The Rise and Fall of Causal Reasoning and Negative Emotion

Before I present these analyses I just need to give a little background information about what the Linguistic Inquiry and Word Count (LIWC) has shown in studies on writing in relation to health. Bear with me because this is relevant to the results.

In studies where participants are asked to write about trauma for 20 minutes a day for 3 consecutive days, the greatest health benefits are shown by those whose writing on Day 1 contains relatively few cognitive words but who increase their use of cognitive words by Day 3 (words like “because” or “although” imply that people are developing a causal understanding and are therefore deemed to be cognitive words) [Footnote 1]. People who start off using a lot of cognitive words and remain high, or who start off using very few cognitive words and remain low, don’t show any improvement in health. Pennebaker’s take on this is that health improvements come about as a result of developing a new understanding or gaining an insight into the trauma, an understanding or insight that they didn’t have before. This is what is sometimes called an “information processing” account.

Although Pennebaker’s studies get participants to write about a stressful or traumatic event they have actually experienced, other studies have shown that the same health benefits can be gained by writing about a trauma that the individual hasn’t actually experienced [Footnote 2]. These authors suggest that an information processing account can’t explain this – how can you gain insight for a trauma that you haven’t experienced? Instead, they suggest that health benefits derive from developing skills at dealing effectively with emotions. This is sometimes called an “emotion regulation” account. Other studies have shown that these health benefits can also be obtained by writing about positive experiences [Footnote 3] and about achieving life goals by writing about their best possible selves [Footnote 4].

In one of our own studies [Footnote 5] we looked to see what aspects of language could predict improved health in people writing about their best possible selves. It turns out that a similar picture emerges as for people writing about trauma except that in our study it was the increase in negative emotion words, rather than cognitive words, that predicted whose health would improve. In other words, people who started off writing very few negative emotion words in Day 1 but who were writing significantly more negative emotion words by Day 3 were most likely to experience improved health over the next three months.

This sounds a bit counter-intuitive, that writing about negative emotions would lead to improved, not worsening, health. However, what we found was that health improved when people wrote about mastering or overcoming negative emotions, not simply having them. It seems that they, too, were actively engaged in a process they hadn’t been engaged in before but, in contrast to the one described by Pennebaker, they were identifying and developing skills in regulating negative emotions rather than gaining an insight into a trauma.

The present analysis

So what we’re interested in here is what pattern of word use exists between the beginning and the end of each Bowie album. Do the psychological processes indicated by Bowie’s lyrics develop and change over the course of an album and, if so, how? What is their flow? Is there a psychological narrative?

To do this I assigned all tracks to one of five segments based on where they appeared on each album. This was a bit complicated because Bowie’s twenty six studio albums contain anywhere between 6 and 19 tracks so I couldn’t simply use the track number – for example, track 6 on Diamond Dogs (“Rebel Rebel”) is smack bang in the middle of the album while track 6 on Station to Station (“Wild is the Wind”) is right at the end and track 6 on 1.Outside (“Hallo Spaceboy”) is barely a third of the way through.

The idea was that I could explore differences between parts of albums to see how (or even whether) the psychological processes indicated by the lyric unfold over the course of an album. For those who really care about this stuff I have put a table in the notes at the end so you can see how I recoded all the albums [Footnote 6]. But all you really need to know is that I created five segments for each album: 1 is at the beginning, 3 is in the middle, 5 is at the end … and 2 and 4 are the intermediate parts between the beginning and middle and between the middle and the end respectively (obviously).

Studies on writing about trauma show that healthy writing increases in its use of cognitive words over the period of writing. On the other hand, studies on writing about best possible selves show that healthy writing increases in its use of negative emotion words over the period of writing. Songs are different in both the form and content of their writing than the two tasks participants were asked to complete in Pennebkaer’s and our studies. It’s also a free choice for Bowie, both in terms of what he writes as well as the order in which he puts the tracks on the album. So who knows what’s going to happen with this analysis - but it could still be interesting to see how the writing unfolds throughout the course of an album.

Results

There was only one genuinely significant difference between these segments for any of the word categories and that was for words indicating causal reasoning. Words in this category include words directly indicating causal reasoning such as “think” and “understand” as well as words that indicate that the process of causal reasoning has taken place such as “although” and “because”.

Segments 1, 2, 4 and 5 did not differ significantly from each other but that the middle segment (segment 3) was significantly higher than the other four segments [Footnote 6]

This is shown in the graph below.

This is averaging across all albums, of course. This means that, on average, it is the middle segment of an album where the main causal reasoning takes place, the part of the album where the real cognitive work is being done. There is a lot of variation so not all albums will necessarily show precisely this pattern. In fact three albums (David Bowie, Black Tie White Noise and Earthling) show completely the opposite pattern with the middle segment having the fewest causal reasoning words. These, however, are in the minority and the albums that show most clearly the pattern where the middle segments contain most causal reasoning words are Hunky Dory, Diamond Dogs, Never Let Me Down, Tin Machine and Buddha of Suburbia.

Next time you’re listening to these albums see if you can spot these changes.

The only other result that was interesting fell just short of what statisticians consider to be “significant”. It was still a strong effect but not quite as strong as that for causal reasoning. It’s what we call “marginally significant” and it was for negative emotion [Footnote 8]. Essentially there was a flat line for segments 1 through 4 and then an increase right at the end in segment 5. This is shown in the graph below.

In fact if we compare segments 1 and 5 directly we do get a difference that is considered statistically significant.

But again, this is averaging across all twenty six studio albums. Although overall there is a marginally significant increase in the use of negative emotion words from segment 1 through to segment 5, if we look at individual albums we get two distinct patterns.

Sixteen of the albums show an increase in the use of negative emotion words from the beginning to the end of the album. These albums are David Bowie, The Man Who Sold the World, Hunky Dory, Pin Ups, Diamond Dogs, Low, Scary Monsters, Let's Dance, Never Let Me Down, Tin Machine, Tin Machine II, Buddha of Suburbia, 1. Outside, Earthling, …hours and Reality. We’ll call these albums SET A.

However for the remaining ten albums there is a fall in the use of negative emotion words from the beginning to the end of the album (i.e. segment 5 contains fewer negative emotion words than segment 1). These albums are Space Oddity, The Rise and Fall of Ziggy Stardust and the Spiders from Mars, Aladdin Sane, Young Americans, Station to Station, "Heroes", Lodger, Tonight, Black Tie White Noise and Heathen. We’ll call these albums SET B.

When we combine the sixteen albums in SET A the increase in negative emotion words is statistically significant while, conversely, when we combine the ten albums in SET B the decrease in negative emotion words is statistically significant. This is shown in the graph below. The solid blue line represents the average change across all twenty six albums – it’s the result we’ve already discussed above but which we’ve now discovered is a bit misleading. The dotted red line with the squares represents the change in negative emotion words for albums in SET A (i.e. increasing from the beginning to the end of the albums) while the dotted red line with the triangles represents the change in negative emotion words for albums in SET B (i.e. decreasing from the beginning to the end of the albums).

So what can we conclude from this analysis?

Well, firstly, cognitive processes (specifically causal reasoning) tend to occur in the middle section of Bowie’s albums – there are exceptions to this but, on average, this seems to be the case.

Secondly, some albums start with few negative emotion words and increase towards the end while other albums start with more negative emotion words and decrease towards the end.

As far as this second finding is concerned, it begs another really important question.

SO WHAT? Some albums increase in their use of negative emotion words, some decrease! Big deal!

One speculation about why it might (and I mean “might”) be a big deal is that our study on people writing about their best possible selves [Footnote 4] found that people whose health improved wrote very few negative emotion words initially but gradually increased over the course of the study. So this seems to be a “healthy” pattern of writing.

The overall picture is that this is the pattern that Bowie follows and certainly the majority of his albums follow this pattern. However, we have no objective measure of Bowie’s health before and after each album, just stories of excessive drug and alcohol use and depression. There was also the relatively recent heart scare and the lollipop-in-the-eye incident. But all of these are difficult to date precisely in relation to writing or recording.

In any case, it’s impossible to know the order in which the songs were written, only the order in which they appear on albums. However, I think it is still quite intriguing that, when it comes to selecting the order of tracks, Bowie seems intuitively to follow this pattern.

However, there is another potentially interesting observation here. In the table below I list all twenty six studio albums, divided into whether negative emotion words increase or decrease over the album, and give the first and last songs of each album (The exceptions to this are (a) where the first or last track is a spoken-word narrative or an instrumental rather than a song, in which case I have gone for first and last actual songs and (b) where the same song appears at the beginning and end of the album as is the case for “It’s No Game” on Scary Monsters and “Buddha of Suburbia” on Buddha of Suburbia. Because the use of negative emotion words for these tracks would be identical it doesn’t really tell us anything so for these albums I’ve listed the second and penultimate songs. These exceptions are indicated with an * in the table).

The first songs on albums in SET A have fewer negative emotion words while the last songs on these albums have more negative emotion words. Conversely, the first songs on albums in SET B have more negative emotion words while the last songs on these albums have fewer negative emotion words.

I’m resisting the urge to make any qualitative interpretations of first and last songs because these would be entirely subjective. Even Bowie says he doesn’t quite know what it is he’s writing about much of the time and expresses surprise at some of the interpretations people make.

However, there are some impressions that occur to me. Regardless of whether negative emotion words increase or decrease, the parts of albums with the most negative emotion words seem to be those that are the least personal, whether they reflect alienation (individual such as “Space Oddity” or cultural/religious such as “Loving the Alien”), impending or threatened apocalypse (such as “Five Years” and “Fantastic Voyage”) or social or political control or corruption (such as “Big Brother”, “Reality” or “Goodbye, Mr Ed”). The tracks with the fewest negative emotion words seem to be stories about individuals (such as “God Knows I’m Good” or “Lady Grinning Soul”) rather than pairs or groups (such as “The Bewlay Brothers” or the mythical “The Supermen”). Tracks with few negative emotion words also seem to be those that are more intimate (such as “The Wedding Song” or “Wild is the Wind”) rather than overtly sexual (such as “Bang Bang” or “Watch that Man”).

To pick out another pair of first and last songs to make this difference between personal versus impersonal clearer, this time from Young Americans, the most negative emotion words are expressed in “Young Americans”, a song describing Bowie’s impressions of others (i.e. Americans) whereas the fewest negative emotion words are expressed in “Fame”, a song Bowie is writing about his own personal experiences of fame. This is doubly interesting because the experiences described in “Fame” are essentially negative so one might have expected more, not fewer, negative emotion words. (This is triply interesting if you consider that the hook is a rip-off of another song called “Shame” … you can’t get a more negative emotion word than that, it’s considered to be one of the most destructive emotions in terms of its effect on psychological illness such as depression and, as I’ve discovered in my own work, eating disorders).

In the end, while it might be difficult to put our fingers on exactly what the nature of the difference is between the beginnings and the ends of the albums in terms of negative emotions, the point is that there are differences. Negative emotion words either go up or they go down. If Bowie was simply deciding at random what the order of songs should be on his albums then on average there would be no change overall, either in terms of causal reasoning or negative emotion words. The fact that there are these systematic patterns of change suggests that the order of songs on albums is not random. Bowie is definitely creating a psychological journey in his albums and it’s not the same journey every time. I also suspect this is done intuitively rather than consciously.

Finally, given that the overall trend is for causal reasoning to increase in the middle of albums before returning to baseline levels and for negative emotion words to increase between the first and last segments of albums, is it possible to pick out one album that follows this pattern most clearly?

Actually it is. And that album is Diamond Dogs.

In terms of the psychological journey that Bowie takes us on in his albums, it could be said that Diamond Dogs is his archetypal album [Footnote 9].

When I first heard this I was in my uncle's car and I thought it was some guy trying to sound like Gary Numan. This is doubly silly because Gary Numan was ripping off the Bowie of the late 70s not the early/mid-70s. I spent the summer of 1983 listening to this and stripping paint off my uncle's windows (I still can't pass the smell of butane gas and burning paint without thinking of Diamond Dogs). I also read Orwell's 1984 off the back of this album. Anyway, these are personal recollections. All you need to know is that when Ryko Disk were re-releasing Bowie's albums in the early 90s they reinstated the dog's willy that had been airbrushed out with the original release.

Footnotes

Footnote 1: Pennebaker, 1993

Footnote 2: Greenberg (1996)

Footnote 3: King positive experiences

Footnote 4: King BPS study

Footnote 5: Winn & Troop, submitted

Footnote 6: Tracks included in each segment for albums with different numbers of tracks

	Segment 1	Segment 2	Segment 3	Segment 4	Segment 5
Albums with 6 tracks	1	2	3, 4	5	6
Albums with 8 tracks	1, 2	3	4, 5	6	7, 8
Albums with 9 tracks	1, 2	3, 4	5	6, 7	8, 9
Albums with 10 tracks	1, 2	3, 4	5, 6	7, 8	9, 10
Albums with 11 tracks	1, 2	3, 4	5, 6, 7	8, 9	10, 11
Albums with 12 tracks	1, 2, 3	4, 5	6, 7	8, 9	10, 11, 12
Albums with 14 tracks	1, 2, 3	4, 5, 6	7, 8	9, 10, 11	12, 13, 14
Albums with 19 tracks	1, 2, 3, 4	5, 6, 7, 8	9, 10, 11	12, 13, 14, 15	16, 17, 18, 19

Footnote 7: A Duncan’s multiple range test was used for the post hoc analysis

Footnote 8: Statistical significance is generally taken to be where there is a less than 1 in 20 probability that the result was obtained by chance. The value of interest is p (for “probability”) and, by convention, statistical significance is generally accepted when p < .05 (i.e. 5% or 1 in 20). When the p-value is just slightly larger than this, psychologists sometimes use the phrase “marginally significant” or state that there is a “trend towards significance”. In this analysis we obtained a p-value of .057 which is very slightly larger than .05. In general, psychologists will only discuss “marginally significant results” when there’s very little that’s genuinely significant for them to discuss. You might wish to argue that that’s what I’m doing here. However, as we’ll see from a more detailed exploration of the data, there is some very interesting information we can gather, even from this apparently “marginal” result.

Footnote 9: I know Bowie likes his Jungian references but it’s a while since I’ve read any Jung so I may be using “archetype” in a colloquial sense rather than as a strict Jungian would

Buy Bowie things

Sign up to the Newsletter

Go to the GADB.com shop

Visit NickTroop.com