Dec 29, 2017

The "Chain of Dominants" Progression

Recently I received a comment on this post concerning the bridge to George Gershwin's "I Got Rhythm." The harmony to the bridge is a classic instance of the "circle of dominants," also called the "chain of dominants":  2 bars each of D7, G7, C7, F7. Each dominant chord resolves into the next, but a stable tonic is not established until the chain comes to rest on a major or minor chord -  in this case Bb, the first chord of the last "A" section.

Here's the comment (thanks, Mark B.!):
I've been reading up on popular music harmony recently. The chain of secondary dominants seems to start with Liszt - at least the earliest example I've seen cited is from him. Ragtime used chains of dominant chords in threes and fours regularly, and they also show up in early Tin Pan Alley tunes from the 1890s, although not in a bridge like this example. The old-time barbershop singers consider this progression their own, calling the dominant chord 'major-minor' - the added b7 being the 'minor.' And I think you'll find them in Sousa's marches as well...

I am not an authority on classical harmony, but I just couldn't believe that the "chain of dominants" started with Liszt. I had always assumed that it was a common classical device. I thought I recalled hearing it in Mozart and Bach, but couldn't cite specific instances. I did a little more internet research on this question, and talked it over with my sister, Dr. Laura Spitzer, a fine classical pianist who teaches at New Mexico State University.

About terminology: I've always called this the "circle of dominants"; I picked that up from my college teachers. "Chain of dominants" is a term that may have been coined by theoretician Allen Forte.

Forte also uses the term "chain of fifths," to describe progressions that may be partially or entirely diatonic. I call that the "diatonic circle of fifths." We are talking here about something different, a device that uses all-dominant chords.

Another term for the same device is "extended dominants." Here's the Wikipedia entry on "extended dominants."
An extended dominant is a non-diatonic secondary-dominant seventh chord that resolves downwards to another dominant chord. A series of extended dominant chords continues to resolve downwards by perfect fifths until they reach the tonic chord. Most common is the tertiary dominant, which resolves to a secondary dominant. For example, V/V/V (in C major, A(7)) resolves to V/V (D(7)), which resolves to V (G(7)), which resolves to I (note that V/V/V is the same chord as V/ii, but differs in its resolution to a major dominant rather than a minor chord). Quaternary dominants are rarer, but an example is the bridge section of the Rhythm changes which starts from V/V/V/V (in C major, E(7)). Though typically used in jazz, extended dominants have been used in other contexts as well.
Chains of three or more dominant chords are common in songs from the "Golden Age" of American standards (roughly the 1920s to 1940s), and in tunes composed by jazz artists. Here are just a few:


Sweet Georgia Brown
A Flower is a Lovesome Thing
Prelude to a Kiss
Gee Baby, Ain't I Good to You
Yesterdays
I Got Rhythm
Five Foot Two
Charleston
Jordu (the bridge consists of two "chains," six dominant chords in each)

After looking a little deeper into classical usage, it appears that secondary dominants and "tertiary" dominants are pretty common, but longer chains are less so - as Wikipedia notes, chains of 4 or more are rare. However, there are instances well before Liszt. Some examples:

Luca Marenzio (late 1500s) - This madrigal cycles through almost the complete circle of fifths, using triads, not dominant chords. There is no reason to think that this was common practice at the time - but yet, there it is.

Rameau (1760) - In a treatise on keyboard accompaniment, he suggests learning to make modulations by practicing playing chains of dominants.

Mozart - Walter Piston, in his 1941 text Harmony, cites an example from K. 283 that cycles through what Piston anlyzes as V of II, V of V, V, V of IV, IV. The piece is in G; in letter names these chords would be E7, A7, D7, G7, C. Piston did not use the terms "chain" or "circle of dominants." Rather, he just thought of this sort of sequence as a series of "secondary dominants." Piston is credited with coining the term "secondary dominant."

Mozart - An instance in K. 586 that cycles through six dominant chords, interpolating a few other chords between the dominants along the way (thanks, Laura!)

The Liszt example that Mark B. mentions may be Nocturne #3, Liebestraum - a dominant sequence begins in bar 2, on a IIIdom chord. You could play the "Charleston" or "Five Foot Two" on top of this progression.

I'm sure someone more knowledgeable than me, and/or with more persistence, could find many more examples in classical music. But for my own purposes, it looks pretty clear that although the "chain" device was known to classical composers, it is employed far more often in American jazz and popular styles.

Getting back to another part of Mark B.'s comment - It does seem to be at least plausible that barbershop singing had something to do with the use of chains of dominants in late 19th and early 20th-century popular music. A similar argument, asserting that barbershop may be one significant reason that blues evolved into a form that uses dominant chords on the I and the IV, is presented in Vic Hobson's book Creating Jazz Counterpoint: New Orleans, Barbershop Harmony, and the Blues.

Here's my review of Hobson's book. Highly recommended!

"Play That Barber Shop Chord": A Case for the African-American Origin of Barbershop Harmony is an influential 1992 paper by Lynn Abbott (click to download). It's a great read - thoroughly researched, and completely convincing. The paper stops short of detailed musical analysis, but it seems quite clear that early barbershop harmony (perhaps as early as the 1880s) had a large improvisational element, and that part of the practice was addition of flat sevenths over major triads. This, of course, could result in secondary dominants, as well as flat sevenths on the I and IV chords.

Dec 5, 2017

"Aquarela do Brasil" and "Song of India"

I'm certainly not the first person to notice similarities between Ary Barroso's "Aquarela do Brasil" and Nikolai Rimsky-Korsakov's "Song of India," but I can't resist yet another tune-detective post.

Here's the first recording of "Aquarela," by singer Francisco Alves (1939), with an arrangement by Radames Gnatelli, using big-band instrumentation with samba percussion:




Many of the accompanying riffs in this arrangement, which I'm guessing were written by Gnatelli, have become an integral part of the song as it is usually performed. Click the link above for more about Gnatelli; he was an accomplished composer, arranger, and performer, who had a distinguished career in both popular and classical genres.

Wikipedia has a nice article on "Aquarela," including Ary Barroso's story of how it came to be composed, its path to success, and some notes on the political aspects of the song:
This song, because of its exaltation of Brazil's great qualities, marked the creation of a new genre within samba, known as samba-exaltação (exaltation samba). This musical movement, with its extremely patriotic nature, was seen by many as being favorable to the dictatorship of Getúlio Vargas, which generated criticism towards Barroso...the Barroso family, however, strongly denies these claims...
Anyway, back to the similarities. Here's Rimsky-Korsakov's "Song of India," more properly called "Song of the Indian Guest," from his opera "Sadko":




1) Compare the theme at 1:00 in "Aquarela" with the theme at 2:54 in "Song of India."

2) Compare 1:22 in "Aquarela" with 3:30 in "Song of India."

The themes in 1) above are more obviously similar, but the themes in 2) show a resemblance also - a high held note on the fifth of the key, chromatic descent of a third, then the held note and chromatic line repeated twice more.

There was an earlier instance of the use of Rimsky-Korsakov's piece in popular music. In 1937 Tommy Dorsey released his big-band version of "Song of India":




In Dorsey's song, the theme I have called "2" is the featured melody; you can hear references to theme "1" in Bunny Berigan's trumpet solo. Dorsey's bridge uses yet another theme from Rimsky-Korsakov's piece (this one is not found in "Aquarela").

This recording, with "Marie" on the flip side, was a major hit for Dorsey.

As a side note, Wikipedia cites "Beautiful Ohio" (1918), the Ohio state song, as borrowing a motif from "Song of India." I hear it, but Barroso's use of the theme in "Aquarela" seems a lot more obvious. On the other hand, the reference in "Ohio" to Stephen Foster's "Beautiful Dreamer" is pretty blatant.




There have been countless versions of "Aquarela," but one of the most subtly perfect interpretations has to be Joao Gilberto's: