[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Jazz chord chaos
Jazz chord chaos
Sat, 31 Mar 2001 21:24:07 -0500
I was debating whether or not to say anything on the issue of jazz chords this
time around, but I will (once) quietly say what I
have said several times on this list on jazz chords:
Jazz chord notation is a godawful legacy due to improvised solos/comping,
transcribed lead sheets, lack of reference to original
(that is, written) sources, and musicians without formal training. Don't get
me wrong by meaning there is something wrong with
jazz voicings. I love them. Just the notation varies so extremely from school
to school (to no school) and from person to person
that to define conventions is meaningless. I even (intentionally) change
systems, even using two different sets of changes on the
same score on different lines in the system.
Think of an analogy with TeX. TeX is the best system of math notation that
anyone has formalized into a typesetting system.
Nobody ever said TeX had to _understand_ its equations. Just typeset them.
There are plenty of programs which are capable of
writing equations, simplifying expressions, coming up with proofs, etc, but
these fall totally outside the abstraction barrier of
The same should be true for a music typesetter _unless_ (and this must be
explicit) you care about MIDI performance of chord
symbols, which I don't, since it's a totally ill-posed problem for jazz chords.
It all boils down to the concept that chords are not voicings. A voicing is
easy to define: a subset of the piano keys, or (if
taking into account orchestration) an assignment of a pitch or pitches to every
instrument. A chord isn't. If you write a symbol,
does it represent the harmonic instrument is playing? Is it a representation
of two or more contrapuntal lines coming together?
Is it the vertical sonorities of a soli on the strong beats? If I write a
chord symbol on one line of a score, do I put the same
chord symbols on all parts when copied, or do I simplify/alter some?
I don't want anybody enforcing by a computer program whether or not I compose
#5 or b13, just as I don't want them telling me to
use C# or Db. Though they are mathematically identical, to a player who is
reading changes they are different _concepts_, not
simply different symbols. How they are different is left to the culture of the
If you don't believe me, look at how the New Real Book contradicts itself by
claiming to _define_ C13 as the pitches (ascending) C
E G Bb D F A, then printing "transcribed" lead sheets with recordings which
clearly don't play that. In particular, it is a very
uncommon thing to play all three 9, 11 _and_ 13. Usually 11 is avoided, or 3
is omitted to make a 7sus. The b9 between them is
generally the most dissonant interval and often a source of "mud" in bad
voicings. Go to a piano and listen to it. By contrast, 9
#11 and 13 coexist beautifully. And often. Take a more basic case: who plays
(ascending) C E G Bb for C7 if a chord tone is in
the melody? Love that root position cheezeball sound. I didn't think so.
My position: no style files. No modern-jazz.ly vs traditional-jazz.ly vs
berklee-jazz vs whatever-jazz. The concept is
ill-concieved. Please just make a basic system that allows one to
parenthesize, alter, invent a few simple chord typographies and
leave it at that. And let MIDI representation rest in peace. Kludging your
system of mapping voicings to chord symbols is a very
Here are the necessary layouts that will please everyone. Rather be formal and
incomprehensible, I will be informal by example.
Make the obvious generalizations yourself.
C C<symbol> C<symbol>(tension) C<symbol> tension
C<symbol>(tension tension tension) C<symbol>(tension) <- also allow
(tension) paren omission
--------- <- full recursion is meaningless here.
where <symbol> can be a taken from a generous selection of choices, including:
triangle, minus sign, big M small m, circle, circle
with slash, "x", plus, and to be safe an arbitrary string typeset distinct from
the root name. If I choose I want to be able to
make up a nonstandard symbol C_happy and not have any trouble.
The only thing the computer needs to understand about these symbols is how to
transpose them. This can be done just by noting
which parts of the symbols are letter pitches with sharps/flats and which are
numeric tensions. Even a regular expression search
would even suffice.
On a closing thought I will say I appreciate people's eagerness to implement
feature like jazz chords. But it nearly disgusts me
to see people "thinking with their fingers" when it comes to program design on
subtle issues. Combined with the chronic problem in
the free software community of code hardly ever being thrown away once
incorporated into a codebase, I think this is harmful.
> -----Original Message-----
> From: address@hidden
> [mailto:address@hidden Behalf Of David Raleigh
> Sent: Wednesday, March 28, 2001 2:00 PM
> To: address@hidden
> Cc: address@hidden
> Subject: Re: To Dave Arnold & others: Jazz chords
> Amelie Zapf wrote:
> > I'd pretty much like to go with the "Old Real Book" standards
> Not to include C- for Cm, I hope. They gave up on that, finally. - and
> +, again, tend to be illegible, especially when handwritten. Also if you
> use b9 and #11 it is desirable to use #5 and b5 to keep it simple, which
> is why you find that in jazz arrangements and not in stock arrangements.
> > 2) Sequence of intervals: (m) (6 or 7) 9 11 13 5 (sus4)
> The traditional thing is the farthest *unaltered* extension, followed by
> the alterations list in numerical order, (b5#11), followed by the
> catchall add9 or addBb and possibly the /bass note.
> > 3) Fifth denoted als +5 if standing alone, so that C(+5) doesn't get
> > confused with C#(5) (parentheses only for clarity).
> > 5) Notate everything if there is an alteration, C7 b9 #11 #5.
> C7(#5b9#11) takes 11 columns. The above takes 12.
> > 6) spaces between added tones.
> In the alterations list? Why?
> I think you want to avoid spaces in chord names because when one chord
> follows hard on another a single space can separate one from the next
> Eb+5 add 9 /G C7 b9 #11 #5 Ab13 b9 #11 G11 b5 b9 /D
> Eb(#5)add9/G C7(#5b9#11) Ab13(b9#11) G11(b5b9)/D
> Gnu-music-discuss mailing list