Level of detail

In this section we discuss the single most pertinent cause for inter-annotator disagreement: The question of how large of a musical chunk we want to compress into a label. There is definitely not one correct answer, instead we ask our annotators to make one high-level decision for a given piece and stick to that consistently. That’s what was meant in the previous section by the assignment

Do a complete annotation of corelli_op01n04b.mscx after deciding on its harmonic pace

What harmonic pace did you decide on and how did you decide? Did you listen to a recording or to a live rendition in your head? In case you didn’t, does listening to a recording make you want to change your mind? In the latter case, please create an alternative set of annotations in a separate file for comparison.

Once you’re settled, let’s walk through some of the possibilities and their implications.

Solution 1: Coarse-grained

Coarse-grained analysis of Corelli op. 1/4, II. Adagio

This solution focuses on chunks of a half note’s length. There are a couple of things to note:

Harmonic pace:

The annotator settled on half measures. The set of annotations demonstrates the consequences of this decision: Many eighth notes are interpreted as ornaments and thus not reflected in the labels. When using the DCML standard we do annotate non-chord tones such as suspensions, retardations, additions (see REFERENCE), but not ornaments.

Indeterminacy:

In m. 2, the new harmony is mainly determined by the pitches F and A. Even if an annotator has a strong opinion, it is a good idea to express the indeterminacy by writing both possibilities in one label, separated by a dash, VI-iv6. A label can include only one alternative (i.e., one dash), and the annotator puts the solution they prefer in the front. The solution to put VI in front probably lies in the absence of a bass figure. (In the present case, don’t worry about the funny dash placement when writing iv6-VI.)

Phrasing:

One’s interpretation of the harmonic pace might correlate with one’s perception of musical phrases. The analysis here expresses fluidity not only through a harmonic rhythm in 2/2 meter but also by having the first phrase end only in m. 8 after the harmonic “double time” and the pronounced cadence.

Solution 2: Fine-grained

Fine-grained analysis of Corelli op. 1/4, II. Adagio. Differences to above highlighted in bold font.

Let’s have a look at the differences.

Harmonic pace:

This set of annotations is susceptible to the change of harmony on the quarter beat level. Whereas in the above version four step-wise descending eighth notes were mostly considered as elaborations of one and the same chord, this one here reveals a tendency to see them as elaborations of chord tones from two different chords or inversions. In particular, the annotator has highlighted implicit 5-6 movements through the pertinent patterns i VI6 (mm. 4 & 9), iv iio6 (m. 3), and VI iv6 (m. 2, that one purely virtual). Where the descending eights motive occurs in the bass, two chords with their bass notes a third apart have been assumed (mm. 2, 6 & 9); and where this was not plausible (mm. 5 & 8), a chord inversion was assumed on the fourth note. This solution seems to suggest syncopated harmonic rhythms in mm. 2-3 & 9.

Indeterminacy:

While being more bold in assuming particular chord tones in m. 2, this solution introduces alternative labels for the dominant chords in mm. 3 & 6; namely labels interpreting the same chord as elaborations of dominant seventh chords. For reasons of consistency, the same interpretation would have applied to m. 8: V}{-V2}{ (every side of the dash needs to be a fully valid reading, which is why the phrase annotations should be repeated). But two reasons speak against this interpretation: (1) it would conflict with the V64 label (because no seventh is included here) on the last eighth which, however, reflects consistency with the above-mentioned analytical decision and m. 5; and (2) the alternative reading V2 would represent a possible harmonic progression iv6 V2}{ which is implausible in this cadential context.

Phrasing:

The more fine-grained harmonic analysis has led the annotator to assume shorter phrases, too: The long first phrase from Solution 1 (mm. 1-8) has been split into two interlocked phrases here.

Hint

On the more abstract level seen in Solution 1 the predominant root progression is by falling fifth (i iv and [VI] ii V i). Theorists know that a scaffold of falling fifths affords movement by falling thirds and it is therefore not surprising that a more fine-grained solution exists which highlights these (i VI iv iio [#viio] V [III] i).

Consistency is watching you

Note how in both examples the analytical decisions were implemented with rigorous consistency which is the DCML annotation standard’s highest maxim.

Solution 3: Indicating replaced chord tones

Too fine-grained and partially inconsistent analysis of Corelli op. 1/4, II. Adagio.

The fictional annotator behind this analysis had just learned about the chord alteration syntax using rounded brackets () and wanted to use it everywhere! Let’s say we leave them the pleasure for now to not steal their thunder, but consistency is nevertheless indispensable. Before you try to spot some of the inconsistencies, it will be useful for you to understand how the chord alterations work, so you can empathize with this greenhorn.

• If an upper neighbour of one of the three chord tones 1, 3, 5 is present within the parentheses, e.g. V(2), V(4), V(6), it means that the respective chord tone is replaced (the fourth suspension we had seen already).

• This is independent of the chord inversion, so iv6(2) still means that the chord root is replaced by its upper neighbour (see the 7-6 suspension in all three solutions, m. 8).

• If the actual chord tone is not replaced, but the upper neighbour is instead added, it is preceded by a +, as is the case in V6(+6) in m. 6.

• The upper neighbours can occur simultaneously (always in descending order, e.g. V(42) in m. 5) and always represent the interval corresponding to the scale. In other words,

  • iv(4) translates to the pitches F B C in a C major context but to F Bb C in a C minor context;

  • to express F Bb C in a C major context we would have to write iv(b4) or IV(b4) accordingly;

  • bII6 translates to F Ab Db in both C major and minor, but bII6(62) is F B E in C major yet F Bb Eb in C minor;

  • in contrast, the same suspension chord F Bb Eb in a C major context would need to be expressed as bII6(b6b2) or ii6(b6b2), depending on the context.

Can you feel the joy that drove our fictitious annotator to use rounded brackets so frequently here? So let’s show them some indulgence—but only where the analytical decisions are traceable and applied consistently.

Review Solution 3 and spot as many places as possible where the annotations are inconsistent or would require justification

When DCML annotations are reviewed, the most important task for the reviewer is to spot inconsistencies. When it comes to analytical decisions that the reviewer does not follow, they ask the annotators for a justification and together they come up with a solution that is satisfying for both. So let’s start with a couple of things that are consistent:

Consistencies

• The reviewer was consistent in the decision to assume a 4/4 harmonic rhythm to a point that, where necessary, the harmony on the third eighth of the descending figures was interpreted as suspension chord resolving on the following eighth (mm. 3, 5, 6, 8). (Consider, however, the inconsistencies mentioned below for mm. 4 vs. 9 and m. 7.)

• G# and C# have been interpreted as thirds of dominant chords (mm. 2 & 9) except where hindered by a pedal note (mm. 1, 2 & 10).

• ii%65 was consistently added as a possible realization of bass note D bearing figure 6 (however, consider the comment on m. 10 below).

Inconsistencies

• 13. 4 vs. 10: iio6-ii%65 vs. iio6 ii%65-iio6

• 13. 4 vs. 9: VI6 vs. i(6) i

• 13. 5 vs. 8: V64 vs. V43

• 13. 6: The alternative reading V65 would be implausible in combination with the subsequent V6.

• 13. 7: To be consistent with the rest, beat 2 would require V(6)/iv V/iv

• m. 9: #viio6 is possible but inconsistent with mm. 2, 5, 6 & 8 where no chord changes were assumed on the second eighth.

Justification requested

• 13. 2: why iv6 VI?

• 13. 4 & 7: What are the arguments for putting the phrase boundaries here (other than a V i progression)?

• 13. 10: iio6 ii%65-iio6: the alternative reading would result in the redundant progression iio6 iio6

• 13. 11, iv7: What are the reasons for considering D as chord root, not a mere transition?

Take-home message

The more consciously and consistently the analytical decisions are made and implemented, the stronger the annotator’s position vis-à-vis the reviewer(s). That’s what is meant by “Consistency is watching you”. But no one should feel intimidated: Those people who joined our team of annotators took this coercion as an incentive to express their own perspective on a piece as precisely as possible within the standard’s expressive power. And some of them reported happily that parts of their conceptual building became visible which had been hidden in their subconscious.

OK, time to harness our expressive power to change some keys!