A colleague just quoted from a website on which genie is pluralized as genii.
No, it’s not correct. Genie does derive ultimately from Latin genius (which can be pluralized as genii or as geniuses), but it came to us by way of French, and it’s an English word now, so it’s genies.
But this reminds me of a much larger issue that needs to be addressed: the idea – a rather common one, it seems to me – that there is a Latin plural ending -ii that should be applied to Latin-seeming (and some other foreign-seeming) words. I see it, for instance, when some people write virii instead of viruses.
To be as plain as possible: in Latin, -ii is not a plural ending. Ever. Nor is it one in English (unless this pseudoplural catches on, I guess…). In fact, I can’t think of a language in which it is a plural ending, though there might be one somewhere. Not English, though!
No doubt some of you are saying, “Hey! That’s wrong! What about genii for genius and radii for radius?” Well, what’s the root? Look at the singular: genius. That’s geni + us. Pluralize -us to -i and you get genii. Likewise radi-us to radi-i and so on. The pluralization, in Latin, of -us masculine nouns is -i. If the root ends in i then you get a double i, but that’s because of the root. The Latin plural of virus would be viri; the singular is not virius.
But also, if it’s not a Latin word ending in -us, don’t assume it pluralizes to -i. (It may, as -i plural marking is also used in some other languages, but don’t assume it.) Moreover, if it’s an English word, especially one originally borrowed some time ago from another language, you’re good with treating it as an English word – for example, viruses. You wouldn’t use Latin genitive formations instead of English possessives; you don’t conjugate verbs borrowed from Latin in the Latin way. Unless there’s a well-established borrowed plural, you’re just fine using the good old English -(e)s.
I’m put in mind of Larry Niven’s novels, in which there is a type of creature called bandersnatch. Most of you will recognize this as a word invented by Lewis Carroll for his poem “Jabberwocky” in Through the Looking-Glass. The word is an English invention. The characters in Niven’s books were given the name by an English-speaking biologist, and the characters all speak English. So what’s the plural of bandersnatch in Niven’s world? Not bandersnatches but the more exotic- or technical-sounding bandersnatchi. Apparently -(e)s is too pedestrian and low-level (I really think many people think it is!).
I would venture to suggest that if Niven were creating the creature just now, he might go one step farther and pluralize it bandersnatchii. Which would, of course, be even more frumious.
Sesquiotica 
There’s also a double error in which not only is the word not given the English pluralization that would be appropriate, it’s given a Latin pluralization when it actually comes from Greek. “Octopi” (Greek “octopodes”) is the one I see most often.
One of my math professors told a story about a fellow prof whose colleagues’ pedantry was getting on his nerves. Thinking to make a point, he made a comment about lemmas, “or I suppose you would say lemmae” he sneered. The pedantic colleagues sneered back, “You idiot — it’s lemmata.”
Ha! Wass lemmata wit dem? 😛
The most egregious example of this trend that I’ve seen is the allegedly-humorous pluralization of Elvis (as in impersonators, which abound in, for example, Las Vegas) as Elvii. *Shudder*
Oh! Yes! I knew there was an example that wasn’t surfacing in my mind! That’s it! Elvii would be the plural of Elvius. I know that it’s meant to be humorous, but I always have the sense that those using it think that it would be a proper plural if Elvis were Latin (and just maybe spelled Elvus).
If, on the other hand, Elvis were a Latin word like pelvis, then its proper Latin plural would be Elves – no surprise that no one uses that!
I have the impression, unsubstantiated and not necessarily accurate, that the number of foreign-style plurals is increasing in English usage, even including Latinate plurals on words that have a long tradition of being used as ordinary English words. To the extent this is the case, I’m sure that it’s linked to the common reflexive preference for more marked forms – the belief that they are somehow more formal, proper, correct, or authentic. I discuss this in my tasting note on kneck.
If that’s the case, then perhaps in a couple of hundred years, the use of (e)s for plural will be archaic, replaced by more marked forms. My (admittedly limited) understanding of linguistics suggests that that is one of the paths of language evolution, along with the seemingly-paradoxically-opposite tendency to simplify things.
However, today I read another Latinate-ization (?) that suggests another reason. In a decent post on science reporting, Carl Zimmer (humorously) says “Leaving tuchuses aside (tuchi?)”. (He’s a good enough writer that I’m fairly certain that he knows that ‘tuchus’, even if it is spelt that way, is certainly not Latin.)
Anyway, reading that, it occurred to me that putting an ‘əs’ sound after an existing ‘əs’ sound, as in Elvises, viruses and octopuses, is at least slightly awkward to pronounce. In English, at least (and I suspect universally), we find it easier and more natural to follow a consonant with a different sound (compare, for instance, how quickly you can say “buh-buh-buh-buh-buh-buh-buh-buh-buh” versus “buddabuddabuddabuddabudda” or “buggadabuggadabuggada”).
So, ending a word with two ‘əs’es ( 🙂 ) is a bit awkward, and may tend, if nothing else, to reinforce the tendency you mention, to look for a more marked form.
I think the phenomenon has a lot in common with the fact that people don’t know how to signify possession with non-plural words that end in ‘s’: “Charles’ Garage” (which is, admittedly, better than the “Charle’s Garage” that you used to be able to find in downtown Vancouver); people just don’t deal well with two ess (or zed) sounds together at the end of a word.
Compare the incidence of pluralizing nouns that end in ‘us’ (or even ‘is’!) with ‘i’ to that of pluralizing nouns that end in ‘a’ to ‘ae’. No one would say, “When we were hiking, we saw 3 cobrae,” or “The waiter brought us our tequilae,” even though those words are pretty borrowed-sounding. (There are other words that people do do this with, such as formula or ulna, but I think that the general population usually sees constructions like formulae as being stuffy and snobby, while the us-to-i pluralization is (generally) still seen as being funny (or trying).
You got me on the reductio ad absurdum there. Yes, more ordinary words won’t likely change – ones that don’t have that kind of foreign sound, et cetera.
I think you have a good insight there with the /s/ endings. We do occasionally see pseudo-Latin plurals on other forms, but it seems quite reasonable to imagine the battling fricatives would be a motivator.
Ah, yes, and I did after all do a note on Elvis that went over this turf once already…
I meant to say this back when you posted this comment: “Battling Fricatives” is my new favourite band name.
Argh, I mean the comment where you used that term. Stupid WordPress doesn’t have a Reply link for it. 😦
See also https://sesquiotic.wordpress.com/2009/11/09/why-fetuses/
Italian “Zio” turns into “Zii” when plural.
Meaning “Uncle” and “Uncles”.