If you look at the transistor as two separate PN junctions, then:
Emitter is forward biased (+ to P, - to N)
Collector is reverse biased (- to N)
Base is just what's common between them.
In practical terms, voltages are usually referenced to the emitter, so while it may be correct to say the above, it's more conventional to say "the base is forward biased" instead of the emitter.
In normal operation, the collector is always reverse biased.
The tricky part about a transistor is, it's not simply two diodes glued together; magic happens when the junctions are brought together, very closely. On the order of microns. Current flow in one junction induces current flow through the other, even if the one is reverse biased.
The transistor is apparently palindromic, so you could interchange C and E; this is called inverted operation. However, it doesn't usually work very well, because transistors can be constructed asymmetrically (and usually are -- typical of the double and triple-diffused processes). Some special purpose transistors are made symmetrically (or nearly so), as well as most antique transistors (which are typically "alloy junction" germanium types). "Not very well" means: lower breakdown voltage (e.g., for 2N3904, Vebo = 7V, versus Vcbo = 60V!) and much lower hFE (typically under 5).
Tim