• No results found

4.   Analytical Findings

4.2   HARMONY

4.2.2   Chord Distribution

would have been played D but raised by either using a capo during recording or increasing the tape speed post-recording.

An interesting point of comparison would potentially be the keys used by male and female lead singers, however, there is little difference to be observed. The 22 songs with female vocalists span all the major keys but Bb, B and Db, hardly noteworthy exceptions. Male vocalists sang in all major keys, while the two songs with male and female vocalists, ‘For Today’ and ‘Poi E,’ are in C and F, respectively. The prominent feature was that four females sang in D minor. This is a high proportion for a minor and a specific key, although, given the small sample size, this may be a coincidence. Further investigation into vocal ranges or keys used by other female singer-songwriters (i.e. Carly Simon, Joni

Mitchell, Stevie Nicks etc.) could be fruitful.

Overall, the conclusions are relatively basic. Furthermore, as Sean Sturm, lead singer of EyeTV, revealed, ‘One Day Ahead’ was recorded in Eb but often performed in D to help the guitarists, or in E. The latter key enabled the bass guitar to hit a deep, rich tonic note in the chorus and helped to raise the vocal intensity.3 Therefore, it seems keys are not only dependent on personal preferences, but those used in the studio may also be somewhat arbitrary choices.

The Roman numerals show the distance from a localized tonic; however,

functionality was also considered. For example, ‘Cheryl Moana Marie’ contains an apparent #V chord in the final bar of the second chorus. Given the

modulation up a semitone in the subsequent bar, it is obvious this harmony anticipates the shift; hence it is marked as “V.” The same can be said of other pivot chords; this step is taken to more closely reflect harmonic contexts. That said, applied dominants were initially related to the tonic, not the tonicized note (i.e. a major supertonic is II, not V/V). It was then possible to return to the data and establish which instances of II, for example, were applied chords and which were not.

De Clercq and Temperley establish their chord distribution as counted instances of each chord; from their corpus, I occurs 3,058 times.4 Their method was modified slightly so that harmonies were measured as proportions of a song’s various sections, such as introduction, verse, chorus, bridge and so forth. For example, ‘Nature’ begins with a four-chord progression heard four times: i-III-IV-VI, without inversions.5 The riff lasts two bars, with two chords per bar.

Under the present method, each chord is registered as “0.25,” signaling that each chord occupies a quarter of the introduction. This approach has advantages and disadvantages.

First, proportions highlight the relative weight, and possibly importance, of harmonies. Over a body of songs, the prevalence of particular harmonies will be evident regardless of the method. But for individual songs, greater accuracy is beneficial. For example, The Mockers’ ‘Forever Tuesday Morning’ opens I-ii-IV-bVI V, where the hyphens indicate barlines. If each chord were counted, bVI and I would be judged equally. Under the revised method, I would score “0.25”

compared to “0.125” for bVI6, better reflecting the latter chord’s role as an embellishing upper neighbour to the dominant.

4 Trevor de Clercq and David Temperley, “A Corpus Analysis of Rock Harmony,” Popular Music 30, no. 1 (2011), 60.

5 The bass line descends from the tonic to the submediant; thus, i-IIIc-IV7b-VI.

6 The introduction also contains twelve bars of drums, i.e. “N.C.” This is counted in the actual analysis but avoided here for clarity’s sake.

Second, counting harmonies can produce skewed results according to the lengths of songs. Chris Knox’s ‘Not Given Lightly’ features a less conventional major submediant, G# major, not G# minor, in B major. It is heard sixteen times during the song. The same VI chord, albeit as a secondary dominant, is present in ‘Bursting Through’ by Bic Runga, but is only heard three times. Is VI in ‘Not Given Lightly’ five times more important than in ‘Bursting Through?’ It is difficult to answer, but the frequency of VI in Knox’s song is predicated upon the frequency of the verse progression, which is only four bars long and played sixteen times in the song. By comparison, VI appears in the chorus of ‘Bursting Through,’ which only occurs three times. The relative proportions, “0.25”

(Knox) and “0.08” (Runga), avoid distorted results, while still reflecting the chords’ prominence within their respective sections.

This method also has disadvantages. The distributions do not account for placement within phrases and songs, an issue discussed below. Second, at times the quantity of chords is an important feature. In Split Enz’s ‘Spellbound,’ the introduction rocks between Bm7 and E7 chords in four-bar phrases. The tension in the introduction derives partly from the lack of harmonic orientation and modal implications, and partly because the section lasts 44 bars before the vocals enter. In this case, the important harmonic feature is not that ii and V each occupy half the phrase, but that they are both heard 22 times, the music meditating on their Dorian inflections. The results should, therefore, be interpreted cautiously.

In many cases, proportions could be calculated as divisions of four- or eight-bar units. When sections varied in length across a song — for example, ‘Don’t Dream It’s Over’ contains choruses of 32, 30 and 26 beats — the length of individual chords was summed and divided into the total length of the section (i.e. proportion of 88 beats for ‘Don’t Dream It’s Over’). Every effort was made to divide bars and phrases accurately, but syncopation and anticipation, such as in ‘Sensitive To A Smile,’ was rounded to the nearest quaver. One option would be to use a computer programme to segment harmonies temporally; that said, the extra precision would affect the results so little that the time spent on programming would be of marginal benefit.

The data for each song were entered into a Microsoft Excel spreadsheet. This enabled straightforward calculations, such as overall and sectional proportions.

Furthermore, it was possible to filter the list of songs according to variables, such as chart position, Nature’s Best position or other musical details. The results were updated automatically, allowing for quick comparisons, some of which are considered below.

The complete distribution of chords is presented in Appendix B. Graph 4.1 shows a condensed distribution of the major primary triads, the flattened leading-note and the minor submediant; Graph 4.2 presents the same distribution according to songs’ sections. Graph 4.3 presents a condensed distribution of the “Other” harmonies from Graph 4.1. The top-right segment in each graph corresponds to the first chord in the legend, before moving

clockwise (graph) and down (legend); thus I occurs 32 percent of the time overall.

Graph 4.1 Overall Weighted Harmonic Distribution

Graph 4.2 Weighted Harmonic Distribution by Section

Graph 4.3 Weighted Harmonic Distribution of "Other" Chords

The “Overall” percentages in Graph 4.1 are the arithmetic means from each song section. The data has been weighted to counter the uneven distribution of sections — 99 songs have choruses, whereas only eighteen have a pre-chorus.

The weighting system is outlined in Appendix E. The “Other” category of Graph 4.1 contains 23 chords ranging from the minor supertonic, ii, to the flattened mediant, bIII, to an augmented chord on the flattened submediant in Citizen Band’s ‘Julia.’ Chord colourations were omitted, however, Isus and IIsus were counted from the introduction and chorus of Split Enz’s ‘History Never

Repeats.’ Here, the suspensions are the essence of the chords; they are non-resolving and contribute to the tonal ambiguity.

Graph 4.4 Harmonic Distribution by Root

When the harmonies are tallied according only to the root note, the majority is built upon the first, second, fourth, fifth and sixth scale degrees, as shown in Graph 4.4. The primary harmonies — I, IV and V — account for approximately 60 percent of all chords. The prevalence of bVII, however, is notable; this chord is comparatively rare in common-practice classical music, both in major and minor keys as it eschews the leading-note. This is not the case for popular music. The bVII chord is used more than vi or ii; its relatives, bIII and bVI, also appear frequently enough to support Moore’s earlier suggestion that flat-side harmonies are engrained in popular music’s harmonic language.

This is complemented by the apparent preference for IV over V. Certainly, if popular music has strong subdominant foundations, then harmonies such as bVII (i.e. IV/IV) and bIII (bVII/IV) will naturally occur. This point is, in part,

corroborated below in Section 4.2.3 regarding modulations; more songs modulate to flat keys than to the dominant and its relatives.

The numbers alone, however, do not explain how the harmonies are used. Two examples should clarify this point. Sharon O’Neill’s ‘Words’ is built upon a I-V-IV pattern, all over a tonic pedal; in the chorus, this idea is varied slightly and placed in F major. By virtue of this harmonic shift, the subdominant occurs more frequently than the dominant. One could argue, here, that IV is structurally more important; this claim is supported by the numbers.

‘For Today’ would return similar figures; the song’s eight-bar progression only uses primary chords in C, I-IV-V-IV, each lasting two bars. Proportionally, IV is twice as common as V. But, compared to ‘Words’ in which IV is emphasised as a harmonic region, the subdominant in ‘For Today’ essentially acts as a passing chord. The phrase is structured so that the bass line arches from tonic to

dominant. The ascent rests on F, while the descent from dominant to tonic is softened by again landing on the fourth degree. The pentatonic melody

undermines the traditional tonic-dominant relationship, but the frequency of IV, in this case, has fewer theoretical implications and indicates little more than a particular progression.

Other trends from Graph 4.2 are relatively straightforward with regards to songwriting techniques. The tonic major is used less in the pre-chorus,

seemingly in favour of the tonic minor and the relative minor harmonies. This is likely to increase harmonic tension. Subsequently, the chorus has the highest sectional proportion of primary triads, which helps resolve harmonic tension and confirm the key. ‘Jumping Out A Window’ by Pop Mechanix and ‘I Hope I Never’ by Split Enz are prime examples in this regard. The proportion of tonic chords also decreases in the bridge, the section which explores new harmonic ground in preparation for the final chorus. Again, this seems to be

predominantly a tension-related device, as demonstrated in ‘Better Be Home Soon’ or ‘Blue Day.’

It is possible to compare the distribution of harmonies according to decade, as shown in Graph 4.5.

Graph 4.5 Harmonic Distribution by Decade

The small sample sizes prohibit sound conclusions. There are only four songs written pre-1970; thus, the high proportion of “Other” results, in part, from the frequent use of bVI in ‘Blue Smoke’ and ‘Let’s Think Of Something’ rather than any idiomatic differences. Likewise, the tonic major occurs slightly more often than the overall average in the 1980s songs, but remove Front Lawn’s ‘Andy’

and Blam Blam Blam’s ‘There Is No Depression In New Zealand’ and the figure falls by several percentage points. Both songs, written by Don

McGlashan, have extended sections with a static harmony. In a small sample, these cases skew the results. Overall, however, there is little change over time.

Graph 4.6 Harmonic Distribution by Nature's Best Position

When the Nature’s Best list is segmented, as in Graph 4.6, harmonies are similarly distributed. Tonic major harmonies spike in the second quartile of songs; however, the difference between songs ranked 1-25 and 26-50 equals approximately one extra bar of I every sixteen bars. In the context of a song, such a variation would hardly be noticeable.

The chord distributions change little according to the peak chart position, dividing between, for example, top twenty hits and those outside the top twenty.

Overall, this supports Moore’s claim that popular music has historically been founded on a “static” musical language7, although greater sample sizes would be beneficial. This observation does not render these results irrelevant; they

provide important insights into pop and rock’s musical language, discussed further in Chapter 5.