I want to compare sharpness of two lenses, consider 50mm f/1.8G vs 35mm f/1.8G from Nikon on a crop sensor camera by taking some photos (from some small size text on a paper) and review the results. Here, I don't care the perspective issues (just sharpness is important).
You have an apple and an orange and you're trying to compare them.
I can't think of a useful purpose in comparing sharpness for two different focal lengths. Two identical focal lenses maybe (as a deciding factor in choosing one lens over another, maybe), but not different focal lengths, as the choice of focal length is decided by the shot, not how sharp you want it.
To have the similar frame on both photos, I need to move closer to the subject as the second one is a wider lens. When I do this, the texts are more clear on the wide angle lens. I'm not sure if this is fair for comparison, because when you are closer to a subject, you can see better details.
Sharpness for any lens will vary depending on aperture and what part of the frame you look at. That will be different for two different lenses.
And here's where the sharpness test is essentially useless.
Not only will two different lens designs have different overall characteristics, but two individual lenses of identical design can have slightly different characteristics - design tolerances are not zero. Worse still, any two lenses combined with any two theoretically identical models of camera will produce four different results. The combined engineering tolerance variations of these will pretty much guarantee they'll be different.
And you're not even comparing two focal lengths that even close to each other. !
They're both sharp lenses, but they're not sharp everywhere in the frame at any aperture. I can't think of a lens that is.
But these are wide aperture lenses. The most typical use of these is to shoot wide open, or close to it. Wide open they have very narrow depth of field. Which in practical terms means that unless you're shooting a perfectly flat target aligned to the focal plane (think wall), you're really going to only have an extremely small region of the subject in focus even in ideal conditions.
So why is sharpness so important ? I'd suggest it simply isn't.
All a sharpness test of these lenses does is distract you from the reality of using these lenses.
So, how should I arrange the scene to be able to test the sharpness of these lenses?
Flat surface, parallel with focal plane (that's fun to arrange !), even diffuse lighting ideally. Tripod, mirror lock up and timed exposure release to avoid vibrations causing blur.
Manual focus, use live view. This should avoid any possibility of focus shift when stopping down because when using Live View cameras typically keep the aperture closed down to the set aperture, whereas when not using Live View they'll open aperture wide for focus acquisition and then stop down to the selected aperture when shooting - in some lenses this can result in a slight focus shift.
How I test lenses is simple : I look for someone else who specializes in this sort of delicate work to do it for me. Photozone.de was/is a pretty good lens test website and I'm certain they've tested those lenses. Lenstip.com was another good one.