March 03, 2007


Mark Liberman has just written about "incorrections" -- corrections that are themselves incorrect (in Bill Safire's phrasing).  In the simplest cases, the advice is to replace a correct variant that is falsely believed to be incorrect by another, also correct, variant; Mark's example was the advice to replace entitled by titled in examples like "a column entitled 'Incorrections'".  In more advanced cases, the advice is to avoid a correct variant, which can lead to self-editing in which the end product is a variant that is awkward, unfortunately ambiguous in a way the original was not, etc.  In the worst cases, the advice can lead to self-editing in favor of things that are just flat unacceptable in the intended sense (or even in any sense at all). 

Today's example of truly unfortunate self-incorrection comes from a news bulletin on NPR's Weekend Edition Saturday, about people whose HIV status was revealed

(1) because of a likely clerical error.

This is almost surely the result of avoiding using likely as an adverb, as in

(2) likely because of a clerical error 'probably because of a clerical error'.

You might not be entirely happy with (2), but (1) won't do at all in the intended sense.

I'll get back to (1) in a little while, but first some background examples.

People have been observing for around a century that avoiding stranded prepositions and split infinitives can lead people into rocky terrain.  Here's a case of simple self-incorrection: avoiding a correct stranded preposition, in a work-around I reported on the ADS-L back on 9/17/01:

... heard on a local show ("Minds Over Matter") on the San Francisco radio station KALW.  Talking about a city street, one of the panelists asked, "Who is it named after?  I mean, for whom is it named?"

What makes the example nice is that the speaker didn't just shift from a stranded preposition to its pied-piped [i.e. fronted] equivalent ("After whom is it named?" -- which to my ear is just awful, though perhaps not so bad as "After which parent does Kim take?" as a substitute for "Which parent does Kim take after?", or "For what did you eat that fish?" as a substitute for "What did you eat that fish for?"), but seems to have unconsciously perceived where that strategy would lead her and shifted the preposition as well, to one that's more acceptable in pied piping.

Whew!  Good save!  But there was nothing wrong with "Who is it named after?" in the first place.

Other times you end up with stuff that's flat unacceptable, as in this passage in LaTeX/dvips documentation (reported to me by Geoff Pullum, 8/28/02):

The graphics backend driver now knows with what you are TeXing the document, so it can go out and look for the file with an admissible extension...

Pied piping is awful with what in an embedded interrogative.  It's also awful in exclamatives, as in the second examples in the following pairs

What a curvy road we are driving on!
On what a curvy road we are driving!

What a dirty room the children are playing in!
In what a dirty room the children are playing!

(These are from a net discussion of preposition stranding I posted on some time ago.  The point there was that some people found preposition stranding so offensive that they couldn't accept either of the alternatives above, even the perfectly correct first alternatives.)

On other occasions, people try to fix stranded prepositions by just omitting them, as Mark noted in connection with this dreadful example, where the (believed to be) offending final to or in has been deleted:

Designed to think the way you do, the technology is smart, simple and invisible. A fresh new contemporary space that takes luxury into the wireless modern world it belongs.

For split infinitives, you don't usually get gross unacceptability through self-incorrection.  Instead, the problem is that moving an adverb out of an infinitival VP (to avoid a split infinitive) can cause it to be parsed, unfortunately, with the higher verb, as in these two examples:

Mr. Blackburn added that the Panhandle Regional Narcotics Trafficking Task Force failed adequately to supervise the agent, Tom Coleman, in its eagerness to win battles in the war on drugs. (NYT 3/11/04, p. A14)

Losses widened on Wall Street after Secretary of State Colin Powell failed totally to account for its weapons of mass destruction. (wire-service article during the run-up to the 2003 war in Iraq, as reported by Bill Walsh, The Elephants of Style (2004), p. 64)

Sometimes, however, infinitives are obligatorily split, and moving an adverbial out will result in unacceptability, though the occasional example occurs, via self-incorrection of the worst sort.

It is rubbing salt into the wound more than to double the charges at such short notice. I appeal to the Minister and to his right hon. and hon. ... (parliamentary debate, from my 2004 posting)

Now to return to likely as an adverb.  The perceived problem here might be that likely is an adjective and so should not be used as an adverb; according to this view, that would be as bad as writing "I feel real sorry" for "I feel really sorry".  As MWDEU reports in its entry for likely, things like "He will likely be elected" have been disparaged for a century now, though the usage is old, dating back to the 14th century.  The matter is complex: many usage writers accept modified likely ("He will most likely be elected") but not unmodified likely, and unmodified likely does seem to have gone out of fashion in British formal writing, though not in American.  MWDEU judges it to be standard usage in the U.S.

Despite this, many people avoid the adverb likely and move it into a position where it's available as an adjective, as in (1) -- which, unfortunately, means that the clerical error is likely, not that the cause was likely to have been a clerical error.  This is the worst sort of self-incorrection.

You can see why the writer of the news bulletin might have wanted likely rather than its closest alternative, probably: likely is stronger than probably.  But then you're stuck -- either go with adverbial likely, or settle for second-best probably, or re-cast the whole business with it is likely that.  Moving likely into adjectival position just won't do.

Another wretched alternative would be to take seriously the idea that likely is only an adjective, so that its related derived adverb, likelily, is what you want.  No, you really don't want likelily, or any other -ly adverb derived from an adjective ending in the suffix -ly (princelily, courtlily, etc.).  They're tremendously awkward, as that famous linguist James Thurber observed wryly in the "Ladies' and gentlemen's guide to modern English usage" section of his 1931 The Owl in the Attic (p. 151):

Another adverbial construction which gives considerable trouble, or will if you let it, is the adverb ending in "-lily."  The best thing to do with the adverb in "-lily" is to let it alone.  "Lovelily" is an example.  You can say "he plays lovelily," but even though the word is perfectly proper, it won't get you anywhere.  You might just get by with it at a concert; but try shouting it at a ball game.  There isn't one person in ten who will go ahead with a friendship in which the "-lily" adverbs are likely to occur.  The possible endings of this sort are numberless: you can even say, and be right, "heavenlily" and "ruffianlily."  It is especially advisable to avoid this construction because of its "Thematic Potentiality."  Thematic Potentiality is the quality which certain words and phrases have of suggesting a theme song--that is, some such thing as "Heaven Lily O'Mine," "Ruffian Lily, Come Back to Me," "Love Vo-deo-do Lily," and so on.  Think of something else.

In a word, the -lily adverbs are ridiculous.  You will likelily want to avoid them.

(Card-carrying linguists have looked at -lily words.  See, for example, Lise Menn & Brian MacWhinney, "The repeated morph constraint", Language, 1984. )

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at March 3, 2007 02:46 PM