Discovery: the most poetic word in the English language

Now that I’ve tagged more than half of the evidence quotations in OED2 for genre [see here and here for a discussion of this process], it’s time to start having a poke around in the data. A question that occurred to me last night was: ‘is there anything interesting about the relative density of poetic vs. other kinds of quotations in OED’s entries?’ So this morning I got the computer to count the number of total quotations for each entry and the percentage of the tagged quotations coming from poetic sources (poetry and verse drama), and then look up the frequency of the headword in COCA. Here’s the top few results, by poetic percentage (of the tagged quotations):

a hall, phr. 2 100% 0
abasedly, adv.2 2100% 0
abash, sb. 1100% 0
abashless, a. 2 100% 0
abdite, a. 1 100% 0
abear, sb. 2 100% 0

So that’s no good. But what if we limit the list to words with a good number of evidence quotations, say 100 or more? We get something with a little more bite:

fang, v.1 109 56.76% 332
worse, adv. 125 55.68% 28,173
rage, sb. 112 54.76% 6,815
rede, v.1 129 52.78% 36
strew, v.1 113 51.76% 33
weep, v. 191 51.08% 800
rede, sb.1 102 50.62% 36
stern, a.4140 49.46% 2,324
shun, v. 105 48.00% 321
smile, v.1 108 47.83% 25,534
sooth, sb. 126 46.94% 29

Or, should I say, something that grabs our attention, since ‘fang, v.’ is an old word meaning ‘To lay hold of, grasp, hold, seize; to clasp, embrace.’ It’s labelled arch. or dial. in OED2, but survives in ‘newfangle’, ‘newfangled’,  fairly unnewfangled words, having been in use before Chaucer was born (and used by him, after).

But back to the list. In addition to observing the seeming predominance of negative affect words, it’s apparent that many of the words are, like ‘fang, v.’, quite old, which may account for some of the bias toward poetic evidence in OED2, since the farther back you go, more and more of the available evidence is from poetry (there were no novels or newspapers to illustrate 10thC uses of ‘strew’).

But even though all the words in the list are first recorded before 1350, and all but ‘rage, sb.’ come out of Old English, several continue to be fairly common and not generally regarded as particularly ‘poetic’.

So another way to filter the list would be to add a restriction on the COCA frequency of the word, to remove archaisms and other very rare words. Note that my COCA count doesn’t match for parts of speech (332 is certainly the count for the noun ‘fang’). Excluding the above results, here are the next few that have COCA frequencies above 900:

May, sb.3 151 46.58% 339,243
steep, a.2 121 46.25% 6,443
sorrow, sb. 173 46.15% 2,918
rosy, a. 116 45.31% 1,381
swift, a. 131 44.90% 2,927
subtle, a. 127 44.68% 8,380
weary, a.1 148 44.55% 3,061
urge, v. 160 44.04% 3,368
up, prep.1 139 43.69% 807,218
dear, a.1 165 43.48% 8,569
must, v.1 146 43.16% 177,306
soft, adv. 132 43.02% 26,730
gate, sb.2 126 42.86% 13,009
us, pers.1149 42.55% 371,502

I’m surprised by some of this. The proportion of poetic quotations for ‘up’, ‘must’, and ‘us’ (we can ignore ‘May’ – the COCA count is for another part of speech) is almost three times the average for all quotations in OED2, and almost seven times the average for all OED2 headwords.

Other observations are that while negative affect words continue to show up (sorrow, weary), there are more positive ones (rosy, dear, soft). And we’re starting to get a few more Latinate words (rosy, subtle, urge) as well.

So, what is the most poetic word in English, according to the OED’s compilers? On numbers alone, ‘worse’ is the clear winner, with ‘fang, v.’ disqualified for archaism. But surely at some point the COCA count must work against a word in this kind of contest – it can’t be ‘up’, surely.

For that reason, I’ve decided ‘rage, sb.’ must take first place. So I was pleased to find under that headword a definition I didn’t know: ‘8. Poetic or prophetic enthusiasm or inspiration; musical excitement.’ And, under another sub-sense, this line from Tennyson: ‘The captive void of noble rage.’

Here’s a plot of all 220,000 entries, with Quote Count on the Y axis and %Poetry on the X:

No Comments