A former classmate of mine e-mailed me to complain about Justice Breyer's recent opinion in Burlington Northern Co. v. White. Among other things, she faulted it for using "at least one split infinitive":

As the opinion wants to split hairs, in a manner of speaking, over Congressional intent and matters of plain usage with respect to the English language, I would ask you to ask Justice Breyer's clerk to go back and to proof the opinion before the Court publishes it. An opinion that scrutinizes language should not have any split infinitives.

I've often come across this assertion that split infinitives are somehow wrong. The modern usage dictionaries that I have seen describe them as fully standard. Webster's Dictionary of English Usage (1989) reports that "the objection to the split infinitive has never had a rational basis." The Harper Dictionary of Contemporary English Usage (1985) calls it a "pedantic bogey." Bryan Garner's Dictionary of Modern American Usage (1998) likewise says that they're in principle just fine. So even if one is a prescriptivist who believes in relying on the prescriptions of the authorities, the weight of the authorities is firmly on the side of splitting being just fine. (There are surely reasons to avoid certain kinds of split infinitives, for instances ones where there are many adverbs between the "to" and the verb, e.g., "to boldly yet carefully and thoughtfully go"; but that's a problem with the particular usages, not with split infinitives generally.)

Nor is there any logical reason to avoid split infinitives, even setting aside the difficulties with using logic to analyze English usage. The origin of the anti-split sentiment seems to be that in Latin infinitives just can't be split. But English isn't Latin. Moreover, sometimes unsplitting will change the meaning or at least the emphasis: Consider Garner's example, "she expects to more than double her profits next year," or Fowler's, "modifications intended to better equip successful candidates for careers in India."

Unsplitting also often makes the revised version sound stuffier, at least to my ears: Consider Breyer's split infinitive, "such a limited construction would fail to fully achieve the anti-retaliation provision's 'primary purpose.'" "Such a limited construction would fully fail to achieve" means something different. "Such a limited construction would fail to achieve fully the anti-retaliation provision's 'primary purpose'" is fine, but it strikes me as clumsier -- a personal judgment that I wouldn't foist on others, but that's more than ample to justify Justice Breyer in writing "to fully achieve."

When I mentioned most of this to my correspondent, she replied, "I disagree on split infinitives. They are bad form." Well, it's hard to argue with a distaste for split infinitives, just as it's hard to argue with a distaste for butter pecan ice cream. Yet it likewise seems to me hard to credibly condemn others for not sharing one's disaste.

I'm often skeptical of claims that some common usage is "wrong," partly because I'm not sure that there is a coherent and useful definition of linguistic "wrongness" other than "divergent from common usage."

But there are often good reasons to avoid certain usages, even if they are technically quite correct. Using "fulsome" to mean "abundant" is one example. The Oxford English Dictionary gives "Characterized by abundance, possessing or affording copious supply; abundant, plentiful, full" as the first definition for "fulsome," attested back to 1250. The definition "Of language, style, behaviour, etc.: Offensive to good taste; esp. offending from excess or want of measure or from being 'over-done'" dates back only to 1663. Neither is listed as obsolete.

Yet while it's hard to say that it's somehow linguistically "wrong" to use "fulsome" to mean "abundant," and while many such usages might not even be ambiguous, they are still likely to be rhetorically ineffective: If you want to convey a positive or a neutral message, you shouldn't use a word that will bring up a negative image in the listener's mind, even if the listener will quickly realize that you're using the term in its positive or neutral sense. For instance, a positive review of the Diablo cigar in Forbes FYI (Feb. 24, 2005) should probably not have said, "It has a nutty aroma and a fulsome flavor that will stand up to the bullying of a big after-dinner Cognac."

Now there may be times that some people might choose to use a word despite the possibly negative reactions that it may evoke in some readers. For instance, I know that split infinitives annoy people, but I think that unsplitting the infinitive often makes the phrase sound stilted, so on balance I'm happy to keep splitting. (I'm also an obstinate fellow who's willing to fight this battle even at some modest cost to the rhetorical effectiveness of what I write.) I use "handicapped" and "rule of thumb" in spite of the fact that some people, who are duped by false claims about the terms' origins, consider them to be offensive; that's just cussedness on my part.

But when a word's unwanted connotation stems from an alternate meaning, and not what I see as an unsound rejection of the word that deserves to be fought, I prefer to avoid conjuring up that unwanted connotation; and I'd counsel others to do the same.

(Thanks to Ben Barros for reminding me about this matter.)

Many usage debates, I think, are actually debates not just about the overtly contested words (e.g., "fulsome") but also about other words: "is," "should," "correct," "error." One commenter on the split infinitives thread, for instance, writes "It is generally bad form, to be sure, and it should be avoided." What I don't quite grasp here is what "is" and "should" mean here. Does that mean that the word is "bad form" in some objective sense, and should be avoided for some reasons that are in some sense obligatory? Or does it just mean "I don't like it, and I like writing that avoids it"?

Another commenter complains about people's use of "octopi" to mean "octopuses":

My favorite [faux language-police correction] was when I was corrected when referring to "the octopuses" by someone who told me that the correct plural was "octopi."

What makes this faux correction even better is that the split-infinitive police are correcting an acceptable option with another acceptable option. The octopus police are correcting an acceptable option with an error.

"Octopus" is not a Latin root, but a Greek one. A proper Greek plural would be "Look at those octopedes!" "Octopuses" is also correct. "Octopi" is not.

I certainly agree that "octopuses" is a perfectly legitimate plural for "octopus"; and I don't like "octopi," partly for the reasons that the commenter mentions.

But what exactly does "error" or "not" "correct" mean, when used about "octopi"? "Octopi" may well be bad Latin and bad Greek. But according to the English dictionaries that I've consulted, "octopi" is a fully standard English word. It may be grating to some people, because of its air of pedantry that proves to be etymologically ill-grounded pedantry. It may not represent "logical" etymology, though the life of the English language has often not been logical. Yet under what coherent and useful definition of "error" or "correct" can we condemn "octopi" as incorrect or erroneous, rather than merely inelegant?

UPDATE: For those who want more sources, the online Oxford English Dictionary reports for "octopus," "Plural octopuses, octopi, (rare) octopodes." Two of the examples it gives, an 1834 edition of Cuvier's Animal Kingdom XII and a 1942 National Geographic article, use "octopi." My New Shorter Oxford lists "octopi" as well; so does my American Heritage (4th ed.); so does the Cambridge Advanced Learner's Dictionary.

While checking still more sources, I have found two contrary ones -- Garner's Modern American Usage calls use of "octopi" "mistaken[]," and the New Fowler's Modern English Usage says it is not "acceptable" and is "misconceived." Webster's Dictionary of Modern English Usage, on the other hand, doesn't condemn octopi. The majority (though not unanimous) view from the sources that I've consulted, then, supports octopi as an acceptable plural, and it's hard for me to see why the usage listed in the OED, the American Heritage, and the Cambridge would be "wrong."