The Lexicographer's Rules

This keyboard plugs into your computer so you can use all your mad texting skillz there, too. This is nowhere on my Christmas list. More here.

It is quite incredible that Corey Kilgannon would write in the New York Times about Daniel Cassidy’s book How the Irish Invented Slang without talking to historical lexicographers, historical linguists, or experts in Irish Gaelic linguistics.

They would tell him that Cassidy’s theories are insubstantial, his evidence inconclusive, his conclusions unlikely, his Gaelic atrocious and even factitious, and his scholarship little better than speculation. In short, his book is preposterous.

Cassidy paints himself as the maligned scholar, the unappreciated genius, the outsider. He may be all of those things, but he is them by choice: his work cannot withstand scholarly scrutiny so he simply cannot afford to join forces with any larger body of experts who do this sort of thing for a living. His book falls apart on first reading by anyone with some expertise in the field.

In October, Michael Patrick Brady did a powerful job of taking down Cassidy’s work in a way that makes it clear that it is poorly conceived, poorly executed, and should be poorly received.

Because people keep asking, I have found myself repeatedly pointing out that Cassidy’s book is not to be trusted. There are too many such works in the world that set a course for factual disaster, so in the areas in which I have some skill—supporting or disproving word histories—I feel I should speak up. I spend my days trolling through the historical record examining word origin stories and every one of Cassidy’s theories that I’ve checked—including those for “jazz” and “bunkum”—are no more believable than leprechauns and their pots of gold.

Daniel Cassidy first appeared on my radar in late 2004 or early 2005 when he began making a series of posts to the email list of the American Dialect Society, a group for which I am a vice president and with which I have been involved in one way or another for 15 years. ADS is a 118-year-old academic organization that has in its ranks a highly qualified group of professional, respected, published linguists, lexicographers, researchers, and grammarians of many backgrounds and specialities. It publishes the learned journal American Speech, co-founded by H.L. Mencken.

Of course, the members of the American Dialect Society email list looked at Cassidy’s messages and judged the worth of his theories.

The theories came up wanting in a bad way. Cassidy’s arguments were then, as they are now, mainly that the English and American lexicographers are biased against the Irish and don’t want to give them their due, and that, in fact, the Irish were the source of much of our most common slang.

So, he decided he would fill the gap by finding obvious phonetic and orthographic similarities between Irish Gaelic and English-language slang. Which is, of course, a big heaping load of hooey.

Cassidy’s response to criticism of his theories on the email list was to unsubscribe and then send me requests in the back-channel to post his content for him. I told him I wouldn’t and tried to strike up a correspondence with him that would help him reach his scholarly goals. It went nowhere.

In January 2005, I challenged Cassidy to present all of his evidence. I told him that I’m the descendant of three strains of Irish, four strains of empiricist, and the son of a bluster-catcher, and I said he was going to have to do better than trot out the same-old “they’re all against me!” argument of every perpetual motion inventor.

To date, what he’s provided as evidence is flimsy and fouled by scholarly incompetence.

“To cry Wolof” is a recent coinage used to describe amateur etymologists who propose absurd theories based upon superficial similarities between different languages. It comes from the widely circulated but false claim that the word “hip” ‘cool, fashionable’ comes from the West African language of Wolof. You can see many such language coincidences here. They are provably, demonstrably, unquestionably coincidences.

This is Cassidy’s primary mistake: he has wrongly assumed that similar spellings or pronunciations between words prove a connection. They do not.

Linguist Bill Poser wrote compellingly about the tendency of amateur etymologists to make these wrong-headed leaps of faith. Poser translates Georg von der Gabelentz from his 1901 book Die Sprachwissenschaft:

“It is terribly seductive to roam the world of languages comparing words from them at random and then to bestow upon scholarship a series of newly discovered relationships. Very many stupidities also result from this; for the most urgent discoverers have unmethodical minds.”

As von der Gabelentz says, spelling and phonetic similarities must be looked at, but they are simply a starting point. They prove nothing. They merely provide a clue to be investigated by gathering evidence for and against the connection.

Evidence. Above all, Cassidy needs to support his claims with published evidence that shows the etymological path. Dated, continuous, in-context quotations from any written source will always be superior evidence over phonetic speculation based upon national, linguistic, or ethnic pride.

The main thing that bothers me about most of his theories, besides his overall unwillingness to express doubt and caveats about them and his apparent inability to do the work required to prove or disprove his own theories, is that in cultural overlaps and linguistic contact situations in which words are borrowed, there tends to be written proof of it.

This happened repeatedly with contact contacts by the English, French, and Spanish settlers with Native Americans in the New World, and it continues to happen where Spanish and English rub up against each other today. We can see in the written record where the languages have loaned words to each other and how those loanwords changed.

In those cases, we find borrowed words set off by quotes, dashes, or italics, or explained as “as my gram used to say,” or “as we used to say,” or even given plainly by a regular person as a word from another language, and so forth. In order to prove Cassidy’s claims, primary source material that might contain these sorts of statements needs to be found and examined: letters, books, diaries, newspapers, what have you. Certainly, across the whole of his book there should be lots and lots of this sort of “language contact” evidence, but there’s little that I can see.

If the words he’s writing about really did come from Irish Gaelic, the only way to prove it is to find those Irish words repeatedly showing up in some form in print in English-language contexts.

To put it another way, he’s failed to find early uses of the transformed or transforming terms. He would need to show a variety of phonetic or Anglicized spellings that resemble the terms as we know them today, i.e., word forms somewhere on the continuum of change that might demonstrate that they were earliest, or nearly earliest, used by Irish-speakers or people of provable Irish heritage or in direct contact with Irish people.

Of course, if no such texts are ever found, or the words are not found in them, then the theories are unproven, and that is that. Cassidy has promoted his unsubstantiated theories so widely that he cannot back down now without looking foolish. It is not in his best interest to do the etymological work properly.

Besides that, substantiation is a lot of work, and as we have seen repeatedly, those would-be scholars who “cry Wolof” have little stomach for the tedium required to prove their theories.

Another problem with Cassidy’s evidence is one of “Irishness.” It doesn’t require a fluent or native understanding of Irish Gaelic, which I do not have and which Cassidy does not have, either—he is usually careful to leave this point unclear—to see that he’s taking words that have complex meanings and cherry-picking the subsenses that most suit his purposes.

He seems to have plundered Irish dictionaries and when it has suited him he has adjusted his plunderings to make the meanings broad enough to support his theories.

On top of that, the Irish definitions he gives are little better than glosses (that is, short one- or two-word definitions more like synonyms presented in a thesaurus) and do not show a complex understanding of context nor frequency, neither presently nor historically. He has played fast and loose with the Irish in the same way he’s playing fast and lose with the English.

To bolster his Irish derivation theories, he found and quoted writers of supposed Irish heritage who have used the English forms of the words, going by surname only in some cases, in others choosing people who live or had lived or could have lived in a region that was widely settled by Irish or Scots-Irish. He’s done little to verify whether those people he is quoting had any knowledge of Irish. He seems to be working under the assumption that some Irish just lingered in the air.

Then, in the cases that I have seen, he has chosen as supporting evidence English-language quotes that contain the English word under discussion. I have yet to see a single one of his quotes include any form of the Irish word in an English-language context, except when he’s quoting from dictionaries which, in all cases I have seen, are talking about an Irish meaning rather than the supposed English meaning.

Even his Irish forms of the word that are cited are usually different from the form that was supposedly transformed into an English word. Many of his Irish forms should be prefixed with an asterisk because he has not found them in the wild but merely postulated their existence.

For an example typical of his scholarship, see his claims about “bunkum.” I choose this word because it’s one of the relatively few of those words not obviously derived from another language for which we do indeed know the origin with near certainty. Dave Wilton has a correct and reliable summary of it at his WordOrigins.org web site which you should read to understand how, exactly, Cassidy has gone off the rails.

Cassidy says that the congressman from Buncombe County lived in North Carolina, which had a Scots-Gaelic and Irish-speaking population. This, somehow supported by information gathered from a 2005 Scotsman newspaper article that said Dizzy Gillespie’s family from North Carolina and Alabama were African-American Gaelic speakers, means that “Buncombe” comes from buanchumadh, which he defines as “a long made-up story, an endless invention.”

His other evidence is three uses of “bunk” in the plays of Eugene O’Neill, one from 1939 which has it as “de old Irish bunk"—the oldest cite he has, 89 years later than Oxford English Dictionary’s first recorded use of the word.

He has no citations spelled buanchumadh at all, neither in English nor Irish. Nowhere does he attempt to explain the early expression “talking to Buncombe,” nor the capitalization or spelling of Buncombe in early uses, nor the existence of Colonel Edward Buncombe for which the county is named. These are all specific historical references that would have to be, so to speak, debunked before his Irish theory can be given any credence.

This is in the same entry in which he casually throws in unsupported Irish etymologies for “swank” and “to dig” ‘understand.’ They are presented with no evidence, not even bad evidence, yet given in Kilgannon’s article as if they are faits accomplis. They are not.

To summarize: Daniel Cassidy’s work is unreliable and not to be trusted.

Chris Wemmer is a Californian, biologist, and retired Smithsonian scientist who keeps an engrossing blog, Camera Trap Codger, about the animals he photographs. His camera traps are set in the woods near water sources and other high-traffic animal zones, where the animals trigger the traps with their activity. Besides interesting photographs, Chris has a pleasant, easy-going manner as he relates tales like those about a mentor and colleague and these memory snips covering more than 50 years.

From a rewritten press release (also picked up by the New York Times) is a passage spoken by a man who’s probably never studied Civil War letters, or any collection of letters written before the typewriter age:

Computers are second nature for today’s college students. They grew up with technology and the Internet. However, they also grew up with e-mail and text-messaging, which have undoubtedly eroded the fundamental writing skills needed in most professional positions. Shortened syntax, incomplete sentences and no punctuation may be acceptable for instant messaging between friends, but most offices require a much higher level of sophistication, even in e-mails between co-workers. It would be in college students’ best interest to enroll in more writing classes.”

I won’t argue the best point here, that students should take more writing classes, since I think everyone should take more writing classes regardless of their natural abilities, but abbreviation and indifferent punctuation have been going on for as long as people could write. As has good old mistake-making.

But that’s mild quibbling, perhaps, when there is a bigger fish to fry: the unsubstantiated claim that technology is to blame. There is no research I know of—and I am currently heavily reading up on this in preparation for some public presentations—that shows language is, has been, or will be eroded by email or text messaging. In fact, heavy users of those forms of communication have been shown to have higher levels of literacy and to practice more nuanced communication than their peers who do not use those modes of communication.

Be very careful if you choose to defend the sentence including “undoubtedly eroded”: “erode” doesn’t mean “to make people use language in new ways that I am uncomfortable with.” It means “to gradually destroy.”

But there’s a still bigger fish: there’s no study to back up the original press release. It’s a survey of 100 “human resources executives.” (A study is not the same as a survey, no matter how often journalists use them as synonyms.) The claim, as expressed by the Arizona Republic, that “Forty-five percent of survey respondents said written communication is where recent graduates are most deficient,” is meaningless. No matter how many HR execs think writing is a problem, it could be that only 5% of entry-level applicants need better written communication skills and that all the other complaints about entry-level applicants represent still smaller numbers of applicants.

I cannot find the press release on the Internet, including on the web site of the head-hunting firm that released it, so I cannot see if there are more complete numbers available.

For enlightenment about online communication, I recommend reading articles in the Journal of Computer-Mediated Communication, which are freely available online.

UPDATE (8/24): The battle is enjoined! Robert Hartwell Fiske calls his readers “to arms”! As it turns out, speaking good English allows you to beg for pocket change.

Choire Sicha at Gawker calls the Barrett-Fiske kerfluffle a cat fight. We’re frienemies, you might say. For the older set who don’t know that word frienemy, just think of the fake feud between Jack Benny and Fred Allen. (I’m Benny: he could make Groucho laugh.)

Angela Gunn at USA Today’s Tech_Space blog is havin mai seatz, readin mai mailz, because she thinks a fight about language is just the cure for a slow news month.

Gizmodo gives the Wall Street Journal a smack with a blackjack, too. Somebody gets called an alarmist idiot. Gizmodo’s take gets play here, here, and here, among other places.

This joker thinks the article is a sign somebody will start teaching l33t in the classrom.

They love me in Canada part zillion: Teacher Lady has my back.

Network Performance Daily takes a cynical view of the article, too: “In this hard hitting expose by the Wall Street Journal, arguably one of the pre-eminent business newspapers of the world, reporter Christopher Rhoads takes a hard look at a matter of vital importance to the world economy. Apparently, gamer’s shorthand, or ‘leetspeak’ is changing the way that human beings communicate.”

Below is my post that set off this limpest of tempests:

....

Since I spoke at length to the reporter who wrote it, I guess I should be relieved that I wasn’t quoted in this asinine story about online language, “What Did U $@y? Online Language Finds Its Voice,” in the Wall Street Journal. However, I would have preferred that Christopher Rhoads get it right, even if he didn’t quote me.

Comments in the article by my colleague Allan Metcalf are correct and were, no doubt, confirmed by the reporter in nearly identical comments that I made.

But let’s talk about the major problem points:

1. The lede: “TEh INTeRn3T i5 THr3@+EN1N9 t0 Ch@n93 thE W4Y wE $p34k. (Translation: The Internet is threatening to change the way we speak.)”

First, it’s the usual cliché: write a line or two in the supposed slang being discussed in the article, then translate it. Lazy journalists and copy desk editors have been using this hoary mechanism since at least the 1930s.

Second, nobody writes all in leet-speak except as a stunt or unless they are exceptionally clueless, both which, I guess, apply here.

Third, why is this a threat? Why is this a negative? The reporter doesn’t say. That sentence is what I call the “confirming cloud of doom”: it satisfies those who are uneasy with the progress of the world around them, while at the same time it offers no data that can be refuted. It’s editorializing in its purest form.

The rest of the article offers more of this: a few people talking from their gut about how everyone who speaks differently is an idiot.

2. “For years, heavy users of Internet games and chat groups have conversed in their own written language, often indecipherable to outsiders.”

Well, no. Studies of online chat corpora show that only about 20% of chat language is abbreviated or innovative, and of that 20%, nearly all of the uses are transparent, fairly standard abbreviations. The intent in using such language isn’t to obscure, it’s to move faster, so chatters gravitate towards language that is both clear and quick. When obfuscation occurs, it is usually later clarified through context and restatement.

4. “As the Internet becomes more prevalent, leetspeak, including acronyms that used to appear only in text messages like ‘LOL’ for laughing out loud, is finding a voice.”

There is sufficient evidence to suggest that the full form of leet-speak is little more than a historical curiosity mainly perpetuated by ignorant journalists and ironic Interwebbers who use it precisely because it’s stale and out of fashion, kind of like saying, “you bet your bippy” or “it’s the bee’s knees!”

Also, LOL isn’t leet-speak. It’s simply an acronym used online.

5. “The words’ growing offline popularity has stoked the ire of linguists, parents and others who denounce them as part of a broader debasement of the English language.”

This is wrong in at least three ways.

First, the offline, oral popularity of online chat language is limited to a handful of words: LOL, BRB, WTF, pwn, and a few others. If its popularity is growing orally and offline, it’s by one or two words every few years. That’s hardly a trend.

Second, no linguist I know is showing ire about this. I know lots of them, starting with my wife and moving on to my colleagues in the American Dialect Society, the Linguistic Society of America, and the Dictionary Society of North America. I read their journals, listen to them report their research at conferences, and read their weblogs.

Third, Robert Hartwell Fiske isn’t a linguist. He’s a self-involved curmudgeon—that’s not a compliment, but a criticism of his intellectual limitations—who is the go-to guy for the same kind of dismissive claptrap you’ll hear from anybody who’s speaking on language outside their area of expertise.

6. “These ‘elite’ users developed leetspeak, occasionally to conceal their hacking plans or elude text filters. (It still has that use for some: ‘pr0n’ is leetspeak for pornography.)”

This defies logic. It assumes that people who are online are idiots who believe that nobody’s on to them. It’s a secret language! It’s safe, like pig latin! Nobody will ever figure it out!

Bunk, I say. It’s simply a playful way of writing that looks futuristic, different, anarchic, challenging. If it was ever used for obfuscation, that obfuscation lasted about ten minutes.

The idea of “elud(ing) text filters” as the source of l33t is inane, no matter where or how often this myth is repeated. It assumes stupidity on the part of the hacker-hunters, that they would not, after the first encounter with a respelling of a keyword (warez or pr0n, for example) also add it to their text filters.

In my experience on the bulletin boards, “elite” has always referred to the posers and the wannabes—better referred to as the “l33t”—not the truly elite. Such leet-speak was used not by the truly elite, who were having their conversations and file-sharing activities in closed or encrypted environments, but by the naive, the n00bs, and the numbskulls.

Nobody who is truly elite considers themselves elite, nor cares about it, nor would they adopt a language to prove it. It really is a nice manifestation of the saying “I wouldn’t want to be a member of any club that would have me.” In fact, the only sure sign of elite status is invisibility to the non-elite. Otherwise, if you showed only the supposed trappings of being elite (like using the leet-speak lexicon) but not the activities of eliteness (having valuable warez, inside connections, special privileges, admin status, proven hacking skills), then you are not elite but only l33t.