Minor Keys, Roman numerals

[trevordeclercq]

Dearest Hooktheory team and users,

The Hooktheory web site does not appear to have a consistent approach with regard to how to handle Roman numerals in minor keys. This is very frustrating, especially if one hopes to explain harmony in a consistent way. I see three main approaches used throughout the site. I list these three approaches below, using the example of the chord progression Am, G, F, and E in the key of A minor.

1. Encode in the relative major, i.e., vi, V, IV, III.
2. Encode using traditional Roman numerals, i.e., i, VII, VI, V.
3. Encode in the parallel major, i.e., i, bVII, bVI, V.

I have searched for guidelines in the help materials and this forum as to which approach is preferred, but I have not been able to find clear guidance. That being said, this web site shows tacet acceptance of the second method, since this is the way that the chords for minor keys are presented as an encoding default. Nonetheless, many users specifically go against this approach, putting the song in the parallel major to get “flat” chords in a minor key.

If the site developers do indeed prefer use of the second approach, I would suggest they strongly reconsider in favor of the third approach (encoding in the key of the parallel major). Although the second approach is commonly used in the analysis of Western art music from the common-practice period (c. 1600-1900) and is therefore taught in most music theory undergraduate classrooms, it has become deprecated within the professional academic music theory community for the analysis of pop and rock music due to the great amount of mixture in this style and general lack of parallel-key chord progression similarlity. In other words, the default approach to minor used on this site is somewhat antiquated, applicable more to the music of Mozart and Beethoven than the music of Led Zeppelin and Miley Cyrus.

The first approach (encoding in the relative major) is viable for ambiguous cases in which it is not clear whether the material is in a minor key or the relative major. But for songs (like Adele’s “Rolling in the Deep”) that are clearly grounded in a minor tonic and avoid the relative major altogether, the third approach has become a standard method.

Overall, I really love this site, and I use it every semester in my teaching. It really promises to revolutionize the study of music theory, especially for pop and rock styles. But if the developers and users of this site are truly attempting to move music theory pedagogy into the future, we should at least move the harmonic nomenclature out of the past.

Best,
Trevor de Clercq
PhD in music theory, Eastman School of Music
Assistant Professor of Recording Industry, Middle Tennessee State University

[Ryan]

@trevordeclercq,

Thank you for your thoughtful post. You are correct that there are multiple approaches being used, and that we need to define some guidelines.

When we first launched Hooktheory, we only supported chords built on the major scale. At the time, we were teaching a course at UC Berkeley on the music theory of popular music, and found that this was sufficient for our purposes.

Many songs, as you mention, are ambiguous as to whether they the key is properly major or minor (or if it really matters at all). Simply because a song starts on a minor chord does not mean that it is necessarily in a minor key. Historically, music theory surrounding the minor mode centered around devices that are less common in popular music. For example, students of music theory are often taught to look for a raised scale degree corresponding to the leading tone of the harmonic minor scale to indicate the presence of a minor mode, whereas in pop music, the natural minor chords VII and v are more common than their harmonic minor counterparts vii^o and V. Therefore, in our class, we gravitated toward approach #1. This approach was effective, because our students then (as well as many of Hooktheory users now) knew very little music theory. Once they became familiar with how the basic chords (I, ii, iii, IV, V, vi) function, it was easy for them to write songs without having to ponder things like, “Is this song technically in the X mode”.

However, as the site evolved, it became clear that we needed to support chords borrowed from other modes in order to analyze a great many songs using a mixture of modes The cadence: IV → iv → I, for instance, is a very popular progression. This led to some users analyzing songs with all chords borrowed from a specific parallel mode (approach #3). This was troubling to us for a number of reasons. First, the analyses looked very sloppy. As you know, we use solid colors to represent scale degrees that are diatonic, and hashed colors to indicate scale degrees that are non-diatonic. Borrowing every chord from the minor mode meant that any chord built on ♭3, ♭6, or ♭7, appeared to be non-diatonic. Melodies containing these scale degrees also became technically non-diatonic. People were confused, because relations like vi → IV which were previously well-understood, now looked like i → ♭VI which didn’t make any sense. Ultimately, we decided that this approach compromised the simplicity of the old system.

In order to simultaneously accommodate songs that appear to tonicize a chord other than the major tonic, while still preserving the simplicity of the original system, we introduced approach #2. By allowing users to change the mode of the analyses, we let them decide which chord “feels” like the tonic, while also having the ability to choose the scale that has the most natural fit. To preserve the relationships that users are familiar with, we change the colors of the chords and scale degrees so that red always reflects the tonic of the relative major mode. We also allow users to view any theoryTab in a relative mode by using the mode selector.

So in the end, I would say that we would prefer that users analyze songs using approach #1 or #2. I personally prefer approach #1, but concede that chord progressions that are convincingly in a different mode, like vi → ii → V/vi → vi are probably more appropriately analyzed as : i → iv → V → i

One possible solution would be to have user settings that allow analyses to be viewed by default in the relative major mode.

[trevordeclercq]

Ryan,

Thanks for the quick reply. I am definitely familiar with the history of the HookTheory site, as I have been loving it for a long time. I definitely remember being frustrated with only being able to encode songs in major keys, and I was elated when you introduced the option for minor.

I have no disagreement that some song sections, which could be charted in a minor key, fair better in the relative major because the song as a whole is clearly grounded in the relative major. The opening to Avril Lavigne’s “When You’re Gone” is a good example. Although it begins on what sounds like a toggle between a bVI chord and the minor tonic, the verse quickly reverts to a clear major key, after which the song is most simply understood as chords related to the major tonic.

When I talk about approach #1, it is not in reference to these sorts of common situations in rock and pop. I think these are not even all that ambiguous. Instead, I am concerned about the cases where a song is solidly in a minor tonality.

Perhaps it is true that using bVI or bVII for chords in a minor key ends up being sloppy in your system, but that is only a function of your encoding system. If bVI in a minor key is considered diatonic, then you do not need the hashed colors. Why can’t the hashed colors be contextual based on the tonality of the tonic?

You say that you prefer that users analyze using approach #1 or #2. I’m not really concerned about approach #1, because it is easy enough to shift the analysis between relative keys based on the drop-down menu. But why have a preference for #2 given the case of a clear minor tonality? Both the academic music world (as evidenced in scholarly publications) and the professional music world (as evidenced in Nashville number charts) prefer approach #3 when charting commercial songs. I think the preferences of those worlds would be more important to represent on the site than the personal preferences of the developers.

It seems like it would be easy enough to have the default chord collections for a diatonic minor tonality be i, iio, bIII, iv, v, bVI, and bVII, with those being solid colors (i.e., diatonic) to the minor key. (Of course, I say that as someone who is not doing the programming work!)

If I may add a final piece of evidence for approach #3: Consider the do-based movable-Do solfege system used across the country in college-level music departments. In this system, the diatonic seventh scale degree in minor is “Te”. Similarly, b7 in major is “Te.” In other words, G natural is sang as “Te” whether tonic is A major or A minor. We want to draw the students’ ears to the fact that G natural has the same quality, irrespective of the modality of the tonic chord. Similarly, we want students to hear the similarity between a G major chord in A major and A minor. That similarity is obscured if one is called bVII and the other VII.

In the end, I think you are confusing more users (especially if they don’t have a classical background) by allowing approach #2. Honestly, it’s an antiquated approach to parallel keys, and it’s frustrating (as someone who teaches many music theory students each semester) that it’s being perpetuated given how great the site is otherwise. (I apologize for putting such a sharp point on it, but it really is a serious issue in my opinion.) If you really hope to have the site make inroads into pop/rock music education (as it appears you do), I really would urge you to seriously reconsider your default labels for minor.

Best,
Trevor de Clercq
PhD in music theory, Eastman School of Music
Assistant Professor, Middle Tennessee State University

[Ryan]

@trevordeclercq ,

I think I understand better now. Essentially you’re suggesting that when modes are switched away from major, that the naming of the scale degrees and chords reflect the naming in a parallel sense. For example, the diatonic scale in the minor mode would be:

1 - 2 - ♭3 - 4 - 5 - ♭6 - ♭7

In this system, my example above (vi → ii → V/vi → vi) in approach #1 would still transpose to the minor mode as: i → iv → V → i. However, the progression vi → IV → V → vi would look like:

In particular, retaining “diatonic” color schemes, but labelled with the ♭s.

This seems reasonable to me. I especially like that it’s easier to figure out the relationship between scale degrees in different modes. For example, right now it’s basically on the student to memorize that scale degree 5 is flat in the locrian mode but natural in the phrygian mode (as compared to the major scale). This is precisely the type of inconvenience that we were trying to eliminate in our system. In your suggested system, this information would be explicitly indicated.

I’m curious to see what other users think about implementing this change. Also, since you are in the academic sphere, could you point us to some references where this issue is discussed?

[HertzDevil]

Both styles are mentioned on the Wikipedia entry for “Universal key” (which by the way has no citations at all). The popular music style for Roman numerals simply assumes that the underlying mode of the Roman numeral chord symbols always be Major / Ionian regardless of the actual mode, and, as pointed out above, this was attempted in some Theorytabs as a compatibility workaround before modes were implemented (cf. "Pushing Onwards (Verse, revision 1); compare revisions 0 & 2). In order to display chords in this style in Hookpad, a feature in the key menu to change to a parallel key might make as much sense as the current ability to change to a relative key, viz.:
C Major -> C Natural Minor (parallel)
i - ♭VI - ♭VII - i -> i - VI - VII - i (common practice)
i - ♭VI - ♭VII - i -> i - ♭VI - ♭VII - i (popular music)
or:
C Major -> A Natural Minor (relative) -> A Major (parallel)
vi - IV - V - vi -> i - VI - VII - i -> i - ♭VI - ♭VII - i (common practice)
vi - IV - V - vi -> i - ♭VI - ♭VII - i -> i - ♭VI - ♭VII - i (popular music)
eliminating or creating borrowed chords whenever needed. In particular, such feature could enable users to easily edit Theorytabs that are in an incorrect parallel key. Switching to the parallel Major in this way, however, involves a change in not only the chord displays but also the chord data, as well as the diatonic notes of the melody track, which means that if the popular music style is achieved through modifying the chords instead of only changing the display / HTML codes, then there would be difficulty drawing an analogue between “6.4.5.6” and “b1.b6.b7.b1” or even the Phrygian “3.1.2.3” and “Y1.Y6.Y7.Y1” in the Trends API since they become different chord prpgressions in the relative major key (this issue already exists in older Theorytabs); furthermore, the current interface for the melody track will look sloppy since, if a melody in minor mode is forced into the major mode, the notes would have uneven vertical spacing regardless of their display colours.

Hence this popular music style for universal key chord symbols should only be implemented as an option on the Theorytab settings page, modifying only the accidentals that appear on chord symbols. The parallel key switch feature could be implemented nonetheless, for purposes other than displaying chords in the popular music style. There are other implications:

• Since the popular music style for chord symbols uses the parallel Major key per definition, the aforementioned instant change to other parallel keys would not be possible (again, that feature could be implemented nonetheless);
• This style assigns the same scale degree to every key up to enharmonicity and irrespective of the Theorytab’s mode, so in practice it can separate the chord scale from the melody scale and, if the melody track hides those pitch degree labels to the left of each row, easily generalize to any scale including ones that do not include 7 notes, as the display of the chords depends only on the root pitch but not the mode;
• For the same reason, the features of the 7 diatonic modes are downplayed, as is the relationship between relative modes of other scales e.g. Freygish and Harmonic Minor;
• This style could introduce double accidentals for supermodal chords. Consider the extreme example: A Theorytab in Major mode borrows chords from the parallel Locrian key, and these chords are i°, ♭II, ♭iii, iv, ♭V, ♭VI, and ♭vii (note that this is also how the Locrian mode’s diatonic chords would already look like in the popular music style); if the Theorytab then changes to the relative Locrian key, these borrowed chords would become ♭ii° , ♭♭III, ♭iv, ♭v, ♭♭VI, ♭♭VII, and ♭i. The repeated accidentals are tantamount to using a key signature of E♭♭ Major;
• To avert double accidentals in the Roman numerals altogether, these Roman numerals could be converted into one single Major scale that is independent of the Theorytab’s mode as well. It is then arbitrary whether to use sharps or flats to display the 5 non-diatonic pitch degrees. (For instance, most DAWs use only sharps when a key signature event is not supplied, thus similarly the key of F Minor, relatively equal to G♯/A♭ Major, would use “I - ♯I - ♯II - IV - V - ♯VI - ♯VII”)

Overall I am against the idea of mandating that the popular music style for Roman numerals be used, but this style should prove beneficial for an increasing number of Theorytabs of non-diatonic songs. It has to depend on the preference of every user (some people would even use only the modern notation below the Roman numerals).

[trevordeclercq]

@Ryan: Yes, you’ve got the idea now. To confirm what you already know, the scale degrees for the Dorian mode would be:

1 - 2 - ♭3 - 4 - 5 - 6 - ♭7

… and thus the diatonic chords in Dorian would be:

i ii bIII IV v vio bVII i

… but I would advise “minor” to default to the Aeolian mode, unless the user wants otherwise. Personally, I think that charting songs in modes like Dorian and Mixolydian is a somewhat misguided practice, but I’ll save that for another post.

As far as academic references go with regard to discussion of this issue, I cannot say that there has been much debate in print per se. The sea change has pretty much just happened over the last decade or two, with analysts tending towards, for example, the use of bVII to represent a G major chord in A minor (if choosing A minor as a i chord instead of a vi chord). I’ll look for something a bit more substantive, but for now, here are some publications (including those of my own) where Method 3 is used.

Temperley, David & Trevor de Clercq. 2013. “Statistical Analysis of Harmony and Melody in Rock Music.” Journal of New Music Research 42/3: 187-204.

Temperley, David. 2011. “The Cadential IV in Rock.” Music Theory Online 17/1 (March).

de Clercq, Trevor and David Temperley. 2011. “A corpus analysis of rock harmony.” Popular Music 30/1 (January): 47-70.

Biamonte, Nicole. 2010. “Triadic Modal and Pentatonic Patterns in Rock Music.” Music Theory Spectrum 32/2: 95-110.

Everett, Walter. 2009. The Foundations of Rock: From “Blue Suede Shoes” to “Suite: Judy Blue Eyes.” Oxford: Oxford University Press.
(see especially Chapter 10).

Capuzzo, Guy. 2009. “Sectional Tonality and Sectional Centricity in Rock Music.” Music Theory Spectrum 31/1: 157–174.

Method 3 is also pervasive within the professional community of pop/rock musicians, as evidenced by these publications:

Riley, Jim. 2010. Song Charting Made Easy. Milwaukee, WI: Hal Leonard. (see especially the transcription on page 38 of the metal song “Before”.)

Kachulis, Jimmy. 2005. The Songwriter’s Workshop: Harmony. Boston, MA: Berklee Press

Wyatt, Keith, Carl Schroeder, and Joe Elliott. 2005. Ear Training. Milwaukee, WI: Hal Leonard.

Perricone, Jack. 2000. Melody in Songwriting. Boston, MA: Berklee Press.

Wyatt, Keith and Carl Schroeder. 1998. Harmony and Theory. Milwaukee, WI: Hal Leonard.

I should say that it is not impossible to find theorists who still use the traditional system (or a variety of other hybrid Roman numeral systems) in their analyses of pop/rock music . But my own perception of the field is that the general trend has been towards adoption of Method 3 as the standard.

[trevordeclercq]

@HertzDevil:

You bring up important points related to the current encodings of many songs. I have been frustrated myself in trying to update a song (especially one with both an encoded harmony and melody) to change the tonic key. You seem to have a better handle on the programming and database issues. All I can say is that, whether we keep the current system or adopt a new system, a fair number of currently-encoded songs are a mess. Perhaps more songs would be a mess with a new diatonic system for minor, but I think it would be worth it to avoid digging deeper into the current mess of having people encode songs in the parallel major because they are trying to get the bIII, bVI, and bVII labels.

As far as the implications you mention, I think you are overstating the importance of modes (other than major and minor, i.e., Ionian and Aeolian) in this repertoire. To be sure, jazz musicians are very concerned with modes, as these are jumping-off points for soloing over a variety of harmonic structures. But my own perception of pop/rock music (as a player, writer, and theorist) is that modes are more coincidental and fleeting than structural properities. Take the chorus to “Lucky” by Radiohead, for example. We could say that it is in Dorian, but then the bVI chord at the end thwarts a pure Dorian mode. Mixture is too strong a compositional element in pop/rock to adopt a primarily modal approach. Even when songs (or song sections) are purely modal, the question of mode is somewhat trivial. Take the apparently Mixolydian verse to “Hungry Like the Wolf” by Duran Duran, for example. I think we’d be better off charting it as I bVII and not worrying about whether it is major (Ionian) with bVII mixture or diatonic chords drawn from the Mixolydian scale. There is really no way to prove either one. Preferably, our Roman numeral system would be as neutral as possible, hopefully avoiding these analytical distractions. It is enough of a quandary already to decide whether a song is in minor or major, let alone all possible modes.

So a lot your implications are somewhat trivial, as far as I can tell. What songs are we charting in Freygish Minor? And in what sort of situation would I be charting something in the parallel Locrian but then want to view the chord changes in the relative Locrian key? (Can you even name a single song where the harmony should be charted in Locrian?) I’m not opposed to thinking big picture and trying to avert possible problems. But I think some of these problems are entirely theoretical, and I think they might result more from someone charting a song incorrectly (i.e., with the wrong tonic) than from a problem inherent in the Roman numeral or scale degree system itself.

Overall, I guess I’m not opposed to having the minor Roman numeral and scale degree labels be a user preference (as you seem to suggest), but it seems to me that if there is a standard method for charting for popular music and we are charting popular music, why not use that method? I realize that this web site does include some common-practice era musical examples, but I think those are not really the primary focus of the web site.

[trevordeclercq]

@HertzDevil, @Ryan:

I sent out a group e-mail to the subscribers of the Pop-Analysis interest group for the Society of Music Theory. All replies have confirmed that Method 3 (i, bIII, iv, bVI, bVII, etc.) has become the standard technique among theorists and pop musicians. To quote a pre-eminent theorist in the field of pop/rock analysis:

“[Method 3] has been used in rock guitar method books and magazines going back to the 1970s, as I recall, and I suspect in jazz improv and theory books going back even further. I’ve always used it as my default with no sense that the major key is anything more than the easiest neutral place for comparison and orientation.”–John Covach, Director; Institute for Popular Music, University of Rochester

No one seems to know of any specific arguments for Method 3 in print, at least in terms of addressing its use over the traditional method (Method 2). But all seem to agree that it has become the de facto standard.

[Ryan]

There is a lot of great discussion here.

It’s clear from @trevordeclercq that the music theory community is moving toward the “popular style” of notation. However, as @HertzDevil has pointed out, there are a lot of implications of switching to this style, and the “common practice style” has its advantages as well.

Pedagogically speaking, I think there are strong arguments to be made for both styles. If we could, I’d like to put this debate on hold for a moment and instead discuss some practical considerations regarding the implementation of a “popular style” notation in the Hooktheory system.

Let’s hypothetically assume that we had a setting that allowed users to switch Hookpad and all theorytabs into the “popular style” notation.

I agree with @HertzDevil’s assessment that the popular style for Roman numerals is fundamentally Ionian-centric. In this system, other modes seem to manifest as altered versions of the major scale rather than shifted versions of the major scale. Furthermore, I agree that transpositions between modes in this style seem to be more natural in the parallel sense than the relative sense.

I discussed this with the Hooktheory team extensively last night, and we feel that the most consistent way for us to capture this style would be color the tonic chord as red, regardless of the mode:

In this color scheme, the hashed colors in the popular style serve to indicate explicitly the differences between the current mode (minor in this case) scale and the major scale, analogous to how the altered scale degrees (3, 6, 7), are prefixed with “♭”.

One nice aspect of this style is that it is much easier to visualize the function of chords across parallel modes. In my view, it would be a shame to have such a system and not also support transposition to parallel modes (e.g., C major → C minor). Such a feature would provide a far richer framework through which the scales of various modes could be studied and understood.

Logistically, of course, this creates a great number of issues that need to be resolved. One issue that I am currently considering is what to do about borrowed chords under a parallel transposition. Borrowed chords in the popular style are in effect absolute (the label of the chord is independent of what the home mode is). In this sense, it may be most reasonable to preserve the borrowed mode during a transposition. E.g., a chord borrowed from the minor mode remains borrowed from the minor mode regardless of the home mode; If the home mode is transposed to minor, then the chord is no longer borrowed. This would allow us to fix those pesky theorytabs where all of the chords are borrowed from another mode.

All of this, of course, is in the planning stage at the moment. This would be a significant change to the site, and we want to make sure that such a change would ultimately enrich the Hooktheory experience for our users.

As always, we are grateful for your feedback.

[trevordeclercq]

@Ryan

One of the issues that seems to be of great importance to you (and maybe also the users of the site) is transposition of songs to other modes. But it’s not entirely clear to me when or why this would be of value, except in the case of going between relative keys. (Although I use the site, maybe I’m not using it in the way others do.) Do you have some specific examples that I could think about? I should also point out that the benefit of the popular style is that you really never have to shift between parallel keys, just between relative keys.

The issue of chord colors is not something I’ve thought about a lot, since it’s not a standard thing in the theory community. But I have to admit that I liked your previous suggestion of having the colors relate to similar functions in relative keys, e.g., tonic in minor would be purple and the vi chord in major would be purple. That way you have the different tonic axes but retain the relative relationships. The color scheme you suggest in your prior post, where III in minor “common practice” is red but tonic in minor “popular” style is also red doesn’t really make sense to me. (I understand the reason behind this, i.e., parallel key relationships, but not the value.) Perhaps I haven’t seen all the consequences. Yes, the popular style of Roman numerals does derive from a parallel-key thinking, but if those are indeed the Roman numerals used in a song, then bIII will still have the same sound and pull as the major tonic (and thus the same aural “color”). Note that the “common practice” style is a parallel-key method as well, in that you are ultimately labeling the minor tonic as a i chord and the dominant as a V chord.

Basically, I see it as a simple solution. The popular-style of Roman numerals allows for easily shifting (and understanding relationships) between parallel keys, and your color scheme allows (or could allow) for easily shifiting (and understanding relationships) between relative keys. I suppose someone could chart a song in Dorian and want to see what the chords would be in Ionian. But while this may be a useful theoretical exercise, I’m not sure it reveals all that much about the harmonic or melodic structure of songs and only makes your web site more complicated. Generally speaking, the use of modes in the analysis of pop/rock music was an idea that was toyed around with in the '90s, but modern theorists have essentially abandoned that sort of thinking. (I’m not trying to close down scholarship or inquiry here, I’m just trying to reflect the field as I see it and how I think your web site could be greatly simplified.)

[Ryan]

@trevordeclercq, thanks for your comments. The issue that we are running into, is that there are fundamentally two different ways of looking at a change of mode. This is, in effect, similar to the la-based minor vs. do-based minor discussion.

Do you:
1) Think of the natural minor scale (or any mode) as a cyclic permutation of the major scale

or

2) Think of the minor scale as an altered version of a major scale (take major scale and flat SD 3, 6, 7)

Each description has its pros and cons that don’t need to be repeated here. Currently the Hooktheory color scheme aligns with the first description. We permute our colors so that they simultaneously preserve their intervalic relations and accurately reflect the intervals defined by the given mode.

The problem presented by the popular style is that it fundamentally describes an alternate way of thinking about a scale (description 2). By using accidentals on scale degrees and chords, we are explicitly contrasting the chords in a given mode with the equivalent scale degrees and chords from the parallel major mode, which is largely inconsistent with the color scheme. The concern is that this may cause a great deal of confusion.

Again, this is purely hypothetical, but we essentially see 3 ways to proceed on colors:

1. Add popular style notation but keep colors the same: (@trevordeclercq’s suggestion) This option would preserve the convention that solid colors represent diatonic scale degrees and hashed notes representing non-diatonic scale degrees. This would also preserve the aural relationship between colors (e.g., green to purple represents a specific interval regardless of mode). The downside to this is that the colors offer very little intuition for different modes. For example, the red scale degree is 5 in the lydian mode, ♭3 in the minor mode, and ♭6 in the phrygian mode. This lack of intuition is somewhat present already, but the addition of the accidentals seems to make this worse.

2. Change colors to reflect differences from the major mode: In this scheme, ♭3 would be hashed orange and yellow, regardless of mode (since it is between the orange 2 and the yellow 3). Here we forfeit the solid/hash colors representing diatonic and non-diatonic scale degrees, but gain intuition and context as to how chords and notes function in other modes. This color scheme would also be consistent with the popular styling notation. Finally, chords borrowed from a mode would have the same color scheme in any mode, including the borrowed mode.

3. Colors represent position in scale: In this scheme, a yellow scale degree would represent the 3rd degree of the scale, regardless of whether it has an accidental or not. Here we lose the aural consistency between colors, however, the relationship to the major scale is still referenced by the labels. Another version of this could also put small gaps between scale degrees on the staff to indicate whether the distance between different scale degrees are whole or half steps.

[trevordeclercq]

@Ryan:

I think that the natural minor scale should be considered both a result of a cyclic permutation of the major scale (or rather, the same diatonic collection with a different tonal center) and as an altered version of the major scale. These two ideas reflect the two common ways of understanding minor, i.e., relative and parallel keys, or (as you say) through a La-based and a Do-based method. Both are valid systems. The nice thing about HookTheory is we have a chance to have both at once. Roman numerals align with the Do-based method (1 is 1), and colors could align with the La-based method.

To address your concerns about colors:

1) I think it is a mistake to chart songs in modes. It is much simpler to say that a song is either in a major key (i.e., the third of the tonic triad is major) or a song is in a minor key (i.e., the third of the tonic triad is minor). Songs may shift between the tonic triad being major or minor, but that represents a modulation, i.e., a shift in the diatonic collection, if the change is signficant and prolonged. (A little bit of mixture, e.g., a minor tonic in a song that is otherwise major, would not require a wholesale key change.) Beware of conflating major and Ionian (and minor and Aeolian); these are similar but very different concepts. The great amount of mixture found in rock makes modes unnecessarily complicated for understanding harmony (even if we see evidence of modal patterns in the melody or riffs of a song). So the downside to Method 1 for colors does not exist for me.

2) This scheme is fine, but it’s basically telling you the same thing as the Roman numeral. As a result, I don’t see any problem with it. It’s a shame to lose the relative relationships that could be reflected, but we aren’t used to seeing them anyway.

3) If I understand you correctly, this method kind of uses the colors to reflect the traditional Roman numeral method (i.e., both iii and bIII would be yellow), right? I think that’s fine, too. On some level, it is better than Method 2 because it doesn’t highlight some inherent oddness of bVII by giving it a hashed color. (More on this below.)

Personally, I don’t really have a preference with regard to which color scheme you use. I think each has it’s benefits. I will say one thing, though.

As founders of this web site, you seem to want to convey to your users the knowledge of certain scales and information about departures from those scales. That is great for beginners, who may just be starting to learn the terminology of music theory. But you are implicitly making a fundamental assumption (especially with the hashed color schemes) about what is normative and what is non-normative based on these scales. I think that is a mistake.

For example, given all rocks songs in a major key, the most common chord built on some version of scale-degree 7 is the bVII chord. Why would we hash the color of that chord – which calls attention to some presumed important depature from some presumed normative structure – when bVII is the most common chord in rock songs after V and IV (or at least, bVII is way more common than viio)? Perhaps we might change our diatonic collection from Ionian to Mixolydian to account for bVII, but V (which includes scale-degree 7) is much more common than bVII. There is no single diatonic scale that accounts for the common chords in rock music. (Note also that b7 is more common as a melodic scale degree than 7, even though all other major scale degrees are more common than those from the parallel minor). As John Covach states, we use the major scale as a convenient reference point. But we don’t want to over-emphasize the role that the major scale plays in rock songs. Or worse, we don’t want to unnecessarily draw attention to something when that something is rather common.

I’m not sure how that affects the use of colors. I suppose that argues more strongly for Method 3 (generic colors) than any other. I still like Method 1, though. Perhaps I might suggest some hybrid:

Method 4: Use generic color schemes, i.e., scale-degree 2 (whether sharp, natural, or flat) would be orange (or whatever). But have the generic color schemes shift based on the tonality of the tonic triad. The tonic of a major key would be red, for example, with vi and bVI both being purple. But if the key is changed to minor, then the tonic would be purple and and the bIII chord (or biii chord even) would be red. I don’t know. I haven’t thought this totally through.

Thanks for listening to me “Ramble On.” I really appreciate the time and care you’re taking with my thoughts and suggestions.

[Turbo]

I’m not sure where I would fall on that dichotomy as I treat each scale is its own beast. If a scale is described as “the third mode of the double harmonic major scale” or alternatively “phrygian b4 bb7,” I don’t get any sense that the internal working of the scale is actually contingent on double harmonic minor or phrygian, but just recognize that the labels are convenient to note a scale that doesn’t have a proper name, or at least a well known name (apparently someone calls this one ultraphrygian.) Two cents thrown, I like option 2.

[Turbo]

@trevordeclercq

It is much simpler to say that a song is either in a major key (i.e., the third of the tonic triad is major) or a song is in a minor key (i.e., the third of the tonic triad is minor).

I’ve only encountered these definitions in association with the phrases “major mode” and “minor mode.” As in: Ionian, Lydian, and Mixolydian are major modes because the major third in their tonic chords. I unfortunately can’t find a way to verify which one is correct however. Have you got a reference for this?

[trevordeclercq]

@Turbo:

My general rubric to put a song in a major key if the tonic triad is major (and minor if the tonic is minor) is not a definition of major or minor keys. Keys are inherently complex and multi-faceted things. But aside from the issue of relative key relationships (e.g., is the song in C major or A minor?), can you give an example in which the key quality (major or minor) does not correspond to the quality of the tonic triad?

[Ryan]

@trevordeclercq, thanks again for your comments. Your voice and expertise is extremely valuable to us as we continue to improve Hookpad and Hooktheory.

To respond to your comments, it is true that Hookpad makes the assumption that diatonic scale degrees are in a sense, normative. Ultimately we must choose some reference, and we feel that the major scale is the preferred choice. As a consequence of this, chords such as ♭VII appear hashed; however, I disagree with you about the importance of ♭VII. We keep detailed statistics about the relative frequency of chords and scale degrees. ♭VII is actually far from a common chord among the 5000 songs currently in the Hooktheory database. The most common chords are all diatonic: I, IV, V, vi, ii, iii (although I agree that vii^o is not too useful), as are the scale degrees. ♭VII ranks 13th, and is far less popular than V⁶, which uses a natural 7 in as its bass. Scale degree 7 is over 50 times more likely to show up in a melody than ♭7. Of songs that do use ♭VII, many of them only contain I, IV, and ♭VII and have no instance of scale degree 7 in a natural sense, so many would argue that these chord progressions are properly mixolydian in nature.

The colors are important to us, because they provide an additional dimension in which harmony can be visualized. And while everybody looks at a theoryTab differently, many of our users (especially those who don’t know any other system) rely on the colors to orient themselves to the scale.

However, as you point out, if we use Method 1 (outlined in my previous post) the colors and labels in the “popular” style notation could serve a dual purpose; they would simultaneously show the cyclic nature of the scale (la-based approach) while also showing the relationship to the parallel major scale (do-based approach). While there are still a downside, I think that ultimately this color-scheme may be the best option if we were to implement this notation style.

Regarding the Ionian/Major issue, I suppose it is understood in modern music parlance that when we speak of modes, we’re speaking about modes of the major scale unless otherwise specified. Please correct me if I’m wrong

[Turbo]

Metal songs using locrian come to mind, like Metallica’s “Sad But True.” Theoretically, locrian should be minor, or diminished if you count that separately. In practice, there’s just no tonic triad, and the closest thing is a plain power chord, omitting the third and using the perfect fifth. The power chord works because it outlines the prominent harmonics of the 8vb from root, to manufacture a phantom fundamental. So the theoretical locrian progressions like io-bII-bV-io can be approximated in play as 1-bII-bV-1 and it has nowhere to fit in your categorization.

[trevordeclercq]

@Turbo:

Your “Sad But True” is a great example of how it’s important to differentiate between keys, harmonies, modes, and scales. I assume when you say that “Sad But True” is “Locrian”, you are referring to the main riff that goes b7-4-b5-4-b3-1. And yes, that could be construed as a subset of the Locrian scale. But this is a melodic figure (a riff), not a harmonic construct. Note especially that when the riff descends to scale-degree 1, we hear a power chord on tonic. That power chord, of course, includes regular scale-degree 5 (not b5). So you can’t really say the song, even in this short span you’ve identified, is really Locrian. You are also correct, of course, that the power chord does not include a third (major or minor). But during that riff, there is no real change of harmony. So in harmonic terms, we simply have one long expansive tonic chord during the riff. Therefore, the b3 in the riff makes our ears interpret the tonic harmony as minor because it occurs during this span of music. Our ears take notes in close proximity and interpret them as harmony and harmonic events. We don’t necessarily have to have all the notes happening at the same time to infer keys and harmony. That’s why a cappella melodies can be heard as major, even though you’re only hearing one note at a time.

So the theoretical Locrian progression you present (io, bII, bVI, io) does not exist in reality as far as I can tell, and thus I’m not sure why we need some complicated system of harmonic modes to track it.

[trevordeclercq]

@Ryan:

Well, the statistics you generate are strongly tied to a number of different factors. For one, they rely on the corpus under study. Your corpus is the collection of songs that are easily encoded by users. Users may choose to encode more diatonic or major songs because your system more easily allows for those encodings. Plus, I would guess many of your users are novices with regard to their understanding of theory, so they are encoding more simple songs and have a more simple understanding of harmony. They also may choose more recent songs because those have current popularity, which may not reflect pop/rock music from a broad perspective. To more clearly show that the statistics you are generating are not related to mitigating factors, you might take a random sample of chart hits over the decades and seeing if those statistics still hold. Davy Temperley and I did that in our rock harmony article (see above), and we found bVII to be very common.

The other factor is the encoding system itself. If users are not able to easily encode minor keys with a bVII chord, then you will not find many instances of bVII in your database. Also, since there was a long time when Hook Theory only allowed major keys for songs, a lot of the encoded songs are probably mapped through the relative lens of major instead of minor.

Another issue is that you only have parts of songs encoded, so when you say “songs”, you can’t be talking about whole songs I don’t think (or at least adjusting your statistics for the number of times the chords occur during the course of a song, i.e., bridge material would be weighted less strongly). The fact that many parts of songs only contain I, IV, and bVII is not surprising. These are Mixolydian-like moments for sure. But I would venture to say that most of these song parts are complemented by a variety of other chords in other sections, such that the song as a whole does not conform very well to a single mode.

You are correct, of course, that modes refer to rotations of the major scale. But we should be very careful to distinguish between a major KEY and a major SCALE. The two are entirely different things. And conflating melodic modes and key tonalities is a dangerous thing to do. (See my post below in response to the “Sad But True” Metallica example.)

[Turbo]

@trevordeclercq Actually the interest is in the verse, stretches of D5, with fleeting C5 and Ab5 providing dominant functioning “push” that would be expected from them in harmony even though the power chords have a more contrapuntal(?) capacity. I’m not sure that should so easily disregarded. The only reason I went looking into the obscure phantom fundamental effect was to try to figure out why this article was dignifying locrian when all I have heard prior to it is “It can’t be done.” I find it well presented other than omitting the explanation of combination tones, but then again I don’t spend much time reading articles on theory since my primary work is in production - you might find it utterly cringeworthy. I’m all for hearing out your criticisms and counterpoints to it. No pressure to spend your time to do so, of course, but I’ve a feeling I wouldn’t be the only one to benefit in getting properly exposed to both sides.