§157

We might imagine creatures trained to be reading machines. We could imagine one that hasn't participated in the training producing sounds when presented with written words. The sounds might occasionally come out right. If someone heard the pupil on such an occasion they might say 'he is reading'. The instructor would respond 'no. He isn't. He just got lucky' or something like that. But what if the pupil were to respond correctly to further words presented to him? The instructor, at some point, will grant that the pupil can read. But what of the first sound/word? - Do we now have reason to say that the pupil read that? - It makes no sense to ask which was the first one read.
It would make sense to ask which the first word read was if the criterion was that it was the first time the pupil had a certain feeling. But according to the way reading was defined in §156 ("rendering out loud what is written or printed...") the concept 'reading' is independent of mental or other mechanisms.
I assume that the same point would hold of our own concept of 'reading' - that we don't have criteria for determining the first word read.