All You've Got To Do Is Win

In the chapter "The Battle for Britain ... and America" I looked at linguistic predictors of the number of weeks spent in the UK and US charts. While this is a useful behavioural indicator of popularity, it is not necessarily the yardstick by which dyed-in-the-wool Bowie fans like us decide on the quality of Bowie’s oeuvre. One way to explore whether Bowie’s use of language relates to the quality of his work is to find which albums Bowie fans love best.

To do this I’ve looked at three polls, one from BowieWonderWorld.com, one from RateYourMusic.com and a poll of 30 celebrity Bowie fans published in Uncut in May 2008. I’ll describe each of these in turn but the reason I’m presenting all three is because of the issue of “replication”. In science generally, and psychology is included in this, there is no such thing as the perfect study. So knowledge isn’t advanced just because of one analysis but, rather, by the accumulated wisdom of a number of studies. Okay, so what I’m posting on this site isn’t exactly science in the sense that it’s being published in a dusty academic journal that only 15 people in the world are going to read but I’m using the same principal. And by using the results of more than one poll we might get closer to an “accumulated wisdom”.

(The obvious thing would be to look at a poll from DavidBowie.com but I could only find the poll for albums released after 1980. I obviously missed the previous poll so if anyone has the results for that I’d be grateful if you could let me know [Footnote 1].)

BowieWonderWorld.com (BWW)
On June 14th 2005, www.BowieWonderWorld.com closed its polls for the top 30 Bowie albums. In fact, once they had identified the Top 5, voting continued for these five albums until a number one could be established definitively. As I recall, the travesty was that 1.Outside just pipped The Rise and Fall of Ziggy Stardust and the Spiders from Mars at the last minute on this second round of voting.

No matter. I’ve taken the poll at the close of voting on June 14th as the chart position to use because re-voting on a small sub-sample limits choice and does not reflect the free voting when the whole sample of albums can be selected. And anyway, the best kind of analysis to do for this chapter [Footnote 2] means we need to compare the top 5 with the rest and so it’s not important to know exactly where an album finishes in the top 5, just that it is in, rather than outside, the top 5.

The full list includes four live albums but these aren’t included in the analysis I’m presenting here because they are largely (but not entirely) made up of songs from the studio albums.

RateYourMusic.com (RYM)
There is another website called RateYourMusic.com. The thing about this site is that is doesn’t ask people to vote for their favourite album. Instead it asks people to give a score for all the albums they feel like rating. Some albums have thousands of people giving ratings while for others only a few hundred bothered. This is useful, though, because it doesn’t force a choice. People can, in theory, given their views on all of Bowie’s albums, not just pick out one. I took the ratings as they stood over the summer of 2008.

Although the way this poll is carried out differs from BWW, once we’ve identified those five albums with the highest average ratings, three of them also appear in the Top 5 for the BWW poll. These are The Rise and Fall of Ziggy Stardust and the Spiders from Mars (which had the highest score), Low (which had the third highest score) and Station to Station (which had the fourth highest score).

The final chart position for Bowie albums on BowieWonderWorld.com (BWW) and RateYourMusic.com (RYM) is given in Table 1. [Footnote 3].

Table 1. Chart position for Bowie albums on BowieWonderWorld.com and RateYourMusic.com

The 26 studio albums do not appear in exactly the same position in both polls but they are highly correlated [Footnote 4]. Several of Bowie’s later albums (Earthling, 1.Outside and Reality) appear in the Top 10 of the BWW chart when in the previous chapter they all appeared in the bottom five (based on commercial chart success). On the other hand, all of the top10 albums in the RYM poll are from the 1970s.

Table 2 gives the correlations between an album’s position in the two polls and the number of weeks in the charts and highest chart position in the UK and US.

Table 2: Correlations between album position in polls and chart success in the US and UK

The position in the BWW poll does not correlate with chart position or the number of weeks spent in either the UK or US charts [Footnote 5]. The position in the RYM chart is also not significantly correlated with chart position in the US or UK or with the number of weeks in the US chart. However, it is correlated with the number of weeks spent in the UK charts. In other words, the people voting on RYM are probably more representative of the general record/CD buying public than the people voting on BWW. This is to be expected since RYM is a general site for all music while BWW is a Bowie fan site and so is inevitably visited (almost) exclusively by Bowie fans.

LIWC analysis
For the BWW poll, if we compare the songs on albums in the top 5 with songs on all the other albums there are four key differences. Songs from albums in the Top 5 use fewer negating words, fewer motion words, fewer school words and fewer words relating to grooming. As you can see, the absolute proportion of words for negating and motion are greater than those for school and grooming. However, these results tell us what BWW visitors don’t like in Bowie’s albums but not what it is that they do like, at least not in terms of linguistic processes.

Turning to the RYM votes, songs on albums in the top 5 use more references to the third person (e.g. he, she, they) but fewer negating, school, death and body words. While we could speculate on the social nature of references to other people and the negative emotions reflected in the use of death and body words (either depression or disgust … see the previous chapter for a discussion on this), the only common category to both the BWW and RYM analyses is that of negation. I’ll say more about this below.

Celebrity fans
In May 2008, Uncut asked celebrity Bowie fans to pick out their favourite Bowie songs. There were thirty altogether although because four of these tracks were only on singles and not on any of the studio albums I’ve excluded them from the analysis (many of these songs have appeared on compilation albums, I hasten to add, but this would be a whole other analysis … I may get around to analyzing singles, compilations and songs written for other people another time). The reason I’ve kept this analysis separate from the above where I looked at BWW and RYM together is that for those analyses BWW and RYM looked at albums whereas Uncut looked at individual songs.

Anyway, here are the results of that analysis comparing the songs selected by celebrities with all of the other songs on Bowie’s studio albums which didn’t make it into their list.

Use of words relating to “I” appear less often in celebrity picks while social words and references to other people appear more often. This is broadly consistent with the greater use of 3rd person pronouns (he, she, his, her etc) in RYM since, although the specific categories are different, the underlying social/non-self theme is still there. On the other hand, this isn’t reflected in the BWW poll at all.

Less reference to feeling (such as itching and rubbing) in celebrity picks is also in the same ball park as fewer references to the body in the RYM ratings although, again, this is not reflected in the BWW poll.

At a stretch one could even make a case for less reference to religious themes in celebrity picks being consistent with fewer references to death in the top 5 albums from RYM since death and religion words are both categorized as “metaphysical” in the LIWC. However, this is a bit far-fetched and, in any case, yet again, this is not reflected in the BWW poll.

Replication
Probably the most interesting thing is that one word category crops up in all three analyses. That of negating words.

As I said earlier, received knowledge or wisdom is not based on the results of a single study but, rather, on the accumulated results of a number of studies. Here we essentially have three separate sets of data all pointing to the fact that Bowie fans (be they celebrity or “ordinary people”) tend to favour albums and songs which are expressed in positive rather than negative terms. This one category is replicated across all three studies and so we can be pretty confident that this is a genuine result and not one that was obtained purely by chance.

Negating words include not only the word “negative” but also words that negate otherwise positively expressed words. For example, “not” as well as the “n’t” suffix that turns “would” into “wouldn’t” and should” into “shouldn’t” also counts, as do words like “no”, “never”, “neither”, “zero”, “nowhere” and “without”.

So it’s important to be clear that negating words do not make an album negative in the sense that it is depressing or pessimistic. If that were the case, Low wouldn’t appear in the Top 5 of both the BWW and RYM polls. It means only that statements are made negatively rather than positively. To give an example from a song with one of the highest percentages of negating words, “I Can’t Read” from Tin Machine includes the line “I just can’t get it right”. This could be expressed positively as “I keep getting it wrong”. It’s still a negative message but in this form it is expressed positively and contains no negating words.

The three songs with the most negating words are "Don't Sit Down" (13.6%), "I Can't Read" (12.9%) and "You Can't Talk" (12.7%) while there are 31 songs that have no negating words at all. These are listed in Table 3 along with the album on which it appears. Where this album appears in the Top 5 of either the BWW or the RYM polls (or both) it is written in red.

Table 3. Songs (by album) that contain no negating words at all

There are 7 albums that appear in the Top 5 of either the BWW poll or the RYM poll (or both) and these 7 albums contain 65 songs in total. Fifteen of these songs appear in the list above which means that 23% of the songs on these seven albums have no negating words at all.

That may not seem like a lot but it leaves 201 songs on the remaining albums, 16 of which have no negating words and appear in the above list. That is, only 8% of the songs that appear on albums that fail to be included in the Top 5 of either of these polls have no negating words.

Put another way, top-rated albums contain nearly three times as many songs with no negating words as do Bowie’s other albums.

So what does it mean?
So we have three separate sets of data that give a range of different word categories that differentiate Bowie fans’ favourite albums/songs from their least favourite. Amongst all these we have one word category that is replicated in all analyses, that of negating words.

But why do we prefer albums and songs that express things in positive ways rather than negative ways?

To begin to speculate on this perhaps it’s worth discussing the significance of negating words in psychopathology and behaviour in general. Negating words are used in suppression, of both thought and behaviour, and attempts at suppression are recognized within psychology to lead to an increased likelihood of engaging in the behaviour or having the thought than if suppression had not been attempted. It’s called “rebound”.

On the other hand, attempts at distraction are generally more successful where, instead of simply trying to remove a thought, the individual replaces one thought with another.

For example, if I tell you not to think about a white bear, the first thing you think of is a white bear. But if I tell you, instead of thinking about a white bear, to think about a green rabbit, the first thing you think of is a green rabbit. It has been proposed that this is caused by monitoring … if you’re told not to think about something then you have to monitor whether you’ve thought about that thing in order to know whether you’ve thought about it. And by monitoring whether you’ve had the thought or not, inevitably you’ve just had the thought (“Don’t think about a white bear!”, “Okay … erm, have I thought about a white bear yet? … Bugger!”)

Obviously white bears and green rabbits are a trivial example but this kind of process has been proposed to play a role in a range of behaviours and disorders:

• Pain – suppressing thoughts about pain during a cold pressor task (essentially sticking your hand in a bucket of very cold water) leads to an increase in the perception of pain in a later unrelated task
• Eating and other appetitive behaviours – suppressing thoughts about eating chocolate leads to eating more of it (the same is true for cigarettes and alcohol and even the perception of attractiveness of other men and women following suppression of thoughts about sex)
• Post-traumatic stress disorder – people try to suppress painful memories of trauma and this type of avoidance leads to intrusive and automatic thoughts and flashbacks
• Obsessive compulsive disorder – people have thoughts that are unpleasant, even abhorrent, such as the death of a loved one and sufferers are motivated to expel these thoughts (either because of an inflated sense of responsibility for the well-being of others where having the thought increases the likelihood of the negative event happening or possibly because of a thing called thought-action fusion where to have the thought is as bad as doing the act)

I wonder, though. Maybe it’s simpler than this. Maybe we are not put off by albums simply because they are expressed in this way just because it speaks to our own pathology. Maybe we are just put off albums where Bowie is in the frame of mind that leads him to write songs for a particular album that are largely negatively phrased.

I’m not saying here that we dislike negative albums because they are depressing - clearly we don’t dislike Low and “Heroes”. Nevertheless we seem to prefer Bowie to express the stories and the points he wants to make in positively worded statements rather than negative. Negating words are so limiting.

Don’t tell us not to sit down; tell us to stand up
Don’t tell us not to look down; tell us to look up
Don’t tell us you can’t read shit; tell us what you can read

Anyway, whatever the reason, the replication of negating words means that we can't all be wrong. Something significant is happening here.

Footnotes

Footnote 1: Although DavidBowie.com is the obvious place to go to find out the opinions of Bowie fans, there was a debate on a Bowie group on MySpace a couple of years ago which suggested an awful lot of Bowie fans weren’t members of DavidBowie.com because it cost money, many of whom seemed to resent the idea. So members of DavidBowie.com may not necessarily be representative of all Bowie fans – some fans are prepared to pay for their fandom and some aren’t so paid sites put of one section of fans while free sites wouldn’t exclude anyone.]

Footnote 2: Independent samples t-test

Footnote 3: For the BWW poll. David Live was at No. 16; Ziggy Stardust: The Motion Picture was at No. 22; Stage was at No. 23; Oy Vey Baby was at No. 28

Footnote 4: Spearman’s rho = .78, p < .001

Footnote 5: Using Spearmans rho