Lisa’s post got me to revisit the name Tverdovsky, apparently the standard Anglicization of the Russian name Твёрдовский, if I’ve got it right. Like Dvorak and Vlasic, Tverdovsky has an initial cluster which is illicit in many non-Slavic languages, and which is often resolved with epenthesis. Further, since the Anglicized form of Tverdovsky carries penultimate stress, the epenthetic vowel is before a weak syllable, so in speech it sometimes is stressed, being realized as [ɛ]. When I first remarked upon it, I did not look for Google misspellings, but I have done so now and can report on the results. Updated 2/7/05.

It turns out that the length of the name is both a blessing and a curse. The benefit is that misspelled forms like teverdovsky don’t end up looking like other real items, so the Google counts aren’t skewed. In contrast, running a similar survey for the shorter name Kvasha is difficult, because the search term Kevasha nets some hits that might not be errors of the same kind:

The 14th century Kevasha temple in Somnathpur is another popular destination, and can make a pleasant afternoon trip.

However, the length of the name adds other junctures at which poor transliteration can occur, notably for the second vowel, the second v, and the final vowel. Here’s a schematic of the combinatorial possibilities.

I haven’t searched all the combinations; it was clear that the o option, as in Tvordovsky, is exceedingly rare. I also left out the r substitution, which is due to a single broadcaster’s error that I reported here and here. Others have noted it on message boards:

i was stuck on thorne still saying TEVERDORSKY. why, after all these years, can’t he pronounce his name right? it’s pathetic.

So the combinations I used are these:

Here’s a sampling of the search results:

Cica vroiau sa-l aduca pa Oleg Teverdovski da nu o sa-l aduca sigut pt ca are mai putin de 35 de ani iar Toronto nu vor sa semneze pe nimeni care sa fie asa…

Je parle Teverdoski on aurait été perdant encore une fois…

… I am sure York and Carney, even Tvordovsky would have fetched some good prospects and or picks. But this can also be done over the summer. …

…Favorite players are Brodeur, Andreychuk, McSorley, Robataille, Hebert, Storr, Vanbiesbrook, Hull, Terreri, Fuhr, Roenick, Khabibulin, Tverdofsky, Irbe, and a…

…Scott Stevens, Easton Synergy, No. Chris Tamer, Sherwood. Yes. Oleg Teverdovsky, Bauer Composite. No. Oleg Teverdovsky, Easton Ultra Lite. Yes. Chris Therien, CCM …

He played along side Superleague greats such as, Oleg Teverdovsky, Teammate Maksim Kondrat’ev, Superleague leading scorer Aleksandr Prokop’ev, and Sergi…

… The other players selected from the domestic league are goaltender Maxim Sokolov, defenseman Oleg Tverdovski and forward Maxim Sushinski, all three from the…

Google also could tell what I was up to:

teverdovski: Did you mean: teverdovsky
teverdosky: Did you mean: tverdovsky

The distribution of hits is below, first for the –sky forms, and then for the –ski forms. In both tables, orthographic epenthesis is overwhelmingly disfavoured, but it occurs enough to suggest it’s entrenched for some users.

Whether this means that these speakers represent the inserted vowel underlyingly is a different question, and one highly parallel to last week’s lengthy flap debate.

For laughs, I also searched using other orthographic vowels as resolvers, with some pretty firm results:

taverdovsky 0
tiverdovsky 0
toverdovsky 1
tuverdovsky 0

It really seems like e is a favoured orthographic representation for inserted vowels, at least for Roman-alphabet webpages that mention Russian defencemen named Oleg. But a broader survey (i.e., of more than one target item) could prove more challenging given the issue mentioned above with Kvasha (also an Oleg, but not a defenceman).

Update: in light of some other posts appearing shortly after this one appeared, I have a minor addendum to make. Eric posted a discussion of intrusive orthographic [o] in Cleopat[o]ra and Sumat[o]ra, a befuddling fact. Philip then offered an account based on Japanese epenthetic patterns; [o] is suitable after coronals (as in these targets) because [u] (the default) and [i] (the next best) both condition palatalization of the preceding [t]. With an extreme caveat about drawing any reliable conclusion from the solitary google hit mentioned here for toverdovsky, I should just add that the site in question was from the domain geocities.co.jp. Nevertheless it’s all consistent; according to Philip’s account, [o] is exactly the vowel you would expect in a Japanese borrowing of Tverdovsky. Why it only happened once in all of the Japanese web space is a different question – maybe nobody in Japan has any time for the New Jersey Devils (or, these days, Omsk Avangard).