I am making educational videos (in maths, science, etc.) that I plan to put on YouTube and perhaps elsewhere.
I am hoping to reach as wide a global audience as possible. Obviously then, if there is any one language I should use, it is English.
My question here is: If there is any one single accent in English I should use, what is it? Is there any one single accent in English that is considered by linguists to be the most widely and easily understood across the world? (Similar Quora question.)
Having taught English for a few years, and being a non-native speaker of English and a native speaker of Czech, I think that if you want to be understood as widely as possible, the variety to choose should be a "rhotic" dialect. Although I generally prefer RP, because most of my teachers were British, perhaps, I have noticed that the disappearing /r/'s often seem quite confusing to the learners, and even if they learn not to pronounce them in (pre-consonantal) syllable codas, they usually keep having problems catching the word. Add to it the "intrusive /r/'s" and "linking /r/'s", and you can see one possible source of trouble.
Frequent disappearance and reappearance of sounds might turn out to be a rather universal complication for most, if not all, English learners. On the other hand, the issue is far from this simple, as the ease of comprehension always depends on the particular learner's native language (especially phonological) features, and on their experience.
If "being understood globally" is really what you are pursuing, then I would recommend something along the lines of Standard North American, perhaps, which most of my own students tend to find way easier to follow than, say, Received Pronunciation (interestingly enough, the relatively great differences between the two vowel inventories don't seem to pose serious problems for most Czech learners). On the other hand, you might want to consider making it just a little more difficult for the learners using a non-rhotic variety instead, as this could prepare them a bit better for the every-day reality of having to cope with the tremendous (especially phonetic and phonological) variability of both standard and non-standard (or native and non-native) Englishes.
I agree that a rhotic dialect would be preferable, and also along the same lines of preserving distinctions used in other dialects, use a dialect where "cot" and "caught" are pronounced distinctly (I used to suggest central Ohio, but the distinction is dying there). However, there are a number of casual-register phonetic reductions that are now fairly standard in the US, which tend to make American English difficult for non-speakers to understand. In actual casual speech, vowels get seriously reduced or deleted, consonants and entire syllables disappear or are radically changed (e.g. k→x in flapping contexts; coda reduction of /t,k/ in the direction of glottal stop). These are mostly register-governed rules, so if you aim for a slower, less-conversational register, that should bring these features under control.
