I haven't even gone to bed last night yet and today's bullshit coronavirus story is already here, courtesy of @pbleic.

Elsevier's journal _Medical Hypotheses_ brings us a paper entitled "The use of aspirated consonants during speech may increase the transmission of COVID-19".
That's right. The hypothesis is that English speakers may be more likely to spread COVID because the aspirated consonants at the end of words (p,k,t) generate respiratory droplets.
The paper draws on an obscure 2003 letter in The Lancet, which made a similar argument about SARS.

That letter is four short paragraphs long; you might as well read the whole thing. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7135525/
The 2003 letter is based on the fact that many US tourists to China, Hong Kong, and Taiwan caught SARS, but no Japanese tourists did.

That's it. The claim is based on that one comparison.
Also odd: the claim that language you speak determines your chance of *getting* the disease in China/HK/Taiwan, not your chance of spreading it. This is blamed on shopkeepers—silly, all the more so given the difficulties in following aspiration patterns in a second language.
(In _Calling Bullshit_, @jevinwest and I write about the importance of considering alternative hypotheses before accepting data as evidence of a claim. Could there be other differences between US and Japanese tourists beyond three aspirated consonants?)
So back to the new paper.

The authors write "In this paper, we support the hypothesis reported in Inouye [the letter discussed above] that aspirated consonants might produce more droplets in comparison to unaspirated consonants."
Let's see how they support it.

They do a manifestly silly t-test comparing COVID incidence in countries with and without aspirated consonants.

But it's not the ridiculous design that is the best part.
The best part is that their analysis provides ZERO support for their hypothesis. Their t-test is non-significant. It's not even close. If I understand what they are doing, their t value of 0.73 with 18 df gives p=0.237 ( @pbleic got the same value).

Here are the data.
But of course we get the old "not significant but in the right direction" chestnut.
Finally, I want to clean up a bit of my own bullshit. Earlier in this thread I referenced aspirated consonants at the end of words. In English we don't aspirate ending consonants.

If you're familiar with the International Phonetic Alphabet, you might say that I fuktʰ upʰ.

/fin
You can follow @CT_Bergstrom.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: