Stephen Carlson wrote:We don't have any way of distinguishing such sub-phonemic details for a dead language. In most modern European languages, the difference is insignificant. (But Spanish is a notable exception.)
Generally, languages with a trilled 'r' also permit a tapped 'r' in certain environments, such as unstressed positions. In Modern Greek, the 'r' is tapped between vowels and trilled otherwise. Koine might have worked the same way, or different regions could have done it differently. We just have no way of knowing.
Yeah that is what I suspected. If you go to Mounce's website that compliments his grammar book, he has a page for each chapter in his book, and each page also has a subpage for vocabulary where he has an audio file of him pronouncing vocab the Erasmian way, along with a woman pronouncing them the Modern Greek way. She taps every single rho, if I recall correctly, except for ῥῆμα which she rolls. But this seems to contradict what you say about Modern Greeks only tapping between vowels, because the woman on Mounce's site taps 98% of the time, between nouns, between a noun and a vowel, etc.
Of course, if it doesn't really matter, then it doesn't really matter. Personally, I find it easier to trill after a vowel or unvoiced stop, and tap after an unvoiced and aspirated stop.