I had this same problem ( not knowing what 2D engine to use ) and I ended up writing my own.
My main issue with most 2D engines is that they don't support shaders.
Even Torque 2D has just some silly blend modes that makes me think that they just use fixed pipeline when rendering everything in there.
Using a 3D Engine for 2D games might give you a lot of work too. You will need to change the way physics works if you wnat to make a sprite based engine ( not just a 3D game where you see the character from the side). This can be easy or not depending on what you choose here. For example for UDK it's nearly impossible to change the physics code and replace it with Box2D since you don't have access to the sources.
In the end you need to ask youself what is the end goal.
Do you want to make a game in order to sell it? I doubt it can be successfull given the huge amount of first-project flops that come even from experinced programmers.
Not wanting to discurage you here, but since you mentioned PASCAL I start to believe that you're not that experienced with game development or programming in general.
If you want to make it for your portofolio, you could impress people more if you do your own engine. It would also be a really nice learning experience.
Well I don't need shaders, I'm not looking to create a game of high-end graphics. I'm looking to create a game that looks like it has been created in '91, they don't have shaders and possibly other features that you needed for your project.
I've mentioned Pascal because I had to work with it at school. It was absolute horror! The primitive way how it works... Horrible experience. So in short, that's unrelated to my own choices, it was an enforced enforced by school, I will never lay my hands on it again. However that is true, I'm not very experienced programmer in general, I am good with the logic, but writing it down is hard for me.
I do understand that I'm not experienced, however when I work on something, I always do my best and then some, I never leave it half-way. I've seen many projects and (no offense), but most of them are horrible. I think the commercial way is the way to go for me, due to the lack of money mentioned in real life. But as the first project, I don't think writing an engine is the right way to go, it is a lot of work and has a great potential for a lot of bugs and inefficiency compared to what's already written and is proven to work.