It generally looks good. If possible shorten it and focus on the main  
result, which I think is the SiII line as diagnostic of AGNs.  So 
change the title to  reflect that: the phrase "new AGN diagnostic in 
the mid-IR"  would be good to have in the title.  Similarly, in the 
abstract one should read about this before turning the page, i.e.
more prominence.

In terms of the actual proof, you present SiII/NeII ratios which  
behave the same way as OVI/NeII in the AGN/starburst mixing ratio. 
Then you present plausibility arguments, that SiII can be a daignostic, 
based on XDR predictions.  My impression was that none of these XDR 
predictions came true. [OI] was supposed to be higher, but that has 
not been seen (correct me if I am wrong: a couple of Dale et al. citations
seem to imply that). The fact that SiII traces X-ray gas should be  
presented as a hypothesis.

Further tests can be done in support of that. X-ray images are  
available for about half the SINGS galaxies. One could check SiII/X-ray 
from the nucleus if possible.  If not, make it fodder for future paper 
or a Chandra proposal.