I think that it would be very dangerous to market TPRS as a guaranteed way to remove fossilized errors.
TPRS is a tool designed for a specific purpose (and with a specific group of learners in mind). The creator of TPRS was a high school teacher in California, teaching adolescents Spanish. That was his target group. Keep in mind that when he started working with all this, he was on the verge of getting fired from his teaching job for gross incompetence and failure to keep the kids in line. He had to do something, so he started looking at the research and experimenting to find something that would work in that situation.
People since then have substantially altered what TPRS is (including its name!) and the groups it will work with. I myself have done a lot of work on expanding TPRS or incorporating it as part of a larger system, so that it will meet the needs of those learning Mandarin, which it does not "out of the box" (IMO). But to my knowledge (probably because of lack of commercial opportunity) no one has really applied TPRS on a large scale in a culturally Chinese environment. Most of the (very few) teachers using TPRS (some or most of the time) in a Chinese cultural environment AFAIK are in international schools, which are by definition outside the norm for that area.
I think the idea about educating the pineapple farmers is a very sound one. The aim of TPRS is to find a method that will get most of the people fluent most of the time. Traditional teaching gets a few of the people fluent some of the time. TPRS doesn't claim to get all of the people fluent all of the time -- (we wish, but it doesn't do that) and even the creator of the method says all the time that if someone shows him something that works better, he'll switch on the spot.
TPRS was never intended to deal with fossilized errors specifically. I'm not sure it would be very effective in so doing within the periods of time usually available for classes or tutoring. Think about it -- if a child takes 10,000 hours of input by age 3 or so to acquire the definite and indefinite articles in English, how many additional hours of input would it probably take to undo problems with that and overlay it with correct acquisition of that feature? (Although to be fair, "the" usually wasn't incorrectly taught, it was just taught, which is the core problem -- every student has a different degree of error with "the" usage, usually depending on how much they have to rely on rules and on support from other languages they speak in the absence of an immediate ability to apply the rules for using "the" in speech.)
Where TPRS really shines is with true (or false) beginners; with disadvantaged language learners (people who are "not academically inclined" in the first place, or are handicapped in some way -- all my special ed kids did great with TPRS and it gave them something they could truly achieve in on an equal footing with the other kids) and for learners who have an environment of some sort that includes the TL, so that they "notice" their progress day to day. TPRS tends to be more difficult to accept for students who traditionally do well in school (since memorization is no longer rewarded over just paying attention), students who are very analytically-minded (these are the 4% who can learn languages through textbooks alone, as I acquired Spanish and can't really explain how I got fluent) and students who are preparing for a specific test or wrestling with a specific glaring error(s).
To me, what this means is that in an ideal world, we would have pure TPRS (CI) classes for beginners through the point where they have acquired all the structure of the language -- probably four or five years for English assuming you start in middle or high school -- more if the start were earlier in elementary school because you can't cover as much structure with the little ones. Then a switch to supported extensive reading, writing improvement (="how to edit yourself" in the linguistic sense) and test prep.
If I were living in Taiwan and had work rights, I'd set up some kind of program called "English for Everyone" or something like that, and target housewives and pineapple farmers. But I'm not sure it would work because I don't look Chinese. If I had a Chinese face, I could become a millionaire in Taiwan doing that, but as a foreign face, it's "well only foreigners can teach that way because they're native speakers and anyway you don't want to take a class with a foreigner because XYZ and the other foreign teacher I know says that method doesn't work at all and..."
I still think we should set up a CI organization for the ROC and think about holding an annual conference -- I'd do it the same weekend as ETA, or one day of ETA at first, maybe Sunday when ETA runs out of steam anyway -- and get some people in to train teachers how to use CI. Strands in going all-CI and also in "sneaking" CI into your classroom practice. Get the whole thing to be reliably bilingual from the start, get the materials translated so that Chinese teachers can really understand them (professional development in English is not always a good idea if the aim is really professional development rather than "I went to see a foreign speaker lecture in English") and offer classes to groups like associations for the handicapped, disadvantaged schools in rural areas, or other "non-achievers". Oh, and we'd also need to produce a set of materials for teaching English to Taiwanese using CI, because the existing materials would perhaps not be optimal for them.