I will illustrate this question with an example, since I think it's much easier to see what I'm asking there, rather than from an abstract question.
I am a native English speaker. I was, for a while, an expatriate living in Luxembourg. On wikipedia articles about Luxembourg and Luxembourg-related things, the adjective "Luxembourgian" was often used. I edited this out whenever I saw it, since it was a word that no one (native speaker or otherwise) ever says in the English speaking community in Luxembourg. (Luxembourgish or sometimes Luxembourgeois are accepted). However, despite my fairly limited efforts, "Luxembourgian" still gets used on wikipedia, and I've heard people (when forced to guess what the adjective might be) plump for this ugly sounding word.
So here's the question: do the native English speakers living in or near Luxembourg have some kind of privileged status when determining what constitutes correct usage for Luxembourg related words? Or to put it another way, if enough people who know very little about Luxembourg think it's probably "Luxembourgian", does that trump the ex-pats considered opinion?
Best Answer
There are a few issues I want to address here. First is the question of distinguishing preferred usages from correct usages. Correctness or grammaticality is a question of overall usage by all native speakers of a language. It is indefensible to label common usages from native speakers as “incorrect”—the closest we can get is to label some usages “nonstandard”, where speakers of the standard language nearly universally condemn a particular usage at a particular formality level. It is important not to confuse informal usage with incorrectness, but it is certainly reasonable to call for example ain’t non-standard for formal English. At the same time, while I believe it is indisputable that using both less and fewer for count nouns is standard English, individuals certainly can have preferences, using only fewer for count nouns and reserving less only for non-count nouns. Deciding on such preferences is up to the individual writer or the organization the writer is writing for, but don’t forget that individual or institutional usage preferences are just that—preferences—and deviation from those preferences does not constitute “incorrectness”, because using dispreferred usages is still grammatical English.
Next is the question of whether the usage of a group which is more closely connected to a topic (“in-group” usage) gets more weight in preferring a usage. I think there is a strong argument against following in-group preferences because often in-group usage is highly jargony and the general speaker or writer will usually want to avoid jargon or group-specific language. For example, the general writer may prefer to use the word hacker to describe people who breach computer security even though the in-group preference is for cracker for the simple reason that the lay audience will find discussion of “computer crackers” and not “computer hackers” to be puzzling. It is my belief that consideration of the reading audience outweighs consideration of the group being discussed. However, not everyone agrees with this philosophy, and some people may feel compelled to respect a connected group’s preferences. An example of this is in this question about the pronunciation of Islam, where some posters expressed a preference for respecting Muslims by pronouncing the word the way they would prefer to; that is, as closely to the original Arabic as possible. I, personally, don’t find such arguments to be persuasive, but many people do use that kind of criteria to guide them in selecting a preferred usage, and I think that is a defensible thing to do, even if it results in confusing or misleading the audience. Remember I am only addressing now the criteria for preference and not the criteria for correctness, so I am careful to distinguish the legitimacy of choosing a preferred usage from labeling dispreferred usages as “incorrect”.
Finally we get to the example posed in the original question. Even if you want to use emulation of in-group preference as a criterion for which usage to prefer, amateur linguistic analysis of the sort “it was a word that no one (native speaker or otherwise) ever says in the English speaking community in Luxembourg” tends not to be very reliable. Such statements are more often wrong than they are right, being victims of multiple cognitive biases, such as confirmation bias and expectation bias. Statements like that are a big red flag—in my experience people only come to such hard-edged conclusions as a defense mechanism for coping with the cognitive dissonance from encountering counter-examples. I generally assume the opposite to be true until I see some kind of authoritative support or statistical analysis.