I recently read Poetic Rhythm: An Introduction by Derek Attridge. It’s an engaging survey of the topic, as informative as it is readable. It even included an illuminating few pages on the rhythmic workings of rap. But what I found particularly exciting was the last thirty pages, presenting the idea of phrasal scansion. The only frustration is that these scansions get rather complicated, so here I’d like to explore one possible way of presenting visually everything they tell us about the structure of a poem. Here’s an example of what I’m going for, and then I’ll explain how we build it and what it means:

Slowly he moves
to and fro, to and fro,
then faster and faster
he swishes up and down

Roughly, the idea of a phrasal scansion is to divide a poem into it’s climax, the part that anticipates the climax, and the part that extends the climax. Those three sections are again sub-divided in the same manner, and that process can continue all the way down to the lexical level, if you wish. These three sections are labeled anticipation (ANT), statement (STA), and extension (EXT). Actually, Attridge uses a fourth category as well, arrival (ARR), which he uses for statements when and only when they preceded by one or more anticipations. No new information is thereby added, so these categories can be conflated without loss.

The best way to show what’s going on is with an example. The symbol ‘>’ means the section continues on to the next line, while the symbol ‘|’ signals the end of a section. This example is taken from Attridge, except, as I said above, he uses ARR for STA that are preceded by ANT:
1 STA____________________________>
2 ANT____________________________>
3 STA____________________________|
4 ANT_________|STA_______________|
Ask me no more where Jove bestows,
1 ________________________________|
2 _______________|STA_____________|
3 EXT____________|
4
When June is past, the fading rose;
1 EXT___________________________>
2 ANT___________________________|
3
4 ANT_______________|STA________|
For in your beauty’s orient deep,
1 ______________________________________|
2 STA___________________________________|
3
4 ANT_________|ANT________________|STA__|
These flowers, as in their causes, sleep.

Awesome. But trying to see everything that we’re being told by these scansions can be difficult. The first thing I noticed is that we are effectively parsing the poem, implicitly using the following simple phrase structure rules:

STA -> (ANT)* STA (EXT)*

ANT -> (ANT)* STA (EXT)*

EXT -> (ANT)* STA (EXT)*

The parentheses mean the category is optional, and the symbol * means there can be more than one. Actually, the main reason I wanted to conflate the STA and ARR categories is because it simplifies these grammar rules. Looked at in this way, we can turn the above scansion into a tree diagram:

Now that we have these trees, we can do some fun stuff. Consider the path from a phrase to the root of the tree (the STA at the top). For example, the phrase “Ask me no more” has the path: text — ANT — STA — ANT — STA. I’m going to use these paths to assign a background color to each phrase. This path tells us all the different functions the phrase is participating in at different levels. In particular, the number of ANTs tells us how much a section of poetry is building up tension, the number of STAs tells us how central it is, and the number of EXTs tells us how much its concluding or cooling things down. Then, we can assign these values colors. The choice is of course some what arbitrary, but I let ANT be green, STA be red and EXT be blue.

The depth of a phrase (the total length of the path to the root) in the tree also tells us something. How deeply nested a phrase is tells us how directly relevant or important it is to the whole poem, so we’ll make the darkness of the background inversely proportional to the depth.

Okay, so here are the formulas I use to assign values to the redness, the greenness and the blueness of each phrase, using the RGB color model:

$\displaystyle\mathop G=450\centerdot\dfrac{freq(ANT)}{height}$

$\displaystyle\mathop R=450\centerdot\dfrac{freq(STA)}{height}$

$\displaystyle\mathop B=450\centerdot\dfrac{freq(EXT)}{height}$

Freq(ANT) is the total number of ANT nodes in the path, and freq(STA) and freq(EXT) are similarly defined. The height of the tree is the maximum depth of any phrase. It, along with the constant 450, makes up the normalizing factor.

Now, let’s see what we can see when we apply this to a poem. First, we’ll try this with another example from Attridge. Based on his phrasal scansion, here’s a visualization of the poem “Boy on a Swing” by Oswald Mbuyiseni Mtshali:

Slowly he moves
to and fro, to and fro,
then faster and faster
he swishes up and down

His blue shirt
billows in the breeze.
like a tattered kite.

The world whirls by:
east becomes west,
north turns to south;
the four cardinal pointsmeet

Mother!
Where did I come from?
When will I wear
long trousers?Why
was my father jailed?

The orange-red of the phrase “he moves” corresponds to how it works as a false center for the poem. At all the lower level, it’s functioning as part of statement (red) and only at the highest (well, second highest, if you count the whole poem) does it serve to anticipate, giving it a bit of green (red plus green is yellow in the RGB color model). All of the purple and blue (extension) of the first three stanzas represents the poem’s creation of misdirection; with all this extension, we feel poem is coming to completion. Then, the phrase “mother!” tells us that in fact something big is coming up. The dark green indicates that’s it’s directly relevant to the whole poem and broadly anticipatory. The subsequent bright green is the height of suspense, followed by the poems true climax – the completely red word “jail.”

Let’s compare this to the poem “At North Farm” by John Ashbery. This one isn’t in Attridge’s book, but I worked out its phrasal scansion in accordance with his guidelines as best I could:

Somewhere someone is traveling furiously toward you,
At incredible speed, traveling day and night,
Through blizzards and desert heat, across torrents,
through narrow passes.

But will he know where to find you,
Recognize you when he sees you,
Give you the thing he has for you?

Hardly anything grows here,
Yet the granaries are bursting with meal,
The sacks of meal piled to the rafters.
The streams run with sweetness, fattening fish;
Birds darken the sky. Is it enough
That the dish of milk is set out at night,
That we think of him sometimes,
Sometimes and always, with mixed feelings?

Without even reading the poem, the colors tell us a lot about its structure. All of the green in the first half reveals its dedication to immediately establishing and building a sense of suspense. Notice how different that looks from the “A Boy on a Swing” poem. The grey in the second half tells us that, overall, these sections are participating equally in anticipating, stating, and extending, which may suggest a sort of neutrality.

So this is fun! And, beyond the visualization, seeing the phrasal scansion as a phrase structure grammar brings up a lot of questions about its relationship to syntax. ANT, STA, and EXT are reminiscent of Spec, X, and Comp in X-Bar theory — how well would they actually correlate if we parsed these poems syntactically? But I digress.