The message

August 27, 2013 @ 7:14 am · Filed by under Computational linguistics, Language and culture, Prosody

« previous post | next post »

This year's Penn Reading Project book is Adam Bradley's Book of Rhymes: The Poetics of Hip Hop.  In my discussion group yesterday afternoon, several participants complained that some important things about the "poetics" of rap are lost in a purely textual presentation of the lyrics. One student observed that in pieces he knows, the rhythm is there in the written form — but the lyrics for pieces that he doesn't know seem flat and lifeless in comparison.

There are good reasons that this is more true for the works of Melle Mel or Jay Z than for Elizabeth Barrett Browning or W.H. Auden, I think.

One of the advantages of the weblog format is the combination of text, images, and audio or video clips, so for this morning's Breakfast Experiment™ I decided to present a small exploration of the "poetics of hip hop" in a multimedia — and somewhat quantitative — framework.

This exercise will clarify why transcriptions of the lyrics, even with bold-face indications of stress, are missing an important dimension. The lines' scansion depends not only on the syllable sequence and on where the performer puts phrasal stresses, but also on the alignment of the syllables with the musical meter. This alignment is not automatic or always obvious — it has artistically-relevant degrees of freedom beyond those available in most other genres of text setting.

For those whose appraisal of Bradley's book was (interpreting freely) "not enough vampires and car chases", this will probably make things worse — you have been warned.

Let's look at a piece from the classical period, Grandmaster Flash and the Furious Five's 1982 "The Message":

The background has a clear and steady beat. The metrical pattern that lines up with a line of the verse lasts about 2.4 seconds — one period of that pattern is shown below. (I've taken it to represent two bars of 4/4 time, though Jonathan Mayhew in the comments suggests that a more standard analysis would be transcribe it as one bar.)

What matters to the analysis is that if we track this metrical pattern through the times when Melle Mel is rapping, we find that he subdivides it into (a maximum of) 16 minimal time units, as in this couplet from the first verse:

The basic pattern of the musical meter is the usual binary hierarchy — to get the 16 units, the line is subdivided into 2, subdivided into 4, subdivided into 8, subdivided into 16.

Above, I've used a couple of Audacity label tracks for the musical meter and for the alignment of the words with the metrical line divided into 8 time-units, which I've notated as 1 through 4 and then 1 through 4 again. As usual, it's the syllable onset — the place where the amplitude is rising most rapidly, known to generations of phoneticians as the "P Center" — that aligns with a given metrical "beat".

Here's the same couplet presented in tabular form, with the off-beat positions (nominally eighth notes in the transcription I've assumed) labelled with "+":

+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1
I can't take the smell can't take the noise
got no mo- -ney to move out I guess I got no choice
+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1

The relation between musical meter and poetic form is more complex here than in (some types of) traditional verse setting. Among other complexities, we need to deal with syncopations of the kind that I discussed in the post "Rock syncopations: Stress shifts or polyrhythms?", 11/26/2007 — thus in phrase "can't take the noise" in the first line of the couplet shown above, the syllables "can't" and "take" are aligned with the offbeats that I've labelled 2+ and 3+, rather than the nominally more prominent positions 3 and 4. Is this just a stylistic syncopation, or is it something more fundamental?

Whatever the answer to this question, a kind of statistical shape emerges if we look at the overall syllable-to-musical-meter correspondence. Here's the whole first verse (of five) in the same format — four couplets, eight lines, 16 bars, 128 minimal time units:

+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1
bro- -ken glass ev- -ry- where
peo- -ple pis- -sing on the stairs you know they just don't care
I can't take the smell can't take the noise
got no mo- -ney to move out I guess I got no choice
rats in the front room roa- -ches in the back
jun- -kies in the al- -ley with a base ball bat
I tried to get a- -way but I could- -n't get far
cause a man with a tow truck re- -pos- -sessed my car
+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1

If we just add up the number of syllables in this verse, meter position by meter position — even without taking account of stress — we get a plot like this:

Except for position 2 in the first bar, which is clearly a kind of upbeat, an alternating binary strong-weak pattern does emerge. If we do the same thing for all five verses (74 lines in all) the pattern is confirmed:

However, in the chorus, something very different is going on. In the opening four lines — treated as a double-length couplet, with rhymes at the end of each pair of lines — each 16-count line is divided as 3|3|3|4|3. Along with Melle Mel's delivery, this shift into a polyrhythm — with the background maintaining the straight 4|4|4|4 — is a sort of metrical icon of the mental instability he's talking about:

+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1
don't push me cause
I'm close to the edge
I'm try- -ing not to
lose my head ha ha
ha ha it's like a jun- -gle some- -times it makes me won-
-der how I keep from go- -ing un- -der
+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1

Here's a graphical comparison of the syllable distribution in the chorus, compared with the syllable distribution in the first two verses:

And as the piece develops, the shift of the ends of lines to a 3|2|3 pattern

x     x   x     x
x x x x x x x x x x
2 + 3 + 4 + 1 + 2 +

becomes more and more common,

Here's a comparison of syllable-counts-by-position in verses 1-2 (24 lines) with verse 5 (28 lines):

For a good example of this effect, see the last two couplets:

+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1
it was plain to see that your life was lost
you was cold and your bo- -dy swung back and forth
but now your eyes sing the sad sad song
of how you lived so fast and died so young
+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1

You could see this polyrhythmic infiltration as a metrical symbol of the personal instability that the lyrics talk about.

As the genre develops, varied and complex polyrhythms become more common and are given a different interpretation, as in Eric B. and Rakim's 1987 "I Know You Got Soul":

+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1
Ra- -kim 'll be- -gin when you make the mix
I'll ex- -per- -i- -ment like a sci- -en- -tist
+ 2 + 3 + 4 + 1 + 2 + 3 + 4 + 1

But that's a story for another time.

August 27, 2013 @ 7:14 am · Filed by under Computational linguistics, Language and culture, Prosody


20 Comments

  1. Jonathan Mayhew said,

    August 27, 2013 @ 9:25 am

    Great post. The beats you label as 3s I take to be 1s. I'm curious why you don't hear those as the beginnings of the measure? Of course, I could be wrong. It wouldn't be the first time.

    [(myl) You might be right. But in this context, what really matters how the verses line up with the 16-unit two-bar sequence, and the choice I made has the lines ending in the 4-to-1 region (around the end of a bar) rather than in the 2-to-3 region (in the middle of a bar), which seemed more appropriate to me.]

  2. Dimitri said,

    August 27, 2013 @ 9:39 am

    I agree with Jonathan. The famous — iconic — keyboard riff starts at the beginning of each measure.

  3. Jonathan Mayhew said,

    August 27, 2013 @ 10:24 am

    Also, as I count it the tempo is about 100 beats per minute. There are about 25 bars in the first minute of the video, for example. You seem to be counting 8ths as quarters to get to 198bpm.

    [(myl) Whatever note denominations we use in a musical transcription, the 16-time-unit sequence that lines up with a line of the verse lasts about 2.4 seconds. If we take that to be two bars of 4/4 time, then that's about 2.4/8 = 0.3 seconds per quarter note, or about 60/0.3 = 200 quarter notes per minute. On that analysis (which is what I assumed), each of the 16 minimal-time-units is an eighth note.

    You're interpreting the same sequence as one bar of 4/4 time, I guess, in which case there are about 100 quarter notes per minute, and the minimal time unit of the 16-unit sequence is a sixteenth note.

    This doesn't matter to my analysis, since all that I care about is the alignments of the syllables with the 16-unit sequences. But if my (mis-?)assumptions about translation to musical notation confuse the picture, I apologize.]

  4. Jonathan Mayhew said,

    August 27, 2013 @ 10:42 am

    Right. It makes no difference to your analysis. It would just be confusing since it diverges from the standard way in which musicians or knowledgeable listeners would normally count.

    http://www.beatport.com/track/the-message-original-mix/1589435

    [(myl) OK, I've modified the presentation in a way that (I hope) relieves the confusion…]

  5. Ralph Hickok said,

    August 27, 2013 @ 11:01 am

    Without regard to the musical accompaniment, I've long felt that rap/hip hop prosidically resembles skeltonics, but i have neither the technical knowledge nor the ability to do an analysis.

    [(myl) Maybe so, in content as well as in form — one example of Skelton's work:

    Youre key is mete for euery lok,
    Youre key is commen and hangyth owte;
    Youre key is redy, we nede not knok,
    Nor stand long wrestyng there aboute;
    Of youre doregate ye haue no doute:
    But one thyng is, that ye be lewde:
    Holde youre tong now, all beshrewde!

    To mastres Anne, that farly swete,
    That wonnes at the Key in Temmys strete.

    Or another:

    Say this, and say that,
    His hed is so fat,
    He wotteth neuer what
    Nor wherof he speketh;
    He cryeth and he creketh,
    He pryeth and he peketh,
    He chydes and he chatters,
    He prates and he patters,
    He clytters and he clatters,
    He medles and he smatters,
    He gloses and he flatters;
    Or yf he speake playne,
    Than he lacketh brayne,
    He is but a fole;
    Let hym go to scole,
    On a thre foted stole
    That he may downe syt,
    For he lacketh wyt;

    ]

  6. J.W. Brewer said,

    August 27, 2013 @ 3:29 pm

    University administrators trying to appear hip-and-with-it-and-relevant by exposing freshman born circa 1995 to a text from the lost and irrelevant world of 1982 seems like something out of a David Lodge novel. Better to have them read Skelton.

    [(myl) Adam Bradley's book takes the story up through the time he sent the ms to the publisher, in 2008. I chose an old-timey example because the freshmen in my discussion group, born in 1994-5, were familiar with recent work but not with older stuff. As for getting 2500 Penn freshmen to read Skelton, good luck with that.]

    Back in the Sixities, I think the Uptight Grownups used to play a parlor game in which lyrics by the Beatles or Dylan or whoever the Young People made extravagant claims for the literary merits of could be solemnly intoned aloud with a po-faced affect, as if they were poems found in a dusty volume of Dickinson or Tennyson, in which context they would typically appear ridiculous. I expect that sometimes this was in fact because they actually were ridiculous as stand-alone texts – they worked in context only because of the musical setting or harmony or whatnot — but in others it was because the way the lyrics worked with the music meant they were sung with some sort of non-intuitive prosody or timing that was at odds with how you would just read them aloud left to your own devices, and thus worked better if you could figure out how to adjust your reading-aloud performance to the timing in the song. I do not think the Ur-rap of 1982 was particularly novel in this regard.

    [(myl) Jazz, blues, R&B, and rock have all had a similar — or even greater — amount of freedom in text setting, and so I'm sure that the effect I wrote about here applies to lyrics in those genres as well. For songs that are metrically more "square", there are obviously other reasons why being familiar with the music adds to the experience of reading or recalling the lyrics. Equally obviously, there are kinds of poetry that are created and interpreted mainly through the channel of text. This post was an attempt to illustrate one aspect of the difference.]

  7. AntC said,

    August 27, 2013 @ 4:25 pm

    @J.W.B. Perhaps you're thinking of Peter Sellers reciting "A Hard Day's Night" po-faced? I found (and find) it riotously funny.

  8. maidhc said,

    August 27, 2013 @ 4:41 pm

    A classic example of what J.W. Brewer is talking about is Steve Allen reading the lyrics to "Be-Bop-a-Lula". You can find it on YouTube. What I wasn't able to find on YouTube, though, is the answer. Phil Spector (I think it was) went on the Steve Allen Show and told Steve that he was missing the point because the point was the beat, illustrating his argument by banging loudly on the desk while singing the lyrics.

    Now that we have a historical perspective, I think Phil Spector (if he was the person) won that argument.

    There's a difference between strict poetic metre and song metre, because in poetic metre every syllable has to be accounted for, whereas in song metre it's the stresses that are important, while the number of unstressed syllables in between can vary. Although some poetry can work like that too.

    I'm a GET up in the MORN-ing I be- LIEVE I'll DUST my BROOM

    (although neither the Robert Johnson or the Elmore James version is that rhythmically simple, they both put the last stress on the upbeat)

  9. maidhc said,

    August 27, 2013 @ 4:48 pm

    AntC: Peter Sellers did a few Beatles songs. The one I hear the most is the Germanic version of "She Loves You" — "Ja?" "Ja!". But the difference is that Peter Sellers was a friend of the Beatles, and they were great admirers of his work. They had the same producer, George Martin. Whereas Steve Allen was just being nasty.

  10. Asa said,

    August 27, 2013 @ 4:56 pm

    I read the first two comments (by Jonathan Mayhew and Dimitri) as pointing out a question not of measure length but of where the measure starts. I agree with what I take those first comments to be saying: that the standard way to parse this would be to start the measure on what you call beat 3. For example, in your first "tabular form" example, I definitely hear "can't" as beat 1, "smell" as beat 3, etc. I think a large part of what has me convinced of this is the snare drum hit that occurs every four beats. In the measure-long scheme of things, such a hit tends to fall on the off-beat (i.e., beat 3).

    [(myl) FWIW, there was also an issue about the right metronome marking, which came down to how long the bars are and how to notate the minimal time units. I rewrote the post so as to be as neutral as possible on that point. As for where the bar lines are, I think it's clearer to have the lines start and end about where bars start and end — the relative strength of the metrical positions. But you can renumber everything shifted over by two, if you want to.]

  11. J.W. Brewer said,

    August 27, 2013 @ 6:45 pm

    I suppose we all develop strong idiosyncratic/extra-textual associations with particular texts depending on particular life circumstances in which we may have encountered them. For me, "The Message" is irrevocably "about" being with an overwhelmingly white and suburban crowd of high school students at a Model U.N. in Hershey, Pa. sometime in the winter of '82-'83 and trying to strike up a better acquaintance with a young lady representing West Germany (by way of some town in New Jersey), even though I strongly suspect that was not what the lyricist(s) had in mind. So I can see the pedagogical point of using a text that may have fallen into sufficient obscurity that students are hopefully coming to it with their minds a blank slate. (Although because of continuing Boomer cultural hegemony, Guitar Hero, and various other phenomena, it can be difficult to predict which hit songs of my childhood can in fact safely be considered obscure for Kids Today — indeed sometimes songs of my youth that only us obscurantist hipsters knew about subsequently became widely known to a mass audience because some hipster who grew up to work at an ad agency used them in a commercial.)

  12. dainichi said,

    August 27, 2013 @ 6:54 pm

    Interesting post!

    I second @Jonathan Mayhew's view that the measures should be shifted by 2 beats. It might not make a difference to the result of the analysis, but it makes it much harder to follow for musicians who are used to 1s being 1s and 3s being 3s.

    Just some observations:
    In "broken glass everywhere", "-ry-" falls on (what you call) 4, not 3+ and "-where" falls on 4+, not 4.
    In "junkies", "-kies" falls on 3+, not 3.
    In "I'm trying not to lose my head", "not", "to" and "lose" fall on 4, 1+ and 3, not on 3+, 1 and 2+.

  13. Layra said,

    August 28, 2013 @ 2:02 am

    While the exact numbering of the beats might not matter to the question of between-beat syncopation, it does have rhythmic implications in that the first beat of the measure is generally the strongest and hence acts as a rhythmic reference point. So the alignment of the lyrical lines with the musical measures becomes a coarser-grained form of syncopation and which beat is considered the start of the musical measure is rhythmically significant here.
    Thus in the interest of allowing the same charts to be used for analysis of lyrical-versus-musical alignment on the scale of beats rather than portions of beats, it would be helpful to indicate the beginning of the musical measure.

  14. Jason said,

    August 28, 2013 @ 3:30 am

    University administrators trying to appear hip-and-with-it-and-relevant by exposing freshman born circa 1995 to a text from the lost and irrelevant world of 1982 seems like something out of a David Lodge novel.

    I'm not so sure about this. There was a Tupac Shakur-centric course at the University of Oslo last year, and in recent years, at University of Washington and UC Berkely (offered on and off since 1997.) They wouldn't be offering these kinds of courses if students weren't interested.

    Which reminds me.. what ever happened to the once-fecund field of Madonna Studies? (The artist, obviously.) I don't see much Gaga Studies. Presumably the successor to Madonna will be studied by the successor to Camille Paglia, but I don't know who that is. Hopefully no-one?

  15. Lance said,

    August 28, 2013 @ 9:58 am

    But the difference is that Peter Sellers was a friend of the Beatles, and they were great admirers of his work. They had the same producer, George Martin. Whereas Steve Allen was just being nasty.

    Even more than that, Peter Sellers was being deliberately incongruous. Allen was presumably trying to point out that the lyrics seem banal when read as poetry; but no one expects the lyrics to A Hard Day's Night to sound right when done in the style of Laurence Olivier doing Shakespeare: http://www.youtube.com/watch?v=zLEMncv140s .

  16. Joe said,

    August 28, 2013 @ 11:15 am

    The punctuation in Emily Dickinson's poetry (em-dashes, capitalizations, semi-colons) had been judged "idiosyncratic" because it did not seem "grammatical" or followed a particular writing convention. However when you try to set Dickinson's poetry to music (typcally in hymn meter) these "idiosyncrasies" start to make sense as rhythmic notations (eg, em-dashes tell you when to pause for a whole beat, captializations indicate stresses). This post reminds me of that insight.

    Because of this, Camille Paglia calls Dickinson "the first modernist master of syncopation and atonality". She calls Dickinson a bunch of othjer names, but I think this particluar one sticks.

  17. J.W. Brewer said,

    August 28, 2013 @ 2:46 pm

    I don't begrudge universities the right to offer courses about contemporary or contemporary-when-the-professors-were-kids popular culture. (A relative of mine in the academic game has taught undergrads about plays first performed in the reign of Charles II and also taught undergrads about movies first released when he and I were teenagers.) I do suppose I am skeptical about the whole let's-make-all-incoming-students-read-the-same-book-over-the-summer-and-arrive-prepared-to-talk-about-it fad, when the book will generally have the usual "suburban book club" problem of being picked for seeming timely and "provocative," but not provocative enough to cause anyone in the administration any real agita, and thus near-inevitably ending up being middlebrow and "safe" in the relevant social context. Either trust the kids to choose electives and corresponding reading lists according to their own interests (which can include exegeting Tupac or Madonna or perhaps someone younger than that), or impose a core-curriculum everyone-reads-the-same-thing syllabus with authors that are safely dead and sufficiently safely canonical that they won't be at risk of being badly dated by the time the freshmen come back to campus as alumni. But given the existing phenomenon, Prof. Liberman has admirably demonstrated how to do cool high-tech prosodic analysis that may be pedagogically useful, or even inspire some kids to check out the Linguistics program.

  18. Jonathan said,

    August 29, 2013 @ 8:05 am

    I love this post! But the first commenter is right. The snare hits show the third beat of the bar, not the downbeat. This mistake makes the whole analysis difficult to follow.

  19. Collier said,

    August 29, 2013 @ 10:36 pm

    Derek Attridge's book "Poetic Rhythm" has a great chapter on this. He points out that scholars can't figure out quite how poetry in Old English should sound, because it's strong-stress verse, basically like rap, instead of metrical verse. Because no recordings exist, all we have is what's on the page, which is impossible to fully scan. He uses EPMD's "It's My Thing" and Tone Loc's "Wild Thing" as examples illustrating how difficult it is to map stressed verse on a page. Here's a link to part of the chapter, available in Google Books: http://bit.ly/14IAKBQ.

  20. C. Scott Ananian said,

    September 5, 2013 @ 4:38 pm

    Count me in with the musicians. Shift things over two. For whatever reason, lyrics are typically set with a partial "pickup measure" so that the unstressed words precede "beat 1", which is the location of the first stress. Even if you look over at hymns or fiddle music, the melody crosses the "measure boundaries" in a way which, I guess, a nonmusical person (or expert linguist?) wouldn't expect. Musical notation has developed mechanisms to handle how this interacts with measure boundaries, eg repeats with alternate endings, dal signo (https://en.wikipedia.org/wiki/Dal_segno), etc.

    FWIW, Square dance singing calls also have a stereotyped "8 lines of 8 beats each" structure: https://www.youtube.com/watch?v=3_TqFOyYYmc

RSS feed for comments on this post