To a pretty close approximation:
P = |Eg| |Vt| / X
where Eg is generator internal voltage (unchanged during the fault).
Vt is terminal voltage which would approach zero in an idealized three-phase short circuit situation.
X is a generator direct-axis reactance for the transient time period in question.
Of course Torque = P/(2pif) which is approx proportional to P assuming speed=freq don't change more than a few percent from their original value.
The surprising result is that the Electric torque and power decrease, meaning that the generator will speed up during the short circuit.
That's a little counterintuitive considering that it's putting out more current. The answer (at least in this idealized model) is that the voltage drop is occuring across an inductance, so only reactive power is being delivered by the generator, no real power.