Poking around the OED today, I came across this entry, which I’ll reproduce here in full:
aˈcrasial, a. rare-1.
[f. acrasy + -al1.]
Ill-regulated, untempered, intemperate.1851 S. JUDD Margaret II. xi. 321 (1871) ‘Acrasial Philogamy? Brother Edward, what is that?’ ‘That,’ replied Edward, ‘is an incurable malady to which young persons are subject.’
I guess that the OED included acrasial because of the association of acrasy with Spenser, who has ‘irregularity, disorder, intemperence’ personified in a character named Acrasy in The Faerie Queen. The etymology, though it appears transparently to be a+kratos (i.e. un + mixed or tempered), turns out to be somewhat more complicated according to OED:
[ad. med.L. acrasia, which seems to confuse Gr. ἀκρᾱσία ill-temperature, badly-mixed quality (f. ἄκρᾱτος unmixed, untempered, intemperate) applied by Hippocr. to meats, with ἀκρᾰσία impotence, want of self-command (f. ἀκρᾰτής powerless, without authority, without self-command, incontinent).
Well, the etymology of philogamy is just as transparent, if not more so, but OED didn’t see fit to include that as a headword, despite it having exactly as much usage evidence as acrasial.
In my OED2 (1989), there are 42,383 headwords which are included on the basis of only one citation. Such terms are sometimes called hapax legomena (Greek again, for “once said”), or “nonce words”, which is a term made up by OED editor James Murray for the particular purpose of describing words made up for a particular purpose (as OED3 glosses it: “a word apparently used only ‘for the nonce’, i.e. on one specific occasion or in one specific text or writer’s works.”)
Of the 43,383 single-quotation headwords in OED2, 15,357 (a little more than at third) are from the 19th century, like Sylvester Judd’s immortal, unreattempted deployment of acrasial.
Now, this is very rough, but within these 15,357, I count about 2,300 quotations which, like Judd’s, also contain another word which is not an OED headword. Many of these are spelling variations of one headword or another, or foreign words, or proper or scientific names.
(Even some of these can be puzzling: s.v. corundic, we get the following quote: ‘to express the relative hardness of other substances, by the following parallel terms: Cretic, Gypsic, Marmoric, Basaltic, Felsparic, Crystalic, Corundic’ — but gypsic, felsparic, and crysticalic aren’t OED headwords; and marmoric is included on the strength of only one quotation, but it isn’t this one! basaltic gets three quotes.) [*edit – see 7/11/21 comment by kts]
Some are interesting, especially when the words are being used in parallel ways. Below is a table of haphazardly selected OED hapaxes, from single quotations which also include words not in OED, on the model of ‘Acrasial Philogamy’. Make of this what you will.
I can’t say exactly how many of the 2,300 are like these – what I take to be true “ghost hapaxes”. Perhaps a few dozen, or a hundred or more – but the lines of inclusion get a bit blurry around the edges.
[…] “Acrasial Philogamy – Ghost Hapaxes in OED”, I documented some words that are included as headwords in the OED on the basis of only one piece […]
The ones marked rare -1 really do show up once. The ones marked rare -0, however, appear only once and only in a dictionary or word list of some sort.
The term “ghost hapaxes” confuses me: a ghost word is a word that exists only in the dictionary and not in actual use, but these are the reverse. “Buried hapaxes”, maybe?
felsparic: Spelling variant of feldsparic, which is a sub-entry under feldspar.
crysticalic: Must be your typo. Crystallic has been there since the first edition, although they don’t record the spelling variant with only one l.
The table is missing. Any chance of getting it back? It’s in the old captures of this page at archive.org, you can reconstruct it from there if necessary.
@kts – tables all disappeared when the plugin was discontinued. I hadn’t thought to recuperate them via the wayback — thanks for the idea.
marmoric is included on the strength of only one quotation, but it isn’t this one! — It’s from the same book. But, really, this whole citation should never have been used, because these words were stillbirths. They did not catch on. The author was trying to invent terminology to rank the hardness of rocks on a scale, but nobody else ever used his system. (Geologists use the Mohs scale of hardness instead.) Google and Hathitrust find zero uses of corundic except for this book, a couple of reviews of it, and the N.E.D. itself! One of the reviewers was wonderfully snarky: “How greatly our language has been enriched by these ingenious adjectives we do not pretend to determine.”
Some of these terms were independently re-invented and have real-life uses, unrelated to this citation — e.g., gypsic is used in soil science for gypsum-containing soils — but cretic also has never been used outside this book in a rock-hardness sense.
Evidently some reader thought “Ooh, shiny vocabulary!” and sent in a slip to the OED, and they didn’t check with any geologists to find out whether these words were used in real life. (Or maybe somebody in charge of G did check and that’s why gypsic didn’t get in. It wasn’t re-invented until the 20th century.) Science was a weak point of the First Edition.
True but isn’t that the case for more or less all hapaxes? They might /just/ be justified for one or two FAMOUS AUTHORS on the basis that people would might them up, but even eg. the Urquhart quots are pretty useless.
I don’t believe it’s anywhere near all. Often a single citation only means the OED didn’t collect more, not that there aren’t more. (Quantifying how often would be an interesting project…) Let’s check some of the examples in your table:
contestational before 1893: At that time it seems to be an idiosyncrasy of Jeremy Bentham, used only by him (multiple times) and somebody quoting him. Plenty of usage more recently.
distrix before 1896: There’s now another citation in the online entry, even though it isn’t fully revised. However, I don’t find it anywhere outside of medical dictionaries and lists of diseases. Could be a dictionary-only word.
garance before 1972: An antedating has been added to the online entry, even though it isn’t fully revised. (We want change-tracking!)
arzica: Italian. May not be sufficiently assimilated to qualify as an English word; probably that’s why they didn’t enter it in the 1972 Supplement along with garance.
raiiform/raiform/rajiform ‘shaped like a ray (fish)’: A few other citations before 1903; well established in the 20th century. See the Third Edition update.
squaliform ‘shaped like a shark’: Plenty of hits going back to 1839.
torpediform ‘shaped like a torpedo’: There are other hits.
Thanks for the interesting discussion! Incidentally, acrasial wasn’t unreattempted forever; it had a few 20th- and 21st-century uses, represented in the Third Edition revision, December 2011. Philogamy has a few recent uses, too, and should have also been included.
And one more nitpick: nonce word is not a synonym of hapax.
True enough. In principle there ought to be at least 3 quots for all but the shortest lived word senses (earliest, latest and middle) but in practice plenty of words were insufficiently researched or underdocumented for some other reason.
In OED2, fwwiw, one can break down the number of quots per entry as so:
0 60431
1 41941
2 34013
3 28894
4 22818
5 17271
6 12846
7 9985
8 7650
9 6203
10 4950
[The zero-quot entries are mostly cross-references.]
Yes, I suppose I have been using “hapax” in different senses, understood from opposing directions. And you’re right to unpick these from “nonce word”. Thought of as single occurrences in the big broad corpus of written stuff, hapaxes will always result in single quotation entries, if they result in entries at all; thought of as single-quotation entries per se, yes of course they will many times be simply under-researched or clipped for some other reason.
Nonce word I take to be different from both of these in principle if often not in fact, being JAHM’s nonce term for a certain type of neologism (most of which end up being hapaxes in both senses).
The cryptic notations rare -1 and rare -0 have just recently been revised, replaced respectively with just plain rare (which they’ve kindly defined as “typically used when there are fewer than three quotation-examples available in the databases available to our editors”) and “Apparently only attested in dictionaries or glossaries.” OK, that’s an improvement. But they’re also replacing the label nonce-word with “Apparently an isolated use” — aw, that was part of the OED’s charm, but I guess it too must have been too obscure for some people. (Yes, pancakewards has the new label; don’t we all have a secret soft spot for that one?)
Do they enter nonce-words anymore? Or is the policy against them now?
kts’s comment prompted some digging around in the archives, written up here.