I have conducted some testing on the effects of weapon proficiency and agility on weapon attack speed. The basic test was conducted by recording the number of two-handed polearm thrusts that could be executed within 60 seconds. I preformed the tests with a number of polearms of different weapon speed and with different polearm proficiencies. All of the reported values were taken from the averages of 3 individual time trials conducted at each condition. The following plot shows the time per thrust verses polearm proficiency. I have also noted the agility that the testing character possessed at the time of the trial.
The most obvious feature of this data is the linear relationship between time per attack and proficiency. The slope of each weapon line gives the reduction in time per attack per point of proficiency while the y-axis intercept of each weapon line gives the time per attack with 1 proficiency (technically 0 proficiency but the difference is inconsequential given the small slopes.) The time per attack at 1 proficiency has a linear relationship with weapon speed as seen in the following figure:
The slopes of each weapon line are such that 100 proficiency results in a ~6.5% reduction in time per attack relative to the 1 proficiency time per attack. Therefore, weapon proficiency has a greater absolute reduction in the time per attack for slower weapons than for faster weapons. Additionally, these results show that agility has no observable impact on the time per attack for a weapon as there is not a significant change in the slope of time per attack verses proficiency between the different agility test conditions.
To put this in perspective, a level 30 character with 3 agility, no weapon master, and 110 polearm proficiency would attack ~5% slower than a level 30 character with 27 agility, 9 weapon master, and 180 polearm proficiency if both characters were to use a speed 94 polearm. The absolute difference in time per attack between these two characters would be ~0.06 seconds. This is significantly less than the 12% difference in attack speed one would expect if agility was providing the purported 0.5% reduction in attack speed per point and the difference of 70 proficiency was ignored.
Another interesting result was that all of the polearm attack animations (thrust, overhand, and slashes) take the same amount of time to execute completely. However, this does not necessarily mean that the time between releasing an attack and landing an effective hit is the same for all of the polearm attack animations.
It's this thread http://forum.c-rpg.net/index.php/topic,2531.msg42737.html