Skip to content

The Myth Of The Autobiographical Lyrics

Elliott Smith


Can we put a rest to the myth ‘it was all written in the lyrics’? I cannot count the times I have seen people interpreting the lyrics of an artist as a biography or worst as a prophecy! There was even someone proposing that Elliott Smith had named his first album Roman Candle because he released only 5 albums when he was alive and Roman candles ‘typically shoot 5 rounds off till they die’… Besides the ridicule proposition, we always read way too much in songwriters’ lyrics, and if I take Elliott Smith as an example it is because he may be the best example of this irritating trend since his tragic death. How many times have you heard that his last album was a long suicide note? And this, despite the fact that many of the lyrics of the songs were written several years before he died.

For what it’s worth I have started a small experiment using a free version of LIWC that you can use online. In case you are not aware, these Linguistic Inquiry and Word Count algorithms are all the rage now and have even teamed up with Facebook to scan posts, comments, and videos of users for indications of immediate suicide risk. Not sure it’s a good thing in spite of the initial ‘good intention’, and experts have agreed that it is unclear whether the company’s new approach is either accurate, effective or even safe.

A lot of research is done regarding poetry, writing samples or song lyrics and what they can actually reveal about the writers, and I even stumbled on a study ‘Using Computational Text Analysis Tools to Compare the Lyrics of Suicidal and Non-Suicidal Songwriters’, considering Stuart Adamson, Adrian Borland, Kurt Cobain, Ian Curtis, Tom Evans, Pete Ham, Shannon Hoon and Phil Ochs for the suicide group and Paul Weller, Mark Smith, Chris Cornell, David Byrne, Joey Molland, George Harrison Scott Weiland, Tom Paxton for the non suicide group,… ironically the study was done before Chris Cornell’s death. if the conclusion of the four authors suggested ‘that suicidal songwriters use words of lower concreteness, fewer words, more future verbs, and fewer death-themed words in their lyrics’ – yes, the opposite of what you would logically expect – a lot of results were not significant: ‘Although there was a trend of suicidal songwriters using more self-references and fewer references to others, the distinctions were not significant. There were also no significant differences in frequency of emotion words, references to time, and communication words.’

This should close the debate, but to pursue the experiment, I have entered the text of 10 Elliott Smith’s songs on LIWC website, one by one, and collected the data that were available. The free version gives you 9 output plus the word count, whereas the actual LIWC2015 version actually produces about 90 different output dimensions, but I didn’t want to publish a thesis! I picked the 10 more popular songs on Spotify (‘Between the Bars’, ‘Say Yes’, ‘Angeles’, ‘Needle in the Hay’, ‘Waltz #2’, ‘Somebody That I Used to Know’, ‘Miss Misery’,’Son of Sam’, ‘Baby Britain’, ‘Pitseleh’) and added ‘Kings Crossing’ to have a song from his last album. Based on what was available to me, the results were given on a 100-point scale, where 0 = very low and 100 = very high.

For many dimensions (I-words, Social words, Positive emotions, Negative emotions, Cognitive processes, Analytic, Clout, Authenticity and Emotional tone), the results were all over the place for the songs considered, but this is what I got:

– The use of I-me-my words went from 1.7 to 12.5 with an average of 7.4 +/- 3.2 whereas the used of Social words (that make reference to other people, e.g., they, she, us, talk, friends) was higher, going from 7.8 to 17.8 with an average of 13.8 +/- 3.1. This tends to show that Elliott was mostly talking about other people although it varies a lot from song to song. LIWC says that people who use a high rate of self-references tend to be more insecure, nervous, and possibly depressed, but they also tend to be more honest, whereas people who use a high level of social words are more outgoing and more socially connected with others…the higher average for Social words surely goes against the preconceived idea of the always depressed Elliott Smith.

– The positive versus negative emotions is interesting because the two averages are almost equal, making an almost perfect balance with 2.9 +/- 1.4 for positive emotions and 2.9 +/- 1.9 for negative emotions! Of course, positive emotion words are correlated with a more optimistic vision, whereas negative emotion words are weakly linked to people’s ratings of anxiety or even neurotic. So again, nothing here that would exacerbate feelings of depression or despair.

– Cognitive processes were relatively low with an average of 11.0 +/-3.7 and Analytic (analytical thinking ) was high (with a large variation) and an average of 43.6 +/- 32.7. The LICW website suggests that this component is linked to people who perform better in college whereas low analytical thinking suggests people using language in a more narrative way, focusing on the here-and-now, and personal experiences,… which seems to correspond to Elliott, who had a college degree and was not writing songs in a storytelling tradition.

– Clout is an interesting component because it had an average of 70.6 +/-23.8, so a high level, and when you know that ‘Clout refers to the relative social status, confidence, or leadership that people display through their writing or talking’.., it is almost laughable, especially when you know that Trump’s speeches get the highest clout rank among presidents. Elliott, who was known for his low self-esteem and humble demeanor in real life, gets this high Clout score, and this may be the best demonstration that he was not talking about himself in most of his songs, after all, the military imagery is a recurrent theme in his songs.

– Unsurprisingly, Authenticity is pretty high with an average of 77.3 +/- 19.4. This is linked to the way people show themselves in an authentic or honest way, revealing more personal, humble, and vulnerable side.

– Lastly, Emotional tone had an average of 34.2 +/- 31.0, which means it is all over the place going from 1 (Miss Misery) to 96 (Waltz #2)! The algorithm is built so that the higher the number, the more positive the tone, whereas numbers below 50 suggest a more negative emotional tone. Again this suggests a large range of emotions depending on the songs.

Of course I would take all this with a grain of salt, but it could demonstrate a few things: the large range of emotional tones, the confidence  (Clout) and the higher use of social words versus I-words in particular strongly suggest that Elliott Smith’s songs were not autobiographic for the most part, without losing any of their authenticity. This should close, once for all, the debate about the so-called self-fulfilling prophecy.

Leave a Comment


Support Let Me Help Inc by shopping at

Scroll To Top