Yeah, you're definitely not alone. I don't substitute similar words that often, but I do garble sentences to an extreme more than I'd like to. It ends up sounding like word substitution, but completely nonsensical.
>I think that when I'm speaking, I want to say words faster than my brain can build a sentence.
That's pretty much what it feels like in my case.
>I'm suspecting that working on a computer for so many years might be the issue, where I spend most of my time speaking via a keyboard.
Same. My history includes fairly extreme real life social isolation for about 15-20 years. Didn't have any friends I'd see in person on any sort of regular basis during that time, including my entire adolescence.
As such, my good friends that I talk to on a daily basis have historically been remote. Communicating with them has always been about 80% keyboard and 20% voice. In terms of voice, it's always been push-to-talk out of preference. Unlike open mics (think phone calls or default Skype settings), PTT allows you to mull over your reply quite a bit more without it being considered awkward. Combine that with typing being the preferred method of communication otherwise—which allows even more time to formulate a reply—and it's little wonder.
Some other components that likely play a part include being less comfortable than an average person in many social situations and at times poor eye contact. It almost feels like the verbal centers of my brain are rendered dumb due to a combination of having to come up with things to say faster, and processing facial expressions/body language (which historically wasn't required very often). It's not surprising that it goes hand-in-hand with not really being able to get the word in edgewise most of the time, nor being particularly good at thinking up new topics of conversation when engaging in small talk.
Despite being extremely social in person the last couple years, it still persists, though usually it's the worst at the beginning of a conversation. That may have to do with difficulty switching from my internal voice (which is usually fast) to actually verbalizing things.
If it's any sort of meaningful conversation in a relaxed environment, there's rarely any problem. Likewise if it's people I'm already comfortable with. Alcohol tends to make the problem evaporate—even with strangers. My theory is it simultaneously slows me down a bit and lowers my speaking inhibitions considerably, killing off the over-analysis and mental/verbal speed mismatch that garbles sentences in the first place.
>So I'm asking for your help, to know if I'm the only one, if this is something known, if this something that can be fixed, and how.
Ultimately I think it's a latency thing. Some people aren't used to the lower reply latency that in-person interactions require.
Practice makes perfect I guess. Just keep talking to people in person more. I'm trying to do the same.