I'll give some ideas that might help other than re-doing or EQ matching as Tetsujin said. Another great piece of advice given by Tetsujin is to eliminate the room factor so you only have one factor to play with.
Let's also assume (as you say) that the talent performed in the "same" way. So that leads us to handle only one factor which is the mic behavior to the placement.
There are 2 major factors here direction of performance (where the mic was "hit") and proximity. For the first factor (direction), you should expect more prominent sibilance and a tad higher definition. For the second factor (proximity) you should expect changes in dynamics and EQ, mostly on the bass regions (the closer the bassier). There also might be proximity curves for your microphone which might help!
My take on the subject would be to create a processing that would brutally eliminate all of these factors. The following ideas are not stated with any order, you can experiment with that.
Keep in mind that these processing techniques will change the fidelity of the audio, so if you can't afford that just skip this answer :)
- Without normalizing try to compress with an aim to match the peak/RMS of each performance, you can use "signature" signals like P's K's and similar stuff that will cause more diaphragm movement. As you do this move in small steps cutting 3-5db's per compression/limiting layer (experiment here). At the same time, you might want to raise the level of the recording of lower amplitude parts so they get a bit compressed as well (but don't just blatantly normalize). You should start with high speeds and high ratios to address higher peaks and as you progress with the compression layers slow the speeds down. As you add more compression layers you should end up at a point where the last compressor has the same behavior regardless of the recording material. Always keep your eye on the peak/rms levels, you want to match that!
- Brutal EQ, the idea here is, why try to match when instead can "sacrifice" some quality and fidelity to "mask" the differences. Take an EQ and kill the regions that hurt e.g: roll off all the bass until it's "Acceptable", same with the hi's. So now you have a band-passed signal, which is better cause you have X percentage less of factors changing in a sample. Now you can try to notch the EQ generously to give a specific and apparent but also acceptable overall "tone" to the material so maybe the brain doesn't focus on the (now less) differences but it focuses on the similarities. Simply, make the forcefully similar rather than trying to match them.
- Multiband compression, you could also run both materials in a group with a multiband compressor, adjust levels and thresholds accordingly so you address only what you want out of the two materials (e.g. closer mic is louder on bass, this material should be the one moving the compressor and not the other). For this one grouping is not mandatory but helps in perspective.
- Phase & Null. Do you have the same words? Do you have excerpts from the material that match in many parts? Null them!, Take 2 channels and put the excerpts as close as possible so they sound like one (in timing). Flip the phase of a channel and play with its fader until you find the quietest spot. Now what you listen to is the "difference" which can help you make better decisions and understand what's going on. Keep a Spectrum analyzer open cause you might miss very highs and lows that actually contribute to the overall signal and different feeling.
I think these are some ideas you could try, not sure if they'd be much much better than just using a matcher plugin and normalizing but if you have problems re-doing it might be worth giving it a shot! Also, keep in mind that there's a lot of experimentation for these techniques (e.g. shelving rather than rolling of etc).