Song Lyrics and “Normal Prose”
In general, song lyrics consist of prose. They are written in a natural language (although mixed languages are permitted). They are constructed from words, the words are put together into sentences, the sentences are syntactically valid, the sentences usually have discernible meanings, and the sentences are strung together in a manner that usually makes some kind of sense.
If song lyrics are “just” prose, then is there anything special about the content of song lyrics?
Are song lyrics prose, that just happens to be accompanied by music that somehow “matches” it – end of story?
Or, is there something else going on in the verbal content of song lyrics, that makes them different to other types of prose?
On the “Meaning” of Song Lyrics
Sometimes people ask “what is the meaning of such-and-such lyrics?”, and the implication is that the lyrics for that particular song have a meaning that is not completely obvious.
Typically, in these cases, the listeners have already decided that they like the song – because usually no-one is interested to talk about the lyrics of a song that they don’t like.
They like the song, but, apparently, they are not sure what the true meaning of the lyrics is.
So our ability to enjoy a song apparently does not depend on understanding what the lyrics mean.
We might think this implies that the lyrics don’t matter – that we could happily listen to music with no lyrics at all. But most of the musical items that most people actively listen to are songs. Almost all of the music on modern popular music charts consists of songs.
So it’s like the song has to have lyrics, even though it doesn’t quite matter whether or not the audience of the song can fully understand what the lyrics are supposedly saying.
In the case of “normal” prose, ie any spoken or written language that isn’t song lyrics, people do not usually tolerate prose where they cannot understand what they are hearing or reading.
This indicates that there is some kind of difference between how we listen to songs and how we listen to normal spoken language, and there is some significant difference between the criteria for what is a “good” song and the criteria for what is “good” prose. And in particular, in the case of song lyrics, the criteria for what is “good” doesn’t necessarily include knowing what the lyrics actually “mean”.
Line-by-Line Mode
I would now like to propose a hypothesis, based on my own personal analysis of song lyrics, in particular song lyrics that I personally judge to be effective in the context of the actual songs as I listen to them.
(I will add a side-note here: because music itself is such a subjective phenomenon, in my opinion it doesn’t make any sense to attempt to analyse any aspect of music in relation to music that you yourself don’t think is musical.)
The main idea of the hypothesis is that when we listen to song lyrics, we are simultaneously listening to the lyrics in two different modes, and these modes can be identified as:
- Prose mode
- Line-by-Line mode
Prose mode is the mode where we are processing the content of song lyrics the same way that we would process the lyrics if they were just spoken outside of any musical context.
If prose mode was the only mode, then song lyrics would just be prose.
It’s the line-by-line mode that changes the criteria for what counts as “good” song lyrics.
In prose mode, we can look at the body of prose, ie all of the lyrics, and we can ask: “what does this mean?”, ie “what do these lyrics mean”?
In line-by-line mode, the listener answers this question separately for each line, ie:
- Given this line, what does it mean?
Of course we can break down the processing of lyrics in prose mode into a line-by-line processing, but then the question is:
- Given the context of what has been said so far, what does this next line mean?
The difference between the two is context. In the case of line-by-line mode, there is no maintenance of the on-going evolving context that you get when procesing normal prose. There may be a general background theme, but this will be something fairly abstract and reasonably constant. For example, in a popular song, the theme might be “I’m worried that my girlfriend is cheating on me”, or, “All the other boys have girlfriends but I don’t” – and all the lines of the song will be evaluated in line-by-line mode against that same unchanging context.
In many cases a line from the lyrics in a song won’t even be a complete sentence, so the meaning of the line won’t even be the meaning of a sentence - it will just be a set of possibilities for what the complete sentence could be.
The other important thing about line-by-line mode is that not only should each line considered separately have an identifiable meaning, but also this meaning should have an emotional impact.
That is, each line, considered separately, should have an emotional impact.
This is different to “normal” prose, where even for types of prose where the author is aiming for an emotional impact on the listener (or reader), it is not necessary for every line of prose to have its own emotional impact.
For example, many of the lines in ordinary prose may serve a function to move the plot forward, and to set up the context for the emotional impact of something that will, eventually, be revealed.
With song lyrics, setups with deferred revelantions are not so much of a thing, and it’s more usual for the lyrics to consist of a relentless sequence of individual emotional revelations.
What is a “Line”?
In stating this hypothesis, I have glossed over one question, which is: what counts as a “line” of a lyric?
Song lyrics are typically written down in lines. If you go to a well-known song lyrics website, the song lyrics will be presented in the form of a sequence of lines.
But how well-defined is the division of song lyrics into lines?
For any given song, how do we know that we have “correctly” broken the song lyric into separate lines?
A similar but different question relates to how normal spoken language is made up of words and sentences.
All spoken human languages have “words” and “sentences”. Individual sounds (or phonemes) are grouped into words. Words are grouped into sentences. (In modern written language, sentences are grouped into paragraphs, but this is not a fundamental feature of spoken language.)
Any native speaker of a language has a strong intuition about what constitutes one word, and what group of words constitutes one sentence. (There can be quite a lot of variation in how different languages form words and sentences.)
For any given language, if we try to formally define exactly what is a word and what is a sentence, we will encounter quite a few complexities and subtleties.
One important aspect of both word formation and sentence formation is that in both cases they can be formed from incrementing or joining together smaller examples of the same thing.
So a word can be extended to make a new word by adding a prefix or a suffix, and a new word can be made by joining two words together.
There are various ways of joining sentences together, ie mostly with conjunctions or conjunctive phrases. (In English there is the famous “Oxford comma”, which is deemed officially incorrect, but people do it, so it is a thing.) A sentence can also be contained within another sentence, where the inner sentence is the subject or object of the full sentence.
So, for example, if we form some hypothesis about how the brain processes words or sentences, such a hypothesis would have to include a description of how that includes the processing of words or sentences that contain other words or sentences (respectively) as internal components.
On the other hand, if we want to test a hypothesis about the processing of words or sentences, we don’t need to depend on any formally specified criterion for what counts as a word or a sentence – it’s sufficient to rely on the intuitions of native speakers of the language in question, for the purpose of choosing a set of valid words or sentences that you are going to use in your experiment or in your analysis.
If I want to extract words and sentences from a body of existing text, then the work is already done for me, because whoever wrote the text has already applied their own intuitions to determine what the words are and what the sentences are, and all I have to do is read the space characters as word separators and the fullstops as sentence terminators.
The equivalent approach with lyrics is that we go to a well-known lyrics website, and in practice that site will display song lyrics as a sequence of lines, and that is probably a good enough answer in most cases, for most songs, to the question of how one should divide the lyrics into lines.
One can also apply various heuristics, which, a lot of the time, give us a “correct” divisions into lines, especially in the case where the different heuristics all agree with each other:
- A line will be terminated by a rhyme
- The different lines will be a similar length to each other.
- Each line corresponds to one melodic phrase. This of course then raises the question of “What is a melodic phrase?”.
Melodic phrases can be identified by their own heuristics:
- Different phrases will be separated by a gap in the melody
- The different phrases will be a similar length
- Different phrases will have noticeable similarities, where one phrase will subjectively be heard as some kind of alteration of a previous phrase
- Sometimes sequential phrases are actually identical (ie a special case of the previous heuristic, with zero alteration).
Empirical Verification
To understand how line-by-line mode works, you have to actually pick song lyrics that “work” for you, in a song that “works” for you, and then think about:
- What, if anything, is the overall theme of the song lyrics?
- For each line, considered entirely by itself, does it have emotional impact?
My claim is that you will find, when you do this, that most (if not all) of the lyric lines in a “good” song will have an emotional impact, even when considered independently of any specific context created by all the other lines in the song.
Observed Variations
My own observations so far are that quite a lot of song lyrics follow this hypothesis for most of the lines, but with some variations:
- Sometimes most of the lines have self-contained emotional impact, but there may be the occasional line that doesn’t.
- Sometimes there is a “clever” line in the lyrics, which, considered by itself, is the least emotional line – because the interpretation of that line is more dependent on everything else in the song.
- Sometimes the emotional impact of one line does depend on context provided by the immediate preceding line, but not more than that.
If indeed there is some dependence on context provided by a previous line, one possible interpretation of this is that processing in line-by-line mode is not completely “context-free”, but can rather be characterised by more limited or simplified context.
In other words the listener is not maintaining a context based on all the details revealed by a complex plot structure, but, at the same time, what has already been said already – or what has just been said previously – may give a very rough indication of what the current line is about, to a sufficient extent that the current line has the desired emotional impact on the listener.
An example
”My momma told me slow down” by Citycreed
https://www.tiktok.com/@citycreed/video/7569778621653519671
This is a reasonably straightforward example of a song where one can break the lyrics into lines, and, on the one hand, the lyrics can be read and understood as coherent prose, but at the same time, each individual line has an obvious emotional implication when considered by itself (ie without reference to other specific lines in the song), especially if considered against the general background theme, which might be identified as something like:
- “My mother is sufficiently concerned that maybe I’m not coping well with my life situation that she feels the need to give me some advice about it.”
The actual lyrics, as I would break them up into “lines” -
- My momma told me Slow Down - Learn to take it slow
- Cause acting like you’re fine - don’t make the hurt a lighter load
- Boy I can hear the pain you hide - it’s slurring through your words
- If you’re drinking just to feel again - you’ll only make it worse
- But Mama I’ve tried and nothing’s healing
- I’m out here barely breathing
- and I can’t take this feeling anymore
In this particular case there is one significant aspect of context, which is to know which lines are spoken by the mother and which are spoken by the son.
Lines 1, 3 and 5 make it directly clear who is saying them, ie via “My momma told me”, “Boy”, “But Mama”.
One could argue that lines 2, 4, 6 and 7 all have sufficient emotional impact regardless of who you think is saying them, so this particular aspect of context doesn’t matter.
Or, one could argue that in the general context of “mum giving advice to son who is ‘hurting’” that it would be reasonably obvious who is saying each of lines 2, 4, 6 and 7, in each case (ie even if that was the only line you happened to hear).
Or, one could argue that this particular component of context is indeed passed from one line to the next where relevant – so it falls into the scenario mentioned above of limited or simplified context that is carried along from one line to the next.
Extra note:
- I’ve hyphenated the first four lines in the lyrics, because they can almost each be identified as a pair of lines each. However it’s hard to identify the individual halves as having emotional impact by themselves, also only the full length lines have the rhymes – apart from a possible weak rhyme between “fine” and “hide” - so my preferred division is into the 7 lines as listed.