase in the visibility of the yellow triangle resulting from the extension emanating from the two centralized stringendo sections continues. Next, the tiny fleck of coloring above the center-most surface-level stringendo—circled by the zoom lenses—increases in intensity from A1-12 to A1-13, lessens slightly at A1-14, and increases almost imperceptibly in A1-15.

Finally, the faint emergence of blue shading along the graph’s right edge is noticeable by A1-15; the box on this latest graph (following Lando Bartolini’s 17 February 1996 performance) highlights this development.

Figure 4-26 reflects the trajectory of the tempo rubato tradition of “Non piangere,

Liù” at the Met through its most recent measurable point. Unsurprisingly, there are no major shifts in the makeup of the Time Scapes representing the most recent trajectory of this tradition. This is in part because of the nature of both averages and tradition: the more constituent parts that exist, the more difficult it is for any one of them to radically alter the trajectory of the tradition. Here, the dot of indigo that emerged in A1-11 – A1-15 with Nello Santi conducting disappears in A1-17. Meanwhile the emergence of the structural link between the two primary stringendo sections becomes more established, with the crossing extensions becoming visible in a color-rich digital environment with accurate color rendering capabilities. Furthermore, the link between the first two

192

ritardando sections and the one between the final two become solid by the time that

Marcello Giordani’s 2009 performance is added to the tradition.

The contour maps of the performances of “Non piangere, Liù” illustrate tradition’s existence as a repository reflecting elements of every performance that created it. As the lower contour map in Figure 4-27 exhibits, overall trends emerge and change even as the numerical values associated with a tradition become more standardized.

Variability within a tradition’s contour thus depends on the novelty or uniqueness of individual performances. In contrast to performances of “Signore, ascolta!” under Santi’s baton, performances of “Non piangere Liù,” also conducted by Santi (P11-P15), exhibit strong independence from each other, as the area delineated by the dashed yellow lines indicates. Even so, only a few moments in these performances are enough to affect noticeable difference in the overall tradition, specifically beats 46-7 and 57-61 in the lower contour map in Figure 4-27. Thus, in order for performances to alter the trajectory or trends in a tradition, they must first be more strongly distinctive than that tradition itself. Second, there must be more than one instance of similarly unique performances in order to affect the course of the entire tradition. Finally, these shifts, at least for performances in the present study, affect only isolated moments within the musical fabric. In other words, no single performance of “Non piangere, Liù” at the

Metropolitan Opera between 1961 and 2009 initiates a paradigm shift for the entirety of the aria. These changes testify to a tradition’s ability to reflect changes resulting from new creative practice while also connecting all performances, even those most dissimilar.

193

Figure 4-26. Time Scapes for or “Non piangere, Liù” tradition evolution 2000s (Top Row, L-R: A1-16, A1-17, A1-18, Bottom Row: A1-19).

194

Figure 4-37. Tempo contour maps for performances (top) and moments in the tradition (bottom) for all recorded and accessible performances of “Non piangere, Liù” at the Metropolitan Opera, 1961-2009.

195

“Tu che di gel sei cinta”

Sometimes, changes in a tradition are gradual and difficult to perceive either in small or larger groupings. In such instances, the most effective means of understanding these changes is to consider a macroscopic and dynamic model of the tradition as it evolves, as object 4-1 demonstrates. Alternatively, any static layout in which large groups of the tradition are visible can be useful. Figures 4-28 and 4-29 illustrate the tradition for Liu’s suicide aria, “Tu che di gel sei cinta,” as it evolved. Here, the two dominating hierarchical features—the faster-than-global beginning and mostly slower- than-average remainder—clearly express the evolution in tempo choices. The size of the slower-than-global region diminishes over the course of the 49 seasons this study represents, indicating a less intense slowness in relationship to the overall amalgamate.

This does not indicate, however, that the tradition reflects less of a slowing, but that the slowing is closer to uniformity as the tradition progresses. In other words, singers and conductors all seem to agree that these moments need to be slower than the global tempo for the aria. The continued dominance of this region can also make musical- dramatic sense. As Liù’s life comes to its close, performers might want to emphasize the moment as it carries the preponderance of the opera’s pathos and serves the dramatic role of (ostensibly) melting the ice around Turandot’s heart.

Object 4-1. Dynamic Evolution of tradition for "Tu che di gel sei cinta." (.mp4 file 8.1MB)

196

Figure 4-28. Time Scapes representing tradition of “Tu che di gel sei cinta,” 1961-87

197

Figure 4-29. Time Scapes representing tradition of “Tu che di gel sei cinta,” 1988-2009

198

The contour map for both performances and the tradition of performing “Tu che di gel sei cinta” shows three distinct shifts in the overall trend, as Figure 4-29 illustrates.

There are clear breaks between the productions (following T1-8 and before P9), during

Nello Santi’s tenure as the principal conductor of Turandot at the Metropolitan Opera

(P11 – T1-15), and in the time since Santi’s last recorded appearance. The large blocks of the same color in Figure 4-29 reinforce the gradual means by which the performance tradition of “Tu che di gel sei cinta” evolved. Perhaps unexpectedly, this moderate shift in tradition does not result from high degrees of similarity amongst these performances, as I will explore in Chapter 6.

199

Figure 4-30. Tempo contour map for performances and moments in the tradition for all recorded and accessible performances of “Tu che di gel sei cinta” at the Metropolitan Opera, 1961-2009.

200

Conclusions

Chapter 4 illustrates how traditions embody mathematical principles associated with averages. Music, with its many mathematically describable phenomena—in terms of tempo (beats/time), dynamics (decibels), or pitch frequency (Hertz)—thus lends itself naturally to quantitative description. Analysis of the quantitative model of tradition for performances of “Signore, ascolta!” reveals how the sometimes minute differences between performances can result in definitive characteristics of a tradition as it evolves from one performance to the next. Analyzing the performance history of “Non piangere,

Liù” affirms the mathematical principle of averages with an expanding dataset and its corollary for traditions. The more data that exists in a dataset makes it more difficult for new, individual entries to dramatically shift the average. Similarly, as more performances occur, each one’s ability to influence the overall shape of a tradition diminishes. Finally, considering the entirety of the tradition of performing “Tu che di gel sei cinta” at the Met, as it appears in Time Scapes, is easiest in a dynamic, moving- image environment. One overall trend that emerged here is that the tradition itself showed a slowing over its fifty-season progression, which results from the performances themselves becoming increasingly slower over this span. Furthermore, frequent vacillations of tempo across sections of this aria result from independent choices by each performer, suggesting the lack of a single convention for performing it. As digital tools continue to develop, more sophisticated representations of traditions will be possible,

Analyzing tempo traditions based solely on their chronological emergence is a necessary part of identifying how or why traditions evolve as they do. This process allows the clear identification of what tempo traditions are and when they emerge in a

201

performance history. The data presented above situates tradition as a resultant entity, rather than a model, guide, or prescriber of musical practice. Such an entity can be used by performers in different ways. If, for example, performers decide to follow the model of a performance or a feature of a tradition, such an act might begin the process of creating a convention within the given tradition. How conventions, which imply expectation and even authenticity, affect onstage musical behaviors remains work best suited for a combination of digital and analog musicological inquiry. The current data- based approach works best as an identifier of traditions and descriptor of practices and traditions. As Chapter 5 demonstrates, other factors, such as singer choice, conductor decision, or production history provide additional information that can help us understand the essence of a tradition.

202

CHAPTER 5 QUANTITATIVE AND DEMOGRAPHIC-BASED MODELS OF TRADITION

Chapter 4’s chronological organization of can illustrate “what” they are, or “when” or with “whom” they emerged. It cannot, however, elucidate “how” or “why” they came to be. Examining these findings in the context of the opera’s production and performance histories—what I term performance demographics—can help further elucidate the opera’s performance traditions. This chapter begins with analysis of the production history and its relationship to the tradition of Turandot’s Act II aria “In questa Reggia,” highlighting the prominence of singers Gwyneth Jones and Ghena Dimitrova, and conductor Nello Santi. Consideration of other conductors and their approach to the riddle scene demonstrates Santi’s clear shaping of the tradition in a way that is markedly different than conductors who preceded and followed him. For “Nessun dorma,” Franco Corelli and Plácido Domingo emerge as the two most influential singers of Puccini’s most famous tenor aria at the Metropolitan Opera. The additional layers of inquiry introduced in this chapter can offer data-based confirmation of conclusions resulting from conventional inquiry, such as mitigating factors in performance practices, and secondary anecdotal evidence about working relationships among performers.

Conversely, comparing the data of a performance tradition against the demographics— the who, what, when, and where—of the performance history that created the tradition can shed new light on assumed or inherited conclusions.

Despite the significance of any performer or production in this model of tradition, attributing influence beyond anything other than a performance’s influence on the overall shape of the tradition is erroneous. In other words, this model cannot determine if the singers who followed Jones used her performances or their sense of the tradition

203

following her performances as models. All reference to influence in this study, and for any discussion of relationships between performances herein, is strictly for the correlative relationship between numerical descriptions. Correlation or similarity cannot determine causation, although they can show instances where one might most fruitfully direct questions of influence.

“In Questa Reggia” in Two Productions

Turandot’s measurable performance history spans two productions. Cecil Beaton designed the first of these, which ran from 24 February 1961 until 28 December 1974, while Franco Zeffirelli’s production debuted 12 March 1987 and continues to the present.1 Figure 5-1 exhibits the headdress, costume, and set as Cecil Beaton designed them for Act II, scene 2 for the 1961 revival. Here, red, orange, pink, yellow, and gold comprise the primary color palette of Turandot’s long-trained dress and heavily ornamented headdress. The set consists of a single canopy atop a large staircase protruding from upstage left to downstage center. The backdrop of stage right shows a few spires from Peking and the mountains further in the background, while the stage left backdrop appears to show a cliff face. The crowd of Pekinese citizens flank the foot of the staircase on both sides, while Turandot assumes her place in the middle of the staircase and her aged father, the Emperor Altoum, occupies a seat underneath the canopy.

Figure 5-2 displays the dress, headdress, and set design of Franco Zeffirelli for

Act II, scene 2. Here, the multi-colored costume for Birgit Nilsson in the Beaton

1 While Joseph Urban’s production of Turandot (1926) predates both the Cecil Beaton and the Franco Zeffirelli productions, no known recordings of performances from it exist. The last performance of Turandot with the Urban production was 08 January 1930, and Roger Flury dates the first recordings of a Turandot performance to 1934.

204

production has given way to bejeweled silver with cream accents. Zeffirelli’s headdress is more evocative of a star burst than Beaton’s bauble-heavy rainbow. The set depicts an inner courtyard within the palace walls, a tri-canopy complex accentuates upstage, while a small pool and pathways over and around it dominate the mid-stage. The crowds gather at the foot of the stage to watch in anticipation.

The Beaton Production

Representations of the initiation and evolution of the performance tradition of as it pertains to Cecil Beaton’s production of Turandot appear as figure 5-3. This production covers eight performances of Turandot. As figure 5-3 exhibits, the tempo rubato tradition pertaining to “In questa Reggia” remains fairly consistent over the fourteen seasons using Beaton’s production. This is Turandot’s entrance aria, and she explains in it the reason for her hatred of men (vengeance for Lou-o-Ling) and her insistence that no one shall ever possess her. Three significant and apparent changes occur over the recorded run of this production, and each pertains to one of the aria’s three major structural sections. First, the introduction—the triangular feature outlined by the colors of faster-than-global in the lower left corner of each Time Scape—becomes faster throughout its structure. While its greatest tempo difference from a global tempo occurs in A1-4, following Birgit Nilsson’s performance on 03 December 1966 with Zubin Mehta conducting, this feature becomes solid yellow and orange by the end of this production in 1974.

205

Figure 5-1. Detail of headdress, dress, and set as designed by Cecil Beaton for Act II, scene 2. This production was first in use beginning 1960-61 season. Images used with permission by the Metropolitan Opera Archives.

206

Figure 5-2. Detail of headdress, dress, and set as designed by Zeffirelli for Act II, scene 2. This production was first in use beginning 1986-87 season. Images used with permission by the Metropolitan Opera Archives.

207

Figure 5-3. Tempo tradition Time Scapes for “In questa Reggia” in Cecil Beaton’s production.

208

The second significant change occurs in the slower-than-global section wherein

Turandot recounts the horror of her ancestress’s rape and murder. Here, the structural significance of this feature grows from Nilsson’s 1961 performance throughout the entirety of the Beaton production. At the beginning, a clear “x” results from the extensions at the beginning and end of this section crossing each other at the apex of the triangle and continuing toward the opposite edges of the graph (P1 in figure 5-3).

The triangle on the bottom of the “x” shape is almost entirely blue, indigo, or purple, while the top triangle of the “x” form here is mostly blue and green. As the performances accrue and create the tradition, however, this top portion becomes more solidly blue and indigo as the total area of the slower-than-global reflected tempo relationships increases toward the upper right hand side of the Time Scapes. The third area of change covers the aria’s second main structural section. Like the first two sections, the change over the duration of Cecil Beaton’s production is one of increasing structural significance. Here, the faster-than-global tempi yellow extensions expand from not reaching (P1) the right edge of the graph to an increasingly solid block of yellow along this edge (A1-8).

The Zeffirelli Production

Figure 5-4 reflects the origin and trajectory of the tradition resulting from performing the Franco Zeffirelli production of Turandot at the Met from 1987 to 2009.2

Here, as for the Beaton production, the most noticeable changes occur in the respective intensities of the tempo modifications as they contribute to the overall tempo hierarchy.

2 The Time Scapes in this figure result from considering the emergence of the tradition solely in relationship to the Zeffirelli production. No tempo data from performances of the Beaton production are part of the data set used to generate these graphs.

209

Unlike the first performance of the Beaton production, the first performance of the

Zeffirelli production establishes a considerably faster-than-global and more structurally significant set of tempo relationships for the aria’s introduction (compare P1 in figure 5-3 to p2P1(P8) in figure 5-4). This region is also unlike the steadily intensifying progression of this region in the Beaton production in that it first lessens dramatically, becoming almost entirely structurally insignificant by the point of p2A1-7. The stark weakening of this region’s significance in the overall tempo hierarchy begins with p2A1-3, which represents the tradition following Ghena Dimitrova’s performance on 13 February 1988 with Nello Santi conducting (P11). Santi conducted each of the following four performances, whose occurrences result in the evolution of the tradition represented by p2A1-3 – p2A1-7. The most dramatic decrease in the intensity and structural significance here occurs beginning with p2A1-4, when Gwyneth Jones sang Turandot, shown in Figure 5-4. Jones was also on stage for the performances that led to the tradition as Time Scapes for p2A1-5 and p2A1-6. The data generated from these performances indicates that Nello Santi was the most influential conductor in shaping the tradition of “In Questa Reggia” and its tempo hierarchy. A significant portion of

Santi’s influence stems from his five recorded appearances on the podium for Turandot at the Met, which accounts for 26% of this opera’s total recorded corpus. Furthermore,

Santi conducted 31.5% of all Turandot’s 311 performances at the Met, which might suggest that his influence in the recorded corpus is not significantly dissimilar from his influence on the entire performance tradition of “In questa Reggia.”3

3 As of June, 2016, the Metropolitan Opera Archives Online reveals 313 entries for Turandot. There are 312 records of performances, with a double entry for Performance 306. There are two different CID

210

Because the singers in these performances were Ghena Dimitrova and Gwyneth

Jones, they emerge as the most important ones in the aria’s tempo rubato tradition in either the Zeffirelli or the Beaton productions. Dimitrova’s performance in 1988 marks a definitive shift away from what Eva Marton had done in this aria with James Levine conducting (P8, P9), resulting in relative temporal stasis between performances and each performance, and the new sub-tradition resulting from the new production. Her

1996 performance lessens the standardizations and intensifications of tempo modification that emerged with Jones’s performances. Jones’s performances both standardize the tempi at the aria’s beginning to near global, and expand the structural significance of the slower-than-global first section and faster-than-global second section, setting her apart from Dimitrova, Dugger, Gruber, Guleghina, and Marton—the other singers recorded singing the Zeffirelli production. Given that her performances accentuate Santi’s tempo choices in a way that Dimitrova’s do not, the creative partnership between Nello Santi and Gwyneth Jones between 1990-1995 stands out as the most influential in the recorded history of Turandot at the Met.

numbers for this performance, but no clear indication in the record as to why two unique CIDs are present.

211

Figure 5-4. Tempo tradition Time Scapes for “In questa Reggia” in Franco Zeffirelli’s production.

212

Figure 5-5. Tempo tradition Time Scapes for Zeffirelli production performances, inclusive of tempo data from Beaton production.

213

The tempo data from performances in the Zeffirelli production along with the

Beaton production results in the Time Scapes appearing in figure 5-5.4 This arrangement reflects the significance of Ghena Dimitrova, Gwyneth Jones, and Nello

Santi on the tempo modification tradition; it occludes the majority of the ways in which they are the foremost shapers of the tradition and its evolution over two productions.

Nevertheless, the small yellow triangle immediately following the dominating blue triangle becomes more yellow than green over the progression from A1-11 to A1-15.

Also, the large blue triangle becomes darker overall, and its upper edge extends further into the Time Scapes’ structural regions. Determining whether or not an audience is aware of this shift seems tenuous at best at present, but tracking such changes in performance, tradition, and an audience’s awareness remains an area for future research.

“Straniero, ascolta!”

Conductors are also performers—those persons primarily responsible for the musical behaviors in the opera house; organizing a review of performance chronology that includes conductors and their roles can provide additional insight into a tradition’s creation. The riddle scene—“Straniero, ascolta!”—is an excerpt that lends itself to an examination of a conductor’s role, as it includes many parts for both soloists and chorus. It also offers a clear structure of the three riddles and their respective responses. Naturally, conductors who led performances of the work numerous times exert a stronger influence on the tradition than those with fewer performances. As mentioned in the foregoing discussion about “In questa Reggia,” Nello Santi conducted

4 These Time Scapes reflect all tempo data from performances in both productions.

214

five of the performances in the current corpus, accounting for roughly 26% of the available and thus measurable tradition. Kurt Adler, Zubin Mehta, James Levine, and

Marco Armiliato each appear twice on record, accounting for about 10.5% each and

44% overall. Thus, repeat conductors account for approximately 70% of the recorded tradition of Turandot and conductors with single appearances account for approximately

30% of the corpus. Conductors who appear only one time in this corpus are, in chronological order of their appearance: Leopold Stokowski, Fausto Cleva, Gabor

Ötvös, Alberto Erede,and Bertrand de Billy.

Distinguishing the effects of conductors who led the work twice is easier if these conductors appear earlier in the tradition’s chronology, as figure 5-6 illustrates. Thus, the respective effects of Adler, Mehta, and Levine on the tradition are clearer than those of Armiliato. Adler’s two performances shape the tradition as it appears in A1-2 and A1-

6. Mehta’s performances result in the tradition as A1-4 and A1-5 exhibit. The similarity within and between these pairs is most evident when comparing them to P1 (Leopold

Stokowski) and A1-3 (following Fausto Cleva conducting in 1965). Here, the first riddle’s delivery (the first slower-than-global feature) becomes successively less solid as the tempi relationships of accruing performances become more complex in and of themselves, and in relationship to the other performances. The response to the first riddle (the first faster-than-global feature) changes rapidly from P1, becoming mostly red in A1-2 and growing in size. Adler and Mehta both set tempi that resulted in the formation of the hierarchical feature associated with the second riddle’s delivery and response, as well as the delivery of the third riddle. The hierarchical features follow the tendencies evidenced by the addition of the performance resulting in A1-2 (Adler in

215

1963), and that each marks a further departure from the shape of the 1961 performance

(P1). This particular trajectory establishes Kurt Adler as the most influential conductor over Turandot’s first ten recorded performances.

The similarity of graphs resulting from the additions of Adler’s first recorded performance (A1-2), Mehta’s two recorded performances (A1-4 and A1-5), and Adler’s second recorded performance (A1-6) to the tradition reflects both a similar sense of pacing, and a clearly defined trajectory for the tradition. The first riddle is substantially slower than the global tempo, but with some moments (beats where Turandot is not singing) closer to global. The response to this riddle becomes increasingly faster, perhaps indicating a stronger desire and push to balance the tempo relationships. The second and third riddles follow a similar pattern. As the tempo relationships continue to clearly define this scene’s formal divisions of riddle and response, the clear alternation between faster- and slower-than-global tempi prevents the tempo trajectory from becoming only a large ritardando. The graphs resulting from Levine’s two conducted performance entering the tradition show little change between them, and also little change from A1-8. This results both from Levine’s similarity in tempo to the tradition at the points he conducted Turandot in 1987, and also to the increasingly large data set compared to when Adler or Mehta conducted (1:9, or 1:10 vs. 1:2, 1:4, 1:5, or 1:6).

216

Figure 5-6. Tempo tradition Time Scapes for Turandot’s riddle scene, “Straniero, ascolta!”, part 1 (1961-1987)

217

Figure 5-7. Tempo tradition Time Scapes for Turandot’s riddle scene, “Straniero, ascolta!”, part 1 (1988-2009)

218

As the most-recorded conductor in this corpus, Nello Santi shaped the tradition more than any other conductor or pair of conductors. Over the course of Santi’s five conducted performances, color intensities and structural areas of tempo modulation increase regularly, as figure 5-7 exhibits. The first riddle’s tempo relationships remain significantly slower-than-global throughout Santi’s cultivation of the tradition, growing slightly larger in area (along the right edge of the first triangular feature) from A1-11 to

A1-15. This indicates that the structural significance of this demarcation grows slightly, and that its growth is partially the result of the diminishing significance of the structural importance of the first response (the first faster-than-average feature in the graph).

200 180 160 140 120 100 80 60 40 20

0

Tempotofollowing beat (in bpm)

1

24

47

70

93

116

139

162

185

208

231

254

277

300

323

346

369 392

Beats (577 total) 415

438

461

484

507

530

553 576 Average tempo performances 1-15 Performance 16 Average tempo performances 1-16 Average tempo performances 1-17 Performance 18 Average tempo performances 1-18

Figure 5-8. Tempi across performances conducted by Marco Armiliato in 2003 (dark blue) and 2007 (light blue), in comparison to tempo relationships as evidenced by the tradition before and after these performances.

219

With this particular data set, and the presence of five recordings featuring the same conductor in sequence, any differentiation from the pattern becomes increasingly difficult to ascertain if looking only at the Time Scapes. The most immediate way to see how and where (musically) Armiliato differs from the norm is to consider the surface level events, as Figure 5-8 illustrates. Here, Armiliato’s conducted tempi at each riddle’s response peaks high above the tempi that the lines representing the tradition might otherwise indicate. Even so, each instance of it does not greatly affect the tradition immediately following it—unsurprising given the size of the tradition’s data set.

However, this occurrence again reflects the parallel natures of both tradition and statistical averages; more constituent parts make discerning a fundamental shift from a single addition in a progressively amalgamated corpus more complicated. Such a property (of both tradition and statistical averages) results in the formation of a tradition with no true outsiders. Every performance contributes to the tradition; none can thus actually exist outside of it. Some performances will have weaker connections to others and to their traditions, but they remain connected by virtue of their having come into being.

“Nessun dorma!”

Another fruitful way of viewing a tradition’s trajectory is to consider its most significant divergences and the performances that cause them. In the current model, a tradition’s trajectory emerges in the changes of averaged tempo values. For example, the change in values for tempo relationships from A1-2 to A1-3 represents the trajectory of development. Statistical regression is one means of characterizing these changes. A danger with a regression metric in artistic endeavor is its application for predictive purposes, which could easily establish a sense of expectation for how a performance

220

“should” be, rather than how it “could” be. Because of this possibility, and the limited scope of the current project, examples of how one might use regression metrics do not appear; this potential application is an area for future investigation.

This view focuses on the musical events that occurred, thus inviting questions about how performers change over time, and what their working relationships with their collaborators are. Over the course of the Cecil Beaton production (1961-74), Franco

Corelli’s influence on the performance tradition of “Nessun dorma!” is indelible, seemingly independent of the various conductors with whom he performed. As figure 5-

9 exhibits, Corelli’s performances—which enter the tradition at Pa, A1-2, A1-4, A1-7, and A1-8—emphasize Calaf’s lines “Tu pure o Principessa, nella tua fredda stanza, guardi le stelle che tremano d’amore e di speranza” and “quando la luce splenderà!”5

These separate surface events form the first mid-level structural crossing of this aria.

This feature then remains present throughout all subsequent performances, though the five performances conducted by Nello Santi (performances resulting in Time Scapes

A1-11 – A1-15) de-emphasize its structural importance within the overall tradition.

The first noticeable difference in these two elements is the overall de- intensification of the first slower-than-global tempo feature. That Corelli was Calaf in both performances might indicate that Stowkowski’s interpretation differed from Corelli’s and how he performed at later dates. Another noticeable change occurs over the central slower-than-global feature, framed in A1-3. Here, the size and color intensity remain similar between P1 and A1-2, but this region expands with A1-3, when Jess

5 “You, o Princess, in your cold room, watch the stars which tremble with love and with hope,” and “when the light will shine.”

221

Thomas sang Calaf under the Baton of Fausto Cleva. The intensifying intersection of these two regions at A1-2, followed by its gradual disappearance between A1-3 – A1-6, and reemergence at A1-7 and A1-8, each of which results from a Corelli performance, implies that Corelli’s pacing over these two sections was the most influential over the fifteen seasons represented in figure 5-9.

The other prominent area that changes is the faster-than-global region at the aria’s end. Here, the interpolated fermata on the aria’s penultimate note (the last area of purple) breaks up the yellow and orange and red areas surrounding them (framed in

P1). In the 1961 performance, the audience’s applause is robust enough to warrant

Stokowski’s cessation of the orchestra until the fervor dissipates. This magnifies the effect of the slow tempi on the surrounding faster tempi. In subsequent performances, audiences hold their applause until after the orchestra plays the aria’s primary theme, thus making the Stokowski performance more of an exception to the tradition than a holistic model to follow. The one noticeable change in the tradition following Plácido

Domingo’s performances under James Levine is the reemergence of the intensified crossing of the two middle slower-than-global areas, as the box in A1-10 frames.

222

Figure 5-9. Time Scapes representing tradition of “Nessun dorma!”, 1961-1987.

223

Figure 5-10. Time Scapes representing tradition of “Nessun dorma!”, 1988-2009.

224

As figure 5-10 illustrates, most of the structural features present following A1-10 remain the same shape and size through A1-19. The two areas that continue to change in the overall tradition are the respective structural crossings resulting from extension from the first three slower-than-global sections. The first—in the upper frame of A1-11— loses its stronger emphasis, which the fading of indigo into the lighter blue indicates.

The second—which the lower frame in A1-11 shows—intensifies gradually, growing somewhat larger and accumulating more dots of indigo until four such dots appear clearly in A1-18 and are still apparent in A1-19. These instances embody the property of tradition wherein disparate surface events between performances alterations to the tradition do not necessarily appear on the surface, but in the hierarchical structure.6

Conclusions

Chapter 5 illustrates various ways of examining these mathematical principles in the context of performance history. For both “In questa Reggia” and “Straniero, ascolta!”, clear changes result from the performances directed by Nello Santi, since he led more of them than any other conductor. Furthermore, and particularly evident in the visualizations, these changes are subtle, owing to the decreasing proportion of single performances to an increasing and collective whole. In other words, the third performance in a history has a 1:3 ratio to the corpus of its tradition while the fifteenth has a 1:15 ratio. This reality reinforces the notion that a tradition is a collective repository for all performances of the same work, just as traditions are additive in nature and expand with every new iteration. Because later performances contribute to a larger

6 Defining the theoretical nature of the hierarchical structure of both performances and the traditions that link them, especially as it pertains to Heidegger’s notions of wesen, Enframing, and poiēsis, remains an area for future consideration.

225

and more diverse tradition than earlier ones, they must be substantially different from the tradition in order to stand out against it.

In the case of “In questa Reggia,” Gwyneth Jones and Ghena Dimitrova exert the greatest influence on tradition, while Nello Santi emerges as the most influential conductor, owing to his numerous appearances in the corpus of this study. Even so, the shift in production from Cecil Beaton’s design to Franco Zeffirelli’s appears to have a more significant role in shaping the tradition of this aria at the Met than any individual performer or group of performers. Whether this significance has to do with costuming— including perhaps the weight of the respective headdresses or clothing—or choreography, or some other factor, remains unknown. Nello Santi’s higher percentage of involvement in the overall performance history, coupled with his unique pacing of the riddle scene, renders him the most significant in the performance tradition of Turandot’s dramatic crux. Furthermore, Santi’s decision to emphasize the silence between Calaf’s answer and the ensuing confirmation of it results in a unique sub-tradition that was a convention of his interpretation of Turandot. The performance tradition of “Nessun dorma!” is perhaps the most dynamic of those studied in this chapter. However, comparatively few of the inherent changes appear on the surface of the tradition; most changes emerge in the hierarchy, especially as the chronology progresses through more performances. Because the structural levels below the surface are home to most changes, one can conclude that traditions, like performances, are not merely one-

226

dimensional entities. Both create and exist with nuanced and hierarchical formal structures.7

The analyses in Chapter 4 and Chapter 5 each track various changes and possible causes in the evolution of traditions surrounding performances of Turandot at the Metropolitan Opera. Some of this discussion has engaged several of the ways in which these traditions have related to each other and to the traditions to which they contribute. Considering how a performance tradition develops in relationship to the people whose performances create it is a vital step towards understanding who its creators, arbiters, and dissidents may be. The next chapter takes an even broader perspective of the history of these performances, using correlation and data mapping to visualize Turandot at the Metropolitan Opera as a dynamic network.

7 These structures merit further analysis; borrowing concepts from Schenkerian analysis of a musical score—especially fore-ground, mid-ground, back-ground—currently appears one viably fruitful path forward.

227

CHAPTER 6 CORRELATION NETWORK ANALYSIS OF PERFORMANCE AND TRADITION

Overview

Collectively, Chapters 4 and 5 identify and examine the traditions of the same

Turandot excerpts whose performances Chapter 3 textualized and analyzed. As such, each excerpt’s performances and resulting traditions share inherent connections. The foregoing analyses are based on numerical data, making statistical analysis the most appropriate method for examining the varying strengths of the network’s interconnectedness. As Chapter 2 states, correlation is a straightforward means of assessing the strength of relationships between sets of data. As Chapter 2 outlined, correlation values, or r- values, reflect the strength of similarity between sets of data, where an r-value of +1 represents an identical match, 0 represents no discernible correlation, and -1 represents an absolute opposite relationship. Most of the correlation values in this study—all for tempo relationships in the performances and points in each aria’s respective tradition—fall in the range 0  +1. As the following discussion will underscore, strong correlation values—those closest to +1 between any two data sets— do not insinuate causation, but do highlight robust similarities, which could result from influence or causation. As there are 666 different relationships (one between each performance or point in the tradition and each of the others), there are eighteen orders of correlation values. Correlation matrices can be cumbersome to navigate, thus visualization of the various relationships within a data set via network graphing is often more useful than examining the correlation matrices themselves.

Network graphs consist of nodes, which represent objects, and edges, which represent any number of connections between objects. Nodes can be either sources

228

(the original point of reference), targets (the point of comparison against the source), or both. Graphing all orders of correlation in a network is possible, though in a performance network, every node relates to every other node at some point. Doing so reveals the complexity of a network, but at the expense of gathering finite, clear information about specifics within it. Analyzing only one order of correlation offers considerable clarity of understanding nodes’ relationships to each other. First order correlation values are the strongest and second order correlations are the next strongest, while eighteenth order are the weakest.1 Thus, an examination of all first order correlations considers only the strongest link from each performance. In a first order correlation network, the likelihood of nodes being only a source or a target is greater, while the likelihood of a node being both is generally less. As a correlation network accounts for more orders of correlation, the chances of a node being only a target or a source diminish. In each of the following graphs, the edges reflect the correlation values between nodes, and each node’s shape reflects whether it represents a performance (diamond shaped node) or a point in the tradition (elliptical shaped node). Furthermore, each edge has a dot touching its source node and an arrowhead touching its target node. Source nodes are the relationship’s origin point and the target node is the one to which the source relates. Different layouts of nodes and edges highlight certain characteristics within each network. An explanation of the kind of layout in use accompanies each graph.

1 A first order correlation value for any node is closest to an r value of “1,” while an eighteenth order correlation value is, for its respective data set, the furthest r value from “1.”

229

These graphs and their accompanying descriptions are not meant as comprehensive analyses of the performances and traditions they represent, but samples of how network analysis might be useful in better understanding the dynamic natures of music performance, tradition, and the networks they embody. This chapter’s main focus, therefore, is to analyze and visualize the data of Turandot’s fifty-year recorded performance history to better understand how its tradition has evolved. This analysis will lead, in turn, to better understand the relationship between tradition and performance, and, moreover, how tradition is similar to and different from convention.

“Signore, ascolta!”

Using Cytoscape’s Prefuse Force Directed layout of the first order correlations— only the strongest similarity of each source node—for “Signore, ascolta!” results in

Figure 6-1. This layout models the nodes as physical objects and the edges as springs connecting those objects together whose length results from the interaction attribute

(correlation value). Thus, the closer nodes appear to each other (via edge length), the stronger their similarity is. This arrangement is useful in that it exposes a high degree of a network’s structure, particularly in revealing which nodes are central connection points and how they form clusters. For Liù’s act one aria, four clusters of performances and their traditions emerge. These clusters reveal the demarcation of four distinct points in this aria’s performance convention, as well as the centrality of tradition itself while performances remain on the periphery. Furthermore, only two performances (P18 and

P19) are target nodes as well as source nodes, and only one point in the tradition is not a first order target node (A1-4). The paucity with which performances are target nodes

(10.5%) and points in the tradition points are not (5.5%) point toward tradition’s connective and intermedial nature for relating performances and those who make them

230

happen through time. In this sense, tradition becomes the result of performances, since it is a connector whose existence relies on a work’s repeated realizations. Specific variances or subgroups of them within a tradition might constitute a convention, and are thus part of the tradition, but are not the entirety of the tradition.

The first cluster reflects the trajectory of the tradition as it resulted from the first ten performances. Its embodiment following Edda Moser’s April 1974 performance (A1-

7) has the highest degree of centrality (four first order connections) in the largest cluster. Centrality here reflects the number of incoming connections a node has. Thus, if these nodes did not exist, this network would look vastly different. Conversely, nodes without any incoming edges are less central to a network’s existence; the network’s overall shape would not change too much if these nodes were absent. Three branches connect to A1-7: the tradition over its first three performances (P1, P2, P3, A1-2, A1-3), the tradition resulting form the next three performances and two of these performances

(A1-4, A1-5, A1-6, P4, P6), and the performance immediately preceding the tradition to this point (P7). One way to think about the prominence of this node’s centrality is that its comparative commonness to the nodes connected to it renders it less distinctive than nodes without incoming edges. The tradition at this point is, perhaps expectedly, most reflective of all performances that have occurred before it. One might also consider

Edda Moser’s performance, which was the final contribution to this evolutionary moment, as indirectly less distinctive than the preceding performances, at least in terms of its tempo relationships.2

2 Moser’s performance is distinctive in its other expressive features, the analysis of which currently remains rudimentary or theoretical at best. This is an area for future research.

231

Figure 6-4. Prefuse Force Directed Layout of first order correlations amongst performances and points in the performance tradition of “Signore, ascolta!” Node size is relative to edge count (incoming and outgoing), with higher edge count nodes being larger. Points in the tradition tend to be central and flow in chronological order, while performances remain on the periphery.

232

This node then connects to the tradition as it emerged following the next three performances (A1-8, A1-9, and A1-10). The reflexivity between A1-9 and A1-10 is unsurprising as the performances leading to theses points occurred only one week apart, and with the same singer (Leona Mitchell) and conductor (James Levine). The first order correlation of Leona Mitchell’s March 1987 performance (P9) to A1-10 rather than A1-9 indicates that the tempo relationships of this performance are distinct from those of the accrued average prior to her performance. They are, however, not strong enough to have shifted the convention with only one performance, but their similar repetition in her performance a week later (P10) cements Mitchell’s pacing as a mildly significant departure from the tradition as she inherited it.

The next cluster accounts for the tradition as conductor Nello Santi shaped it

1988-1995 (A1-11 – A1-14). Santi applied his unique approach to dramatic pacing to

“Signore, ascolta!”; his approach was distinctive enough to create a dramatic shift in the overall trajectory, which is why A1-11 is the only node of tradition which is not also a target node. This means that, at the first, strongest order of correlation no performance or other point in the tradition’s development refers to this one. In one sense, this is unsurprising as it represents a clear break from conventions that preceded it. Even though this point is an origin for a new trajectory, A1-11’s source-only-status reflects an expected and primarily forward momentum for a dynamic tradition as it results from continued and varied performances. This node’s second order correlation is a reflexive one with A1-10, which aligns to the chronological accumulation of tempo relationships.

As P10, P12, and P11 all relate most closely to a point in the tradition beyond the point when they enter it (at A1-10, A1-12, and A1-11, respectively), one can consider each of

233

their shaping effects latent. This quality reflects another characteristic of a tradition in that its constituent parts’ effects on its overall shape, trajectory, and identity are not always immediately apparent. Sometimes, conventions can seem subtly out of place at the moment of their performance; these can emerge later as interpretations shape more realizations, or they can be testaments to creative enterprises’ inherent individuality.

The smallest cluster in this network consists of only three nodes: A1-15, P16, and A1-16. The reflexive first order relationship between A1-15 and A1-16 is a natural result of their sequencing in the data pool, as is A1-16’s centrality in this cluster. A1-15, which represents the tradition at the conclusion of Nello Santi’s time on the podium, connects forward in the chronology at first (to A1-16), second (A1-17), and third (A1-19) order correlations. This is partially due to the latency of the effect of the performances before this point; moreover, it reveals the effect of the combination of Nello Santi and

Angela Gheorghiu on the tradition of this aria following her 1996 performance (P15).

Alternatively stated, tempo relationships in Gheorghiu’s performance of “Signore, ascolta!” under Santi’s baton do not imply a reflective or historically oriented inclusion of tradition. Rather, that the first five orders of tradition point correlation for P15 are A1-15

(r = .7419), A1-19 (r = .7511), A1-16 (r = .7495), A1-17 (r = .7446), and A1-18 (r =

.88168) demonstrate Gheorghiu’s and Santi’s lasting effect on the tradition’s shape. In relationship to other performances, Gheorghiu connects most strongly to Marina

Poplavskaya’s 2009 realization (P19, r = .7976). Her next strongest performance correlations are to Adriana Maliponte in 1974 (P8, r = .7135) and Lucine Amara in 1965

(P3, r = .6968). In fact, eight of Gheorghiu’s ten strongest correlations are performances

234

that predate hers, potentially reflecting an awareness of what her artistic predecessors had done, if not some degree of influence or inspiration.3

The four closest correlating performances to P15 are Leona Mitchell in 1988

(P11, r = .7958), Maria Spacagna in 1995 (P14, r = .7452), Aprile Millo in 1990 (P12, (r

= .7444), and Teresa Stratas in 1992 (P13, r = .7320). That these performances punctuate the correlation values of Gheorghiu’s performance to the tradition is likely the result of Nello Santi having conducted all performances from 1988-1996 (P11-P15). The varying degrees of similarity among these performances could of course be coincidental given the dynamic nature of human experience and the creative process, but the noticeably strong connection to Mitchell’s 1988 performance with Santi than to all others could also reflect some level of influence, whether deliberate, conscious, or neither.

Highlighting this strength of similarity, especially given the common conductor, can help to focus inquiries about how and why Gheorghiu’s performance is more similar to

Mitchell’s than to any other singer who performed with Santi.

The final cluster is unique because it contains only three points in the tradition

(A1-17, A1-18, and A1-19), but nearly half of the performances of “Signore, ascolta!”

A1-19 is the central node here, serving as a primary first order target for five performances (P5, P8, P14, P18, P19), a secondary first order target for two (P13,

P15), and a tertiary first order target for one (P17). This high connectedness is unsurprising, at least in part, due to A1-19’s data sequence being the average of all nineteen performances in this study and reinforces the reality of tradition as an inherent

3 Should Gheorghiu ever grant an interview, attaining such knowledge would be a primary goal, though all attempts to contact her prior to 2016 for an interview have been unsuccessful.

235

and interpolated connection between performances. In addition, the early performances—Martina Arroyo’s 1969 and Adriana Maliponte’s December 1974 performances—for which A1-19 is a first order target (the strongest correlation for each) might indicate their roles as forerunners or models for later incarnations of the tradition, if not also for performers. The strongest correlations between Arroyo’s performance

(P5) and all performances (i.e. – excluding points in the tradition’s development) are to those that follow her chronologically. Aprile Millo in 1995 (P12, r = .7996), Hei-Kyung

Hong in 2007 (P18, r = .7967), Leona Mitchell in 1998 (P11, r = .78569, Norah

Amsellem in 2003 (P16, r = .7843), Edda Moser in 1974 (P7, r = .7672), Teresa Stratas in 1002 (P13, r = .7664), and Marina Poplavskaya in 2009 (p19, r = .7583). As was the case in the preceding analysis of Gheorghiu’s performances, such comparative strengths of similarity indicate areas of significant potential for further investigation in the creative process for performers whose performances share high correlation order connections with other performances rather than points in a tradition.

“Non piangere, Liù”

One useful way to visualize the first, second, and third order correlation (each node’s three strongest correlations) network between performances and tradition for

“Non piangere, Liù” is a modified yFiles hierarchical layout, which appears as Figure 6-

2.4 The yFiles hierarchical layout algorithm is useful for representing main direction or

"flow" within a network mapping evolving chronological events, such as a performance history.5 The unmodified graph positions the moments in the tradition’s evolution

4 yFiles is a set of layouts designed by www.yworks.com specifically for use with Cytoscape. 5 http://wiki.cytoscape.org/Cytoscape_3/UserManual/Navigation_Layout. Due to license restrictions, detailed parameters for yFile layouts are not available.

236

vertically and in chronological order while it places the performance nodes horizontally across the top of the graph with respect to edge strength (correlation value) and without concern for chronology. The modification here is to arrange the flow of the network in two streams—one for performances and one for tradition’s evolution—by making the performance stream’s layout mirror that of the tradition’s evolution.

Also in this graph, larger nodes have higher in-degrees (more incoming edges), while the smallest nodes are those with no incoming edges. Thus, the largest nodes

(A1-14, A1-6, and A1-17) are target nodes more often than any other nodes in this particular graph. This layout is useful in that it retains a chronological trajectory for both performances and points in the tradition, making clear when any of a node’s three strongest correlations contradict notions of a straightforward evolution through time.

Furthermore, node sizing according to in-degree demonstrates clearly what points in a chronological series of events are the most direct interconnectivity. Respectively, these kinds of highlighting can indicate chronological reflexivity within data sets, which might mirror similar tendencies or considerations in a creative process. Chronological reflexivity is a term I use for describing the phenomenon when two subsequent nodes relate most strongly to each other (first order correlation), such as is the case for A1-4 and A1-5, P9 and P10, A1-9 and A1-10, A1-12 and A1-13, A1-15 and A1-16, and A1-17 and A1-18 in Figure 6-2. This occurrence for the points in the tradition is unsurprising, especially since the latter point in each of these examples relates to its successive one in the second order correlations for this network. The second strongest connections from A1-5, A1-10, A1-13, A1-16, and A1-18 are, in respective order, A1-6, A1-11, A1-

14, A1-17, and A1-19. All points in the tradition for “Non piangere, Liù” except for A1-3

237

relate in a first, second, or third correlation order to their respectively immediate predecessors and successors. This occurrence reveals tradition’s centrality, and therefore intermediality in this performance history.

Figure 6-5. Variation on a yFiles hierarchical layout for first, second, and third order correlations among performances and the tradition for “Non piangere, Liù.” Node size reflects in-degree from highest (largest) to lowest (smallest). Here, tradition’s existence and evolution as a result of performance emerges and performances that might be significantly “ahead of their time” in respect to the tradition (P1  A1-5, P3  A1-9, A1-12 A1-14) or “reflecting back” on it (P19  A1-5) become apparent.

238

Perhaps expectedly, most performances here have target correlations of points in the tradition nearest to their occurrence, and most are only source nodes. In other words, most performances only refer to points in the tradition at their strongest levels of connection, while only three are also referents or targets. Two of these latter type are performances 9 and 10, both of which occurred within a week of each other and highlighted Plácido Domingo as Calaf and James Levine conducting. The span of time between these performances is the shortest between any two subsequent iterations in the current study, and both included identical casts, which is why the chronological reflexivity between them, while unique, is relatively unsurprising. These performances are the same ones that shared first order correlations with Leona Mitchell portraying Liù for “Signore, ascolta!”. The other performance that is a target node occurred on 1 March

2003 (P16), with Richard Margison as Calaf and Marco Armiliato conducting. This connection is also unsurprising as Margison and Armiliato were the two performers responsible for the performance occupying both the source and target (P18) nodes in this pairing.

One conclusion to draw from these examples is that performances in close chronological proximity, and with the same cast, are more likely to share high degrees of similarity than disparate ones. This does not mean, however, that all such performances will behave this way, just that they are more likely to do so. This kind of connection bears further investigation in contexts where the audience has a prominent role in the art world that brings an opera to the stage, such as appears to be the case at

Milan’s Teatro alla Scala. Another possible, if not likely, reason for the reflexivity

239

between these performances for both “Signore, ascolta!” and “Non piangere, Liù” is their role in the musical structure of Act I’s conclusion.6

The source-target relationships that do not fit the norm of chronological nearness highlight the most interesting aspects of this aria’s performance tradition. The final performance in this study (P19), for which Marcello Giordani sang Calaf and Andris

Nelsons conducted, expectedly connects to the tradition just before and after its occurrence. The unexpected connection is the only edge that moves significantly backwards in the chronology further than to its immediate predecessor. This connection is the second order connection for this performance (second strongest), with a correlation value of .7707, only slightly less strong of a connection than to its first order connection (A1-19, .7912), and stronger than its third order correlation (A1-18, .7687).

In other words, Giordani’s 2009 performance is closer to the tradition as it stood at the end of the 1960s than to any other point preceding it. Whether this means he or

Nelsons listened to earlier recordings and allowed anything he or they heard influence their interpretation remains an area for future consideration.7 Alternatively, this could indicate the beginning of a new cycle within the tradition, which further and future analysis will be able to determine.

The other asynchronous relationship for “Non piangere, Liù” exists between Jess

Thomas’s performance with Fausto Cleva conducting on 16 January 1965 and three points in the tradition during the decade Nello Santi was the principal conductor. The

6 Liù’s aria flows immediately into Calaf’s, which in turn flows immediately into the concertato finale. The only break between these arias in performance occurs because of applause at the conclusion of “Signore, ascolta!” 7 Giordani has expressed a strong interest in this project as a means of better understanding his own evolution as a performer. This question will arise in a future interview with him.

240

first order correlation is to A1-11 (.9138), the second is to A1-12 (.9137), and the third is to A1-14 (.9124). Unsurprisingly then, its fourth and sixth order correlations are also to points in the tradition while under Santi’s baton (A1-13, .9111 and A1-14, .9084). Such a strong correlation to the tradition with as constant and, at times, enigmatic figure as

Nello Santi seems perhaps the strongest indication of influence. However, influence is impossible to ascertain only quantitatively, thus requiring corroboration from qualitative investigation. Nevertheless, such an occurrence here highlights this possibility and can serve as a guide in subjective (rather than objective) interrogations of the creative process and its influences.

“In questa Reggia”

Using Cytoscape’s edge-weighted spring embedded layout on the first and second order correlations for “In questa Reggia” results in Figure 6-3.8 The spring- embedded layout is a variation on a force-directed layout which treats network nodes like physical objects that repel each other. The connections between nodes act like metal springs attached to each pair of nodes. These springs repel or attract their end points according to a force function. The layout algorithm sets the positions of the nodes in a way that minimizes the sum of forces in the network. One visual result of this is that the spatial relationships between nodes approximate the strength of attraction between them. In this model, this strength of attraction is the correlation value of each pair of nodes.

8 www.cytoscape.org

241

Figure 6-6. Edge-weighted spring embedded layout on the first and second order correlations for “In questa Reggia” showing a clear delineation between performances and traditions associated with the Cecil Beaton production (top) and the Franco Zeffirelli production (bottom).

242

This layout reveals two distinct groupings. First, the group at the top of the map consists almost entirely of the tradition’s origins and earliest versions of “In questa

Reggia.” Beginning with P1 (which is the origin of any tradition), the tradition follows an unbroken sequence through P2, A1-2, A1-3, A1-4, A1-5, A1-6, A1-7, and A1-8, with several early performances around the periphery. The one exception in this grouping is

P15, which featured Ghena Dimitrova as Turandot with Nello Santi conducting. Given that the tradition under Santi’s guidance for “Non piangere, Liù” relates so strongly to a past performance, P15’s presence in this grouping is not entirely unexpected, as the data indicate at least a correlation-based reference.

Following the expected reflexive link between A1-8 and A1-9 leads to the second, larger grouping of nodes. Here again, one can easily trace the uninterrupted sequence from A1-9 through each point in the tradition’s evolution until A1-19. A1-9, A1-

10, A1-16, and A1-17 all have low degrees of centrality as they have the minimum (4) or only one more connection than the minimum. Nodes A1-11, A1-12, A1-14, and A1-15 all have more than two connections, indicating some degree of centrality for them. This is perhaps expected since these moments in the tradition are those under the influence of

Nello Santi’s conducting. A1-18 and A1-19 are the most central nodes in this larger cluster, which is a result of their position in the tradition’s chronology and the mathematical and traditional principles attending to their status as accrued averages.

Such a distinct division into these groups is striking, heretofore unseen at any level or layout with other arias in this study, and thus invites further investigation as to its possible causes. Of course, this bifurcation could be purely coincidental. Cross- referencing Turandot’s performance history with the layout as it appears here can reveal

243

smaller clusters of performers, but ultimately cannot explain the larger scale structure of this network. Turandot’s production history at the Met includes three distinct productions, as we have seen in Chapter 5: Joseph Urban’s 1926 version, for which there are no known extant recordings, Cecil Beaton’s 1961 contribution, and Franco

Zeffirelli’s 1987 offering. As recordings of performances of only the Beaton and Zeffirelli productions are extant and accessible, and that the production switched between A1-8 and A1-9 strongly suggests that the opera’s production history influenced performance practices and their resultant tradition for Turandot’s lone aria.

Each of the clusters in Figure 6-3 has outliers—performances whose chronology does not match either their first or second order correlation. While an absolute and comprehensive explanation for them likely remains elusive, their existence highlights tradition’s ability to connect performances across any span of years. Ghena Dimitrova’s performance of “In questa Reggia” on 17 February 1996 relates most closely to Birgit

Nilsson’s performance on 21 February 1970 and the tradition as it resulted from its occurrence. This could mean that Dimitrova intentionally took Nilsson’s performance as an exemplar, or that she perhaps heard a recording of what Nilsson did on the stage 25 years earlier without deliberately choosing it as a model. Here again, definitively ascribing influence is erroneous, but the strong connection between them is not readily apparent in listening repeatedly to each. As such this instance is an example of how this mode of analysis can highlight perhaps otherwise occluded similarity, which opens an avenue of inquiry into influence and modeling in the creative process.

One might consider the outliers attached to the cluster of the Zeffirelli production

(P5, P7, P8) as perhaps unwitting forerunners of the performance traditions that

244

developed around the latter production, or perhaps as attempts to go against the conventions that had developed around this aria between 1961 and 1966 (which covers

P1 – P4). The nature of any connections could, of course be something else altogether, or pure coincidence. While a definitive conclusion about the relationships between these outliers and the two productions remains elusive, their existence highlights potentially fruitful avenues of enquiry for future consideration of the creative process.

In addition, the comparative centrality of nodes representing points in this tradition’s evolution reflects its connective nature in Turandot’s performance history.

That performances mostly fill the periphery indicates the indirect paths that exist between them at the highest degrees of correlation. Coupled with the highest degrees of correlation among the points in the tradition (see line types and correlation value ranges), the evolution of the tradition for “In questa Reggia’ is fairly static, resulting directly from the performances that create it. Performances themselves have more multifaceted origins and are more independent of each other. They often relate more strongly through tradition than directly to one another. Thus, it is not possible at present to use this method to draw conclusions about any performance or point in a tradition directly influencing another. However, it does underscore that, together, all performances result in tradition and, as such, tradition is different from convention in that it reflects a collective history rather than sets forth expectations. Furthermore, being able to follow sequentially this tradition’s evolution from A1-2 through A1-19 in an unbroken line emphasizes tradition’s centrality in this network. Centrality in this particular network—where performances interrelate far more often through tradition than

245

directly to each other—accentuates tradition’s intermedial nature in this particular performance history.

“Straniero, ascolta!”

Considering the first and second order correlations between performances and tradition highlights the two closest other performances and points in the tradition to each. Using the yFiles orthogonal layout, bundling edges, and sizing nodes according to betweenness centrality results in Figure 6-4. The orthogonal layout organizes nodes in such a way as to clarify pathways between them by limiting edge bend (curvature of the edge) and overlap. Edge bundling, which has recently “become a common clutter reduction technique in information research,” deforms and regroups similar edges, which provides “an abstract and uncluttered view of a visualization.”9 This technique is particularly useful when the number of edges overwhelms a visualization, thereby obscuring underlying patterns. Betweenness centrality is the degree to which a given node lies on the shortest path between other nodes; it thus is a means of quantifying a performance’s or tradition point’s intermediality. Cytoscape’s included network analysis package calculates betweenness centrality among other properties. Highlighting this property of each node in a performance network provides a convenient means of understanding each performance’s or tradition point’s intermedial centrality. A performance’s or tradition point “is central to the extent that it falls on the shortest path between pairs of other points.”10 In other words, any performance or point in the tradition that lies on the shortest relational path between others is central to them. This

9 Hong Zhou, Panpan Xu, Xiaoru Tuan, and Huamin Qu, “Edge Bundling in Information Visualization,” Tsinghua Science and Technology 18 (2): 145.

10 Linton C. Freeman, “A Set of Measures of Centrality Based on Betweenness,” Sociometry 40 (1), 37.

246

combination of analytical steps also reveals neighborhoods of nodes which can otherwise be difficult to discern apart from reading a correlation matrix in its entirety.

Figure 6-7. y-Files Orthogonal layout for 1st and 2nd order connections from each node to a performance and a point in the tradition’s evolution for “Straniero, ascolta!”. P4, P6, and P7 are potential reference points more often for other performances and points in the tradition than any other performance, thus making them significant in the performance history and tradition of “Straniero, ascolta!” Node size reflects Betweenness Centrality.

247

The highest degree of betweenness centrality for the performance history of

“Straniero, ascolta!” belongs to Birgit Nilsson’s performance on 03 December 1966 with

Zubin Mehta conducting. This means that each of the twelve nodes connecting to P4 also connect most directly to each other through P4, accounting for seventy-eight possible combinations of nodes whose edges (arrows) target P4. The point in the tradition’s evolution with the highest degrees of betweenness centrality is A1-18, which is intermedial for seven different nodes and their twenty-eight possible combinations. Of the nodes targeting A1-18, four are performances, which still suggests tradition’s primary intermediary nature between performances. A performance’s position as the prevailing intermediary of the development of a tradition is unexpected, given the discussion and description of tradition with which this document began, and the other examples in this chapter, each of which asserts tradition’s centrality. Such an occurrence offers fertile ground for future investigation.

Relative ease in discerning the most influential singers, conductors, and how production history affects a performance network is another advantage of considering betweenness centrality. All but two of the nodes connecting to P4 are points along the tradition’s evolution subsequent to Nilsson’s 1966 performance and only one other performance node is the target of more than two other performance edges. Elinor

Ross’s performance as Turandot on 27 April 1974 (P7) is the target node for performances by Adrienne Dugger (P16) and Andréa Gruber (P17, P18). This gives

Ross’s performance the highest degree of betweenness centrality among performances in this tradition. No point in the tradition is a target for more than two other points in the tradition, as is the case for A1-8, A1-11, A1-15, and A1-18. This delineation marks the

248

three main demographic divisions in Turandot’s performance history at the Met. A1-8 covers the tradition over the entire duration of the Cecil Beaton production. A1-11 reflects the tradition after the introduction of the Franco Zeffirelli production, but before

Nello Santi’s comparative multitude of appearances could definitively augment the tradition’s trajectory. A1-15 reflects the tradition as it stood following Santi’s decade of conducting Turandot at the Met. A1-18 serves as the marker for the tradition in the post-

Santi era rather than A1-19 as points in the tradition tend to relate most strongly to subsequent points first, and then to immediately preceding points and there is no A1-20 yet discernible.

Figure 6-5. yFiles Orthogonal layout for 1st and 2nd correlation orders of performances only for “Straniero, ascolta!”, revealing three distinct neighborhood clusters of more than two nodes. Node size reflects Betweenness Centrality. The cluster containing Nello Santi-led performances (P11, P12, P13, P14, P15) also includes P5, which Zubin Mehta conducted, perhaps indicating possible influence.

Several neighborhoods of nodes are also evident. Expectedly, some neighborhood clustering occurs within segments of the tradition’s chronological

249

trajectory (e.g. A1-2 – A1-4, A1-5 – A1-9, and A1-10 – A1-13 in particular). Less expected perhaps are the neighborhood clusters of performances, especially ones that do not follow chronological groupings as neatly as points in the tradition do. These performance neighborhoods have less to do with direct connection between each of the nodes within them than they do with their interconnectedness via betweenness centrality. To clarify the neighborhoods of performances, Figure 6-5 provides a network of only the performance nodes from Figure 6-4, also utilizing the yFiles orthogonal layout and with node size here reflecting betweenness centrality.

On the right edge of Figure 6-5, P1, P2, P3, P4, P6, P8, and P19 form a network where P6 (Birgit Nilsson and Plácido Domingo on 21 February 1970) has the highest degree of betweenness centrality in this group. In this cluster, only P8 (Ingrid Bjoner and

Franco Corelli) and P19 (Maria Guleghina and Marcello Giordani) do not feature Birgit

Nilsson in the title role. Thus, the clustering of P1, P2, P3, P4, and P6 is somewhat unsurprising. P1’s low degree of correlation (r = .5256) to P6 and absence of a second order correlation to another Nilsson performance perhaps indicates that its eccentricity in relationship to Nilsson’s other performances is the result of Leopold Stokowski’s conducting and how it affected the creative process for those onstage.11 Alternatively, this comparative eccentricity could result from the newness of the production and P1’s status as the first broadcast of Turandot in the Met’s history, or from other factors still awaiting discovery and consideration. Another consideration for future inquiry is why the most recent, measurable point (P19) in this opera’s performance history at the Met

11 P1’s second order performance correlation is to P7, with Elinor Ross and Franco Corelli in 1974, with r = .5204.

250

relates most strongly to performances in 1974 (P8, r = .6525) and 1970 (P6, r = .6082) rather than to any points chronologically closer or with more cultural esteem.12

Beyond the small, expected cluster of P9 and P10 (one week apart in 1987 with the same cast), two other neighborhoods are apparent. First, P7, P17, P17, and P18 form a network that highlights the Met’s 1974 spring tour in Boston (P7) strongest connections to performances in 2003 (P16) and 2005 (P17) respectively, and two indirect connections to this study’s 2007 performance (P18), also coming through P16 and P17. This chronological clustering between P16, P17, and P18 might be unsurprising, but the changing personnel within these performances limits the extent to which similarity results from common casting. The other prominent neighborhood contains performances P11, P12, P13, P14, and P15, all from Nello Santi’s time conducting Turandot at the Met (1988-1996), along with one from 1969 (P5). Santi’s unique approach to the musical structure and pacing of the Riddle Scene easily accounts for the clustering of performances he led. The first and second order connections of Marion Lippert’s performance of the Riddle Scene with James

McCracken in 1969 under Zubin Mehta’s baton (P5) to P13 and P14 could establish this performance as a quasi-prototype for Santi’s approach to this scene. Of course, tempo relationships are only one component of the expressive palette available to performers, so these connections could also be coincidental.

12 As with other seemingly anomalous strongest relations between performances, such information might be possible to discern in interviews with Maria Guleghina and/or Andris Nelsons, should either party consent.

251

“Nessun dorma”

Figure 6-8. Chronologically arranged yFiles hierarchical layout with only performances of “Nessun dorma” as target nodes for first, second, and third order correlations. Node size reflects Indegree (incoming connections), edges bundled to clarify connection paths, revealing the prominence of P1 and P2 (Franco Corelli) and P9 and P10 (Plácido Domingo) in the performances history and tradition of performing “Nessun dorma” at the Met.

252

Another organization considers only performance nodes as targets. This is useful because it positions performances as the source of traditions in the data pool and the resulting visualization, thus revealing which performances appear most significantly in various embodiments of the tradition as it evolves. Such an arrangement for Calaf’s Act

III aria “Nessun dorma” appears as Figure 6-6. Node sizes reflect the number of incoming connections—indegree—from smallest to largest, and edges are bundled to de-clutter connection paths. This layout makes it immediately apparent which performances are most often referents either from other performances or from points in the tradition. Another way to consider these connection paths from each of the points in the tradition is that their connecting edges are like the roots to the sources that enable their existence. Thus, the more incoming edges from a tradition point that a performance node has, the greater one can consider the overall importance of that performance to the evolution of the tradition.

The first sixteen points in the tradition’s evolution all have a connection to

Corelli’s 1961 performance, which is at once a performance and the earliest embodiment of this tradition, covering 1961-2005. In addition, all points in the tradition’s evolution from A1-10 onward connect to P9 and P10, which were the first two recorded performances in of the Franco Zeffirelli production, which featured Plácido Domingo as

Calaf and occurred one week apart. As Corelli’s (P1, P2) and Domingo’s performances

(P9, P10) have the highest degrees of connection in their respective productions, they, in many ways, are the two most prominent singers in the relationship between performance practice, production history, and the tradition of performing “Nessun dorma” at the Met.

253

Except for Corelli’s performance on 03 December 1966, the performances in the

Cecil Beaton production (which covers P1-8 and A1-2-A1-8) are target relationships for instances of the tradition nearest them in the chronology. Only one point in the tradition’s evolution connects to a subsequent performance. While this connection, A1-2 to P9, is the third order connection to a performance from A1-2 its presence links the performances practice traditions as they pertain to each production. Understandably, all three of the connections with P1 as their source connect to future performances; however, only one of these three connects to another of Corelli’s appearances in this chronology. This performance is from April 1974 (P7), while the Met was on tour in

Boston. As the second order correlation for this performance (with a value of .6912), this is only slightly less similar than P1’s connection to its first order correlation, P6), Plácido

Domingo’s performance from 21 February 1970 (with a correlation value of .7035). The third order correlation for this performance is P10, which was the video recorded performance on 04 April 1987, also with Plácido Domingo as Calaf. That two of the strongest correlations from this performance are performances by Domingo, one possibility is that he, in his only recorded appearance in the Beaton production and when he reprised the role for the Zeffirelli production, could have found influence or inspiration of the first recording of Turandot at the Met (P1).

The last performance to which any point in the tradition’s development connects is P16, which featured Richard Margison as Calaf and Marco Armiliato conducting on 01

March 2003. Being the latest first order correlation of any point in the tradition makes it the most influential performance on the tradition in its current measurable state. In addition, P16 is a target node for P9 and P10, and each of its three subsequent

254

recorded performances (P17, P18, P19), thus covering all performances of the Zeffirelli production except for the five that Nello Santi conducted. This highlights both the divergence from the convention that Santi’s conducting provided and Marco Armiliato’s and Richard Margison’s comparative return to the trajectory set by Domingo and Levine in 1987.

“Tu che di gel sei cinta”

The yFiles orthogonal layout for the first, second, and third order correlations of

Liù’s Act III aria “Tu che di gel sei cinta” appears as Figure 6-7. Node sizes in this figure reflect the stress distribution on each. “The stress of a node is the number of shortest paths passing through [it]. A node has a high stress if it is traversed by a high number of shortest paths.”13 By contrast then, if a node has a low stress, and fewer transections by shortest paths, its ties to the overall network are less strong, making that node more extraordinary. Stress is then one way to gauge a node’s individuality. In the current study, all performances of “Tu che di gel sei cinta” are highly individualistic in the context of their respective chronological happenstances and, moreover, against the backdrop of a developing tradition.

13 “Network Analyzer Online Help,” last modified 20 June 2013, http://med.bioinf.mpi- inf.mpg.de/netanalyzer/help/2.6.1/.

255

Figure 6-7. yFiles orthogonal layout of 1st, 2nd, and 3rd order correlations for “Tu che di gel sei cinta.” Node size reflects stress distribution. Edges are bundled, emphasizing coalescence around late points in the tradition. Edge thickness also reflects strength of correlation, except edges between P6/P4, P11/P12, and P13/P14, whose thickness emphasizes the rarity of strongest correlation between performances.

256

This graph also features bundled edges for path clarification and varying edge widths based on the correlation value the edge represents. Generally, thicker edges have the highest correlation values, while thinner edges have weaker values. As in all previous graphs, line type also reflects each edges’ r value. The only exceptions to this are the edges that exist between P4-P6, P13-P14, and P11-P12. These thickest edges emphasize the few first, second, or third order correlations between any performances.

The bundled line P11-P12 is thickest because it reflects two reflexive edges between these two nodes.

For Liù’s suicide aria, P2 serves as a germinal node for the development of the tradition, which mostly follows an expected development of closest correlation to its nearest predecessor and its two immediate successors, reinforcing tradition’s chronologically bounded evolution. Expectedly, and in line with Nello Santi’s presence as conductor in over more than one quarter of all performances, the tradition following these performances has a comparatively high degree of reflexivity. All points of the tradition associated with Santi’s conducting except for A1-11 relate to each other as first or second order correlations. A1-11 relates as a first order correlation to A1-12, and as second and third order relations to A1-10 and A1-9, respectively. This behavior is consistent with Leona Mitchell having sung Liù in P9, P10, and P11.

The performance network for “Tu che di gel sei cinta” is unlike the other performance networks this chapter has examined. As the bundled edges emphasize, most performances in the present network coalesce around the last four incarnations of the tradition, A1-16, A1-17, A1-18, and A1-19, which emerged between 2003 and 2009.

Broadly speaking, performances shaped the tradition in its most recent incarnations

257

rather than earlier points in the tradition’s development molding any performances, thereby emphasizing its resultant nature from performances over time. Furthermore, the centrality of late embodiments of the tradition is indicative of Puccini’s desire for performers to bring their own understanding of the drama to bear upon the music of “Tu che di gel sei cinta” within the framework of the score. The aria’s homophonic orchestral texture, doubling of the vocal line, and lack of countermelody are among the most inviting compositional choices among all musical examples in this study for singers to individualize their performances. This aligns to precepts that Luigi Ricci relayed about

Puccini’s sense of dramatic motion as it relates to tempo as mentioned above: “La gazzella e l’elefante si muovono con la deambulazione che è loro propria.”14 In other words, every Liù must die in her own way.

Conclusions

The relationships existing among performances of the same music and the traditions they create is measurable via quantitative means, most notably correlation.

Cytoscape is a valuable tool for visualizing and analyzing each network of performances and its attending traditions. Furthermore, these relationships create networks whose density and complexity expands as more performances join them. Varying layouts reveal different aspects of each network, and the analysis package available through

Cytoscape offers numerous ways of interrogating how individual nodes function within a network.

“Signore, ascolta!”, demonstrates the centrality and intermedial nature of tradition in connecting performances and highlights anomalies that are useful for framing

14 Ricci, 12. “The gazelle and the elephant move with the steps that are proper to them.”

258

questions in qualitative investigations of performance. The network for “Non piangere,

Liù” reveals how some performances can correlate most strongly to points significantly asynchronous to them. When these tradition target nodes are in the future in relationship to their source, they demonstrate how some performances can be ahead of their time. When in the past, they suggest influence and offer a clear focus for further investigation.

Analysis of the network for “In questa Reggia” reveals significant groupings of performances and their traditions with high degrees of correlations to the production history of Turandot at the Met. Using edge betweenness centrality to examine the network of performances of “Straniero, ascolta!” reveals how a single performance is the primary interlocutor for several points in its tradition. Furthermore, it emphasizes the unique influential role Nello Santi had in its performance history. Analysis of “Nessun dorma” reveals the influence the first two performances in each production had on the tradition developing in each. Finally, hierarchical analysis of the performance network for “Tu che di gel sei cinta” highlights the individuality evident over this aria’s performance history, which is in keeping with significant conceptions Puccini appears to have had about performances of his works.

While data analysis in each of the preceding instances highlights or suggests possible influence, actual influence detection is only feasible in conjunction with qualitative approaches—interviews, memoirs, correspondence, or autobiographies. It is especially useful for cases where the surface of two performances—what one hears in listening to them a few times each—differs to the point where similarity detection is convoluted at best. This approach generates results faster and at a level of specificity

259

not possible with listening alone. Furthermore, this process is repeatable and therefore confirmable or debatable according to principles of scientific research. Thus, one benefit of the current methodology is to highlight instances where the data of sonic events suggests similarity strong enough that influence might be a factor. By incorporating this kind of process and its resulting information into its standard, analog investigations, musicologists can more accurately and more efficiently seek to trace influence in creative processes.

260

CHAPTER 7 CONCLUSIONS AND FURTHER CONSIDERATIONS

Conclusions

Turandot stands as what Alfred Loewenberg described in 1940 as “so far the last world success in the history of opera.”1 While other operas have certainly garnered world-wide acclaim since then, none has as broad a performance history as Puccini’s final work. This premise frames discussions of Turandot’s musical language, and has its origins in the reception of Turandot at its premiere in 1926. Furthermore, that The Jazz

Singer followed Turandot’s premiere by only one year signifies sound film’s emergence as the most visible touchstone of the shift in multimedia entertainment away from opera.2

Turandot’s tradition continues to grow and evolve, so long as opera companies continue to perform it. Mezzo-soprano Joyce DiDonato cited a central risk and its solution for performing operas with long traditions: “As an opera singer, in a way, we’re in competition with the ghosts of the past. It’s almost impossible for opera lovers to come to the theater and not hear Maria Callas or Beverly Sills or Leontyne Price. . .The audience [has] to listen with very fresh ears.”3 Continued and frequent presentation of this opera by The Metropolitan Opera—as the United States’ preeminent opera company—constitutes this country’s most visible, and therefore analyzable, contribution to the legacy of Italian opera. This contribution depends upon all producers of the

1 Alfred Loewenberg, Annals of Opera, 1597-1940 (London: Calder, 1978), 1388.

2 Daniel Snowman, The Gilded Stage: A Social History of Opera (London: Atlantic, 2010), 286.

3 National Public Radio, “In 'Great Scott,' Joyce DiDonato Leads An Opera Within An Opera,” article posted 31 October, 2015, http://www.npr.org/2015/10/31/452909764/in-great-scott-joyce-didonato- leads-an-opera-within-an-opera (accessed 15 June 2016).

261

operatic art world working together in the sense of Heideggerian technē to reveal this work, which exists as a dynamic entity in the sense of Heideggerian poiēsis.

Each of Turandot’s three primary roles—Liù, Calaf, and Turandot—requires its respective singers to communicate a dramatic arc of grand opera proportions through some of Puccini’s most famous, exhilarating, and heart-rending music. For Marcello

Giordani, Calaf is a character who is more than the stereotypical sexual predator of a tenor; he exudes a love based on a concern for all characters in Turandot’s Peking.

Franco Corelli, one of the most known singers of Calaf, portrayed a character whose ardor and confidence in his own abilities could help the Princess Turandot transcend from abject cruelty into fiery passion. Andréa Gruber presented a Turandot who is not merciless by her own design, but rather one who is a victim of her father’s attempts to maintain power and control over her as his age and infirmity inhibit his ability to do so to the people of Peking. For her, Calaf’s arrival is a rescue, albeit one fraught with trepidation over the dual unknowns of a life outside her father’s dominion and of being able to satisfy her latent and suppressed sexuality. Birgit Nilsson’s projection, like

Corelli’s of Calaf, could emanate from a conception of Turandot’s unyieldingness and desire for domination as outgrowths of her aspirations for autonomy. Every singer who portrayed Liù, and whose recorded performance was available for analysis in this project, projected unique iterations of Puccini’s final tragic ingénue. Of the many conductors who have guided performances of Turandot at the Met, none has had a greater influence on the shape of this opera’s tradition than Nello Santi. Similarly, even the two different productions for which recordings exist—one by Cecil Beaton and one

262

by Franco Zeffirelli—have connections to how performers have realized the story of

Puccini’s most fantastical antagonist-turned-heroine.

By their very existence, each interpretation and performance in Turandot’s history at the Metropolitan Opera, and at any opera house, has contributed to this opera’s ongoing tradition. Future investigations focused on the world’s other preeminent opera houses, and using the methods that this dissertation has developed, will establish what the performance history has been in terms of sonic events, and how each new instantiation shapes these houses’ respective traditions. Furthermore, a comparison of these performances and traditions against each other will generate better understandings of how the producing companies have each presented unique versions of Turandot. Taken together, all of these performances enrich Turandot’s existence, regardless of how audiences or critics respond to them. Performances of this, or any, opera, while connected by their being technē—in working to reveal a work as poiēsis— and the traditions they create, ultimately serve as testaments to the diversity and vibrancy of the human experience.

In returning to the definition of tradition that Chapter 1 established, the ways in which digital tools and the techniques of this document promise to enhance it become clear. First, an understanding of a performance tradition must begin with the ability to understand performance itself. While performances of any work could have sufficed as the initial case study, Puccini’s Turandot occupies a unique position in the operatic world. As the first part of Chapter 2 establishes, it is the last Italian opera to have regular performances in the world’s prominent opera houses, and “opera is a terrific site for the interrogation of what tradition means and how its different discrete dimensions

263

interrelate.”4 Furthermore, its status as the only major Italian opera with a premiere after the advent of electronic recording technology significantly simplifies the path toward accurate analysis of its performances.

Chapter 2 also explains the process for using Sonic Visualiser to extract features and data pertaining to a performance and the basic method for creating and interpreting

Scape Plots: precise visual texts of the events in a performance. As opera is inherently more complex than the music of the original case study for this method (solo piano performance), and as tempo and pacing are of paramount importance in operatic performance, tempo relationships are a natural first inroad for this mode of analysis.

Furthermore, by focusing on thinner moments in the opera’s texture—arias and scenes dominated by solo singing—tracking the pacing of individual voices is easier, and consistent with what current technology enables (extracting individual voices from a thick polyphonic texture is prohibitively complicated without recording on multiple channels). This facilitates an understanding of the temporal emphases performers place on different phrases, both musical and textual.

Chapter 3 demonstrates how to adapt the data extraction and visualization for opera performance. Examining all of the musical excerpts from the first and last performances in the recorded performance history of Turandot at the Met identifies the bookends of this tradition. Time Scapes identify each performance’s unique tempo, timing, and therefore pacing structure. These analyses, as well as the other examples in

Chapter 3, adapt the technique of tempo tracking as the Centre for the History and

Analysis of Recorded Music for operatic use by incorporating textual underlays with

4 Victoria Johnson, Jane F. Fulcher, and Thomas Ertman, Opera and Society in Italy and France from Monteverdi to Bourdieu (Cambridge: Cambridge University Press, 2007), xxix.

264

Time Scapes. This adaptation facilitates understanding of temporal-textual emphasis, which is promising for better understanding of the creative process. By comparing textual-temporal emphases evident in Time Scapes against the structure of the music and text, it is possible to work backwards from the sounded event to a performer’s initial dramatic impulse. This is especially effective in combination with interviews of performers and promising for future insights into performance psychology. Ultimately, the method Chapter 3 espouses and exhibits shows the unique identity of each performance.

Chapter 4’s central premise is that the behavioral patterns of tradition, as

Chapter 1 catalogues them, are nearly identical to those of statistical means. As music performance offers a multitude of quantifiable data, calculating a progressive accruing and chronologically ordered average creates a mathematical embodiment of a tradition.

Moments in such a development lend themselves to mapping with Time Scapes, just as individual performances do. Like performances, one can group them in many ways to see how the tradition interacts with a performance history’s demography. As such, the tradition connecting each of the performances is not itself a performance; nor is it

Gadamer’s conception of a “right representation.” Rather, each tradition serves as a sort of aesthetic repository that accounts for all performances.

Each of the excerpts has a unique trajectory for its tradition. Analysis of the tempo tradition attending to “Signore, ascolta!” offers a view into the incremental and periodic nature of its (or any performance tradition’s) evolution. Furthermore, traditions tend to result from performances rather than influencing them. As such, later performances change a tradition in nuanced ways rather than obvious ones.

265

Examination of the tempo tradition for “Non piangere, Liù” by decade allows one to garner a sense of it on a slightly broader scale, but not at the level of generalizations.

This allows one to discern trends that emerge and fade as they occur over time.

Arranging the Time Scapes for the tradition of “In questa Reggia” according to the two productions of Turandot appearing in this study demonstrates how each production can form its own subset of traditional practices while continuing to contribute to the total tradition. Furthermore, considering the singers associated with significant shifts in the tradition’s evolution highlights possible inroads for further study about a tradition’s origins, intentions, and how singers perceive of them and interact with them.

Considering the trajectory of the tradition of “Straniero, ascolta!” against the conductors who led performances of Turandot reveals Nello Santi as the singularly most influential one. Additionally, various kinds of mapping of tempo data, both performed and theoretical (tradition), beyond Time Scapes can show how individual performances differ from the tradition. As the discussion of the traditions developing around “Nessun dorma” demonstrate, one can also consider a tradition’s most divergent moments and the performances that cause them. This approach can track how an individual performer’s practices change over time and open avenues of inquiry into the opera’s collective creative process and the relationships amongst performers onstage and in the pit. Because traditions are dynamic, their dynamism is predominantly incremental rather than sudden, and the documents of performance and tradition that this document creates are visual, a dynamic visual projection can offer the most useful means of illustrating its overall development. Ultimately, considering any of the changes along a tradition’s path against a variety of factors of its cultural context offers a more thorough

266

understanding of that tradition than examining either the hard data or qualitative conclusions on their own.

Fashioning texts of both performances and the points along a tradition’s evolution facilitates analysis of the relationships that exist among them. Chapter 5 demonstrates a variety of ways network analysis can be useful for illustrating and interrogating them.

The use of a prefuse force directed layout on the network of performances and the tradition of “Signore, ascolta!” highlights its overall structure and features instances of the tradition as central, intermedial components. Thus performances relate more closely through them far more often than directly to each other. Separating the performances from points along the tradition and arranging each chronologically reveals isolated instances of connections between performances and points in the tradition distant in time from them. Such occurrences are promising initial sites for investigations of influence in qualitative analyses. An edge-weighted spring embedded network confirmed Chapter 4’s hypothesis about the correlation of performance practices in “In questa Reggia” and the productions of Turandot at the Met since 1961.

Sometimes, a performance is a more involved interlocutor for tradition points and performances than is any point in a tradition. Analysis of the betweenness centrality in the network for “Straniero, ascolta!” highlighted Birgit Nilsson’s 1966 performance as the prevailing interlocutor for most of the tradition surrounding it. This is one way that a performance can be the most influential in a corpus. Setting only the performances as referents in a network can reveal which of them are most significant to each point along the tradition’s development, offering a focusing mechanism for questions about the nature of their importance. Applying this mode of analysis to “Nessun dorma” revealed

267

distinct demarcations within the overall tradition based on production history, cultural affirmation of certain performers over others, and the influence of certain conductor over others. Finally, a hierarchical network organization, such as analysis of “Tu che di gel sei cinta” used, revealed the strong cohesion that a tradition has with itself as it develops. Furthermore, the apparent lack of chronological connections from performances to their near- and co-occurring traditions highlights the individuality that singers have brought to this aria over the course of Turandot’s recorded performance history at the Met.

Tradition is perhaps ultimately unknowable in a comprehensive way. However, the analyses that this document contains and exhibits strengthen the qualitative definition of tradition at its outset. Tradition is first and foremost, derivative of the instantiations of a specific work as they occur through time (See Chapter 3).

Furthermore, tradition is not authoritarian or exclusionary; because it results from performances, it must reflect all of them, poorly and excellently executed alike (See

Chapter 4). It is also historical and relational, connecting to itself as it develops and to its germinal performances at varying degrees (See Chapters 4 and 5). Finally, tradition depends on the contexts of the performances that create it, and because these contexts vary and depend on the decisions and actions of humans, it is also enigmatic (See

Chapters 3 and 5). Above all, a distinctly useful way of considering tradition is to acknowledge that, primarily, it is a means of understanding how, even across time,

“poetically dwells man upon this earth.”5

5 Heidegger, 34.

268

Further Considerations

This project has laid a foundation for investigation of the expressive palette for both vocal and instrumental music. Future developments will build on current means of analyzing dynamics across a performance and refine or develop techniques for examinations of: pitch modulation (including vibrato, shading, and portamento), articulation, phrasing and breathing, timbre (including dark and light vowel sounds, and vocal quality color associations), and diction. Figure 7-1 below offers one model of the ontology of a performance’s expressive features. The ability to interrogate the pitch related elements of the expressive palette will require collaboration with computer scientists specializing in digital audio signal processing to streamline the extraction of individual voices from operatic textures. Another benefit of this is that it will open an avenue to examining both the whole of a thick texture and the sum of its individual parts, which will also aid in examinations of the social aspects of ensemble performance. This will, in turn, aid in the identification and study of the tools and means that individual performers use in a single instantiation or as they develop their musical interpretations and behaviors over the course of a career. Additionally, these developments will allow a multivalent mode of evaluation of a given performance’s unique identity in any number of traditions or their subsets. The methodologies for these modes of investigation will also require continual adjustment of parameters to consider individual moments and phrases in relationship to larger-scale formal divisions. Future work based on the methods and structure of this project will provide the means to fill knowledge gaps and heighten precision of current modes of analysis of the creative process.

269

In addition, the human-curated data that these modes of analysis will generate are useful as ground truth or control sets for use in machine learning. By comparing the results of musical feature extraction algorithms (how a machine identifies parts of the expressive palette) against human curated data, more accurate algorithms are possible, thereby speeding up the data extraction process. Potential partnerships with the Music

Information Retrieval Evaluation eXchange (MIREX) at the University of Illinois, the

Center for Computer Assisted Research in the Humanities (CCARH) at Stanford

University, the Oxford e-Research Centre, and the Semantic Media Network help foster further collaborations for continued knowledge generation.

270

Figure 7-9. Map of Performance Features Ontology. Line types reflect different kinds of interactions between points.

271

Figure 7-10. Revised map of the musical work ontology, emphasizing the role of performance as a realization made by artists, and of the score as an encoding of the work.

272

Finally, adopting a semantic web structure for the data of music performance and its analysis makes it possible to arrange and link the data describing various features of a musical performance into the music ontology.6 Supplementing the feature ontology enhances understanding each of its upstream segments. Understanding the features of a musical performance (event) fosters better understanding of each event

(instantiation). In turn, improved knowledge of an instantiation enriches comprehension of a musical work in a variety of historical or cultural contexts and in relationship to its encoding in notation. Similarly, enhanced awareness of a musical work shines new light on genres and the composers who wrote works in them. Figure 7-2 illustrates the musical work ontology, which can link from the point of the composer, the genre, performers, publishers to their individual ontologies and to the overall ontology of music.7 The additions present are the inclusion of the performance ontology, which links to “work” by acknowledging that each performance is a realization thereof, and to “artist” by identifying them as those responsible for making a realization. In addition, the amendment to the link between musical work and “score” is an indication that rather than being the source of the musical work, as the current ontology indicates, the score is simply an encoding of it, which performers often use to aid in realization, but not always. This adjustment reinforces Chapter 1’s application of Heidegger’s notions of technē to both a score and a performance, and of poiēsis to the work. Ultimately,

6 Yves Raimond, Samer Abdallah, and Mark Sandler, “The Music Ontology,” http://ismir2007.ismir.net/proceedings/ISMIR2007_p417_raimond.pdf. The expressive palette analysis would assist in developing the features ontology. Owing to the specificity of the data describing each feature, identification and analysis of them is latent in its development.

7 This version is my own creation, though I have modeled it on the Music Ontology Refactored: http://www.menthor.net/music-ontology.html.

273

identifying and establishing the data of the features of a musical performance facilitates deepened understandings of various cultures and the people who create and inhabit them.

274

APPENDIX A TIME SCAPES OF “SIGNORE, ASCOLTA!”1

24 February 1962 Liù: Licia Albanese; Conductor: Kurt Adler; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù non col nome sulle destino…strada sorriso! (46- Ah, pietà! (57-72) regge più! (1-10) labbra! (17-30) (31-45) 56) Coda

1 This appendix excludes Time Scapes for those performances examined in Chapter 3 (04 March 1961, 14 April 2007, and 07 November 2009).

275

16 January 1965 Liù: Lucine Amara; Conductor: Fausto Cleva; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

276

3 December 1966 Liù: Mirella Freni; Conductor: Zubin Mehta; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

277

22 March 1969 Liù: Martina Arroyo; Conductor: Zubin Mehta; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

278

21 February 1970 Liù: Lucine Amara; Conductor: Kurt Adler; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

279

27 April 1974 Liù: Edda Moser; Conductor: Gabor Ötvös; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

280

28 December 1974 Liù: Adriana Maliponte; Conductor: Alberto Erede; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

281

28 March 1987 Liù: Leona Mitchell; Conductor: James Levine; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

282

4 April 1987 Liù: Leona Mitchell; Conductor: James Levine; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

283

13 February 1988 Liù: Leona Mitchell; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

284

13 January 1990 Liù: Aprile Millo; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

285

1 February 1992 Liù: Teresa Stratas; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

286

11 February 1995 Liù: Maria Spacagna; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

287

17 February 1996 Liù: Angela Gheorghiu; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

288

1 March 2003 Liù: Norah Amsellem; Conductor: Marco Armiliato; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

289

29 January 2005 Liù: Krassimiri Stoyanova; Conductor: Bertrand de Billy; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 Signore, Si spezza… Ma se il tuo Del’esilio… Liù non regge più! ascolta!...Liù col nome destino…strada sorriso! Ah, pietà! (57-72) non regge più! sulle labbra! (31-45) (46-56) Coda (1-10) (17-30)

290

APPENDIX B TIME SCAPES OF “NON PIANGERE, LIÙ”2

24 February 1962 Calaf: Franco Corelli; Conductor: Kurt Adler; Production: Beaton

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride più Liù… fanciulla (25-30) portalo via lui le strade! Liù… (fermata) (4-24) (19-21) con te Questo, (66-83) (84-88) (38-44) questo (58-65)

2 This appendix excludes Time Scapes for those performances examined in Chapter 3 (04 March 1961, 21 February 1970, and 07 November 2009).

291

16 January 1965 Calaf: Jess Thomas; Conductor: Fausto Cleva; Production: Beaton

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

292

3 December 1966 Calaf: Franco Corelli; Conductor: Zubin Mehta; Production: Beaton

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

293

22 March 1969 Calaf: James McCracken; Conductor: Zubin Mehta; Production: Beaton

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

294

27 April 1974 Calaf: Franco Corelli; Conductor: Gabor Ötvös; Production: Beaton

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

295

28 December 1974 Calaf: Franco Corelli; Conductor: Alberto Erede; Production: Beaton

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

296

28 March 1987 Calaf: Plácido Domingo; Conductor: James Levine; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

297

4 April 1987 Calaf: Plácido Domingo; Conductor: James Levine; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

298

13 February 1988 Calaf: Nicola Martinucci; Conductor: Nello Santi; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

299

13 January 1990 Calaf: Vladimir Popov; Conductor: Nello Santi; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

300

1 February 1992 Calaf: Vladimir Popov; Conductor: Nello Santi; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

301

11 February 1995 Calaf: Lando Bartolini; Conductor: Nello Santi; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

302

17 February 1996 Calaf: Lando Bartolini; Conductor: Nello Santi; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

303

1 March 2003 Calaf: Richard Margison; Conductor: Marco Armiliato; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

304

29 January 2005 Calaf: Johan Botha; Conductor: Bertrand de Billy; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

305

14 April 2007 Calaf: Richard Margison; Conductor: Marco Armiliato; Production: Zeffirelli

Non dolce Il tuo Non lo Dell’esilio O mia che non piangere mia Signore… lasciare addolcisci povera sorride Liù… fanciulla (25-30) portalo via lui le Liù… più (4-24) (19-21) con te strade! (66-83) (fermata) (38-44) Questo, (84-88) questo (58-65)

306

APPENDIX C TIME SCAPES OF “IN QUESTA REGGIA”3

24 February 1962 Turandot: Birgit Nilsson; Conductor: Kurt Adler; Production: Beaton

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel tempo…Il Mai Nessun No, no!... …Gli enigma reggia…un grido serena… regno vinto!...dove m’avrà! rinasce…di sono tre, la traverso… Lo-u- oggi rivivi in si spense…Quel L’orror di tanta purità morte è ling (1-41) me… (41- grido e qulla morte! luccise vivo (144-156) una/una è la (1-41) 65) (67-113) nel cuor vita! (161- (114-138) 199)

3 This appendix excludes Time Scapes for those performances examined in Chapter 3 (04 March 1961, 29 January 2005, and 07 November 2009).

307

16 January 1965 Turandot: Birgit Nilsson; Conductor: Fausto Cleva; Production: Beaton

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

308

3 December 1966 Turandot: Birgit Nilsson; Conductor: Zubin Mehta; Production: Beaton

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

309

22 March 1969 Turandot: Marion Lippert; Conductor: Zubin Mehta; Production: Beaton

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

310

21 February 1970 Turandot: Birgit Nilsson; Conductor: Kurt Adler; Production: Beaton

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

311

27 April 1974 Turandot: Elinor Ross; Conductor: Gabor Ötvös; Production: Beaton

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

312

28 December 1974 Turandot: Ingrid Bjoner; Conductor: Alberto Erede; Production: Beaton

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

313

28 March 1987 Turandot: Eva Marton; Conductor: James Levine; Production: Zeffirelli

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

314

4 April 1987 Turandot: Eva Marton; Conductor: James Levine; Production: Zeffirelli

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

315

13 February 1988 Turandot: Ghena Dimitrova; Conductor: Nello Santi; Production: Zeffirelli

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

316

13 January 1990 Turandot: Gwyneth Jones; Conductor: Nello Santi; Production: Zeffirelli

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

317

1 February 1992 Turandot: Gwyneth Jones; Conductor: Nello Santi; Production: Zeffirelli

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

318

11 February 1995 Turandot: Gwyneth Jones; Conductor: Nello Santi; Production: Zeffirelli

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

319

17 February 1996 Turandot: Ghena Dimitrova; Conductor: Nello Santi; Production: Zeffirelli

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

320

1 March 2003 Turandot: Adrienne Dugger; Conductor: Marco Armiliato; Production: Zeffirelli

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

321

14 April 2007 Turandot: Andréa Gruber; Conductor: Marco Armiliato; Production: Zeffirelli

1 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 In questa ava dolce Pure nel Mai Nessun No, no!... …Gli reggia…un serena… tempo…Il regno m’avrà! rinasce…di enigma grido oggi rivivi vinto!...dove si L’orror di tanta purità sono tre, la traverso… Lo- in me… spense…Quel luccise vivo (144-156) morte è u-ling (1-41) (41-65) grido e qulla nel cuor una/una è (1-41) morte! (67-113) (114-138) la vita! (161-199)

322

APPENDIX D TIME SCAPES OF “STRANIERO, ASCOLTA!”4

24 February 1962 Turandot: Birgit Nilsson; Conductor: Kurt Adler; Production: Beaton

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

4 This appendix excludes Time Scapes for those performances examined in Chapter 3 (04 March 1961, 01 February 1992, and 07 November 2009).

323

16 January 1965 Turandot: Birgit Nilsson; Conductor: Fausto Cleva; Production: Beaton

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

324

3 December 1966 Turandot: Birgit Nilsson; Conductor: Zubin Mehta; Production: Beaton

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

325

22 March 1969 Turandot: Marion Lippert; Conductor: Zubin Mehta; Production: Beaton

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

326

21 February 1970 Turandot: Birgit Nilsson; Conductor: Kurt Adler; Production: Beaton

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

327

27 April 1974 Turandot: Elinor Ross; Conductor: Gabor Ötvös; Production: Beaton

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

328

28 December 1974 Turandot: Ingrid Bjoner; Conductor: Alberto Erede; Production: Beaton

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

329

28 March 1987 Turandot: Eva Marton; Conductor: James Levine; Production: Zeffirelli

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

330

4 April 1987 Turandot: Eva Marton; Conductor: James Levine; Production: Zeffirelli

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

331

13 February 1988 Turandot: Ghena Dimitrova; Conductor: Nello Santi; Production: Zeffirelli

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

332

13 January 1990 Turandot: Gwyneth Jones; Conductor: Nello Santi; Production: Zeffirelli

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

333

11 February 1995 Turandot: Gwyneth Jones; Conductor: Nello Santi; Production: Zeffirelli

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

334

17 February 1996 Turandot: Ghena Dimitrova; Conductor: Nello Santi; Production: Zeffirelli

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

335

1 March 2003 Turandot: Adrienne Dugger; Conductor: Marco Armiliato; Production: Zeffirelli

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

336

29 January 2005 Turandot: Andréa Gruber; Conductor: Bertrand de Billy; Production: Zeffirelli

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

337

14 April 2007 Turandot: Andréa Gruber; Conductor: Marco Armiliato; Production: Zeffirelli

1 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 577 Intro (1-4) Answer 1 Riddle 2 Interje- Answer 2 Riddle 3 (397- Answer 3 (115-143) (191- ctions (313-342) 497) (526-553) Riddle 1 (11 – 268) (269- 106) Validation 312) Validation Waiting for Calaf (154-164) (355-373) (498-525) Validation (554-577)

338

APPENDIX E TIME SCAPES OF “NESSUN DORMA”5

24 February 1962 Calaf: Franco Corelli; Conductor: Kurt Adler; Production: Beaton

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba vincerò! dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun (106-111) (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) Vincerò! (112- (13-36) 82) Tramontate 115) Vincerò! stelle! (83-106) (116-120) Coda (121-131)

5 This appendix excludes Time Scapes for those performances examined in Chapter 3 (04 March 1961, 13 February 1988, and 07 November 2009).

339

16 January 1965 Calaf: Jess Thomas; Conductor: Fausto Cleva; Production: Beaton

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

340

3 December 1966 Calaf: Franco Corelli; Conductor: Zubin Mehta; Production: Beaton

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

341

22 March 1969 Calaf: James McCracken; Conductor: Zubin Mehta; Production: Beaton

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

342

21 February 1970 Calaf: Placido Domingo; Conductor: Kurt Adler; Production: Beaton

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

343

27 April 1974 Calaf: Franco Corelli; Conductor: Gabor Ötvös; Production: Beaton

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

344

28 December 1974 Calaf: Franco Corelli; Conductor: Alberto Erede; Production: Beaton

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

345

28 March 1987 Calaf: Plácido Domingo; Conductor: James Levine; Production: Zeffirelli

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

346

4 April 1987 Calaf: Plácido Domingo; Conductor: James Levine; Production: Zeffirelli

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

347

13 January 1990 Calaf: Vladimir Popov; Conductor: Nello Santi; Production: Zeffirelli

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

348

01 February 1992 Calaf: Vladimir Popov; Conductor: Nello Santi; Production: Zeffirelli

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

349

11 February 1995 Calaf: Lando Bartolini; Conductor: Nello Santi; Production: Zeffirelli

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

350

17 February 1996 Calaf: Lando Bartolini; Conductor: Nello Santi; Production: Zeffirelli

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

351

1 March 2003 Calaf: Richard Margison; Conductor: Marco Armiliato; Production: Zeffirelli

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

352

29 January 2005 Calaf: Johan Botha; Conductor: Bertrand de Billy; Production: Zeffirelli

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

353

14 April 2007 Calaf: Richard Margison; Conductor: Marco Armiliato; Production: Zeffirelli

Nessun Tu pure, o Ma il mio Ed il mio (coro) Il nome All’alba dorma! Principessa…e mistero…splenderà! bacio..fa suo nessun vincerò! (1-12) di speranza! (37-64) mia (65- saprà…(Calaf) (106-111) (13-36) 82) Tramontate Vincerò! stelle! (83- (112-115) 106) Vincerò! (116-120) Coda (121- 131)

354

APPENDIX F TIME SCAPES OF “TU CHE DI GEL SEI CINTA”6

16 January 1965 Liù: Lucine Amara; Conductor: Fausto Cleva; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

6 This appendix excludes Time Scapes for those performances examined in Chapter 3 (04 March 1961, 24 February 1962, and 07 November 2009).

355

3 December 1966 Liù: Mirella Freni; Conductor: Zubin Mehta; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

356

22 March 1969 Liù: Martina Arroyo; Conductor: Zubin Mehta; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

357

21 February 1970 Liù: Lucine Amara; Conductor: Kurt Adler; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

358

27 April 1974 Liù: Edda Moser; Conductor: Gabor Ötvös; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

359

28 December 1974 Liù: Adriana Maliponte; Conductor: Alberto Erede; Production: Beaton

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

360

28 March 1987 Liù: Leona Mitchell; Conductor: James Levine; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

361

4 April 1987 Liù: Leona Mitchell; Conductor: James Levine; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

362

13 February 1988 Liù: Leona Mitchell; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

363

13 January 1990 Liù: Aprile Millo; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

364

01 February 1992 Liù: Teresa Stratas; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

365

11 February 1995 Liù: Maria Spacagna; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

366

17 February 1996 Liù: Angela Gheorghiu; Conductor: Nello Santi; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

367

1 March 2003 Liù: Norah Amsellem; Conductor: Marco Armiliato; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

368

29 January 2005 Liù: Krassimiri Stoyanova; Conductor: Bertrand de Billy; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

369

14 April 2007 Liù: Hei-Kyung Hong; Conductor: Marco Armiliato; Production: Zeffirelli

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 93 Tu che di gel sei l’amerai prima… perchè Egli… Prima… io chiudo… Parla! Il cinta, da tanta anche tu! gli occhi vinca ancora. Per aurora non vederlo nome! (85- fiamma vinta (1- (18-28) (31-40) non vederlo più. (61-68) più! (71-79) 92) 17) (40-57)

370

APPENDIX G “SIGNORE, ASCOLTA!” CORRELATION MATRIX

371

372

373

374

375

376

377

378

379

380

APPENDIX H “NON PIANGERE, LIÙ” CORRELATION MATRIX

381

382

383

384

385

386

387

388

389

390

APPENDIX I “IN QUESTA REGGIA” CORRELATION MATRIX

391

392

393

394

395

396

397

398

399

400

APPENDIX J “STRANIERO, ASCOLTA!” (RIDDLE SCENE) CORRELATION MATRIX

401

402

403

404

405

406

407

408

409

410

APPENDIX K “NESSUN DORMA” CORRELATION MATRIX

411

412

413

414

415

416

417

418

419

420

APPENDIX L “TU CHE DI GEL SEI CINTA” CORRELATION MATRIX

421

422

423

424

425

426

427

428

429

430

LIST OF REFERENCES

Adami, Giuseppe, editor. Letters of Giacomo Puccini. Ena Makin, trans. Philadelphia: J.B. Lippincott Company, 1931.

Adorno, Theodor W. Essays on Music. Susan H. Gillespie, trans. Berkeley, CA: University of California Press, 2002.

-----. “On Tradition.” Telos. 94 (Winter 1993-94), New York: Telos Press, 75-82.

Alonso, Miguel, Gaël Richard, and Bertrand David. 2007. "Tempo Estimation for Audio Recordings". Journal of New Music Research. 36, no. 1: 17-25.

Antonopoulos, Iasonas, Aggelos Pikrakis, and Sergios Theodoridis. 2007. "Self- Similarity Analysis Applied on Tempo Induction from Music Recordings". Journal of New Music Research. 36, no. 1: 27-38.

Ashbrook, William, and Harold Powers. Puccini's Turandot: The End of the Great Tradition. Princeton, N.J.: Princeton University Press, 1991.

Attali, Jacques. Noise: The Political Economy of Music. Minneapolis, MN: University of Minnesota Press, 1985.

Badal, James. Recording the Classics: Maestros, Music, and Technology. Kent, OH: The Kent State University Press, 1996.

Baragwanath, Nicholas. The Italian Tradition and Puccini: Compositional Theory and Practice in Nineteenth-Century Opera. Bloomington, IN: Indiana University Press, 2011.

Barnes, Jennifer. Television Opera: The Fall of Opera Commissioned for Television. Woodbridge, Suffolk, UK: Boydell Press, 2003.

Barthes, Roland. Image, Music, Text. Translated by Stephen Heath. London: Fontana Press, 1977.

Beardsley, Roger. “High Quality Recording and Reproducing of Music and Speech.” http://www.charm.rhul.ac.uk/history/p20_4_1_3.html.

-----. “Speeds and Pitching of 78rpm Gramophone Records.” http://www.charm.rhul.ac.uk/ history/p20_4_3.html.

Becker, Howard Saul. Art Worlds. Berkeley: University of California Press, 1982.

Benzecry, Claudio E. "Becoming a Fan: On the Seductions of Opera." Qualitative Sociology. 32, (2009): 131-151.

Berger, William. Puccini Without Excuses. New York: Vintage Books, 2005.

431

Bianconi, Lorenzo, and Giorgio Pestelli. Opera in Theory and Practice, Image and Myth. Chicago: University of Chicago Press, 2003.

Bianconi, Lorenzo, Giorgio Pestelli, and Lydia G. Cochrane. Opera Production and Its Resources. Chicago: University of Chicago Press, 1998.

Bianconi, Lorenzo, Giorgio Pestelli, and Kate Singleton. Opera on Stage. Chicago: University of Chicago Press, 2002.

Blaukopf, Kurt. “The strategies of the recording industries”. In Music Industries and Creativity. Strausbourg: Council of Europe, Publications Section, 1983.

Boerner, Sabine, and Johanna Jobst. 2008. "The Perception of Artistic Quality in Opera - Results from a Field Study". Journal of New Music Research. 37, no. 3: 233-245.

Bowen, José Antonio. “Tempo, Duration, and Flexibility: Techniques in the Analysis of Performance.” The Journal of Musicological Research. 16 (1996): 111-156.

Brown, Howard Mayer and Stanley Sadie, editors. Performance Practice: Music after 1600. New York: W.W. Norton & Co., 1990.

Budden, Julian. Puccini: His Life and Works. Oxford: Oxford University Press, 2002.

Cannam, Chris, Christian Landone, and Mark Sandler. Sonic Visualiser. 2010. http://sonicvisualiser.org/.

Cannam, Chris, Mark Sandler, Michael Jewell, Christophe Rhodes, and Mark d'Inverno. 2010. "Linked Data and You: Bringing Music Research Software into the Semantic Web". Journal of New Music Research. 39, no. 4: 313-325.

Centre for the History and Analysis of Recorded Music. The Mazurka Project. http://mazurka.org.uk/.

Chanan, Michael. Repeated Takes: A Short History of Recording and its Effects on Music. New York: Verso, 1995.

Christensen, Thomas, editor. The Cambridge History of Western Music Theory. Cambridge: Cambridge University Press, 2002.

Chusid, Martin. Verdi's Middle Period, 1849-1859: Source Studies, Analysis, and Performance Practice. Chicago, IL: University of Chicago Press, 1997.

Citron, Marcia. Opera on Screen. New Haven, CT: Yale University Press, 2000.

-----. When Opera Meets Film. Cambridge: Cambridge University Press, 2010.

Cone, Edward T., and Robert P. Morgan. Music, a View from Delft: Selected Essays. Chicago: University of Chicago Press, 1989.

432

Cook, Nicholas. Cook, Nicholas. Beyond the Score: Music As Performance. Oxford, Oxford University Press, 2013.

-----. The Cambridge Companion to Recorded Music. Cambridge: Cambridge University Press, 2009.

-----. Personal correspondence, Fall 2013 onward.

Cook, Nicholas, and Mark Everist. Rethinking Music. Oxford: Oxford University Press, 1999.

Cook, Nicholas, and Craig Sapp, “Purely coincidental? Joyce Hatto and Chopin's Mazurkas,” http://www.charm.rhul.ac.uk/projects/p2_3_2.html.

Cooke, Mervyn. The Cambridge Companion to Twentieth-Century Opera. Cambridge: Cambridge University Press, 2005.

Costello, Stephen. Telephone interview. 19 January 2015.

Cowgill, Rachel, and Hilary Poriss. The Arts of the Prima Donna in the Long Nineteenth Century. New York: Oxford University Press, 2012.

Cytoscape 3 User Manual. Accessed on 15 December 2015. http://wiki.cytoscape.org/ Cytoscape_3/UserManual.

Davies, Stephen. Musical Works and Performances: A Philosophical Exploration. Oxford: Clarendon Press, 2001.

Davis, Andrew. Il Trittico, Turandot, and Puccini’s Late Style. Bloomington, IN: Indiana University Press, 2010.

Derrida, Jacques. Of Grammatology. Translated by Gayatri Chakravorty Spivak. Baltimore: Johns Hopkins University Press, 1976.

Dunstan, Harry Nicholas. “Performance Practices in the Music of Giacomo Puccini as Observed by Luigi Ricci.” PhD thesis. Catholic University, 1989.

Dixon, Simon, Mark Sandler, Mark d'Inverno, and Christophe Rhodes. 2010. "Towards a Distributed Research Environment for Music Informatics and Computational Musicology".Journal of New Music Research. 39, no. 4: 291-294.

Eaglestone, Robert. Contemporary Fiction: A Very Short Introduction. Oxford: Oxford University Press, 2013.

Ellis, Daniel P. W. 2007. "Beat Tracking by Dynamic Programming". Journal of New Music Research. 36, no. 1: 51-60.

Epple, Moritz. "Between Timelessness and Historiality: On the Dynamics of the Epistemic Objects of Mathematics," Isis 102, no. 3 (September 2011): 481-493.

433

Fabian, Dorottya, Renee Timmers, and Emery Schubert. Expressiveness in Music Performance: Empirical Approaches Across Styles and Cultures. Oxford, Oxford University Press, 2014.

Fairtile, Linda B. Giacomo Puccini: A Guide to Research. New York: Garland Publishing, Inc., 1999.

Fawkes, Richard. Opera on Film. London: Gerald Duckworth and Co., 2000.

Fazekas, György, Yves Raimond, Kurt Jacobson, and Mark Sandler. 2010. "An Overview of Semantic Web Activities in the OMRAS2 Project". Journal of New Music Research. 39, no. 4: 295-311.

Flury, Roger. Giacomo Puccini: A Discography. Lanham, Md: Scarecrow Press, 2012.

Frampton, Hollis, and Bruce Jenkins. On the Camera Arts and Consecutive Matters: The Writings of Hollis Frampton. Cambridge, MA: MIT Press, 2009.

Freeman, Linton C. "A Set of Measures of Centrality Based on Betweenness.” Sociometry. 40, (1977): 35-41.

Freitas, Roger. 2002. "Towards a Verdian Ideal of Singing: Emancipation from Modern Orthodoxy". Journal of the Royal Musical Association. 127, no. 2: 226-257.

Friedberg, Anne. The Virtual Window: From Alberti to Microsoft. Cambridge, MA: MIT Press, 2006.

Gadamer, Hans Georg. Truth and method. London: Bloomsbury Academic, 2014.

Gelatt, Roland. The Fabulous Phonograph, 1877-1977. New York: Macmillan, 1977.

Giordani, Marcello. Interview. Torre del Lago Puccini, Tuscany, Italy, 11 July, 2013.

Gossett, Philip. Divas and Scholars Performing Italian Opera. Chicago: University of Chicago Press, 2006.

Gronow, Pekka, and Ilpo Saunio. An International History of the Recording Industry. London: Cassell, 1998.

Grover-Friedlander, Michal. Vocal Apparitions: The Attraction of Cinema to Opera. Princeton, NJ: Princeton University Press, 2005.

Gruber, Andréa. Interview. New York, NY, USA, 29 January 2015.

Heidegger, Martin. The Question Concerning Technology, and Other Essays. Translated by William Lovitt. New York: Harper and Row, Publishers, Inc., 1977.

434

Hein, Serge F. “Thinking and Writing with Ontological Time in Qualitative Inquiry.” Qualitative Inquiry XX(X) 1–9. Thousand Oaks, CA: Sage Publications, 2003. .

Horowitz, Joseph. Understanding Toscanini: A Social History of American Concert Life. Berkeley: University of California Press, 1994.

Hudson, Richard. Stolen Time: The History of Tempo Rubato. Oxford: Clarendon Press, 1994.

Innaurato, Albert. Telephone interview. 4/30/2013.

-----. Personal correspondence. 2013-14.

Izzo, Francesco. Laughter between Two Revolutions: Opera Buffa in Italy, 1831-1848. Rochester, NY: University of Rochester Press, 2013.

Jackson, Paul. Saturday Afternoons at the Old Met: The Metropolitan Opera Broadcasts, 1931-1950. Portland, OR: Amadeus Press, 1992.

-----. Sign-Off for the Old Met The Metropolitan Opera Broadcasts, 1950-1966. Portland, OR: Amadeus Press, 1997.

-----. Start-Up at the New Met: The Metropolitan Opera Broadcasts, 1966-1976. Portland, OR: Amadeus Press, 2006.

Johnson, Victoria, Jane F. Fulcher, and Thomas Ertman. Opera and Society in Italy and France from Monteverdi to Bourdieu. Cambridge: Cambridge University Press, 2007.

Katz, Mark. Capturing Sound: How Technology Changed Music. Berkeley, CA: University of California Press, 2004.

Kerman, Joseph. Opera as Drama. Berkeley: University of California Press, 1988.

Kivy, Peter. Authenticities: Philosophical Reflections on Musical Performance. Ithaca: Cornell University Press, 1995.

Kochman K., Coussement P., Moelants D., and Leman M. 2012. "Using Narrative Analysis to Map the Control Parameters of Enhanced Operatic Singing". Journal of New Music Research. 41, no. 1 (April 2012): 121-136.

Kosman, Joshua. “She Can Sing, but How Does She Look? Deborah Voigt’s Firing Show’s How Opera is Becoming Like Hollywood.” San Francisco Chronicle, March 15, 2004. Accessed February 10, 2013. http://www.sfgate.com/entertainment/article/VIEW-She-can-sing-but-how-s-she- look-Deborah-2781090.php.

435

Lang, Paul Henry, Alfred Mann, and George J. Buelow. Musicology and Performance. New Haven: Yale University Press, 1997.

Latham, Alison, and Roger Parker. Verdi in Performance. Oxford: Oxford University Press, 2001.

Leech-Wilkinson, Daniel. The Changing Sound of Music: Approaches to Studying Recorded Musical Performances. London: Charm, 2009. .

-----. Personal correspondence, Fall 2013 onward.

Lerch, Alexander. An Introduction to Audio Content Analysis Applications in Signal Processing and Music Informatics. Hoboken, N.J.: Wiley, 2012. .

Lewis, Kristin. Interview. Torre del Lago Puccini, Tuscany, Italy, 9 July, 2013.

Liebman E., Chor B., and Ornoy E. 2012. "A Phylogenetic Approach to Music Performance Analysis". Journal of New Music Research. 41, no. 2: 195-222.

Mallach, Alan. The Autumn of Italian Opera from Verismo to Modernism, 1890-1915. Boston: Northeastern University Press, 2007.

Marvin, Roberta Montemorra, and Hilary Poriss. Fashions and Legacies of Nineteenth- Century Italian Opera. Cambridge, UK: Cambridge University Press, 2010.

Maxfield, J. P., and H. C. Harrison. High Quality Recording and Reproducing of Music and Speech: Methods of High Quality Recording and Reproducing of Music and Speech Based on Telephone Research. New York: Bell Telephone Laboratories, Inc, 1926.

McKinney, M. F., D. Moelants, M. E. P. Davies, and A. Klapuri. 2007. "Evaluation of Audio Beat Tracking and Music Tempo Extraction Algorithms". Journal of New Music Research. 36, no. 1: 1-16.

Metropolitan Opera Archives. http://archives.metoperafamily.org/archives/frame.htm.

Mitchell, Helen F., and Raymond A.R. MacDonald. 2011. "Remembering, Recognizing and Describing Singers' Sound Identities". Journal of New Music Research. 40, no. 1: 75-80.

Moelants D., Demey M., Grachten M., Wu C.-F., and Leman M. 2012. "The Influence of an Audience on Performers: A Comparison Between Rehearsal and Concert Using Audio, Video and Movement Data". Journal of New Music Research. 41, no. 1 (April 2012): 67-78.

436

Music Ontology Refactored. Accessed 20 January 2016. http://www.menthor.net/uploads/ 4/6/7/4/46744983/158347_orig.png.

National Public Radio. “In 'Great Scott,' Joyce DiDonato Leads An Opera Within An Opera.” Article posted 31 October, 2015. http://www.npr.org/2015/10/31/452909764/in-great-scott-joyce-didonato-leads- an-opera-within-an-opera (accessed 15 June 2016).

“New Music Machine Thrills All Hearers at First Test Here.” New York Times, 7 October 1925. Accessed 31 August 2014, http://search.proquest.com/docview/103526509/ 6731A389EBD44538PQ/1?accountid=10920.

Philip, Robert. Performing Music in the Age of Recording. New Haven, CT: Yale University Press, 2004.

Poriss, Hilary. Changing the Score: Arias, Prima Donnas, and the Authority of Performance. Oxford: Oxford University Press, 2009.

Puccini, Giacomo, Giuseppe Adami, Renato Simoni, and Franco Alfano. Turandot. Milano: G. Ricordi, 1926.

Puccini, Giacomo, and Eugenio Gara. Carteggi pucciniani. Milano: Ricordi, 1958.

Puccini, Simonetta. Interview on 10 June, 2013.

Radbourne, Jennifer, Hilary Glow, and Katya Johanson. The Audience Experience: A Critical Analysis of Audiences in the Performing Arts. 2013.

Raimond, Yves, Samer Abdallah, and Mark Sandler. “The Music Ontology.” Paper presented at the 8th International Conference on Music Information Retrieval, Vienna, Austria, 23-27 September 2007. http://ismir2007.ismir.net/proceedings/ ISMIR2007_p417_raimond.pdf.

Ricci, Luigi. Puccini interprete di se stesso. Milan: Ricordi, 1954.

Ricci, Luigi, and Harry Nicholas Dunstan. Performance Practices in the Music of Giacomo Puccini As Observed by Luigi Ricci. Dissertation (Ph.D.--Musicology)-- Catholic University of America, 1989.

Rink, John. Musical Performance: A Guide to Understanding : Edited by John Rink. Cambridge: Cambridge University Press, 2003.

-----. Personal correspondence, fall 2013 onward.

-----. The Practice of Performance: Studies in Musical Interpretation. Cambridge: Cambridge University Press, 1995.

437

Rolland, Pierre-Yves. 1999. "Discovering Patterns in Musical Sequences". Journal of New Music Research. 28, no. 4: 334-350.

Rosenthal, Harold D. Two Centuries of Opera at Covent Garden. London: Putnam, 1958.

Rowe, Robert. 2005. "Real Time and Unreal Time: Expression in Distributed Performance". Journal of New Music Research. 34, no. 1: 87-95.

Royal Opera House Collections Online. http://www.rohcollections.org.uk/Default.aspx.

Sapp, Craig Stuart. Computational Methods for the Analysis of Musical Structure. PhD diss., Stanford University, 2011. .

-----. “Hybrid Numeric/Rank Similarity Metrics for Musical Performance Analysis.” Presentation at the Ninth International Conference on Music Information Retrieval, Drexel University, Philadelphia, PA, September 14-18, 2008.

Schonberg, Harold C. “Opera: ‘Turandot’ Back.” New York Times, 25 February 1961. Accessed on 14 September 2014. http://search.proquest.com/docview/ 115347153/1D2B9BB753B84EDDPQ/ 2?accountid=10920.

Schubin, Mark. “The Fandom of the Opera: How Opera Helped Create the Modern Media World.” Presentation at the Library of Congress, Washington, DC October 06, 2011.

-----. Interview with the author, October 6, 2011.

-----. Personal communication with the author, Fall 2011 – Winter 2012.

Schwartz, Arman. "Mechanism and Tradition in Puccini's Turandot". The Opera Quarterly. 25 (2010): 28-50.

Snowman, Daniel. The Gilded Stage: The Social History of Opera. London: Atlantic, 2010.

Stiegler, Bernard. Technics and Time, 1: The Fault of Epimetheus. Translated by Richard Beardsworth and George Collins. Stanford, CA: Stanford University Press, 1998.

Stravinsky, Igor. Poetics of Music in the Form of Six Lessons. Cambridge, Mass: Harvard University Press, 1970.

Taruskin, Richard. Text and Act: Essays on Music and Performance. New York: Oxford University Press, 1995.

The Metropolitan Opera: The Radio and Television Legacy, September 19 - November 22, 1986. [S.l.]: Museum of Broadcasting, 1986.

438

Thomas, Christin. “When Opera Met Film: Casa Ricordi and the Emergence of Cinema.” Paper presented at the Transnational Opera Studies Conference, Bologna, Emilia-Romagna, Italy, 30 June – 2 July 2015.

Tommasini, Anthony. “How to View 26,000 Operas at Once.” New York Times, 31 July 2005. Accessed 02 September 2014, http://www.nytimes.com/2005/07/31/arts/music /31tomm.html?_r=0.

Tuuri K., and Eerola T. 2012. "Formulating a Revised Taxonomy for Modes of Listening." Journal of New Music Research. 41, no. 2: 137-152.

Veronesi, Alberto. Interview on July 23, 2013, Torre del Lago Puccini, Tuscany, Italy.

Wagner, Richard. Wagner on Conducting. New York: Dover Publications, 1989.

Weaver, William. Seven Puccini Librettos. New York: Norton, 1981.

Westphal, Matthew. “Covent Garden’s Music Director Calls Deborah Voigt’s Notorious Little-Black-Dress Tale ‘Rubbish’ (But Evidence Suggests Otherwise).” In PlaybillArts. Accessed April 5, 2013. http://www.playbillarts.com/news/article/6271.html.

Widmer, Gerhard, and Werner Goebl. 2004. "Computational Models of Expressive Music Performance: The State of the Art". Journal of New Music Research. 33, no. 3: 203-216.

Wilson, Alexandra. The Puccini Problem: Opera, Nationalism and Modernity. Cambridge, UK: Cambridge University Press, 2007.

“Wireless Melody Jarred.” New York Times. 14 January 1910. Accessed on 6 September 2014. http://archives.metoperafamily.org/archives/frame.htm.

Wlaschin, Ken. Encyclopedia of Opera on Screen. New Haven, CT: Yale University Press, 2004. yWorks. Accessed on 15 January 2016. https://www.yworks.com/products/yfiles.

Zhou H., Xu P., Qu H., and Yuan X. 2013. "Edge Bundling in Information Visualization". Tsinghua Science and Technology. 18, no. 2: 145-156.

439

BIOGRAPHICAL SKETCH

Joshua Neumann pursued the PhD in Historical Musicology from the University of Florida in 2016, where he held teaching appointments as an Instructor in the School of Music and as in the College of the Arts as a Humanities Teaching Fellow. He also holds a graduate certificate in the Digital Humanities. Mr. Neumann earned a Master of

Music in Music History at the University of Nebraska – Lincoln, where he wrote a thesis investigating the different contexts of recording and their effects on musical behaviors in the early 20th century. Before pursuing a career in musicology, Mr. Neumann earned a

Bachelor of Music in Percussion Performance from Gordon College, in Wenham, Ma.

His diverse research interests include music performance analysis, Italian opera, Tudor church music, film music, computational musicology, and the digital humanities. He is also active as a performer, conductor, and opera director at UF and in the local community.

While at the University of Florida, he served two terms as President of the UF

Graduate Student Council, in the UF Faculty Senate, and on various committees in the

College of the Arts, earning him the UF Presidential Service award. Neumann also won a Foreign Language and Area Studies Fellowship, funded by the United States

Department of Education, and a UF Graduate School Dissertation Research Travel

Award in support of his dissertation research. Mr. Neumann has published on music and gender representations in Alfred Hitchcock’s 1956 remake of The Man Who Knew

Too Much with Oxford University Press, as well as reviews of 18th century music and conducting texts. Before matriculating to the University of Florida, he taught high school music and worked as an opera conductor and director in Philadelphia, Pa.

440