Idiosophy

When I started automating detection of alliterative verse, the original hypothesis was to watch the phrases that look like lines of Old English poetry spike up dramatically when the story got to Rohan. Well, that’s not what happens. J.R.R. Tolkien uses alliteration so much that even when we restrict him to Sievers’s five patterns, there’s a constant stream of alliterative lines.

Here’s the density of alliterative lines in the text, as a function of the number of words since the beginning of the Prologue.  (I used a 100-word bandwidth, for those who care.) You can see a rise in the frequency of alliteration in Book III, where I expected it.  There are also some nice spikes in Book 5, when the Rohirrim make their entrance and when the bard sings a long song in honor of the fallen.The highest density-spike of alliterative lines comes in “The Uruk-Hai” (III,iii, in dark grey) when Pippin is talking to himself.  That was unexpected. In the next chapter, though, Treebeard comes through.  He doesn’t provide any high spikes, but the low troughs disappear. Ents never stop alliterating, murmuring in their slow musical voices. (This sentence shows that I allow quite a few unstressed syllables; Type E can have four in a row.)

Book IV also starts out with a lot of alliteration. Sam is talking to himself, this time. For some reason, writing out Sam’s dialogue with formal line breaks and caesurae is hilarious to me:

Numbskulls! You’re nowt but / a ninnyhammer, Sam
Gamgee; that’s what /  the Gaffer said.

The lowest density of alliteration is in the chapter “Minas Tirith” (V, i). I noticed this the second time I read the book – the tone changes abruptly from the previous volumes. When they encounter a line like, “For the fashion of Minas Tirith was such that it was built on seven levels…” even teenaged Idiosophers are brought up short by the new voice.

Alliteration is not just part of the story, though. The Prologue has a solid population of alliterative lines.  Those bits are Bilbo’s and Merry’s voices, so from this we can derive a consistent theme: The constant background of alliteration is there because the book is narrated by hobbits.  When they’re talking to Ents or Rohirrim the rate ticks up about 5%.  Elves or Gondorians drag the rate down, roughly the same amount. With a larger bandwidth, smoothing over 2000-word intervals, we lose the poem-spikes, but it’s easier to see the overall changes.

Surprisingly to me, the field of metrical research in Anglo-Saxon poetry is thriving. It’s not like any new Anglo-Saxon verses have been discovered recently, but we do have some new alliterative poems. I just found a paper by Nelson Goering [1] (one of the lecturers in my Anglo-Saxon class) that applies analytical techniques developed for Old English to J.R.R. Tolkien’s recently published verse.

The thing that jumped out at me was what Goering calls “crossed alliteration”. Where Anglo-Saxon verse alliterates on one sound per line, crossed alliteration takes advantage of the four stresses to alliterate twice: either A-B-A-B or A-B-B-A.  Examples he gives from The Fall of Arthur are “Fiercely heard she / his feet hasten” (II, 111) [2] and “of south Britain / booty seeking” (I, 9). He notes that 7% of the lines of the poem have crossed alliteration, which is “nearly double the rate of a classical Old English poem such as Beowulf.”

That’s the sort of thing that really gets an Idiosopher (provided that he has spent a year or so on text preparation and code-building) rolling. In this case, rolling right into a brick wall.  To order a computer to look for crossed alliteration, we’d need a regular expression that looks for (string), (any string but that one), (the first string) (the other string).  I’m ashamed to admit that I spent a month trying to turn that into a regular expression.  It just can’t be done in any language I speak. Back-referencing the complement of a single character is possible, but the CMU phoneme set can have two characters, and trying to kluge together a fix got totally out of control.  The reason I’m ashamed is that there’s no reason to use a powerful tool like regular expressions — I don’t need any wild-cards in this simple case so I can just use equals-signs for the test. Most likely you saw it some time ago, and have been laughing at me.

Here are some good ones:

He wandered in loneliness, weeping a little.  
'Spoons? Fiddlesticks!' He snapped his fingers.
Galadriel stood, alone and silent.

The sun was warm and the wind was in the south.
Boromir seemed to be swimming or burrowing...
Then suddenly Frodo fell asleep.

Here’s one that’s bogus: “‘…Black Riders.’ ‘Black Riders!'” JRRT repeats a phrase for emphasis fairly often, and I really think that ought not to count.

Overall, the numbers for LotR are higher than for The Fall of Arthur. A-B-A-B crossed alliteration is 10% as frequent as simple alliterative lines. A-B-B-A crossed alliteration is about the same, 10.5% as common as the simpler case.  Note that the latter was not included in the original computation, but the former was.

To wind up, here’s one that’s intriguing: “Again she fled, but swift he came / Tinuviel! Tinuviel! / He called her by her Elvish name;’   Cross-alliterative enjambment in ann-thennath sounds obscure enough for a whole dissertation.

Notes

[1] Goering, Nelson. “The Fall of Arthur and The Legend of Sigurd and Gudrún: A Metrical Review of Three Modern English Alliterative Poems.” Journal of Inklings Studies 5.2 (2015): 3-56. Preprint here.

[2] Goering has the caesura in a different place from the text, which has been corrected here. Fly-specks like this are of no interest to anyone, but they are commonly pointed out in the literature. Were I to forbear to mention it, this blog would never be taken seriously by any community of scholars.