[I recently posted an email with similar content on the L-TEST mailing list, in response to a claim that “native speakers” don’t exist.]
It is sometimes claimed that “native speakers” do not exist. But perhaps it is better to see “native speakers” as being (like a lot of social concepts) a fuzzy set.
There are people who are definitely in the set of native users for a language; there are people who are definitely not in the set of native users; but then there are people in a grey area, for whom it isn’t clear whether they’re a native users or not – making “native users” a “fuzzy” rather than “crisp” set.
Davies’ (2003) list of six criteria (childhood acquisition, intuition about grammar, high fluency, etc.) seems a useful rough guide. If you meet all six criteria for a language then you’re probably a native user (NU); if you meet few or none of them then you’re probably a non-native user (NNU) for that language. Of course, each of those criteria can be met to varying extents. For example, for a child born in Germany to German-speaking parents who then moves with her family to Japan, so that she completes the rest of her education in Japanese, it is not clear whether she meets Davies’ “childhood acquisition” criterion as a native user of Japanese or not. And this gives rise to a grey area of people who are neither definite-NU nor definite-NNU. – However, just because there are people who do not fall neatly into either category, does not mean that definite native users and definite non-native users of a language do not exist. Personally, I feel quite sure that I am a native user of English, and a non-native user of Mandarin.
As English continues to be used as the global lingua franca, and many regions increase their English language education at ever younger ages (such as is happening in China), the grey area of people who cannot be easily classified as English NU or English NNU will grow. Meanwhile, the proportion of definite-NNU will decrease. So it is likely that the NU-NNU distinction will become gradually less useful for English, as a higher number of unclassifiable users emerge. – Nonetheless, this won’t render the distinction completely useless, and some definite NU and some definite NNU will still exist.