That is very close my marginally educated guesstimate of 80%. I concede never researching the topic before now, but skimming online sources revealed few surprises. He further claims that, counting non-natives, India has more English speakers than ANY other nation; I found both claims on Wikipedia, which cited his book English as a Global Language as source of the first, and the below linked article as source of the second. http://www.theguardian.com/education/2004/nov/19/tefl

As preface to what (d)evolved into a long response (sorry) and FWIW, part of the problem seems to be that other languages with large numbers of non-native and/or international speakers have official standarization organizations to determine what is/not legitimate, but English does NOT. Consequently, all attempts at defining all- or even MOSTLY encompassing standards are necessarily ad hoc.

Checking shows “dialect” is essentially defined as “morphology+vocabulary+grammar+accent.” That is very unsatisfying, because it leaves no term for the first three exclusive of the last. Yet its practical effect is that ONLY the first three distinguish dialects, because “accent” fully covers the sole remaining criterion (i.e. pronunciation.) In terms of dialect, distinct accents are a case of an attribute “necessary but insufficient” to establish separation

Yet which NATIVE dialect is largest is irrelevant until we answer that critical question of whether to include the 75% of NON-native English speakers. English is the “primary” language of 22 nations, where populations of the 6 “Anglosphere” nations are roughly equlled by that of NON-Anglosphere nations (mainly Nigerias 220 million residents.) It is an (if not THE) official language in 15 more—including India and Pakistan. So restricting the question to the 6 “Anglosphere” nations is like Major League Baeball annually crowing a “World Series” champion despite ignoring all teams in Japan and the Caribbean nations that provide a growing number of MLBs best players. Wikipedias page on global English-speakers relies on Crystals stats extensively (though not exclusively; its Indian figures are from Indias millennial census, and about a third of Crystals estimate of Indian English speakers) and so ends with an estimate similar to his: 350 million native speakers; 850 million non-native speakers: https://en.wikipedia.org/wiki/List_of_countries_by_English-speaking_population

Yet even if we DID dismiss 3 of every 4 English speakers, and even bearing in mind multiple accents are typically grouped under a single dialect, there is no single “Southern dialect” as such: Wikipedias English dialect article identifies no less tha SIX distinct Southern dialects (the four I previously mentioned, but with Louisiania and the Appalachian dialects each split into two separate dialects.) If forced to guess, I would say either the Texan or Tidewater (which I referenced as Piedmont) dialects are largest, but neither numbers >30 million speakers at the very most: https://en.wikipedia.org/wiki/List_of_dialects_of_the_English_language#United_States

I was not aware grammar and words differ within (for want of a better term) "African English," and that does make a difference. I never noticed it among coworkers, but they provide most of my experience with English-speaking Africans. I presumed colonialisms legacy included a largely uniform Commonwealth dialect with many accents, but if those dialects vary as much or more than others it is as difficult to precisely estimate their speakers. How great is the difference? The Midwestern dialect is distinct from others in the US, but Wisconsites referring to drinking fountains as “bubblers” does constitute an accent distinct from the rest of the Midwest; one term does not a dialect make: Are African dialects more comparable to that, or to the half dozen distinct and bona fide Southern dialects?

Precisely what I meant, and one would expect that effect to be even more pronounced among separate but interacting nations using English as a lingua franca. After all, much of continental Europes uniform English is because Western European ALSO often uses it as a lingua franca (if often grudgingly.) According to what I first read when preparing to move here, Norwegian visa applicant are required to speak Norwegian OR ENGLISH, lucky for me since my Norwegian remains awful despite living her five years: Because it is so absurdly easy to speak English exclusively.

The Danish equivalent of "Cops" airs here, and a couple years ago the commercial for it included one of the cops, frustrated by inability to communicate with someone he was arresting, said, "You should learn English better before you travel in Europe." In fact, last years ranking of English Proficiency (apparently there is such a thing) rated Denmark #1 and Norway #5; for a reference frame, Belgium was 9th, followed by Germany at 10th, while France was all the way down at 29th—four spots below India, which has twenty times the population (so proportionately more English speakers.) https://en.wikipedia.org/wiki/EF_English_Proficiency_Index The list of nations by English speaking populations notes that, in 2012, 51% of EU residents spoke English, a total second only to the US on that list (though it falls to third if one accepts Crystals estimates of Indias English-speakers rather than that of Indias 2001 census.)

I would wager kroner to smultringer he never speaks any Southern US dialect. New England may be less geographically mobile than the South, but is also far less xenophobic, so far more New Englanders live/work/vacation overseas.

That is a frankly awful and implausible estimate: Only 10% of the world can carry on a conversation in English? Even though ~70% of all internet exchanges are in English? The US, UK, Canada and Australia account for >400 million themselves: Is it plausible Asia, Europe and Africa COMBINED have no more than 300-400 million people who speak English well? Again, a 2012 survey of the EU alone found >250 million people spoke “conversational” English even if some were not perfectly fluent; there are only 50-150 million conversational English speakers among the 6 billion people outside North America, Australia and Western Europe? That is very dubious; Crystal may (or may not) overestimate English speakers, but Kachrus Three Circles Model underestimates them at least as much with far less cause; after all, Braj Kachru HIMSELF is an Indian native.

I worked with an Italian and Pole here who (embarrassingly) spoke three languages better than I speak two; we conversed almost exclusively in English and almost never had any problem fully understanding each other. Likewise (ironically) on breaks during the multiple Norwegian classes here: Even when NO participants were NATIVE English speakers, conversation was almost exclusively in English unless participants were all from the same country. In fact, my Norwegian society/culture class was TAUGHT in English (by one of those native African English speakers, though the fact he lived in Texas for two decades makes him unrepresentative.) All the various Europeans, Africans and Asians in each of those classes displayed a firm grasp of English. It is possible many of those people were the most educated citizens of their respective countries; I know some were. Yet even so: Among 6½ billion people, even the 10% most educated is DOUBLE the estimate of non-US/Commonwealth English speakers.

There is no way the above estimate can be remotely accurate.

It would certainly not be easy, in any case, but HOW difficult depends a great deal on how we define “dialect.”

Fair enough then, and a good argument for Kachrus “International English:” If such an erratically fluid neutral dialect is the norm among the 75% of conversational but non-native English speakers, that “Dialect That Is Not Dialect” could ITSELF be the largest.

No more (possibly less) than others: Wikipedias English dialects list has <10 for Africas 1 billion residents, but THIRTY-FIVE for 320 million US residents.

It is the case in much of the South, because mobility presupposes movements affordability, and even when it exists does not imply WILLINGNESS to move: Much of the South is impoverished and MOST of it is rural, provincial and insular; historically, those factors do not exactly correlate with numerical or cultural dominance. The biggest reason the Southern dialects are so numerous are 1) defeated and disenfranchised Confederates migrating to western territories where they could assume untainted new identities and 2) the Great Migration of Southern blacks who fled Southern segregation for the North (and its segregation) a century ago. Both groups took their dialects with them, passed them on to their kids and consequently formed large Southern dialect salients in the West and North. Yet there are six dialects even within the Souths own historical borders (of which the northern one has been steadily moving south at least since the Civil War) and >30 throughout the whole US; that is only slightly less than the UKs >40, even if each dialects speakers generally include far more people given a US population roughly five times Britains.

Ultimately, it is very hard to believe the upper limit of <30 million speakers of any given Southern dialect is anything like a plurality of the 1 billion+ conversational speakers of various English dialects. 3% is rarely a plurality of anything, and the proving the case for it here certainly requires more than a lone Southern partisan proudly waving the Confederate Battle flag and conflating accent with dialect to defy emeritus linguists and factual global surveys.