Speed.
You want to spin a shaft, then measure the rate of rotation. To spin it, you could
1. Attach an axial flow fan blade (propellor like) which you will have to maintain pointed into the wind. OR
2. Make a cup anemometer - four small light, rigid cups (bottoms of coke bottles ?) attached to the shaft by four identical arms. Less sensitive to orientation. Axis of rotation at right angles to airflow, which on the ground generally means vertical.
In each case you will need to minimise friction.
To measure its rate of rotation, either
1. Connect a small model maker's DC motor to the shaft. Motor will output DC current. Will have some AC ripple which may need to be filtered out with a capacitor. Output is DC. If you want to digitise it, get an Analog/digital converter device from your local electronics store. Match up the output of the motor (now generator) to the ADC if necessary, and display the digital output. OR
2. Attach a disk to the rotating shaft, with a hole on it near the edge. Build an optical sensor around that hole so that in conjunction with a timer circuit, you can count the rate at which the hole appears at the sensor.
Calibrate against a car speedo on a no wind, no police day by holding the device clear of flow distubances from the car.
Direction
You need a digital shaft encoder. Instrument suppliers like Honeywell make them to allow remote sensing of the position of things like valves, crane boom angles etc.
You could make a very rudimentary one to at least show you understand the principle as follows
Imagine a disk with a radial reference line drawn on it, which we will call zero.
On the outside, inscribe it with an arc, starting a 180 degrees, which goes exactly half way around the circle back to zero. Suppose we call that circle "3", with the bit of the circle which is black called "1", and the other bit "0"
Come in a bit and get ready for another circle called "2". Start inscribing more alternating arcs on the disk, starting at 90, but this time each arc subtends only half the angle of the previous arcs. So you draw an arc 90 to 180 degrees, then skip to 270 where you draw another from 270 to 0. Black bits are "1", others are "0".
Come in again for another circle called "1". Start at 45, and draw alternating arcs, but only half the angle subtended by the last set of arcs. That is, the black bits go 45 to 90, 135 to 180, 225 to 270 and so on.
Come in again for another circle called "0" which has black bits going 22.5 to 45, 67.5 to 90, 112.5 to 135 etc.
Each new circle doubles the number of alternating arcs.
Now if you can make up a detector which can read through a radial window what bits of drawn arcs (the 1s) are displayed in what circles, you can now establish the position of the disk to within 360/16 or 22.5 degrees.
More circles - higher resolution.
So 0 to 22.5 reads 0000 binary = 0
22.5 to 45 reads 0001 binary = 1
247.5 to 270 reads 1011 binary = 11 and so on
And the angular position is in the range (22.5 x output) to (22.5 x (output + 1))
All you have to do now is attach such a disk to a wind vane.
If you don't want to do the mechanicals and electronics, you could dummy up a pretty good cardboard replica with neat little windows displaying the binary output, a mock wind vane attached to it and so on.
Lots of other analog and digital alternatives if you think about it.
Hope that prompts some ideas for you.