Appears in the Spring 2013 issue of California Linguistics Notes.
A Preliminary Survey of Some Uralic Elements in Costanoan, Esselen, Chimariko and Salinan, by one A. Fournet, proposes the possibility of a relationship of the Uralic languages — especially the more Western ones, though I imagine this focus could be due to Fournet’s familiarity with Mokša in particular — with a number of languages spoken in northern and central California!
A curious side point is an assertion that Yukaghir and Mongolian are also to be related to the languages in question, though more distantly so. (The main value of this to the proposal seems to be in suggesting that these groups help in estabilishing whether Uralic *k represents original *k or *q, a distinction that would also surface in Costanoan — most strange, since I believe this distinction is already thought to be secondary in both Yuk. and Mong., conditioned by the vowel backness as usual.)
As the hypothesis appears to be very much a work in progress, I will refrain from shooting it full of holes, even though it does not appear too promising yet… a few points stand out immediately for critique, though. As one of the comparanda is a clear Indo-Iranian loanword (“man”: Sanskrit mṛta “mortal” ~ Mordvinic *mirďe, Permic *moort ~ Mutsun mirṭe), Fournet dates the supposed intrusion of “Cal-Uralic” into the Americas as relatively recent (≤ 4Ka). I wonder what pointing out that the data also contains three even more recent loanwords from Germanic and Baltic will do, then?
- Finnic *käü-, *käve- “to walk”, a known loanword from Germanic *skeewjan-.
- Finnic *liiva “sand”, from Baltic (cf. Lithuanian glyvas); probably loaned thence into Permic (as covered in J. Saarikivi (2006), Substrata Uralica: Studies on Finno-Ugric Substrate in Northern Russian Dialects), and further from Komi into Kazym Khanty.
- Finnic *kurkku “throat”, from Scandinavian *kʷerku-. (Mordvinic *kərga probably rather comes from Slavic *kъrkъ “neck”.)
Of course, the “Cal-Uralic” proposal can also be read chiefly as a critique of the idea of Sapir’s Hokan and Penutian “stocks”, to which the four American language groups under consideration have been assigned: if this affiliation is based on no stronger evidence than what is being currently brought forward for Uralic affiliation, the idea definitely should be questioned…
Freelance reconstruction 
Wow, that’s bad.
Penutian is being worked on; it does look real. No idea about Hokan.
Tietty koko juttu on roskaa. Kiinnostavaa, että tyyppi kommentoi sci.lang:ssa, joskus ihan ok jopa. Joskus tulee aivopieruja: