Okay, time for further !!SCIENCE!!. I set up a bunch of 5v5 human battles, the only skill being Proficient swordsman, and no items other than what's in their hands. All equipment was iron. Also, note that most of these were only one round, sometimes two, but the surviving side usually won by a large margin.
Dual-wielding, longsword vs. 2-hander: longswords won
Sword and shield, longsword vs. 2-hander: longswords won
Single sword, longsword vs. 2-hander: longswords won
Single sword, shortsword vs. 2-hander: shortswords won
Single sword, shortsword vs. longsword: shortswords won
Okay, so clearly I can't conclude much from mixed-weapon testing. Time for some single sword, vs. dual-wielding, I think. Also, I'm going to increase the number of rounds, and record win/loss ratios more accurately.
2-handers, round 1: single sword won, 4 survivors in good condition
2-handers, round 2: single sword won, 4 survivors in good condition
2-handers, round 3: single sword won, 3 survivors in good condition, 1 survivor who can't stand
longswords, round 1: dual-wielders won, 1 survivor in good condition, 1 winded crawling survivor
longswords, round 2: single sword won, 3 survivors in good condition, 1 crawling survivor
longswords, round 3: dual-wielders won, 2 survivors in good condition
longswords, round 4: single sword won, 3 survivors in good condition, 1 crawling survivor in pain
longswords, round 5: single sword won, 1 survivor in good condition, 2 crawling survivors
So, conclusion: there DOES seem to be a real penalty to wielding a multigrasp weapon in one hand, since the single sword victories were more decisive (and consistent!) for the 2-handers than for the longswords. Also, in the AI's hands, dual-wielding is not effective.