Picture a very lightweight cricket bat, mounted, a couple inches up from the end of the handle, upright in a 2D gimbal, such that it can wave forward and then return to the upright position, and can twist along the handle axis. Attach a string with a weight on it to the end of the handle.
Now picture
something that looks vaguely like that but is large and has a rope and a pump (or linear motor or baby-swing or something) where the string and weight are on the model.
Wind strikes the flat of the bat, pushing it down and pulling the rope up, doing some work.
At the bottom of the wave the bat twists 90 deg; since it's a thin crossection now the weight pulls it back to the upright position (thus supplying the 2nd half of the work cycle)
At the top, the bat twists again to face the wind.
Rinse and repeat.
Advantages over a windmill:
*Much* easier to build and maintain; no "industrial society" parts, machining or for that matter design, required.
The machinery or whatever the force is applied to is at ground level (as opposed to say a generator on a propeller) making maintenance easy.
In prohibitively high winds it just stays down by itself.
Less dead birds littering the area.
Notes:
You'd want the gimbal to be mounted high enough off the ground the bat doesn't smack you in the head on the downstroke.
A small fin on the back of the handle ensures that it faces the wind when it starts to blow, and the bat will position itself after that since it acts as a "weathervane" as it's coming back up.
The rope can be manually attached at various points on the "work" side of the handle depending on how strong the wind is blowing (or some cool mechanical automatic thing that does it based on how far down the previous stroke went)