Rapid Digit Naming. A phonological processing measure? Uta Frith Research

Today I was listening to Uta Frith on the radio. 

 Twelve years ago, on the suggestion of a colleague of hers. I used rapid digit naming as an alternative to rapid automatic naming of text to find out how the dyslexic students were performing compared with non-dyslexic adults. As Uta reported the mean for the dyslexic students was around 135  digits per minute.

The same as for rapid  word naming.  The inference in her research was that this was evidence for a phonological deficit.

However when the students were asked to undertake RDN using their optimal settings on the computer screen their performance average 184 digits per minute which was the same s the RAN mean of non-dyslexic students.

 It was also the same as reported for RDN by Uta for the non-dyslexic students at University College.

The question I ask is this.

If the Phonological processing is apparently controlled by the visual image, which one is controlling the reading performance?

Anyone has any answers other than the obvious one which I inferred?