There's that risk, yes, if the beginner doesn't follow instructions and copy the material first. On the other hand, this way forces the writer to recognize words that often begin phrases and wait a bit. It doesn't let them rely on the dictator to do the phrasing for them. It's easy enough to mark phrases — just replace spaces with dashes in the text file; I'm not sure what it would do to the (admittedly rather rough) calibration if there are too many. (Then again, with the longer pause after each period, and even longer after each paragraph, the calibration is off the other way if the sentences are short.)

What do successful teachers usually do? Does it vary with level?

I think it’s easier to remember things by phrase, too—that it’s easier to “chunk” things into frequently-encountered units than to remember seemingly-unrelated, individual words. So things such as “I hope that you will” take up only about as much “space” in memory as one or two words, rather than five.

Basically, things that make sense to us are easier to remember than things in isolation.

(I remember reading about how chess grand masters were asked to remember the positions of pieces on the board. If the pieces were in a position that could have occurred during a normal game, it was a lot easier for them than if the pieces were simply places randomly. I think that’s related.)

So ideally, I guess that dictation would be in short chunks, slightly longer when there’s a phrase, slightly shorter if not, but not single-word units.

Disclaimer upfront: I have no idea as to what would be optimum here; this is just from my feeling.

I would guess that one list would be suitable for all versions – that one should base the phrasing on what is natural to one's mind, not what phrases happen to have fixed outlines in this book or the other.

I'd count things such as article+noun as well, perhaps even article+adjective+noun ("the house", "a man", "the big house") – those are not phrased in Shorthand, I think, but they "chunk together" in your mind. Also pronoun+verb ("I think", "he hopes").

I think the list can probably stay the same for all speeds – my first impulse was that phrasing is most valuable in lower speeds since there the diction is most unnatural and it's harder to retain the sense of a sentence – but then I thought that at higher speeds, having phrases clump together a little more than the rest of the text might help you grasp and retain it better even if you're lagging a little behind.

This one doesn't stretch to 40 until lesson 6/13. The final test is 40-50-60-70. So 55 would be comfortable and 70 a stretch. I thought we were to be at 85 (at least for a minute or two) by the end of the theory.

(Eeps. I only remember 3 speeds for the final test. Ah, well, it's good enough for studying.)

Perhaps this author preferred to cover the theory more quickly and then build speed? Is 30 fast enough to ensure you know the theory?

Eventually, I'll run the same text with faster speeds. (Delete explanation of technical reason and swearing at Cepstral — "rate" is absolute in the command line but relative within a file. Can you say "recalibration"?)

What three speeds for each chapter?

(Ooooh, another feature: Tell the program comfortable speed. Include "+1" and "-1" in the text file — each step is 10wpm relative to comfortable speed, or maybe 5wpm. Or maybe it divides the file into 3 equal parts for you. Not today.)

I wonder if any teachers started dictation with "triangle, circle, square", just to give the students an early challenge and show them they'll have to work (before the course-change deadline).

Three seconds to a word is roughly talking-in-your-sleep speed, or a bit slower. Yikes!