The Life of Words

In July 2023 OUP launched a brand new design and search/browse interface to the Oxford English Dictionary Online at OED.com. Long in the works, the new site presents the venerable dictionary, which is now about halfway revised from its previous incarnations (be they 1928, 1933, 1972, 1989, etc.), in a radically new way.

It’s… not great.

It’s not easy for me to say this, since I know a lot of hard work by good people who care about the dictionary has gone into this redesign. And listen – it may well be that I’m wrong and this site is the best way to bring the dictionary and its treasures to future readers. But from my perspective the new OED Online tosses a lot that was great, and unique, and real, about OED, while offering little new of value.

I’m not the only curmudgeon out there. On The Site Formerly Known As Twitter, the new-look OED is getting rolled, most savagely by one @Blythe_Gryphon, who called it not just a “downgrade” but “nothing short of a gift to rising dictators, autocracies, and all forms of pathocracy”, making of it an “introductory dictionary for primary schools.”

A bit much? Maybe, just about everyone has been complaining similarly about the burying of critical OED research under a mass of mass-market hooey. A recent private FB thread with contributions by seasoned OED users had several threatening to dust off their PRINT COPIES of OED rather than use the new site! Paper volumes, some said, were easier and faster to use, more complete, and more historical.

So, is this the new New Coke? The new 𝕏?

What the new-look OED.com has in common with these repackaging failures is that in reaching for new audiences, it devalues and diminishes the very things that make it different and great.

That is, I think one strong motivation for the redesign has been commercial, or at least popular, in the sense that much of the information redesign appears to be geared towards distillation and simplification of complex lexicological research into something a search engine or SEO can easily serve.

Right now if I search up “meaning of ‘WORD’ ” in google I get results from Merriam-Wesbster.com above the fold, and then stuff from the Cambridge dictionary, CollinsDictionary, Dictionary.com, and Vocabulary.com. OED.com doesn’t appear on the first couple of pages, and I’m guessing the new site will bump it up (when it suits OED to do so), and facilitate APIs and other automated interfaces. Add to that eminently 𝕏able “content” packaged in “discoverability hubs” and you have something a marketing person can pitch.

The costs are, in my view, too high. Here are five ways in which I think the new-look OED devalues itself:

It simplifies. The new OED word “factsheet”, which is the first thing that comes up when you look up a word in OED.com, seems especially designed to “serve” in this sense. Indeed it is already “served” in that it pulls certain information from the entry proper, as well as other sources, and wraps it up in easy-to-understand boilerplate, with titles like “What does the noun NOUN mean?” and so on. These factsheets are outside the OED paywall, and so are accessible to all, which is good per se. What they tell you is minimal: the number of senses, pronunciation, date and author of first use, language of origin, and frequency (which drives me nuts – more on that later). All (except frequency) good things for everyone to access, but very much surface-level information which in many cases it would be irresponsible to use without consulting the full entry.

It reduces. “Reduce” in one sense describes the simplifications mentioned above. But in an even more literal sense the new OED reduces, everywhere displaying less and less information with less and less context. Nowhere is this more inconceivably applied than in the sense sections themselves, where the quotation paragraphs, the heartwood of the dictionary, have been truncated to display the first and last few quots. Who thought it was a good idea to elide the one great and unmatched feature of the OED? Must I speak in the language of USPs and barriers to entry? Millions of quotations collected over 150 years–the source and lifeblood of the dictionary–and you want people to click through to display them?

Yes, click past the factsheet, get rid of the “tabular view” (do these things each time you look up a new entry) until finally you have reached what used to be the entry for the word. Find the sense you are interested in, and you get this:

*Mind boggles* … A century ago JAH Murray knew that “what makes the dictionary unique is its historical method” and that the quotation paragraphs were already as concise as they possibly could be to illustrate that method. I just.. can’t even.

It incorporates. OED has always been a compendium, basing its science on millions of quotations from hundreds of thousands of works. It has also always borrowed knowledge from other sources, including etymologies, definitions, and so on, sometimes presenting it as its own. That’s all well and good, but in most cases those other data have been weighed and synthesized by actual lexicographers. Here a bunch of stuff is systematically integrated by algorithm, nothing more annoyingly that the frequency stats for words.

These take up a lot of space in the new-look OED, appearing in the initial search results (next to year of origin), and then in a whole tab in the entry, including two impressive and quantitatively authoritative looking graphs.

Forget that no one who turns to the OED cares much about word frequency (save perhaps a couple corpus linguists who have their own better sources for this), or has a native understanding of what it means or how to use it (pick a word with a similar order of frequency to, say, applesauce–bet you can’t.)

Forget that, as the entries point out in long notes (room for these in every entry!), the historical data is sourced from the Google Ngrams corpus, which is rife with problems well documented and (ok, in my opinion) has had net negative research value since it was introduced (unless of course you count debunkings of Ngram based claims as net positives).

Forget all that — the new OED can’t disambiguate between homographs well enough to make any of this relevant, even if it was useful. Here for example is the top line freq data for 8 different words spelled CAT, which I guess speaks for itself:

Or maybe it isn’t quite self-explanatory, since frequencies are unintelligible anyway. What this says is that CAT v. through CAT n.4 are roughly as common as, e.g., prelapsarian, dirt-cheap, or badass; and CAT (variant of can) as common as all the most common words in the language. Somehow I cat’t believe it.

It atomizes. The complaints I’ve gathered here mainly tend to focus on the ways in which the new-look OED makes OED information hard to access, especially perhaps for academic or research purposes, but I’d argue also for learners such as my poetry undergraduates and maybe even general audiences. All the “good stuff” now seems to be several clicks away, whereas we used to have it all on one page in one place.

These are valid complaints born of understandable frustration, but I think they are a symptom of a deeper and more fundamental problem with the new-look OED. This is that the category of the dictionary ENTRY, as we have understood it for centuries, has been abandoned for an atomized presentation of dictionary DATA. Factsheets, tabs, collapsible quotation paragraphs, imported data from other sources, etc. etc. — all these “upgrades” degrade the integrity of the ENTRY as a lexicological document of interrelated and interdependent parts. Entries are what lexicographers work up (or have worked up), not data. Lexicographers put etymologies and word forms first in an entry because these helped to read the quotation paragraphs that came after (new-look puts these sections lower–I guess based on some idea of what “users” want to “access”). It was a whole, even if it could be read in pieces.

I think this shift is a catastrophic mistake which seriously jeopardizes the value (intellectual, cultural, but also brand value, if that makes a difference) that has accrued in the OED over time. I think, also, it makes ominously easier, and harder to detect, the creeping piecemeal approach to dictionary revision and updating that many of us have been worried about for several years now.

It erases. All of this might be fine (maybe) if a button or a tab or a cookie perhaps would allow us to revert to the entry-based OED we have known and loved. Forever I’ve been advocating for a Variorum or Track-changes OED, where one could see OED as others have seen it. I think now in addition to preserving older lexicography we also need to preserve older presentations of and interfaces with that lexicography.

Especially because of the denaturing of the ENTRY, if the current format persists, we risk losing twenty years of what consulting OED.com was like. We have the print dictionaries. Some of us have CD-ROMs we can retrofit. But no-one even now can replicate fully what OED.com looked like in the first half of 2023, or in 2009, or in 2000. This is a real loss in knowledge, I think, or a loss in knowledge about knowledge.