You are interested in converting individually crafted emails, not 'newsletters'?
There are plenty of 'usability tests', but they are almost always subject or task specific. What are you trying to do, sell something, interest a member of the opposite sex or get yourself a job? There are plenty of studies of those types of emails/cover letters.
If you cannot find a study for your topic, do you really need someone to tell you that spelling, grammar, P+Q's and one's command of language matter?
Obviously txt speak with mishtakes is not a good idea and neither is over enthusiastic promulgation of big words. People get scared of people on many ends of the literacy scale. Less is more and a lot harder to write.
The most important thing to get right is that the email actually arrives in the first place, i.e. the inbox and not in the spam folder. This is the first hurdle and where emails are most likely to fail. Even people with SPF records, the correct reverse-DNS and/or DomainKeys implemented can still have it setup wrong, typically with their MS Exchange server (that they actually use for sending the hand-crafted emails) not being on the SPF record for their domain. Testing, testing and more testing is really important to make sure your emails get delivered to whomever and where-ever.
I am surprised that this is not given a mention in your linked article. The letter equivalent is 'writing the address on the envelope clearly, adding the correct postage and making sure the letter gets posted'.
From a customer service point of view the most important thing is to respond in a timely fashion. If you do not reply in 24 hours people will think bad things of you and the organisation you work for/represent. You can win them back with a really good, considered reply some days later, but, by then they may have had a response from another rival of yours. It would be nice to provide better than anecdotal evidence of this, however, you can do that yourself - write to half a dozen companies wanting help with whatever you need and see who gets your follow-up. The first impression from the person that writes back with a useful answer will invariably win.
As for A/B testing, this is well established in newsletter mailouts. It is also sophisticated in that you can test click throughs and find out who read what. If the delivery to a large group of people is first A/B tested on a random sample then you can get the right version sent out to the group as a whole.
If you want to have a go at learning from the newsletter 'best practice' have a go at civiCRM civiMail module.