Yeah, I would have to say don't run before you can walk. If you can't get a simple "Hello World" program (literally, a program that prints the words "Hello World" on the screen) running on a PC and your target platform, you haven't even *begun* to understand what's involved - seriously, have a shot and try it. If you can't get a "Hello World" program, written in any language, running on the machine you want you haven't even got the knowledge to *start* programming, let alone porting. And if you're not put off by the process and the code you write to make that work, then you'll probably be okay with learning a language. From there, you would then have to actually learn (parts of) a programming language - there are entire units on every university computing course devoted to *just* that task. They attract quite high drop-out rates but it's perfectly possible to teach yourself if you want (I started programming when I was 10, never really had any formal tuition until I knew how to program myself, and have never really stopped - I'm 30 now).
From there, you have to be able to create your own programs from scratch - that's much harder than it sounds when you are handling *every* aspect of the program, from the algorithms that govern gameplay up to the hideous hacks to actually make it work on a real computer. From there, it's feasible to look at other people's code and work out what needs doing to change it for another platform.
I ported Simon Tatham's Portable Puzzle Collection (the clue is in the name - portable - so it's been *designed* so that you can move it between platforms, hence the Windows, Linux, Mac, even Java and iPhone versions of the same puzzles) to the GP2X. It took me 3-6 months (depending on your definition), an intimate knowledge of C programming and computer architectures, not to mention practicalities like getting compilers running, sourcing and adding in libraries (and learning the functions and limitations of those libraries), getting the programs to compile etc. But once the port was done, someone changed one line related to the size of the font I used, and one line related to a video resolution switch, run it through a different version of gcc and produced the Dingoo A320 version. So porting can be as simple as recompiling and running and fixing crashes (which in itself probably requires a good knowledge of debuggers and compilers and being able to understand what the code does - which of the 6000 lines of code actually caused the crash, changes the resolution, etc.) or porting can involve having to write 6000 lines of original code over the course of months in order to introduce things which weren't present in the original but were necessary in the port. For example, 300 lines of that code is just related to getting the GP2X joystick to "pretend" to be a mouse so that I didn't have to rewrite every puzzle in the collection.
To get STPPC2x to come alive, I had to write 6000 lines of C (that's just the part that *I* wrote from scratch - I needed knowledge of about as many again and that was only a small proportion of the code I was porting), include and understand about a dozen SDL libraries, include and understand and INI-file writing library (I could have done it myself but that could have been another 3000 lines of C, easily), understand aspects and limitations of the GP2X hardware (including things like touchscreen when I don't have a touchscreen to test it on). And a substantial proportion of that time and effort was just spent getting the *existing* code to run properly on the GP2X - the customisation to actually make it work with a joystick instead of a mouse, to use the smaller screens, to be able to change settings (by programming my own menu system), to make it "nice" by including artwork, menus, sounds, etc. was more than just running an automated process on some existing code.
Go have a look at http://code.google.com/p/stppc2x/source/browse/#svn/trunk - that's the code. It's for the GP2X / Dingoo (but probably compiles straight into a Wiz version too, without any changes to the code at all) but it'll give you an idea of the amount of understanding involved.
Try something like: install Cygwin with the necessary libraries on your Windows PC (or most Linux distributions hav gcc and the libraries bundled with it, if you run that), "checkout" that code using subversion, and try to get a working Windows program out of it. Then try to get a GP2X version to compile (it doesn't matter if you don't have a GP2X, you can still get it to compile). All the coding work is already done for you, because I know it compiles and runs on Windows and Linux and that both versions can produce a GP2X or Dingoo version. All that's left are the practicalities of sourcing, installing and testing a compiler - that's like saying the house is built, all you have to do is work out how to open the front door. I'll even give you hints to things, like it uses the Open2x toolchain and libraries. If you can't get past that stage, it wouldn't be sensible to start thinking about porting something or learning a programming language.
Porting something is like taking that house, moving it to another country brick by brick, and compensating for the new slope of the land, the new climate, etc. A damn sight harder than just opening that front door. But the front door is your starting block - do that, and the rest will follow naturally if you work at it. I guarantee you that if you're not already programming, you'll find it quite overwhelming to even get a compiler working for cross-compilation. It's not impossible - everyone on these boards that has written a program has had to do it in the past, and probably does it several times a year - but that's the *easy* stage and 99% of the people who come to me wanting to "learn programming" can't even get a working compiler on their system to even start. That's like wanting to learn to drive, but not being able to work out how to book a driving instructor.
Seriously - give it a shot. Almost every programmer starts as a complete amateur. But the drop-out rate on my university course for Java programming (something considered quite simple back in those days because it didn't have quite so much junk in the language back then) was about 70-80% before the second lecture even started.
