From a language teachers’ list:

TPRS is used for teaching fluency.  We use it whenever students don’t have “ease of expression”.  Ease of expression means they speak with confidence, accuracy and without hesitation.  It is been my exprience that “ease of expression” does not happen in just four years of Spanish.  I agree that TPRS works best [ed: meaning “better than communicative teaching”] in upper levels.

I think there is a distinction to be drawn, however, between TPRS used for teaching “initial fluency” and “target fluency” (to make up two terms…)

“Initial fluency” for me is being able to use ALL the major structures of the language automatically, correctly, and without hesitation or thought. It doesn’t involve being able to use a vast amount of vocabulary. The result is a person who is something like an extremely socially isolated little native speaker — he knows what “sounds right” but he is not yet able to sound like a native speaker (or even a very fluent speaker) across a wide range of topics.

“Target fluency” (if there even is such a thing — since we all continue learning for life in languages!) could be defined as the ability to use ALL the structure of the language automatically, correctly, and without hesitation or thought, while ALSO using a sufficiently broad base of vocabulary and collocations so as to be able to deal with most subjects of daily life and with any professional use subjects the student might need to deal with. No one will ever acquire all of any language — including native speakers. There are plenty of words in English that I don’t know, yet I would describe myself as having target fluency in English. I can easily “pick up” individual words or expressions as I need them.

These two (theoretical) stages or types of fluency fit with what Krashen is working with lately AFAIK, in terms of getting students to fluency with academic language (major expansion of vocabulary through reading), and it makes sense if you think of the process that goes on when a child acquires his first language as well. A six-year-old knows most/all of the grammar, but wouldn’t be able to speak fluently about many topics (due not only to lack of vocabulary but also to lack of world experience…but that’s another question.)

I also think that reaching initial fluency means different practical timelines for different languages. The key is how complicated the structure of a language is — or more accurately, how many sessions it would take to circle everything that needs to be circled. To give an example, Spanish is going to take longer to achieve initial fluency in than Chinese because one has to acquire five (or six) different persons for each verb tense in Spanish, and also acquire the forms for multiple verb tenses, while in Chinese all those are just a single form. (I am talking only about language, not literacy…that’s another issue in Chinese, of course!)

So logically (to me at least), there is more reading during phase two (from initial to target fluency) because extensive reading is a great way to expand vocabulary, AND because the more structure of the language a learner already has in his head, the more he can benefit from input — and the “faster” he can pick up new items. There is more known language to connect new things to. The pace is different, and since we are no longer teaching the highest-frequency items in the language, the way we repeat them is going to be different, since it’s just difficult to find or construct readings that are even close to “natural” yet repeat lower-frequency words or expressions many times. (Note I’m not worrying about repeating structure — the student has already acquired that.) So I think we need to ask: what is TPRS™ and where are its boundaries? What does it do well, and what would it be better to adopt a different (though still CI-based) strategy for?

Lately I am thinking that it is possible that TPRS™ in its strict sense (the three steps, go slow, pause and point, personalize) may not be the most efficient way to take a student from initial to target fluency, since those steps are designed to ensure sufficient repetitions for a beginner who is acquiring structure, not an intermediate (? not sure what to call the person who is at initial fluency level — IMO ACTFL’s proficiency standards do not fit very well if a student has been taught all through CI). If we are focusing on the move from initial to target fluency, the focus should be on expanding vocabulary, with repetitions but not as many, slowed down but not as much. That student probably doesn’t need all the support and all the repetition that strict TPRS™ provides. You could say “Well, that’s just adjusting the amount of circling and the speed”, but where do we draw the line? When is it no longer TPRS™, and how do we know when someone is doing (or not doing) TPRS™? (If that matters — in the real world it probably doesn’t, since both would still be CI-based, unless you’re hiring demonstrators or something.)

For Spanish, students are mostly still seeking initial fluency (because of the breadth of grammatical forms needed) in the fourth year. In Chinese, they can reach initial fluency in two years. Then it’s all about expanding the vocabulary and knowledge of collocations and getting them really literate.

This is my take as a person who teaches both, and spends (probably too much time) a lot of time thinking about the theory underlying TPRS™ and classroom practice.