Journal Article - Billie Holiday the Art of Paraphrase

29
CYNTHIA FOLIO ROBERT W. WEISBERG (Temple University, Philadelphia) 1 Billie Holiday’s Art of Paraphrase: A Study in Consistency Introduction Billie Holiday’s style of singing has captured the imaginations of singers, instrumentalists, and listeners alike. Among the vocalists who singled her out as their ideal are Dinah Washington, Anita O’Day, Sarah Vaughan, and Carmen McRae (Gourse, 1997, p. 150). Frank Sinatra claimed that “it is Billie Holiday, whom I first heard in 52 nd Street clubs in the early 30’s, who was, and still remains, the greatest single musical influ- ence on me” (Pleasants, 1997, p. 139). The composer Ned Rorem acknowl- edged her influence: “In bending a phrase, stretching a melody, delaying the beat so as to ‘come in wrong’ just right, she forever influenced my approach to song writing...” (Rorem, 1997, p. 176). This study compares three versions of Holiday singing the standard tune, “All of Me” from 1941, 1946, and 1954. (See the list of recordings in Table 1.) This not only provides an opportunity to examine general features of her style, but allows a comparison of aspects of her style as they devel- oped through her short career. This span of thirteen years covers much of her recording career (which began in 1933 and ended with her death at age 44 in 1959). A number of jazz critics have discussed Holiday’s gift for paraphrasing the tune, for example, the following description by Giddens (1997, p. 91): 1 Authors’ addresses: C. Folio, Department of Music Theory; e-mail: cfolio@ temple.edu. R. Weisberg, Department of Psychology, Temple University, Phila. PA 19122; e-mail: [email protected]. C.F. received support from a summer research grant for initial research and transcriptions for this chapter. She wishes to thank Temple University for this support.

description

sdc

Transcript of Journal Article - Billie Holiday the Art of Paraphrase

Page 1: Journal Article - Billie Holiday the Art of Paraphrase

CYNTHIA FOLIOROBERT W. WEISBERG (Temple University, Philadelphia)1

Billie Holiday’s Art of Paraphrase: A Study in Consistency

IntroductionBillie Holiday’s style of singing has captured the imaginations

of singers, instrumentalists, and listeners alike. Among the vocalists who singled her out as their ideal are Dinah Washington, Anita O’Day, Sarah Vaughan, and Carmen McRae (Gourse, 1997, p. 150). Frank Sinatra claimed that “it is Billie Holiday, whom I fi rst heard in 52nd Street clubs in the early 30’s, who was, and still remains, the greatest single musical infl u-ence on me” (Pleasants, 1997, p. 139). The composer Ned Rorem acknowl-edged her infl uence: “In bending a phrase, stretching a melody, delaying the beat so as to ‘come in wrong’ just right, she forever infl uenced my approach to song writing...” (Rorem, 1997, p. 176).

This study compares three versions of Holiday singing the standard tune, “All of Me” from 1941, 1946, and 1954. (See the list of recordings in Table 1.) This not only provides an opportunity to examine general features of her style, but allows a comparison of aspects of her style as they devel-oped through her short career. This span of thirteen years covers much of her recording career (which began in 1933 and ended with her death at age 44 in 1959).

A number of jazz critics have discussed Holiday’s gift for paraphrasing the tune, for example, the following description by Giddens (1997, p. 91):

1 Authors’ addresses: C. Folio, Department of Music Theory; e-mail: [email protected].

R. Weisberg, Department of Psychology, Temple University, Phila. PA 19122; e-mail: [email protected]. C.F. received support from a summer research grant for initial research and transcriptions for this chapter. She wishes to thank Temple University for this support.

Page 2: Journal Article - Billie Holiday the Art of Paraphrase

250

Despite a thin voice and a range of about 15 notes, she overpowered musicians and listeners with multilayered nuances: She embellished melodies, tailoring them to her own needs and limitations; lagged behind the beat, imparting sus-pense; harmonized well beyond the ground chords of the composition, project-ing a bright authority; and infl ected words in a way that made even banal lyrics bracing... Hers was the art of refl ection.

Many critics have discussed the diff erences between Billie’s early and late recordings, claiming that her voice quality deteriorated, that she lost her vocal control, and that her level of performance became inconsistent. In an article entitled, “Lady Gets Her Due/The Complete Lady Day”, Giddins (1997) compared the early Columbia recordings (1933-42) with the Verve collection (1952-59), claiming that Billie had basically two styles: early and late, with 1942 (the year she apparently began to use hard drugs) as the turning point. He argued against the common criticism of Holiday’s late style: “Can we agree that the Billie Holiday controversy, in which her early recordings were used to trash the later ones, has been put to rest?” He claimed that the later recordings were qualitatively diff erent from the early ones, and in some ways superior (Giddens, 1997, pp. 91-3):

The early records expel a golden-age sheen of sunny rhythms and instrumental bravura; later records are built entirely around the singer. The tempo is slower, the ambience more conversational. But her alterations remain provocative and full of surprise; her enunciation is, if anything, more compelling, the emotions urgent....whereas once she transcended silly lyrics with the intensity of her rhythmic and melodic skills, now she makes them work for her. Every stanza seems autobiographical.

Had she sung “What a Little Moonlight Can Do” at the end of her life as she did at 20, she would not have been much of an artist – she would have been what Lester Young contemptuously called a “repeater pencil” (Giddens, 1997, p. 91). The present analysis initially set out to examine how much Billie’s interpretations changed over her career, and to show how much more “mature” the late versions were in relation to the early versions. However, the analysis resulted in our concluding the opposite. On the sur-face her style appeared to change drastically; changes of key, voice quality, rhythmic feel (e.g. “swing”), tempo, instrumentation, all contribute to what seem to be very diff erent versions of the tune. But after further analysis, we encountered a high degree of consistency among interpretations from diff erent times in her career, and even among versions that were in diff er-ent tempi and that featured contrasting textural settings. The apparent changes in the ways that she departed from the tune were more a matter of degree. Billie’s technique of paraphrase remained highly consistent, espe-

Cynthia Folio and Robert W. Weisberg

Page 3: Journal Article - Billie Holiday the Art of Paraphrase

251

cially in the area of timing but also in pitch, not only from chorus to chorus within a performance, but also from interpretation to interpretation, some-times over relatively long periods of time. This consistency in Holiday’s art, which may be surprising to scholars as well as to more casual listeners, is the main subject of this chapter.

This study combines a qualitative discussion of major points in Billie’s performances with quantitative analysis of “data”2 derived from those per-formances. There is some precedence for this in the music-theoretical litera-ture. For example, Huron (2001) analyzed Brahms’s Op. 51/1 using a rigorous method of comparative analysis to illustrate the distinctness of a motivic feature. As in our study, the statistics provide formal support for his informal intuition. Cognitive psychologists have applied quantitative techniques to the analysis of the creative process in the arts. Weisberg (2004) examined the preliminary sketches for Picasso’s great painting Guernica, in order to cast light on the structure of the thought processes that brought it about, and also examined the development of other creative products (Weisberg, 2003). Martindale (e.g., 1990) has examined changes in French Romantic poetry using quantitative methods (see also Simonton, 1998). In our view, quantita-tive analysis of the creative process in music provides a valuable supplement to the points raised in the more traditional music-theoretical analysis, as well as sometimes providing the basis for new discoveries.

Three Interpretations of “All of Me”: Qualitative AnalysisThe three recordings of “All of Me” off er an opportunity to com-

pare Billie’s paraphrase of the tune, not only from three diff erent times in her life spanning a total of 13 years, but also from performances at widely varying tempi. The fact that the tempi become faster with each version is probably a result of specifi c circumstances, rather than a refl ection of a general trend toward faster tempi in her style. In addition, each of the three recordings has two choruses (off ering six diff erent interpretations of the tune), and the tune is an AB (or AA’ form) in which several phrases are repeated, off ering a total of twelve diff erent interpretations of some phrases.

2 These “data” consist of details of rhythmic placement and pitch, derived from C.F.’s transcriptions of recorded performances. While these transcriptions are not as precise as, say, a computer analysis of a sound fi le, C.F. is a professional jazz performer, with extensive experience. Furthermore, the results are so strong that diff erences of a half-step or a portion of a beat would not have any signifi cant eff ect on our fi ndings.

Billie Holiday’s Art of Paraphrase

Page 4: Journal Article - Billie Holiday the Art of Paraphrase

252

Needless to say, none of the six (or twelve) interpretations stays very close to the “tune” (as one might fi nd it on a lead sheet or in a fake book3). The one that comes closest is the fi rst recording of the tune, which also happens to be the slowest. This is the only one of the three versions in which Billie actually sings the descending arpeggios of the tune that fall on the words, “All of me” in the fi rst four bars of the tune: F-C-A and E-C4-B (the B is a departure from the original A). Notably, she does so only in the fi rst chorus (see Example 1). The fi rst version is also the only one that contains the altered note, C4 (the third of the A7 in that second arpeggio), so characteristic of the tune; subsequent versions all avoid this note. This early recording also comes closest to the rhythm of the original, although even here Billie is anywhere from an eighth note to two quarters behind the beat much of the time. As the tempi get faster in the other two record-ings, she “lags behind” progressively more. In the 1954 recording, she is sometimes as much as four or fi ve beats behind the expected articulation of a word!

This penchant for singing behind the beat is a trait for which Billie was famous. Her timing was so fl awless that she could sing behind the beat without losing her place, only to pounce back on the beat when she saw fi t. Whitney Balliett’s description (1997, p. 100) of Billie’s rhythmic sense serves as an apt introduction to her interpretations of “All of Me”, especially the later versions:

Billie was a rhythm machine. No jazz instrumentalist has had a more fl exible sense of time, and it was infallible... she disconnects each song from its chump-chump-chump rhythm, and, for the two minutes or so that her vocal lasts, makes the song fl oat along somewhere behind the beat, thereby setting up an irresist-ible, swinging tug-of-war between the original tempo and her version of it.

An intriguing study by Huang and Huang (1994-95) of Billie Holiday’s tempo rubato uncovers patterns in the way that Billie lagged behind the beat. The authors call it dual-track time, or “the simultaneous presentation of two diff erent, independent frameworks regulating the passage of time”. They illustrate how she kept an independent steady beat that moved at

3 The notion that there is a single “original” version of the tune is an ephemeral concept in the world of jazz and pop music. Lead sheets and fake books off er only an approximation of the tune, especially in the rhythmic dimension. No singer would ever sing the tune exactly as it appears on the sheet music. For purposes of this study, we compared several fake book versions and picked the most common notational representation of “All of Me” as the starting point for comparison. If another version had been selected as a reference, the overall conclusions would have been the same.

Cynthia Folio and Robert W. Weisberg

Page 5: Journal Article - Billie Holiday the Art of Paraphrase

253

a diff erent tempo from the accompaniment. This separate beat, which they call her “recitation beat”, accounts for our sense of security and order in her line; according to the authors, the perception of the two tempi contributes to Billie’s “fl oating, dislocated quality”.

As an example of Billie’s impeccable timing, there is one place in all three versions in which Billie is not behind the beat: that is on the word “on” in the B section (see Example 2). This usually requires that she accel-erate on the words before “on” since she is behind at this point. This is the only word that is consistently “on” – a brilliant example of text paint-ing! She also breaks slightly between the words “on” and “dear,” which has the eff ect of emphasizing both words, and gives the word “dear” an ironic emphasis. In most versions of the tune, the words “go on dear without” are all grouped closely together rhythmically. The space between these words becomes greater with each successive recording, and even becomes humor-ous in the last version (discussed below).

Other general features of Billie’s vocal style also appear in all three ver-sions of “All of Me,” including her uncanny ability to slide into, out of, and between notes. Note that she slides into the word “All” at the beginning of all six choruses (refer to Example 1). Some of her more expressive slides include the minor seventh interval on “lose” in the 1941 version, 1st chorus, and a similarly expressive slide in the 1954 version (see Example 3). One can hear Billie literally crying on the half-step slide on the word “cry” in the 1941 version, 2nd chorus (look ahead to Example 4). In the 1946 version, she frequently fi lls the half-step from C to D2with slides, even making the D2 a quarter-tone fl at, which accentuates the bluesy feel (refer to Example 2). This expressive feature of her singing is a hallmark of her style.

Billie’s careful control of consonants at the ends of words is an interest-ing feature of her style. She often deliberately places the strong consonants like “s” and “t” on the beat. The transcriptions refl ect this in the use of an “x” as note head at the point in time that the consonants occur. Bauer (1993) discusses this feature in his analysis; his transcriptions employ a phonetic notation of the lyrics in order to show subtle infl ections in the text as they relate to the music, revealing a close relationship between emotion and style. Billie makes a special eff ort on the words “lips” and “arms” in all three versions and emphasizes the words “lose”, “best”, and “rest” in the 1946 ver-sion (refer to Examples 3 and 5 for some of these). Another trait that Bauer is able to show in his transcriptions is her way of opening up the vowels on some of the sustained notes. For example, she often sings the word “you” as “you-ah”.

One other mark of Billie’s style is the signature motive that she fre-quently uses to end tunes – the 1946 and 1954 versions both end with

Billie Holiday’s Art of Paraphrase

Page 6: Journal Article - Billie Holiday the Art of Paraphrase

254

a ̂ 2 to ^1 appoggiatura on the last word, “me”, preceded by a leap of a 4th from ^5. She uses this ^2 - ^1 ending so commonly that Giddens (1997) refers to it as “her most overworked tic”. This signature motive forms set-class (027).

The three recordings diff er in key, tempo, and range. While the fi rst recording is in F major, the other two are in E2. This is probably due to the change in her voice as she matured (or, as some say, her voice degenerated). The total range of each successive version becomes a half-step smaller and the lowest note is a half-step lower: the range of the fi rst version is A3 to C5 (or a m10); the range of the second is A23 to B24 (or a M9); and the range of the third is G3 to A24 (or a m9). As mentioned before, the tempi are quite diff erent, progressing from slowest to fastest. The band that backs her also creates a diff erent setting; for example, the second version has a ragtime fl avor since members of the band improvise behind her to create a busy texture.

The more subtle diff erences among the three recordings are manifest in the general rhythmic feel and melodic contour. Stated in a general way, the second version from 1946 has a more “bluesy” feel, because Billie sings in a swing rhythm; this is refl ected in the transcription by its notation in 12/8 meter instead of common time. For an eff ective example of how she lags just behind the beat in this version, note how she swings on “so why not take the rest” (see Example 5). The 1946 recording best illustrates Billie’s technique of “fl attening” out a melody (see Example 6). For example, the fi rst line uses only three notes – E2, F, and D – as opposed to the six notes and many leaps in the original tune. The setting of the words “All of me” is turned from an arpeggio into a series of repeated notes, and the rest of the phrase is entirely stepwise motion. The second phrase rocks back and forth on a half-step – C to D2– until the last word, “you” (Ab), and again uses only three notes as opposed to the six in the original tune. Note that the three pitches used in this phrase – A2– C – D2 – are the same three notes used in the second phrase of the B section, and they are in turn an inversion of the three notes in the third phrase of the B section – D – E2– G.

The 1954 recording is by far the most unsettled rhythmically, creating a “breathless” feeling as Billie gets further behind the beat. In addition, she replaces the arpeggio of the opening with a sophisticated development of the motive of a P4/P5, for a slightly more angular version of the tune. The 1954 recording is the fastest of the three and also best illustrates Billie’s penchant for lagging behind the time, only to recover at critical moments in the form. Most phrase endings are behind the beat (especially on the word “cry”), but become closer to the “correct” placement toward the end of the chorus. Contrast this with the beginnings of each section in which

Cynthia Folio and Robert W. Weisberg

Page 7: Journal Article - Billie Holiday the Art of Paraphrase

255

Billie “pounces” on the beat – she is consistently right on (or when she fi rst comes in, she is early).

While Billie plays with the time, the pianist in her group, Carl Drinkard, provides humorous Monk-like piano punctuations that provide a playful dia-logue. These dissonant chords appear in the transcriptions as stemmed “x’s” above the staff . Particularly clever is the second phrase in each section, both choruses (refer to Example 2). When Billie sings “I’m no good” Carl’s accent on the fourth beat coincides with Billie’s “no”, giving the word special emphasis. The word “no” is actually one beat late, making it syncopated.

When she sings “go on dear” in this same version (the parallel passage in the B section), she sings “on” a beat earlier than she sang “no” (which is actually where it would be if she had sung it straight). Therefore, Carl’s accent is a beat late, as if he expected her to be late but she wasn’t; she was “on”! The eff ect of the silence between “on” and “dear” is to give the impression that Billie has stopped in her tracks: “How can I go on?” This emphasis on the word “dear” also gives the word an ironic twist. Note that Billie and Carl do the same thing in the second chorus, suggesting that Carl knew that she would be “on” every time. As noted above, Billie is “on” on the word “on” in both choruses of all three versions; this kind of consistency allowed Carl to make accurate predictions.

Carl’s accented chords also support Billie’s interpretation at the begin-ning of each B section, on “Your goodbye.” While Billie “pounces” on the beat to begin the section, Carl, whose punctuations are usually on the second or fourth beat of the measure, joins her on the fi rst and third beats. In the measure that follows, he shifts back to the second beat and in the next bar to the fi rst beat again, coinciding with Billie on the word “left”. The placement of all of these accented chords enhances the playfulness and the meaning of the text, while it propels the time forward and creates fur-ther rhythmic excitement.

One other aspect of Billie’s mature style that this 1954 version illus-trates is the way she develops motivic ideas in the manner of a so-called “classical” composer. She takes the opening P4 idea from the tune – ^1 down to ^5 – and uses this as a springboard for further ideas. The original tune goes from ^1 to ^5 but then completes an arpeggiation of the tonic triad. Instead, Billie springs back to tonic to create a P4 in both direc-tions (see Example 7). In the second half of the same phrase, she expands this P4 into a P5, bouncing down and up on D-G-D. The parallel is clear, since both the P4 and the P5 set the words “all of me”; she has simply replaced the arpeggio with this simpler motive. In the second phrase, the two intervals are combined on “can’t you see” (D-G-C-G); then they are altered to become a tritone from G to D2on “I’m no”. This phrase is all

Billie Holiday’s Art of Paraphrase

Page 8: Journal Article - Billie Holiday the Art of Paraphrase

256

conjunct in the original version. She uses the same basic idea on the words in parallel phrases in both choruses, including “eyes that cry” (D-G-D, with the added E2-D half-step to emphasize “cry”) and “How can I” (D2-G-C). She also ends with a P5 descent on “baby” and combines the P4 and P5 into a three-note group for her three-note “signature” cadence (B2-F-E2).

One signifi cant diff erence among the three versions of “All of Me” is the narrative aspect. The fi rst version is matter-of-fact; it is the younger, some-what naïve Billie. The second version is sad and bluesy, like a woman who has recently lost love and has felt the pain of exploitation, a theme that the words of this tune clearly suggest. The last version is more ironic and reveals a cynical Billie who has been through several disastrous relation-ships: while she sings, “how can I go on dear without you, she knows very well that she can. Her setting of “eyes that cry” in this last version is not as mournful as the fi rst two. The year 1954 marks her European tour, in which she received an abundance of admiration and respect from European fans that she did not receive in the US.

There are many signifi cant similarities among the three transcribed ver-sions that go beyond the general features of her style. For example, all three recordings exhibit the same basic interpretation of the words “All of me” and “Your goodbye” in the second chorus: Billie “fl attens out” the tune – turning the opening arpeggios into stepwise lines, specifi cally ^3-^2-^1 in the key. Another common thread is Billie’s setting of the word “cry” through the evocative half-step motive (see Example 4). In the fi rst record-ing she uses a whole step (C4-B) in the fi rst chorus and a half-step in the second (B2-A), but in the other two recordings she uses a half-step in both choruses (E2-D and A2-G).

What is most notable is that in each recording the two choruses are remarkably similar to one another, as if Billie had worked out her concept of the tune each time. A comparison of parallel passages within each ver-sion (especially apparent in Example 2) reveals how similar the rhythm and contour of the tune are from the fi rst to the second chorus (with the excep-tion of the ^3-^2-^1 change at the beginning of each section, mentioned above). As further evidence that Billie “worked out” her interpretations, one can compare this early version of “All of Me” with another recording that is not included in this chapter (CRCD11102, A Fine Romance 2 March 21, 1941). The reason for not including it is that it is remarkably similar to the one discussed above; in fact, it is diffi cult to tell the two versions apart, except that Lester Young improvises a new solo. It is interesting that Lester’s “break” (after the word “them”) begins with the same initial idea in both recordings, but ends diff erently – showing that he must have also worked out some of his ideas.

Cynthia Folio and Robert W. Weisberg

Page 9: Journal Article - Billie Holiday the Art of Paraphrase

257

One other feature of Billie’s three interpretations that further demon-strates her consistency and at the same time her uniqueness is the specifi c way that she varies the second chorus. She sings diff erent words at the end of the second chorus, words not found in any lead sheet or heard by other singers. Instead of “You took the part that once was my heart; so why not take all of me?” she sings “You took the best, so why not take the rest; baby, take all of me”. She also varies the melody at this point, but she varies it in basically the same way each time, by going higher than the normal climax, up to the fi fth degree of the scale. The combination of the new words and the higher climax create more intensity the last time around.

Three Interpretations of “All of Me”: Quantitative AnalysisThese similarities from chorus to chorus and among the three

recordings suggested a possible hypothesis: despite the surface diff erences among these interpretations, might they demonstrate an underlying con-sistency? We used quantitative methods to examine that question further, concentrating fi rst on Holiday’s use of timing, and then on pitch.

Timing. Figure 1A-C presents Billie’s timing for each chorus for each recording (1941, 1946, 1954, respectively), for a total of six choruses. The positive numbers indicate the number of beats (or fraction of beats) that Billie is behind; the negative numbers indicate the number of beats she is ahead. (The reason for making “behind the beat” positive and “ahead of the beat” negative is that Billie is behind more than she is ahead.) It is obvious from Figure 1A-C that there are many consistencies in Billie’s timing across the six choruses: the peaks and valleys occur in many of the same places, and for all the years, but especially for 1954, the curves are superimposed at many points. Note that the peaks (at 2-4 beats behind and more) occur on the words “me”, “-out” (from “without”), “lose”, “cry” (the highest peak of all), “-out” (again from “without”) and “all” at the end (from “all of me”). Every verse of every version is ahead of the beat (or right on in one case) on the word “them” – the last word of the fi rst part; and there is always more time in the second chorus than in the fi rst. One reason for this might be that Billie wanted to leave time for a “break” between sections; if she ended early, one of the band members could take a short solo. Most versions are also ahead of the beat on the word “not” (or “-by” of “baby”). Note that Billie is very close to the correct placement for the beginnings of verses: “All” at the beginning and “Your” in the middle. Also, the word “me” at the very end is close to its “correct” place in every version. There are only two

Billie Holiday’s Art of Paraphrase

Page 10: Journal Article - Billie Holiday the Art of Paraphrase

258

places where she is right on the expected beat in each and every chorus of all three versions: on the words “on” and “heart” (in chorus 1) or the cor-responding word “rest” (in chorus 2).

We carried out two overall analyses on Billie’s timing. We fi rst examined the consistency of her rhythmic pattern from the fi rst chorus to the second within each performance, as well as her consistency from one performance to the next. Second, we also examined whether her rhythmic “freedom” changed over the two choruses within each performance, and whether she became more free rhythmically over the three performances of “All of Me”.

We examined the consistency in the patterns shown in Figure 1A-C by calculating coeffi cients of correlation, examining the relationship between the two choruses within each year and between pairs of choruses across years. Coeffi cients of correlation (in this case, Pearson r ) provide a sum-mary of the relationship between two variables. A value of 1.0 indicates a close relationship between them; in this case, the timing patterns for any two choruses would be identical. That is, the peaks and valleys would be in exactly the same places and of the same magnitude in the two choruses. A coeffi cient of 0 indicates no relationship; that is, the timing patterns for any two choruses would look totally diff erent. A coeffi cient of -1 would indi-cate a complete reversal in the two patterns: when one was at a peak, the other would be at its lowest. The total of six choruses over the three per-formances allows us to calculate 15 correlations, taking all pairs of choruses. These correlations are shown in Table 2, and each is numbered for ease of exposition.

A comparison of choruses 1 and 2 for each year (#1, 10, 15) indicates a high degree of consistency. All three within-performance correlations were positive and highly signifi cant statistically (p < .01 in each case), mean-ing that it is very unlikely that the patterns shown in Table 2 occurred by chance. Note also that, the later the version, the more consistency there is in the timing from chorus to chorus; that is, the correlation coef-fi cients between chorus 1 and chorus 2 for a given year increase over years (#1<#10<#15, and that correlation is especially high for 1954 (#15: .907). This is consistent with observations in the above music-studies analysis, in which the pianist appeared to predict what Billie was going to do. One can also see visual evidence for this in the high degree of overlap in the two curves in Figure 1C, mentioned earlier.

Not only is there consistency in Billie’s timing within a performance, but there is also a high degree of consistency from performance to performance, which is a more signifi cant fi nding. Returning to the correlations in Table 2, there are highly signifi cant correlations among choruses from diff erent years. Furthermore, the versions that are closer together in time (1941 and

Cynthia Folio and Robert W. Weisberg

Page 11: Journal Article - Billie Holiday the Art of Paraphrase

259

1946) are not more highly correlated than are the other versions (compare #2 and 3 to 4 and 5; and 2, 3, 6, and 7 to 11, 12, 13, 14). This indicates that Billie’s timing pattern was consistent throughout these three recordings. In conclusion, these statistics demonstrate a high degree of consistency in timing from chorus to chorus and from performance to performance. This consist-ency is seen in spite of the time lapse between performances (14 years), the changes in her accompanying musicians, and the wide disparities of tempo and arrangements.

The correlational analysis just presented demonstrated that Billie’s over-all rhythmic pattern was highly consistent from one chorus to the next within each performance, and also from one performance to the next. This means, once again, that the peaks and valleys in each chorus tended to occur in the same places. However, it is also possible that, even with this high degree of consistency, Billie might have shown more freedom in one perform-ance or another. That is, she might have been farther away from the beat in one chorus or in one performance – although the peaks and valleys might be in the same places, they might have been more extreme. This possibil-ity, which we could call the question of “rhythmic freedom,” was investigated by analyzing changes in the absolute deviation of Billie’s timing from the song as written in most fake books and as sung by most artists. Absolute deviation is simply a measure of how far away she was from the rhythm at any point, ignoring whether she was ahead or behind. These results are summarized in Table 3. Examining fi rst the degree of absolute deviation as a function of the chorus in each performance, for none of the separate years was the diff erence in absolute deviations between chorus 1 and chorus 2 signifi cant by t- test (all t values < 1.0). Thus, the absolute deviation from timing was similar for both choruses for each year. The second analy-sis of absolute deviations from timing examined the changes across the three performances, averaged across choruses for each year, as shown in the last column of Table 3. Billie’s absolute deviations from the rhythm of the song increased signifi cantly over the three versions, by analysis of variance (F(2,80) = 25.36 p < .001), and the average overall absolute value for each year is signifi cantly diff erent from each of the other years by t-test, p < .007 in each case. Thus it is highly unlikely that the diff erences in absolute devia-tions from one year to another occurred by chance. These results indicate that Billie’s rhythm became more free over the three performances.

In explaining these diff erences, one possibility is that, over the years, as Billie became more experienced in her style, she became more free in her timing. This would explain the increasing absolute deviations from timing over the three versions of “All of Me” seen in Table 3. However, there is an alternative interpretation of those data: the tempo of the performance

Billie Holiday’s Art of Paraphrase

Page 12: Journal Article - Billie Holiday the Art of Paraphrase

260

increases over the three versions, and it is possible that that increase in tempo caused her to lag further behind the beat. (It should be noted that fewer than 10% of the deviations are ahead of the beat.) The present data do not allow us to diff erentiate between those two explanations for the pat-tern of deviations shown in Table 3. In order to make such a diff erentia-tion, we require songs with multiple performances in diff erent tempi, but in which the tempi do not change systematically over the years.

Pitch. Quantitative analysis was also carried out on the pitches created by Billie over the six choruses of the three performances. Figure 2A-C presents Billie’s deviations from the melody as written in most fake books (in number of half-steps), for the 1941, 1946, and 1954 recordings, respec-tively, for each syllable in “All of Me”. Three points are seen in the fi gure. First, as we saw with timing, there is consistency within each year in Bil-lie’s melodies from the fi rst chorus to the second: one sees a similar pattern within each pair of choruses, although the patterns are by no means identi-cal. Second, within each year, there is more melodic “freedom” in the second chorus than the fi rst: in Figure 2A-C, when the points for the two choruses do not coincide for a given syllable, the value for the second chorus is usu-ally farther away from the X-axis. This pattern was diff erent than that with timing, where there were no signifi cant diff erences from the fi rst to second chorus within a year. Third, comparing Figure 2A-C, one also sees similar patterns across years, indicating that Billie’s interpretation of the melody did not change drastically over the years. Statistical analyses supported those conclusions.

As we did with Billie’s timing, we fi rst examined the consistency of the pattern of melodic deviations shown in Figure 2A-C. We computed coef-fi cients of correlation comparing the patterns of deviations from the melody over all pairs of choruses from the three performances. These correlations are shown in Table 4. Looking fi rst within each year (#1, 10, 15), all three correlations are positive and highly signifi cant, meaning that her pitches were consistent from one chorus to another for all three performances. In addition, examining the other correlations, we see that they are also all positive and highly signifi cant, indicating, as with timing, that Billie’s melodic interpretation of “All of Me” was similar from performance to per-formance, even when 13 years intervened (#4, 5, 8, 9).

We then examined changes in Billie’s melodic freedom, as we did with rhythm, by calculating her absolute deviations from melody, as shown in Table 5. We fi rst compared the absolute melodic deviations for chorus 1 and chorus 2 for each performance, and in each case, the diff erence was signifi -cant by t-test (all t values > 3.5, p < .001 in each case). This indicates that, contrary to timing, Billie fi rst sang the melody relatively conservatively

Cynthia Folio and Robert W. Weisberg

Page 13: Journal Article - Billie Holiday the Art of Paraphrase

261

(for her), and then went farther from the tune the second time around. We also examined the degree of mean absolute deviation in pitch from year to year, as shown in the last column of Table 5. As can be seen, there was an increase from 1941 to 1946, but a decrease from 1946 to 1954. These results were analyzed statistically using analysis of variance, which indicated that the overall diff erences among performances were not signifi cant (F (2, 198) = .94, p > .05). Thus, in contrast to the consistent increase in rhythmic freedom across the three performances seen earlier, there was not a consist-ent increase in Billie’s melodic freedom over the years.

Relationship Between Timing and Pitch Deviations. The fi nal statistical analysis examined the relationship for each chorus between the amount of timing deviation and melodic deviation across the words in the song. We calculated coeffi cients of correlation between the timing deviations and the melodic deviations for each chorus. If these correlations were close to +1.0, it would mean that the words that were sung by Billie farthest away from the expected place in the timing were also those that were farthest away from the expected melody. This could be taken to indicate that she coordi-nated her sense of timing and sense of melody. Correlations of 0 would indi-cate that there was no consistent relationship between timing and melody; the timing and pitch of a given word would be independent of each other. Such a fi nding would indicate that timing and pitch were under independ-ent control as she sang. Finally, if those correlations were close to –1.0, it would mean that the words most deviant rhythmically were the least devi-ant melodically, and vice versa. This would indicate that when she was deviating along one dimension, she might compensate by not deviating in the other.

These correlations are shown in Table 6, and there is only one signifi -cant correlation (for 1941, chorus 1), and it was marginal. In carrying out multiple statistical tests, it becomes likely that a signifi cant fi nding will occur by chance, so one must be cautious in putting too much emphasis on that single barely signifi cant correlation. The others are not signifi cant, and those for 1954 are very close to zero. One pattern that is seen in the cor-relations in Table 6 is a decrease in absolute value of the correlations over the years. If we assume that by 1954 the correlations are essentially zero, it indicates that Billie was in control of two systems by then, and could vary them independently. This conclusion, although it must be tempered, is interesting because it could not have been derived from a qualitative analysis of Billie’s performances; it thus demonstrates the potential useful-ness of applying quantitative methods to the arts.

Billie Holiday’s Art of Paraphrase

Page 14: Journal Article - Billie Holiday the Art of Paraphrase

262

Conclusion

As illustrated earlier, the three versions of “All of Me” show a maturation in her style at the same time as they reveal signifi cant con-sistencies. While all three are remarkable interpretations, the general trend is toward a less literal version of the actual pitches of the tune and a more innovative approach to timing. While her timing became more “deviant” from early to middle to late, it also demonstrated a certain consistency in exactly where these deviations occurred. The latest versions of both tunes also reveal a sophisticated development of motivic ideas, as Billie literally recomposes the tune and creates her own integrated structure. While the tune studied in this article represents only a small portion of Billie Holi-day’s total output, our combination of quantitative and qualitative analysis identifi es specifi c traits that can be studied in her other recordings and opens the door to understanding her very personal art of paraphrase.

BibliographyBalliett, W.

1997 Lady Day, in: L. Gourse (Ed.), The Billie Holiday Companion: Seven Decades of Commentary, Schirmer, XIX-XXII.

Bauer, W.1993 Billie Holiday and Betty Carter: Emotion and Style in the Jazz Vocal

Line, “Annual Review of Jazz Studies” 6: 99-149.Giddens, G.

1997 Lady Gets Her Due/The Complete Lady, in: L. Gourse (Ed.), The Billie Holiday Companion: Seven Decades of Commentary, Schirmer, 89-98.

Gourse, L.1997 There Was No Middle Ground with Billie Holiday, in: L. Gourse (Ed.), The

Billie Holiday Companion: Seven Decades of Commentary, Schirmer, 139-52.

Hodeir, A.1962 Eight Measures of Billie Holiday, in: Toward Jazz, trans. By Noel

Burch, Grove Press, Inc., 191-95.Huang, H. & Huang R. V.

1994-5 Billie Holiday and Tempo Rubato: Understanding Rhythmic Expressivity, “Annual Review of Jazz Studies” 7: 181-99.

Huron, D.2001 What is a Musical Feature? Forte’s Analysis of Brahms’s Opus 51,

No. 1, Revisited, “Music Theory Online” 7/4, <http://www.society-musictheory.org/mto/>.

Martindale, C.1990 The clockwork muse. The predictability of artistic change, New York: Basic.

Cynthia Folio and Robert W. Weisberg

Page 15: Journal Article - Billie Holiday the Art of Paraphrase

263

Pleasants, H.1997 The Great American Popular Singers. The Billie Holiday Companion:

Seven Decades of Commentary, Schirmer, 131-38.Rorem, N.

1997 Knowing When to Stop, in: L. Gourse (Ed.), The Billie Holiday Companion: Seven Decades of Commentary, Schirmer, 176-80.

Simonton, D. D.1999 Origins of genius. Darwinian perspectives on creativity, New York:

Oxford.Weisberg, R. W.

2003 Case studies of innovation: Ordinary thinking, extraordinary out-comes, in: L. Shavinina (Ed.) International Handbook of Innovation, Hillsadale, N.J.: Earlbaum.

Weisberg, R. W.2004 On Structure in the Creative Process: A Quantitative Case-Study of the

Creation of Picasso’s “Guernica”, “Empirical Studies in the Arts”, 22, 23-54.

Williams, M.1973 Liner notes to The Smithsonian Collection of Classic Jazz, Columbia

P6 11891, Smithsonian Institution: 25.

Table 1. Sources of the fi ve transcriptions:Three interpretations of “All of Me”

#1 January 21, 1941 Smithsonian Collection of Classic Jazz (LP’s); from Columbia K9 32124 Eddie Heywood and His Orchestra: Shad Collins, trumpet; Leslie Johnakins,

Eddie Barefi eld, alto sax; Lester Young, tenor sax; Eddie Heywood, piano; John Collins, guitar; Ted Sturgis, bass; Kenny Clarke, drums

Tempo: Quarter = 112-116#2 April 22, 1946 Best of Verve Years: Jazz at the Philharmonic; Verve 849 434-2; rec. at Carn-

egie Hall Joe Guy & probably Buck Clayton, trumpets; Tommy Stark, trombone; Willie

Smith, alto sax; probably Lester Young & Coleman Hawkins, tenor sax; Milt Raskin, piano; probably Irving Ashby, guitar; Red Callender, bass; Dave Cole-man, drums

Tempo: Dotted quarter = 150-154#3 January 2, 1954 Billie’s Blues: Jazz Heritage, Inc. 312947A; recorded live in Köln, Germany Carl Drinkard, piano; Red Mitchell, bass; Elaine Leighton, drums Tempo: Quarter = 190-194

Billie Holiday’s Art of Paraphrase

Page 16: Journal Article - Billie Holiday the Art of Paraphrase

264

Table 2. Correlations Among Billie Holiday’s Timing Deviations Across the Three Recordings (Six Choruses) of “All of Me”

Chorus 1 Chorus 2 Chorus 1 Chorus 2 Chorus 1 Chorus 2 1941 1941 1946 1946 1954 1954Chorus 1 X 1) .583* 2) .548* 3) .413* 4) .491* 5) .523*

1941Chorus 2 X 6) .642* 7) .659* 8) .546* 9) .655*

1941Chorus 1 X 10) .656* 11) .770* 12) .759*

1946Chorus 2 X 13) .647* 14) .760*

1946Chorus 1 X 15) .907*

1954Chorus 2 X1954*Correlation is signifi cant at the .01 level (2-tailed).

Table 3. Absolute deviations in timing (in beats) across the three perform-ances (six choruses) of “All of Me”

Performance Chorus 1 Chorus 2 Mean 1941 .68 (.74)* .64 (.76) .66 (.66) 1946 1.40 (1.01) 1.27 (.93) 1.34 (1.38) 1954 1.63 (1.26) 1.71 (1.16) 1.67 (1.28)*Standard deviation in parenthesis.

Cynthia Folio and Robert W. Weisberg

Page 17: Journal Article - Billie Holiday the Art of Paraphrase

265

Table 4. Correlations Among Billie Holiday’s Pitch Deviations Across the Three Performances (Six Choruses) of “All of Me”

Chorus 1 Chorus 2 Chorus 1 Chorus 2 Chorus 1 Chorus 2 1941 1941 1946 1946 1954 1954Chorus 1 X 1) .728* 2) .793* 3) .632* 4) .672* 5) .627*

1941Chorus 2 X 6) .871* 7) .844* 8) .715* 9) .750*

1941Chorus 1 X 10) .784* 11) .813* 12) .782*

1946Chorus 2 X 13) .594* 14) .750*

1946Chorus 1 X 15) .782*

1954 Chorus 2 X1954*Correlation is signifi cant at the .01 level (2-tailed).

Table 5. Absolute deviations in pitch (in half-steps) across the three per-formances (six choruses) of “All of Me”

Performance Chorus 1 Chorus 2 Mean 1941 2.35 (3.80)* 4.59 (4.08) 3.47 1946 3.72 (3.81) 4.88 (3.97) 4.30 1954 3.62 (3.82) 4.59 (3.77) 4.11

*Standard deviation in parenthesis.

Table 6. Correlations between timing deviations and melodic deviations within each chorus for the three performances of “All of Me”

Year and Chorus Correlation 1941 Chorus 1 -.305*

1941 Chorus 2 -.168 1946 Chorus 1 +.121 1946 Chorus 2 -.085 1954 Chorus 1 +.026 1954 Chorus 2 -.085 * p < .05 (2-tail)

Billie Holiday’s Art of Paraphrase

Page 18: Journal Article - Billie Holiday the Art of Paraphrase

266Fig

ure

1A. “

All

of M

e” –

Dev

iation

s fr

om R

hyth

m –

194

1Cynthia Folio and Robert W. Weisberg

Page 19: Journal Article - Billie Holiday the Art of Paraphrase

267

Fig

ure

1B. “

All

of M

e” –

Dev

iation

s fr

om R

hyth

m –

194

6

Billie Holiday’s Art of Paraphrase

Page 20: Journal Article - Billie Holiday the Art of Paraphrase

268Fig

ure

1C. “

All

of M

e” –

Dev

iation

s fr

om R

hyth

m –

195

4Cynthia Folio and Robert W. Weisberg

Page 21: Journal Article - Billie Holiday the Art of Paraphrase

269

Fig

ure

2A. “

All

of M

e” –

Dev

iation

s fr

om P

itch

– 1

941

Billie Holiday’s Art of Paraphrase

Wor

ds

Page 22: Journal Article - Billie Holiday the Art of Paraphrase

270Fig

ure

2B. “

All

of M

e” –

Dev

iation

s fr

om P

itch

– 1

946

Cynthia Folio and Robert W. Weisberg

Page 23: Journal Article - Billie Holiday the Art of Paraphrase

271

Fig

ure

2C. “

All

of M

e” –

Dev

iation

s fr

om P

itch

– 1

954

Billie Holiday’s Art of Paraphrase

Page 24: Journal Article - Billie Holiday the Art of Paraphrase

272

Example 1: “All of Me” – First four measures of all six choruses

1941 version / A section / fi rst chorus

1941 version / A section / second chorus

1946 version / A section / fi rst chorus

1946 version / A section / second chorus

1954 version / A section / fi rst chorus

1954 version / A section / second chorus

Cynthia Folio and Robert W. Weisberg

Page 25: Journal Article - Billie Holiday the Art of Paraphrase

273

Example 2: “no good” and “on dear”

1941 version / A section / fi rst chorus / second phrase

1941 version / B section / fi rst chorus / second phrase

1941 version / A section / second chorus / second phrase

1941 version / B section / second chorus / second phrase

1946 version / B section / second chorus / second phrase

1946 version / B section / fi rst chorus / second phrase

1946 version / A section / second chorus / second phrase

1946 version / A section / fi rst chorus / second phrase

Billie Holiday’s Art of Paraphrase

Page 26: Journal Article - Billie Holiday the Art of Paraphrase

274

1954 version / A section / fi rst chorus / second phrase

1954 version / B section / fi rst chorus / second phrase

1954 version / A section / second chorus / second phrase

1954 version / B section / second chorus / second phrase

Example 3: 1954 version of “All of Me” / A section / fi rst chorus / third phrase

Cynthia Folio and Robert W. Weisberg

Page 27: Journal Article - Billie Holiday the Art of Paraphrase

275

1941 version / fi rst chorus

Example 4: “All of Me” / B section / fi rst phrase: settings of “eyes that cry”

1941 version / second chorus

1946 version / fi rst chorus

1946 version / second chorus

1954 version / fi rst chorus

1954 version / second chorus

Example 5: 1946 version of “All of Me” / B section / second chorus / third phrase

Billie Holiday’s Art of Paraphrase

Page 28: Journal Article - Billie Holiday the Art of Paraphrase

276

Example 6: 1946 version of “All of Me” / fi rst chorus (examples of “fl attening out” the melody)

A section / fi rst phrase

A section / second phrase

B section / second phrase

B section / third phrase

Cynthia Folio and Robert W. Weisberg

Page 29: Journal Article - Billie Holiday the Art of Paraphrase

277

Example 7: 1954 version of “All of Me” (P4 / P5 motive)

First chorus / A section / fi rst phrase

First chorus / A section / beginning of second phrase

First chorus / B section / end of fi rst phrase

First chorus / B section / beginning of second phrase

Second chorus / B section / last phrase

Billie Holiday’s Art of Paraphrase