Imagine a reading automaton which has to be trained. There is a stage when he ‘accidentally’ says the right word & a stage where he can read. Suppose he says the right word & you say “oh he isn't reading” then he goes on right, should you say “I was wrong he did read” or not? At which moment does he really begin to read? Should we say we can never know? Compare this with a case, in which there is a mechanism I can see & say now he reads now this connection is made.
     Again we are inclined to say that what distinguishes reading from not reading is the personal experience. This is when we say “only he can know whether he is really reading”. Imagine a case where we are doubtful whether he knows by heart or gets a peculiar experience of reading or whether he goes through certain conscious processes or not. On the other hand we can easily imagine a case in which with the experience of knowing by heart we should still talk of reading.
Reading a real language, reading an imaginary language.¹