How one pronounces "Shenandoah" is irrelevant to one's performance of this song. That is, if one imagines one is singing a traditional and/or historical song. The reason being, it's unknown whether the word "Shenandoah" exists at all in the song. There's plenty of evidence to suggest that "Shenandoah" was a mondegreen. If this was the case, the rendering as "Shenandoah" probably (my logic) first appeared in print as someone(s) tried to rationalize the word/name they heard and, later, singers of the song who were exposed to printed versions created their pronunciations under the assumption that "Shenandoah" was the word.

~"Shenandoah" (or whatever word was original/intended) doesn't even figure prominently in the song. If we follow the logic of the vast majority of chanties, the refrains are the only fixed portion, and it is customary to know the songs from the refrains. Hence, the song in question is "Rolling River" (or "Across the Wide Missouri," etc.). "Shenandoah" is not an essential component of the chanty. And yes, it was a chanty— it was mentioned in many testimonies of the 2nd half of the 19th century as and ONLY as a chanty sung in a labor context. NOTHING has been turned up in the 19th century to support the idea that "Rolling River" was sung in any other context. Whall (above) made a claim about it being a non-chanty song a half century before his writing, but no empirical evidence supports that. Whall also made the claim that chanties date back to The Complaynt of Scotland—a work he dated as a century older than it was. Subsequent writers lapped up his statements. It's fair to speculate on the point of when and in what context "Rolling River" was created, but to claim it was anything other than a chanty is pure bullshitting or the repetition of a bullshitter.

There are threads here discussing many songs—all within the ambit of chanties—some of which are clearly variations of "Rolling River" and some which are different songs but which all appear to invoke a variant of the ~name ~Shenandoah. That is, they are variations of that which print authors came to standardize as "Shenandoah." In the majority of these examples, and more so the closer to the source one gets, the variant is something different than "Shenandoah," whether Shanadore, Shannydo, Sunnydo, Salambo, Shanado... perhaps even Shallow Brown and Sally Brown, ... or Seven-long-years... which often fit the same paradigm. One thing that strongly links all these data points is that the songs tend to be associated with people of African-descended cultures of the Western Atlantic. This would not be the first time that writers/singers of one culture misheard an unfamiliar word in another culture's song. (Other candidates for possible mishearings in the chanty repertoire include "ranzo," "hilo," and "rolling king".)

Think that the word _must_ be "Shenandoah," because that is the name of a known thing in the U.S. (as opposed to an unknown thing in the mouths of people from Africa)-- and that all these variations emerged from the lips and pens or people who didn't know the Shenandoah River and, thereby, corrupted the word? I suppose it's possible. But when you put aside all the bullshit narratives of the song in the tradition of Whall (a racist British captain who thought the songs he perceived to be of Black American origin were trash, and who inspired subsequent writers and "researchers" on the topic to ignore the vast wealth of activity of African-Americans STILL singing chanties in favor of meeting decrepit English retired seamen vaguely recalling petrified chanties)... and you put aside what you THINK "the lyrics" are (because some writer standardized some words in a particular volume and then people copied it verbatim, in facsimile after facsimile)... and you also look at the actual lay of the data points-- their appearance in time, who is reporting, etc -- I think you might agree that "Shenandoah" is the product of *standardizing an assumption*, and that you'd have to agree that there is no one correct way to pronounce the word that often crops up in these *songs of African-American folklore*.

(I hope you see what I did there. Framing this set of songs as "African-American folklore" -- something I doubt can be refuted entirely -- would tend to change the way people seek answers about them, as compared to framing them as "sea shanties.")