Not necessary, because the argument concerns the specific issue of whether a language is a sequence of separate words. What is necessary to “understand” a sentence is a separate issue.

(B), if anything, would hurt the argument if assumed. If those who don’t know a language cannot hear how speech actually sounds, that’s a reason to find the native speakers’ perception more accurate than the perception of someone who doesn’t know the language.

Not necessary, because the argument doesn’t concern the amount of attention people pay to language. What matters is whose perception of the sounds is more accurate. Whether they pay close attention or not is a separate issue. (Don’t assume that you need to pay close attention to a language in order to accurately assess how it sounds.)

Not necessary, because the argument relies on the perception of people who don’t know a local language. That doesn’t require the author to believe anything about “accomplished non-native speakers of a language.” That group is different from people who don’t know a language.

Necessary, because if it were not true — if native speakers’ perceptions of their own languages ARE more accurate than those of people who don’t know the language — then we have no reason to think that native speakers’ perception is false simply on the basis of the perception of those who don’t know the language. Maybe the perception of the people who don’t know the language is wrong, and the native speakers’ perception is right.