Linguistic Features Characterizing Metal Lyrics

I was taking a look at my lonely blog today & noticed that it’s been since the middle of September that I last made a “real” update, so I decided to at least make a preliminary analysis of language in metal music. This seems like a good idea because I’ve thoroughly exhausted myself making revision after revision to my statement of purpose for my upcoming grad application deadlines. I need a mental break & I want to listen to some loud music.

If you remember way back to September (or if you scroll down a little ways), you’ll find an entry I wrote about Pronouns & Personality that was based on Dr. Pennebaker’s investigation of the relationship between the use of function words & our affective states. The theme of this blog, epic language in metal music, seems like an apt time to revisit his research. Specifically, I will be using his text analysis software LIWC (the free version available on his website because I’m poor), to measure emotional & some structural aspects of metal lyrics. Read more about how it works here.

The software was developed to measure changes in writing style to track improvements in mental well-being — the idea is that you see an increase in positive emotion words, a decrease in negative emotion words, & an increase in cognitive or “thinking” words when comparing pre- to post-therapy writing samples. Obviously this has nothing to do with metal music, so why am I doing this? Well, I’ve been listening to more of it as of late & noticed that the lyrics have a sort of ineffable “epicness” to them & I’m hoping that this text-analysis software can help me better understand it (in short, I’m a huge nerd & this is what I do in my free time). It’s difficult to define what “epic” means in terms of lyrics, but I’ve had some time to think about it, & here are some features I’ve begun to vaguely outline:

  • An abundance of low-frequency words. Unfortunately LIWC doesn’t measure word frequency against a corpus, but it does measure “big words” (i.e., greater than six letters in length). After doing a few trial runs, it doesn’t seem like this will be an accurate measure of word frequency, so I’m going to choose the 8 “biggest” words from each song & check their frequencies using COCA. Maybe when I’m not poor I’ll buy the full version of LIWC & upload a dictionary from COCA.
  • A theme that may be actually epic in nature. There’s not a way I can use LIWC to give me an “epic quotient,” so you’ll just have to judge for yourself/trust me. See the OED definition of epic below to know what I mean:

3. fig. A story, or series of events, worthy to form the subject of an epic.

  • Brutal or angry lyrics with themes of violence. I’m going to predict that the lyrics will feature a higher rate of negative-emotion words.

So I picked a couple of songs that I felt were particularly representative of “epic” language in metal music. You’ll get to hear Mastodon’s Blood and Thunder & Quintessence, & the Human Abstract’s Digital Veil. And of course I had to pick a few controls. I decided to go with music from my favorite artist, Porcupine Tree. First because they’re not metal so you won’t be judging me for listening to ridiculous music. Second because they’re progressive, which I believe is also a genre that carries with it some of the same features as metal music without being necessarily epic. The other control I chose is Rebecca Black’s Friday because it’s terrible music & that’s all there is to it.

First, I’ll present you with my rough word-frequency analysis. Next to each of the eight words, you’ll see in parentheses the number of hits in COCA. I promise I didn’t make any post-hoc changes to the words chosen to fit my hypothesis that metal lyrics contain more low-frequency words. Then I’ll give my brief analysis of the song’s theme & just exactly how epic I find it to be. Finally, I’ll take a look at the output from the LIWC analysis to determine whether or not metal lyrics contain more negative-emotion words than my controls. I’ll do this for each of the four songs & hopefully we’ll be able to come to some conclusions in the end.


Split your lungs with blood and thunder
When you see the white whale


Remorseless (139), immersed (1566), oars (516), ivory (3543), prevail (2701), propels (312), harpoons (68), brow (3236).

Total Word Frequency Quotient: 1510.125


This song is so epic in nature that is in fact what inspired me to write this blog on the linguistic features of metal music with special attention to “epicness.” You might be able to guess that it’s about Moby Dick, the white whale. As Mastodon notes, “the fight for this fish is a fight to the death.”


LIWC dimension Your data Personal texts Formal texts
Self-references (I, me, my) 9.30 11.4 4.2
Social words 9.30 9.5 8.0
Positive emotions 0.00 2.7 2.6
Negative emotions 3.88 2.6 1.6
Overall cognitive words 7.75 7.8 5.4
Articles (a, an, the) 6.20 5.0 7.2
Big words (> 6 letters) 12.40 13.1 19.6

First I’ll note that I’m more interested in comparing “your data” across each of the songs and between metal & my controls. So here are my variables of interest for Blood and Thunder:

Positive Emotions: 0.00

Negative Emotions: 3.88

Big Words: 12.40


Punching these holes in my head
Losing my skin to the landslide


Demon (2339), mist (3818), omnipresence (111), primal (1222), instincts (2871), quintessence (123), landslide (996), farewell (2442).

Total Word Frequency Quotient: 1740.25


This song is a bit more surreal than most metal lyrics might be (completely my judgment based on not much). It seems to be about astral projection or dreaming in general. But I’d still contend that ideas of letting go of yourself, space, & reality are together a little more epic than what Britney Spears sings about.


LIWC Dimension Your
Self-references (I, me, my) 5.66 11.4 4.2
Social words 5.03 9.5 8.0
Positive emotions 0.63 2.7 2.6
Negative emotions 2.52 2.6 1.6
Overall cognitive words 5.03 7.8 5.4
Articles (a, an, the) 5.66 5.0 7.2
Big words (> 6 letters) 21.38 13.1 19.6

Positive Emotions: 0.63

Negative Emotions: 2.52

Big Words: 21.38


Have we all become voyeurs
covered by this digital veil?


Gallows (392), fiber-optic (173; fiberoptic), compute (755), mediocrity (571), fantasy (8305), voyeurs (106), veil (2541), comprehend (2333).

Total Word Frequency Quotient: 1897.00 (without fantasy: 981.57)


Digital Veil has a definite theme to it: the negative influence of technology on our (social) lives. While this doesn’t fit the traditional definition of epic as you find in the OED, it certainly does express a lot of emotional force, frustration, & anger. Hanging from “the gallows of this fiber-optic nation” doesn’t sound very pleasant. So I don’t find this song to be epic in any other way than its expression, but it does exhibit other features I find to be typical of metal lyrics. One additional thing that I want to note is the abundance of imperative structures in the lyrics; this is another trend I’d be interested in examining, but it is a bit beyond the amount of work I’m ready to dedicate right now.


LIWC dimension Your data Personal texts Formal texts
Self-references (I, me, my) 12.88 11.4 4.2
Social words 13.64 9.5 8.0
Positive emotions 0.76 2.7 2.6
Negative emotions 1.52 2.6 1.6
Overall cognitive words 9.85 7.8 5.4
Articles (a, an, the) 6.06 5.0 7.2
Big words (> 6 letters) 21.21 13.1 19.6

Positive Emotions: 0.76

Negative Emotions: 1.52

Big Words: 21.21


Moments are drowning in mantras of rain
Smiling unsmiling there’s no need to explain


Vistas (965), engulfing (253), waxing (605), nil (471), trigger (7265), inclined (4528), mantras (143), unsmiling (267)

Total Word Frequency Quotient: 1812.125


I don’t think anyone would argue that this song sounds epic or has epic lyrics. There’s not much else to say than that. Porcupine Tree songs usually feature psychedelic/surreal lyrics & I think this song is a good representation of that. I could go ahead & try to determine what linguistic features define psychedelic lyrics, but not now.


LIWC Dimension Your
Self-references (I, me, my) 2.44 11.4 4.2
Social words 3.66 9.5 8.0
Positive emotions 1.22 2.7 2.6
Negative emotions 1.22 2.6 1.6
Overall cognitive words 7.32 7.8 5.4
Articles (a, an, the) 12.20 5.0 7.2
Big words (> 6 letters) 29.27 13.1 19.6

Positive Emotions: 1.22

Negative Emotions: 1.22

Big Words: 29.27

REBECCA BLACK – FRIDAY from The Internet

Got to get down to the bus stop
Got to catch my bus, I see my friends


Fresh (39 796), rushing (7348), weekend (31 204), highway (16 702), cruising (2518), kicking (4818), afterwards (4338), partying (967).

Total Word Frequency Quotient: 13 458


Not epic.


LIWC dimension Your data Personal texts Formal texts
Self-references (I, me, my) 5.57 11.4 4.2
Social words 9.35 9.5 8.0
Positive emotions 8.69 2.7 2.6
Negative emotions 0.67 2.6 1.6
Overall cognitive words 3.12 7.8 5.4
Articles (a, an, the) 6.24 5.0 7.2
Big words (> 6 letters) 22.27 13.1 19.6

Positive Emotions: 8.69

Negative Emotions: 0.67

Big Words: 22.27


So I feel like I’m cheating by using Rebecca Black’s Friday for my control (definitely cheating), so I’m going to analyze a guilty-pleasure song of mine that is much more representative of popular music.

Watch me on your video phone, on your video, video
If you want me you can watch me on your video phone


Approach (55 910), convince (8810), cologne (1085), hustlers (259), film (55 279), cameo (620), hollering (599), popping (2525).

Total Word Frequency Quotient: 15 635.87


Again, these lyrics are nowhere near epic. However, actually looking at the lyrics did remind me of a paper talk given at the NorthEastern Evolutionary Psychology conference last year at SUNY-Binghamton called Songs as a Medium for Embedded Reproductive Messages. They analyzed the content of charting music from R&B, Country, & Pop & found that, surprise, the most popular, best-selling music has more reproductive messages. Anyway, that’s entirely tangential, but still interesting. The full article by Dawn Hobbs & Gordon Gallup Jr has since been published in the Evolutionary Psychology journal & you can find it here if you care to read more.


LIWC Dimension Your
Self-references (I, me, my) 12.57 11.4 4.2
Social words 23.70 9.5 8.0
Positive emotions 3.05 2.7 2.6
Negative emotions 0.72 2.6 1.6
Overall cognitive words 6.10 7.8 5.4
Articles (a, an, the) 3.05 5.0 7.2
Big words (> 6 letters) 6.28 13.1 19.6

Positive Emotions: 3.05

Negative Emotions: 0.72

Big Words: 6.28


I’m never one to be satisfied by descriptive statistics, so I ran a simple independent-samples t-test to see if there were significant differences between metal lyrics and pop lyrics (don’t worry, I only looked at Beyoncé) among the variables of interest – word frequency, positive emotion words, negative emotion words, & “big words.” Of course I’m only comparing lyrics from 3 metal songs of my choosing & 1 pop song, so this isn’t really all that compelling, but it’s still a fun way to spend the day after Thanksgiving. Also, though I gave two word frequency quotients for Digital Veil, I used the greater statistic in data analysis.

As expected there was a significant difference between metal lyrics and pop lyrics in their word frequency quotients, t(2) = -31.669, p = .001. So out of the eight words chosen from each song, those from metal lyrics had significantly fewer hits in the Corpus of Contemporary American English. The only other significant difference was in LIWC’s positive emotion score. Beyoncé’s Video Phone had a significantly higher positive emotion score than the average score for the three metal songs, t(2) = -5.511, p = .031. I will also mention that Rebecca Black’s Friday had a very high positive emotion score, so had I included her song, this difference would have been even greater. Maybe that’s part of what adds to the absurdity of her music.

Perhaps it means good things for my mental state that there was no significant difference between metal lyrics and pop lyrics in their negative emotion scores, t(2) = 1.404, p = .296, or maybe it means I need to do a more in-depth analysis. And as I had suspected, the “big word” score was not at all predictive of word frequency, so I’m glad I decided to rely on COCA to test that hypothesis.

Finally, when I compared Porcupine Tree’s lyrics to metal lyrics, I found no significant differences on any variable. So perhaps there is more in common between progressive/psychedelic rock & progressive metal lyrics than I would have previously guessed! It would at least contribute to an explanation of why I enjoy both so much. At the same time, I don’t see Porcupine Tree’s lyrics being at all epic in theme, but this is of course a subjective judgment based on a pretty fuzzy definition.


Doing this analysis did a bit in the way of answering my question of whether or not there are quantifiable linguistic features that define the epic nature of metal music. I think one of the most defining characteristics would be its use of low-frequency words (especially when compared to pop music). One thing that this analysis lacks is of course an explanation for that phenomenon. What do these low-frequency words lend to the epicness of the lyrics? Or is it instead more a matter of the theme? That is, battling Moby Dick is not what one would consider a quotidian occurrence & neither is the vernacular detailing it. In any case, this was a lot of fun & I may even try to gather more data to take a deeper look. Rock on, everyone!

Megan L. Risdal



Filed under Linguistics

2 responses to “Linguistic Features Characterizing Metal Lyrics

  1. I’m so glad you are blogging again. I loved this post!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s