Charlie Parker and “Honeysuckle Rose”: Voice Leading, Formula, and Motive

Henry Martin

KEYWORDS: Schenker, Steve Larson, bebop, formula, thematic improvisation, motive, structural paraphrase, structural framework

ABSTRACT: One of the most intriguing items in the Charlie Parker discography is his first recording, a medley of “Honeysuckle Rose” and “Body and Soul” performed as an alto saxophone solo. Probably recorded in 1940, “Honeysuckle Rose” is unique: a multichorus solo from early in his career. For the first time we hear Parker confronting the problem of embedding a well-known swing standard inside the personal network of improvisational formulas necessary for multichorus fluency at a bright tempo. While previous studies have differentiated between formula and motive in Parker improvisation, this paper investigates the roots of the distinction in his first recording. The paper begins with two readings of the chorus of “Honeysuckle Rose” to introduce elements of voice leading and motive in the tune itself. The first reading, based on Schenkerian principles, is one I think Steve Larson would have agreed with. I then present a second reading with a modified Schenkerian approach and compare it with the first reading. I continue with a general discussion of formula in improvisation and its relation to motive and voice leading. The final part of the paper relates these issues to Parker’s solo recording of “Honeysuckle Rose.”

Introduction

[1] One of the most intriguing items in the Charlie Parker discography is his first recording, a medley of “Honeysuckle Rose” and “Body and Soul” performed as an alto saxophone solo. Probably recorded in 1940, “Honeysuckle Rose”⁽¹⁾ is unique: a multichorus solo from early in his career.⁽²⁾ For the first time we hear Parker confronting the problem of embedding a well-known swing standard inside the personal network of improvisational formulas necessary for multichorus fluency at a bright tempo. Clarence Davis, a Kansas City trumpet player and associate of Parker’s in the late 1930s, recorded him on amateur equipment.⁽³⁾ Parker also recorded “Honeysuckle Rose” in his first professional recording at radio station KFBI with the Jay McShann band on November 30, 1940,⁽⁴⁾ probably about ten months after the solo recording. Surprisingly, Parker never recorded the piece again professionally, nor have other live performances turned up from his later career, suggesting that the altoist dropped it from his repertory.⁽⁵⁾

[2] This paper begins with two readings of the chorus of “Honeysuckle Rose” to introduce elements of voice leading and motive in the tune itself. The first reading, based on Schenkerian principles, is one I think Steve Larson would have agreed with. I then present a second reading with a modified Schenkerian approach and compare it with the first reading. Next, I continue with a general discussion of formula in improvisation and its relation to motive and voice leading. The final part of the paper relates these issues to Parker’s solo recording of “Honeysuckle Rose.”⁽⁶⁾

“Honeysuckle Rose” Chorus: Original melody

[3] The first analysis of the “Honeysuckle Rose” chorus appears in Example 1. As I have argued in previous publications,⁽⁷⁾ ambiguous primary lines are not uncommon in the repertory of jazz tunes and popular standards; indeed, an improvised solo can often show how the improviser interprets the piece. Parker seems to hear as the primary tone of the song, which he then duplicates in his solo. Whether the primary line proceeds – or – – is trickier, however. Example 1 proposes a more orthodox – – reading of the piece.

[4] In Example 1, the A section appears on the first system and the bridge on the second system. The chorus follows conventional 32-bar AABA form with no variants in the A sections. The original melody appears on level e. Level a posits the piece as based on a – – primary line. To save space, level a compresses the overall background (i.e., that which would extend over the entire 32 bars) so that the opening F-major chord of the first A should be thought of as connected to the /V and /I at the end of the last A. Each A section then features a – – middleground descent, as shown at level b.

[5] At level b of Example 1 (second system), the bridge is conceived as generated by a neighbor motion from the /I chord from the beginning of the piece to /IV at the bridge’s beginning. The of the IV chord is then suspended to become the seventh of the C⁷ (V⁷) chord for the last four bars of the bridge. This as the seventh of the C⁷ then resolves back to A4 as /I at the beginning of the final A3 section.

[6] At level c of the A section of Example 1, we see important details added. The opening F-major chord over measure 1 proceeds to a C⁷ (V⁷) chord with the A4 suspended as a thirteenth. This A4 eventually resolves to over the V⁷ chord (over measure 4) as the structural dominant of each A section.

[7] For the bridge at level c, again on Example 1 (second system), we see the IV and V chords at level b delayed to the last two bars of each of their four-bar units, as these chords are each preceded by their secondary dominants. In the second four bars at level c, the B as the seventh of the C⁷ chord becomes an inner voice and is delayed until measure 15 with the B of the G⁷ chord continuing a rising line to C5.

[8] Complications arise in the generation of level d in the A section of Example 1. The opening F tonic triad is suppressed. The C¹³ chord, which appeared in measure 1, beat 3, at level c, is moved back conceptually and is preceded by its ii⁷ (that is, Gm⁷) supporting C5 as an appoggiatura to B 4. The B 4/Gm⁷ then proceeds to the A4/C¹³ to complete the three-note descent of the song’s principal motive (C5–B 4–A4). The thirteenth, A4, does not resolve to the fifth, G4, until the “and” of the third beat of measure 4. This structural dominant (measure 4) then resolves to /F at measure 5. On the second beat of measure 1 (level d), the melodic D4–F4 rise to the A4 is conceived of as arpeggiating the Gm⁷ chord; it resolves harmonically to B 3–E4 as part of the C¹³ chord. Measure 6 adds something of a tag by repeating the Gm⁷–C⁷ motion of the opening three bars, and in particular repeating the cadencing beats three and four of measure 4.

[9] Regarding the bridge at level d of Example 1, we see that the opening secondary dominants of each four-bar unit are developed via passing motion through their respective chords. The chords of resolution on the third bar of each phrase are prolonged by motion up to and down from the blue thirds of each chord: D 5 at measure 12 as blue third of B (measure 11) and E 5 at measure 16 as blue third of C⁷ (measure 15). The blue thirds are then harmonized with chromatic neighboring dominant-seventh chords (G ⁷ at measure 12 and A ⁷ at measure 16), which slide down to the applied dominants of each chord.

[10] Example 2 presents an alternative analysis, its two systems showing the A and B sections respectively. This analysis conceives of the chorus as generated by a – primary line, which appears at level a of the first system. Because the chorus begins on the dominant, the A4 is supported as a thirteenth of the V⁷ chord (rather than the I chord, as in Example 1). The A4 skips over the scale degree to the /I, which appears over measure 5. In keeping with the piece’s association with the swing era,⁽⁸⁾ this chord appears as an added sixth, a characteristic resolution of the C¹³ that begins the primary structure.

[11] Although the background of the chorus as a whole resolves to F4/F in the A3 section (level a), level b shows that the primary tone A4 is prolonged through the A1 and A2 sections. This A4/F (over measure 5) then connects to the B 4/IV at the beginning of the bridge, as shown by level b on the second system of Example 2.

[12] The analysis of the bridge at level b (second system of Example 2) is similar to Example 1. Level b of Example 2 shows the opening B chord as an added sixth, however, rather than a triad as in Example 1. The C⁷ chord at the end of the bridge at level b features C5 as its highest supported note rather than the B 4 as in Example 1. This C5 is the culminating note of the bridge, which arpeggiates down to the A4/C¹³ chord at the beginning of the final A section.

[13] For level c of the A section (first system of Example 2), a Gm¹¹ chord is prefixed to the C¹³. This chord contains C5 that, at the start of the A3 section, is heard as a suspension of the C5 from the end of the bridge; it then passes through B 4 to A4, the primary tone supported by V⁷. Also at level c, note how differently this analysis treats the third beat of measure 6. In Example 1, the A4 on beat 3 was an appoggiatura to the G4, which itself was an upper neighbor to F4. In Example 2, the A4 is taken again as primary, the thirteenth of the V⁷ chord, which then returns to F4 supported by the tonic. This prioritizing of the C¹³ accords with the song’s basic motive, as seen at level e, measures 1–3. At level c in the bridge (second system), the IV and V⁷ are preceded by their secondary dominants. As in the more traditional analysis of Example 1, the C5 at the end of the bridge connects to the C5 at the A3 section.

[14] At level d of Example 2, additional passing tones are added in the A section. For the bridge, each of the four main harmonies is elaborated directly via melodic motions.

[15] The first, more traditional analysis (Example 1) privileges the underlying triadic basis of tonality and the stepwise primary line of Schenkerian theory. Although one must infer a tonic triad at the beginning of the chorus to complete the Ursatz, that F-major tonality is clear and would be made even more explicit in a performance preceded by an introduction.⁽⁹⁾ The analysis assumes that swing-style harmonies (added-sixth chords, dominant-thirteenth chords, etc.) are foreground events that derive their functions from triadic tonality. Even the prolonged C¹³ chord over the first four bars (Example 1, level c) is effective because the thirteenth is substituting for the fifth of the dominant harmony, as shown by its eventual resolution as a large-scale suspension to the /V over measure 4 at level c. While a swing performer would decorate the harmony in a stylistically consistent manner (added sixths, etc.), any such embellishments do not undermine the piece’s harmonic clarity because the underlying basis of swing tonality, as viewed through traditional Schenkerian theory, enables the kinds of embellishments seen in swing style to work. Thus, Example 1, by maintaining rather than modifying Schenker’s approach, highlights and even helps clarify elements of swing style by showing their relationship to and dependency on the syntax of traditional tonality.

[16] In the second analysis (Example 2), the added-sixth and dominant-thirteenth chords appearing in various levels accord with how a jazz musician might harmonize the piece in swing style. Rather than viewing swing style as depending syntactically on triadic tonality, this analysis views its features as in themselves sufficient for understanding tonality in a swing-style context. Hence, the analysis adopts modifications to show how the piece can be understood hierarchically through typical features of swing style. In the second analysis, therefore, the – primary line invokes a common swing cadence with the structural dominant supporting the chordal thirteenth in place of the . Scale degree (in both measure 4 and measure 6 on the “and” of beat 3) appears not as a resolution of a suspension, but rather as part of an enclosure figure (G4–F4–D4) that leads to . Example 2 also infers fewer notes and harmonies in positing its structural levels because it does not require a tonic harmony to initiate the Ursatz. Rather, the C¹³ chord that is motivically, rhythmically, and tonally critical to the piece’s first four bars appears at the beginning of level a, where it is also seen as generating the piece’s primary motive. The second approach thus posits structural levels that try to reflect characteristics of the relevant jazz style and implies fewer notes, while the more traditional approach—evident in Steve Larson’s work—takes the paradigms of tonal music as interpreted through Schenkerian theory as fixed because they empower swing style in the first place.⁽¹⁰⁾

Formula and Motive in Improvisation

[17] Before turning to Parker’s improvisation on “Honeysuckle Rose,” I would like to discuss formula and motive in the analysis of jazz improvisation—issues that have been central to Steve Larson’s work throughout his career. Schenkerian analysis is able to shed light on both concepts, and indeed Steve’s publications—complementing many discussions of these issues during the years of our friendship—have been a source of continual enlightenment to me. Recent scholarship on the analysis of jazz, including Stefan Love’s article in this issue of MTO (2012), has sharpened our understanding of formula in improvisation. When formulaic analysis is allied to motivic analysis, the resulting approach can yield, via the interpenetration of both concepts, a more fully comprehensive examination of the solo in question. I try to show something of this reconciliation in my discussion of the Parker solo on “Honeysuckle Rose” that concludes this paper.

[18] In discussing jazz improvisations analytically, we often contrast improvising motivically with improvising formulaically. Motivic improvisation relates to the idea of compositional development in Western music, in which “motives”—typically conceived of as brief musical ideas exhibiting some combination of melodic, rhythmic, or harmonic characteristics—undergo a process of transformation that provides an underlying organization to the music. When listeners or analysts claim that a jazz soloist is “improvising motivically,” it usually means that the solo unfolds as an audible process of motivic elaboration assumed to be engaged in consciously by the player. In practice, the degree of intentionality is usually indeterminate, and so I’ve argued that the listener’s inferences are more germane than trying to guess what the soloist was thinking.⁽¹¹⁾ In any event, the motives being improvised on may relate to an underlying melody or they may be of the soloist’s own invention. When a solo relates to a melody, analysis is likely to cite overlapping motives in both melody and solo as germane to the listener’s experience, whether such connections were intended consciously by the player or not. Hence, in such instances I’ve often begun with a discussion of the original melody before turning to the solo in question. Steve felt similarly; see, for example, his extensive discussion of Monk’s “’Round Midnight” as preliminary to his discussing the various performances (2009, 33–50).

[19] Also significant to a solo—and to a player’s overall style—are formulas: note patterns prepared in advance by the player for improvisational fluency. They vary from three or four notes to “licks” up to several bars in length and applicable to recurring harmonic and formal situations.⁽¹²⁾ One learns to recognize a soloist’s formulas through extensive listening. Longer formulas emerge at first, but comparison of different performances will show that brief, seemingly insignificant patterns may also be important constituents of a player’s style. Soloists’ reliance on formulas varies from player to player and may evolve over time, as do the formulas themselves. We tend to think of solos that relate just to their chord changes while avoiding internal development as formulaic, while solos that relate to their underlying melodies are motivic ipso facto. How and when should formulas be distinguished from motives?

[20] Owens (1974) pioneered the study of formula in jazz improvisation in a seminal study of Parker. Through the analysis of some 250 solo transcriptions, Owens compiled a list of formulas that detailed important tendencies in Parker’s melodic style. Studies of formula in jazz improvisation later appeared in Gushee [1981] 1991, Smith 1983, and Kernfeld 1983. Gushee and Kernfeld tackle the problem of the missing prototype, for in fact “a formula” often consists of numerous instances more or less similar to one another with no one particular instance taking precedence. For example, Kernfeld shows that an enumerated list as seen in Owens may not account sufficiently for formulaic variation. His alternative model (1983, e.g., 40–41) captures the intricate possibilities of a given formula, none of which is the prototype, but is visually complex and hence difficult to apply to a given musical situation. Gushee treats formulaic complexity by invoking the idea of a “superformula” ("1991, 240), in which particular formulas are grouped into families based on widely varying criteria.

[21] Givan grapples with formula types in his study of Django Reinhardt, ultimately grouping them into four categories: variable, stable, superformulas, and context-specific (2010, 75–121).⁽¹³⁾ One of his important points, also emphasized by Gushee, is that formulas may be instrument-dependent. That is, for many instruments a formula often occurs in a single key, as its fingering may be less congenial in another key. In guitar performance, however, formulas that do not use open strings are readily transposed. Hence, Reinhardt’s stable formulas occur in a single harmonic area and may involve open strings, whereas variable formulas are not restricted by harmonic area. Givan’s superformulas differ from Gushee’s, as they comprise variable formulas linked together. Interestingly, most of Givan’s superformulas are themselves stable, occurring with little variation, thus equating to what jazz musicians call “licks.” Givan’s variable formulas are often very short, similar to lower-numbered formulas in Owens’s list, i.e., “pathways,”⁽¹⁴⁾ small-scale ways of moving around the instrument that are readily adaptable to different harmonic contexts. Givan’s context-specific formulas occur regularly at formal points in improvisations, such as the beginnings of choruses or 8-bar sections. With sensitivity to the complexities of improvisational formula, Givan combines the insights of Kernfeld and Gushee with the simplicity of Owens’s list.

[22] Larson 1987 and Martin 1996 critique the analysis of improvisation that relies heavily on formula identification.⁽¹⁵⁾ Briefly, each feels that cataloguing formulas reveals little of what makes a jazz improvisation compelling and that a more Schenkerian approach (where applicable) provides greater insight into its beauty. I also point out that formulaic analysis can be integrated into a motivic approach, i.e., that via “thematic improvisation” Parker’s improvised lines can be parsed according to the Owens list, while relating cogently to their underlying melodies (1996, 54–57).⁽¹⁶⁾

[23] Larson in several publications adopts a traditional Schenkerian approach that highlights the interactions of rhythm and meter with voice leading.⁽¹⁷⁾ Larson’s readings also invoke the idea of “confirmation,” which is a form of motivic parallelism. When a voice-leading idea in a passage appears at different structural levels and ends on the same note, Larson views the shorter (generally lower-level) appearance as “confirming” the former (2009, 24–30). Also significant to Larson’s work are his studies of how a soloist develops an overall trajectory to an improvisation in a multichorus solo.⁽¹⁸⁾

[24] If note patterns can be both formulaic and motivic, can we then account for all the notes in a jazz solo as members of one, the other, or both? Or must we admit a third category, perhaps called “running the changes” or “neutral”?⁽¹⁹⁾ We usually cite motives in relation to a melody being improvised on, but what if the solo does not relate to an external melody? What if a soloist simply improvises on a harmonic-formal scheme and has no interest in a theme, possibly even wanting to avoid suggesting any connection? Does a “no-theme” improvisation change the relationship between formula and motive? We might call recurring patterns in a no-theme solo “internal motives,” but how would they differ from formulas? What happens when the same internal motives appear in other solos by the same performer? Do they lose their motivic essence and become formulaic?⁽²⁰⁾ If so, should we restrict formula identification to relationships among solos and restrict motive identification to relationships between solos and original themes? Differentiating between motive and formula can be complex and permitting a catchall category of “neutral” may not provide clarification.

[25] Gross 2011 demonstrates cogently how formula and motive can interrelate in solos based on an underlying melody. He proceeds as Larson and I do by linking tune and solo via voice leading and motive, but brings a fresh approach to the issue.⁽²¹⁾ Analyzing a group of Bill Evans piano solos across the same tunes, i.e., a “performance family” (2011, 6), Gross shows that “structural frameworks” (2011, 117), which are recurring patterns at the phrase level that often underlie Evans’s melodic lines, can duplicate the voice leading of the original melody as “structural paraphrases” (2011, 131). That is, the voice leading of the melody at the phrase level generates a group of formulas, which may then recur within the same solo or across a performance family. Gross’s work confirms the potential overlap of formula and motive, and in particular how a note pattern may originate motivically, but take on a formulaic identity with a wide range of specific possibilities.

[26] Stefan Love in his Parker article in this issue of MTO (2012) shows another refinement of the concept of formula. Unlike Gross, who is concerned with hierarchies of voice leading, Love shows that Parker has developed a series of “schemata,” which are largely stepwise melodic ideas that thread through the chord changes. These melodic paths sometimes follow the voice leading of the chord changes, but often do not, instead proceeding slower or faster than, say, guide-tone lines and often with multiple notes within a given chord change. These schemata are more general than the traditional formulas catalogued in Owens 1974, but on occasion they may duplicate those formulas.

[27] Summing up, we note these (sometimes overlapping) possibilities:

• Brief formulas (pathways): greatly variable; applicable to numerous situations; usually not relevant motivically; seemingly generic, but possibly important to an improviser’s identity
• Licks: perhaps analyzable into briefer formulas; often fixed harmonically to specific chord progressions; possibly stable, i.e., occurring with little variation
• Superformulas: perhaps greatly variable; perhaps instrument-specific; perhaps functioning as licks
• Context-specific formulas: linked to specific formal moments
• Melodic paths as improvisational schemata that may include or overlap more conventional formulas
• Formulas generated by structural frameworks: greatly variable; possibly based on the voice-leading properties of and motivically relatable to the original melody

This last type retains a formulaic essence because of the clear similarities among the possible melodic lines, but also remains motivically related to the melody that stimulated the formula group in the first place.

[28] I hope that this overview has shown how different theorists have approached the concept of improvisational formula and that a formula may not only be stylistic but also may involve motivic and thematic considerations. Let us now turn to the interaction of voice leading, formula, and motive on Parker’s “Honey.”

Parker’s Solo on “Honey”

[29] Because the three A sections of the “Honeysuckle Rose” chorus are identical, only sixteen bars of material appear in its 32 bars. These sixteen bars comprise four 4-bar phrases with each presenting different formal and harmonic opportunities to the soloist:

• The first phrase of the A section [Gm⁷–C⁷ / Gm⁷–C⁷ / Gm⁷ / C⁷ /] lacks an initial tonic and prolongs the dominant.
• The second phrase of the A section [F / Gm⁷–C⁷ / F / F /] prolongs the tonic emphatically, balancing the missing tonic in the first four bars.
• The first phrase of the bridge [F⁷ / F⁷ / B / B /] tonicizes the subdominant via a new and chromatic harmony (F⁷).
• The second phrase of the bridge [G⁷ / G⁷ / C⁷ / C⁷ /] sequences the first four bars up a whole step. The C⁷ as V⁷ (rather than the sequential V) provides a return to the A section. The G⁷ as V⁷/V is the piece’s other non-diatonic harmony among the principal chords of the piece.

[30] In improvising, the soloist can approach each phrase formulaically, motivically, freely, or in some combination, and must link the phrases. Parker occasionally makes clear motivic reference to the melody.⁽²²⁾ He also seems to be crafting the improvisation as a single unit rather than as a series of unrelated practice choruses on “Honeysuckle” changes, so he ends the solo conclusively in the fourth chorus before proceeding to “Body and Soul.”

[31] Woideck’s transcription of “Honey” appears as Example 3 (1996, 229–33).⁽²³⁾ It begins in the midst of what is probably the first chorus, three bars from the end of the second A section.

[32] Example 4 correlates Parker’s improvisations on the first four bars of the bridge. Staff a shows the original melody, with the tonicization of IV. Staff b shows the voice leading projected by the original melody in the bridge’s first four bars—a structural framework, to use Gross’s term. The original melody targets B 4 via rising stepwise motion from F4 with one chromatic note, G .

[33] Staff c shows a voice-leading model related motivically to the bridge of the original melody. The voice-leading line in staff c descends from E 5 to B 4 with, again, one chromatic note, D . Thus, in staves b and c chromatic tetrachords span either an upper or lower perfect fourth to B 4 with a single missing note. The voice-leading model in staff c, by descending, also traces the D –C–B motion from the original song, as shown by the connecting lines drawn between the staves.

[34] The remaining staves in Example 4 show the beginnings of all the bridges in Parker’s solo, chorus by chorus. Note how the basic model given in staff c is generally followed. Choruses 1–3 lack the initial E 5; choruses 1 and 3 add a motion through C 5, completing the chromatic tetrachord. Parker’s fourth-chorus bridge, shown in staff h, not only traverses the E 5-B 4 tetrachord, but also, at the phrase’s end, the lower E 4–D4–C4–B 3, i.e., a diminution similar to a Larsonesque confirmation of the voice-leading idea.

[35] I suspect that in practicing the tune, Parker worked up lines such as the ones he plays here. Each performance of the bridge’s first four bars relates via staff c to the original melody, and is thus motivic along the lines of thematic improvisation. Staff c also resembles a structural framework, and so Parker’s realizations can all be considered structural paraphrases. If staff c had not inverted the voice leading of the original melody, thematic connection to it would be lacking and staff c would simply represent a formula freely applicable to F⁷ followed by B .

[36] Let us now consider Parker’s treatment of the A sections. Example 5 shows the first four bars of each A1 section; i.e., the first four bars of each chorus. Staff a shows the original melody and staff b its voice leading. Because the recording omits the beginning of Parker’s first chorus, staff c shows the second chorus, staff d the third chorus, and staff f the fourth chorus. As is evident in comparing them, Parker’s chorus beginnings show considerable variety.

[37] The lines between staves c and d of Example 5 mark off a notable correspondence between Parker’s second and third choruses. However, the differences are also acute: in the second chorus, at staff c, Parker reaches the primary tone A4 at measure 3; but in the third chorus, at staff d, Parker does not reach A4 until measure 4, and he does so with a different melody. Staff d echoes the melody in its contour in the first bar, bu