Normally, it is advisable to go with a better lens even if that means a cheaper body (due to budget constraints), as lenses make much more a difference than most modern DSLRs.
Since you are going for primarily for video, you need to consider audio as well. There is no built-in microphone that is good enough for serious documentary, so you have two choices:
- Buy a camera with a mini-jack so that you can plug in an external microphone.
- Buy an external sound recording device and record sound separately. Merge the sound track from your recording with the video track using video editing software.
The second option is more flexible and also more costly. As usual, you need to spend enough on a sound recording to get something good. Don't both using you cellphone's sound-recording function ;)
There are very few DSLRs which will autofocus while recording video. It is often a concern for people buying cameras for home videos but for anything serious it does not matter.
If you do not go with one of the sound options above, then you will have to worry about the camera recording the sound of autofocus. On the cameras I tried (Nikon D3100 and A55), it will make your sound track annoying to your audience.
Even if you do use external audio, the contrast-detect autofocus system used by the D3100 causes the lens to move back and forth while focusing. This is easy to spot in videos and just does not look right. The Sony Alpha SLT-A55 and its sibling the A33 are the only cameras that use phase-detection during video recording and avoid this particular problem.
If possible for the subjects you are documenting, you should use Manual Focus while filming. This will let you decide precisely on what to focus, when to change focus and how slowly to do it. It will take lots of practice to get right, so try it on squirrels or whatever animal you have roaming around where you live.
If you are filming things that move a lot or live-events (in which case retakes are not possible), then you may decide to rely on autofocus. You will occasionally end up with shots that are partially out of focus since sometimes the camera simply goes after the wrong thing. You can edit such parts out and move any relevant audio.