In my last post [“Why we need a Variorum OED: ‘Transgender’ ” 9/12/2020], I pointed to the OED entry for TRANSGENDER as it appeared in December 2020 as a good example of the need for a Variorum OED, which would label all elements (etymologies, definitions, quotations) with their individual revision histories.
Well, in the last three months, OED.com has implemented a change in this direction in their entry presentation, which I’m sorry to say is a big fail:
OK, so the edition date has been updated from March 2003 (entry creation date) to March 2018 (latest full update), which is what I had surmised had occured in my previous post. That’s all good, but now we also have an even more recent update listed, the “latest version published online”, which is July 2020.
What is the difference between the edition date and the “latest version published online”, you ask? Does it mean definitions have been updated or more recent or older quotation evidence adduced? OED gives this boilerplate explanation in the Entry History link:
But, even if the link is clicked-through (most users will not) the information is so general as to be meaningless (since the “may include” includes essentially all categories of change) or even misleading, as one will naturally note that definitions may have been updated, while glossing over “styling of citations” (which, like, who cares)?
So what has in fact been updated about TRANSGENDER? You don’t know, and there’s no way to tell. Unless, that is, you painstakingly compare the three versions, which I just happened to have archived here. [tl;dr: nothing much — I challenge you to spot the change!]
So instead of providing useful additional information about the version history of any particular entry, the “latest version published” date further muddles the provenance of an entry’s various elements. That’s because any change whatsoever in an entry, down to the smallest correction or formatting change, will trigger a “latest version” update to the entry.
This was brought home to me earlier this week, when an ESL trans rights activist from the Middle East DMed me on Twitter to ask (having read my previous post) whether the OED definition had indeed been updated in 2020, a perfectly reasonable hypothesis, given the info on OED.com. Now you might say that two years isn’t really a long time in lexicographical matters, but this person had a number of critiques and complaints about the way OED had phrased its definitions, which in the context of a rapidly changing public discourse would be much more acute for a definition updated eight months ago than it would for one drawn up over three years ago (and so perhaps be expected to change some time soon).
That is, it’s not only dictionary data nerds like me who have a stake in understanding the history of OED entries, down to the minutiae of spot edits and updates. It matters to people.
And it’s not only recent entries like TRANSGENDER that are showing even more recent version dates when not very much has changed. Consult virtually any OED.com entry, and you will find a latest version date some time in the last year or two or three, often because of changes in presentation style applied algorithmically across the dictionary. The impression is of a very up-to-date dictionary, which at least half of OED.com is very much not. To take one example, a favorite bugbear of mine, consider the entry for DINNER as it now appears on OED.com:
This is no doubt leading to all sorts of confusion and consternation across the English speaking and English learning globe, though perhaps in a less pressing way than for my recent trans correspondent.
The dinner entry also misspells Mittagessen.
As I’m sure you know, Charlotte Brewer’s Examining the OED project has been calling out the problem with untracked changes for years. In the page on Continuous change, she makes a comparison to Wikipedia, where every change is tracked in a way that’s easy even for beginners to understand. And Wikipedia is a lot bigger and more frequently updated than the OED.
Why can’t the OED do that? I’m just half-assed guessing here, since I haven’t read even one of the many books about the OED. Probably money is a reason. But that’s just another way of saying that culture is a reason: change-tracking is less important to them than whatever they do spend money on. I don’t want to accuse them of an attitude of “we’re the experts, just trust us”, I don’t know enough to say that, but it does seem like they don’t expect their customers to care about this. That’s where they’re wrong. The culture has changed around them since the OED3 was first planned in the 1990s. We now have a culture that knows that websites are impermanent, and expects archiving and change-tracking in response.
John Simpson wrote a blog post about how they’re re-checking quotes from Early Modern authors such as Spenser, reverting 19th-century editions to the original spelling and punctuation and correcting publication dates, so that the quotes appear as they did to their original audience. He’s proud of that work and thinks readers will be interested, and well he should. That work would be visible, if they had a system to track changes.
Keep the pressure on. I think this is a standard that a 21st-century reference work should meet.
We now have a culture that knows that websites are impermanent, and expects archiving and change-tracking in response.
Yes, good point, and arguably with something like OED, revision history is even more crucial than with purportedly up-to-date references such as wikipedia, or the Oxford Dictionary of English etc.. OED is historical in two important ways — both in its coverage and in its own development. Increasingly the difference (and the interaction between these aspects) need to be made apparent, even highlighted. How a word was defined, etymologized, cited etc. in 1888 or 1933 or 1972 or 2000 (or 2003, 2018, 2021) is important lexicological and sociological information in and of itself.
Why can’t the OED do that?
My guess is similar to yours: the underlying system doesn’t keep its own revision history (not so all-singing, all-dancing, then…) and so these would have to be reverse-engineered from static versions (which I believe are kept), and there’s no room in the budget for such a project (not to mention redesigning the online skin to accommodate it etc.).
I just now noticed a ⓘ symbol on a couple of sub-entries under yard n.1, with a mouse-over label “Updated September 2021”. And the entry history box now has a note: “The following senses or subentries (marked with ⓘ) have been recently updated: yard boy at Compounds 2 (September 2021), main sense (September 2021).” (The “main sense” that’s been modified is actually under Draft Additions 1993, a group of Caribbean senses that were part of a major update of Caribbean English this quarter.)
This is a step forward! Or it will be, once it’s rolled out — I haven’t found the ⓘ symbol on any other entries yet, even though I checked some that I thought should have it, such as “to catch one’s arse”, new sub-entry under catch, v..
Interesting; nice find. Unicode ⓘ is, at the moment, searchable (distinctly from “i”) via the online interface, and turns up the following 24 instances, all apparently from the latest (Sept ’21) update:
Advanced search results
1. bollard, n. 1844
… traffic from entering an area.ⓘ 1925 Daily Tel. 10 Jan. 9/3 The Office…
2. conch, n. ?1527
…inhabitant of the Florida Keys.ⓘ 1804 D. M’Kinnen Tour Brit. W. …
3. eh, int. 1569
…xpressing sorrow. Cf. a 2, ah 2a.ⓘ 1569 W. Haywarde A. Guarna Bellum…
4. flagon, n.1 1470
…e than an ordinary wine bottle.ⓘ 1872 Standard 16 Dec. The sale of…
5. -ize, suffix 1594
…f the suffix see the following:ⓘ 1594 T. Nashe Christs Teares new…
6. jump, v. 1511
…larly associated with carnival.ⓘ 1952 S. Selvon Brighter Sun xii….
7. jump-up, n. 1927
…larly associated with carnival.ⓘ 1948 E. Leaf Isles of Rhythm viii….
8. lap, n.1 c897
…ied around the waist. Cf. queyuⓘ 1769 E. Bancroft Ess. Nat. Hist….
9. salt water, n. and adj. a1000
…lt-water Creole salt-water Negroⓘ 1708 J. Oldmixon Brit. Empire in…
10. six, adj. and n. 835
…ation with nouns ending in -er.ⓘ six-bitter six-bitters bit 10a 19…
11. snowball, n. c1400
… count noun) a portion of this.ⓘ 1894 San Antonio Texas Daily L…
12. squirt-, comb. form 1632
…quirt oilcans also squirt oilcanⓘ 1861 Rep. Secretary of Navy U….
13. stage, n. a1300
…y goods and passengers by road.ⓘ 1681 Bp. G. Burnet Hist. Rights…
14. tie-, comb. form c1525
…to a sticky, chewy consistency.ⓘ 1879 Daily Gleaner Kingston, J…
15. unadvisableness, n. 1771
…ular person dead. unadvisabilityⓘ 1844 Hampshire Advertiser 10 Fe…
16. unbooked, adj. 1587
…warehouse, No. 4, Queen-street.ⓘ Not booklearned. 1859 D. Masson Brit….
17. univalent, adj. and n. 1865
…lents are positioned. univalenceⓘ 1865 N. S. Maskelyne C. B. Mans…
18. univalve, adj. and n. 1661
…ing or consisting of one valve.ⓘ 1810 Encycl. Londinensis I. 801/…
19. unsettle, v. 1598
…so rapidly. unˈsettler unsettlersⓘ 1840 J. Perceval Narr. Treatm. Gentleman…
20. unsnubbable, adj. 1847
…nsnubbableness unsnubbablenessesⓘ 1901 Goose-quill Chicago Nov. There…
21. unveracious, adj. 1845
…ver again entered. unveraciouslyⓘ 1850 Bell’s Life in London 28 A…
22. unwarranted, adj. 1577
….they were insane. unwarrantedlyⓘ 1805 Farmer’s Mag. Nov. 397 Common…
23. Victoria, n.2 1846
…tart of summer: see quot. 1988.ⓘ 1901 Scotsman 28 Feb. 7/4 A bill was…
24. yard, n.1 OE
…r general labourer; a gardener.ⓘ 1776 W. Marshall Minutes Agric. 10…
The impression is of a very up-to-date dictionary, which at least half of OED.com is very much not. — Or the reverse! For example, here Language Hat dismissed the pronunciation given for bilious because “entry is from 1887”:
https://languagehat.com/pausing-over-pronunciation/#comment-2372523
In fact, the revision of pronunciation has been running well ahead of everything else, with audio clips added to most entries, and the discussions of pronunciation (for British, American, and World Englishes) on the help pages are quite current, with scholarly references as recent as 2017 and 2018. But all that hard and valuable work is invisible at the entry.
When your page design misleads a reader as sophisticated as Language Hat, your page design is bad!
13 more ⓘ’s appeared in October with the update on climate-related words. But they’re not using it consistently! The blog post on this update discusses a lot of updates, but only some are marked with ⓘ. Some others are listed under “Draft additions October 2021”, e.g. extreme weather, new sub-entry under extreme. And some don’t get any indication except the boilerplate “most recently modified version published online October 2021”, e.g. food insecurity, new sub-entry under food.
Come on, guys, stop making us guess! Show us the sub-entry date!
Hm. I don’t have an inside track on this, but I wonder whether there might be older, pre-ⓘ drafts (or drafts that were started pre-ⓘ) that are coming online at the same time as some newer ones.
The wait is over: dinner was updated in the December 2021 release!
Just in time for Christmas
din..supper.New update (Dec 2021), new set of ⓘ’s. Most of the previous ⓘ’s have disappeared from the entries (but not all: probably sloppiness), but at least they’ve left them in the Entry History pop-up. These new ones seem to include error corrections, re-wordings, and some new additions:
blain, n. New subsense, promoted from ghost word to real word: the old edition said “¶Jamieson’s sense ‘A mark left by a wound,’ is apparently erroneous,” but a real-life Scottish citation has now been found (probably by cross-checking with the Dictionaries of the Scots Language).
boiler, n. Sense 2 (vessels for boiling) seems to have been fully revised, with more subsenses distinguished and many more quotations. I wonder why this sense was published now, without waiting for the revision of the other senses?
contegulate, v. Previously, the headword was cotigulate and the definition was [Erron. for contegulate.] — but there was no entry for contegulate! Now, contegulate is the headword and cotigulate is an alternate spelling. But it’s still a ghost word, no known appearances outside one dictionary. I wonder why this doesn’t qualify as a full revision?
deliriate, adj. Previously dileriate was the headword, and it was defined as “[Erron. for delirate]” — but there were no entries for delirate and deliriate as adjectives, only as verbs. Now deliriate is the headword, and two additional citations have been found with that spelling. Why doesn’t this qualify as a full revision?
epipastic, n. The supporting quotation has been re-checked and found to be an erroneous transcription of epispastic in the source, and it’s been added to the epispastic entry. I bet that sooner or later the other quotation will be reclassified as a misspelling of epispastic too. Much work remains to be done on whether epipastic has any independent existence!
forked, adj. “Forked lightning” sub-entry. This is a byproduct of the full update of lightning this quarter. Why isn’t this a Draft Addition?
frequent, adv. Label revised from “quasi-adv. (Also, in illiterate use, as a real adv.)” to “(nonstandard in later use.)”
heat, n. Sub-entry for phrase “if you can’t take the heat, get out of the kitchen.” Why isn’t this a Draft Addition?
truss, n. New sense added: “… harness worn by an actor, by which they can be safely suspended …” This is a byproduct of the full revision of trousers/trouse/trews: the supporting quote used to be under trouse, but that was a misreading.
It’s still very murky: why does corn chip qualify as a “Draft Addition” and get listed on the New Words List, while forked lightning gets an ⓘ and isn’t on the list, and millennial sense B.2 (the generation) gets no markup at all even though it has obviously been revised?
March 2022 update: All the previous ⓘ’s are now visible again, good! And there are new ones in 13 entries (sometimes more than one per entry).
Revisions of a sense: apart, blowfly, devolution, grope, inflation, inquisition, ride, sanction, tie.
Revisions of compounds: Collop Monday in collop, Diners’ Club in diner, ride-share and ride-sharing in ride, smoke annihilator in smoke.
Missing definition filled in for collep, a hapax; they’ve guessed it’s “A kind of drinking vessel (unidentified).”
For tie, n., the only change is a note added to the sense of a tie in sports or games, comparing the usage of tie and draw. This is a byproduct of the full revision of draw this quarter. I like how they can publish this cross-reference without waiting for the full revision of tie.