Here’s the problem with an all-TPRS Chinese program: there’s no reason why students would not be able to acquire virtually all the structure in modern standard Chinese by the end of the second year. Unlike the FIGS, where there are six forms to be taught for every tense (not to mention different endings within a set, like -AR verbs versus -ER in Spanish) Chinese doesn’t conjugate. That speeds things up hugely in terms of presenting sufficient repetitions of structure to allow generalization by the brain for acquisition.
While the rules-and-output FIGS classes (and even the TPRS ones) are typically still teaching grammar (inputting items with new structures, for TPRS) in the third and even fourth years (although of course everyone is trying to stuff more and more into less and less time these days), Chinese after the second or potentially even the first year is usually focused on expansion of vocabulary and usage, and perhaps the acquisition of an additional syntactic system to be used for formal authentic readings like newspapers and Classical Chinese. [This second syntax is used only for reading; relatively few programs would expect students to write in newspaper style, let alone Classical (the professor we had senior year of college being solidly in the minority on this one, but he was amply repaid for his pains by our going through an unconscionable amount of red pens as he tried to correct our struggling written Classical Chinese output.)]
So I would propose that for Chinese, as taught using Comprehensible Input, we need to view the path to full fluency (control of all the structures of the language over a wide range of vocabulary covering most areas) as consisting of two distinct parts. The nature of these parts is different, and it may be that TPRS or CI, being tools for concentrated, rapid acquisition, are not the best thing to use for that second phase, the one that stretches from limited fluency (control of all the structures of the language over a limited range of vocabulary) to full fluency.
This is probably just as true for the FIGS languages too, but the problem isn’t as obvious, since the three or four years normally spent on getting enough repetitions on the structure to get students to limited fluency often means that enough additional vocabulary gets in there to bring them near to full fluency anyway. And in any case, if it takes three or four years to acquire all the structure of the language, usually that’s enough time to fill a typical high-school language sequence, and everyone “knows” that college language classes are different anyway.
The question is a simple one: if there is no more structure to acquire, and TPRS is focused on structure and getting students to acquire the structure of the language, dressed appropriately in the highest-frequency words because, well, we have to use SOME words to give that structure form and meaning — why use a method that has this focus?
If structure is fully acquired, the situation would be roughly analogous to a native speaker who suddenly needs to “sound smart” in a field that is not his own. A little reading, a list of vocabulary, and it’s possible to sound pretty good even talking about a field you didn’t even know existed the day before. Trust me, I’ve done this over and over as an interpreter. True, many of those words didn’t “stick” long term, because they were so infrequent in the language; but there was absolutely no problem with the acquisition of such words among them that were frequent enough to appear in print and listening opportunities on a more or less regular basis.
So what seems to be going on here is that, after structure is acquired, there is a “fast track” to get words in. (It’s probably a gradual and natural speeding up and lowering of the number of repetitions needed to acquire a new word that comes as more and more structure is acquired, rather than a sudden bump.) It certainly doesn’t take the 50-70 repetitions that we need to give beginners to a language, and since the words that are “left over” for classroom work after structure has been acquired in a classroom or curricular sense are still reasonably frequent (unlike the interpreter acquiring a set of detailed jargon to deal with a conference job with experts talking about their own fields), we would expect that they would appear naturally in authentic texts, whether written or spoken.
So the question becomes: if TPRS is for acquisition of structure, and this isn’t acquisition of structure, but rather expansion of vocabulary, what is the best tool to make that happen for the upper levels? And how great would it be if (character) literacy could be delayed in large part until the acquisition of structure was complete, and then attacked in a coldly principled way? How much more fluency could we build, and how much faster, if we focused the right tool on each phase of the process?