In junction gate FETs, the channel is created by doping the region between the terminals with carriers of the required type.
In MOSFETs, the channels are the inversion layers in their MOS capacitor substructures. When a voltage is applied across the MOS capacitor, the energy bands in the semiconductor near its interface to the oxide bend and their position with respect to the Fermi level may change, so that the surface semiconductor layer may be inverted from n-type to p-type or vice versa. The operation of MOSFETs and characteristics depend on the possibility of inversion and the inversion charge density. As the voltage varies, the voltage at which the n-type and p-type densities at the surface layer become equal, i.e., the inversion layer is about to be created, is called the threshold voltage. 
The conducting channel is the region where the carriers actually flow when a source–drain voltage is applied. In some texts the term "channel" actually means "conducting channel". When the conducting channel exists, then it is said that the FET is turned on.
For a given transistor, the actual geometry of the conducting channel depends on the combination of the voltages applied to the pairs of transistor terminals.
Channel length is basically the distance between source and drain. The smaller the channel length, the faster FET switching is possible.
- "Semiconductor Physics and Devices", by Donald A. Neamen, p. 449
- "Semiconductor Physics and Devices", by Donald A. Neamen p. 483