Turingtest: cd computers EVER understand words like transcendental,mystical, not just operationally but really?

Perhaps this boils down to: how cd the human makers pin down intrinsically vague, gauzy, ill-defined words like these which, despite such characteristics, do have meaning or meanings to humans? There are a lot of words like these, many of them emotively charged & inseparable in significance from the experience & emotional "universe" of humans. Unless we make computers as robots, living, feeling, & emotively thinking like humans, can we make computers that cd really understand humans? Even if we could, wd we want that? It wd be like creating another species of intelligent beings. Cd we do it w/o giving them access to & the feeling of "free will?" But going back to my original question, wouldn't "pinnning down" the meanings I've refrred to involve scientists/linguists/philosophers in "vicious circles" that cd be "broken" only by - partially - real human existence? I don't think AI experts have truly grappled w/ these questions, & yet in their statements it seems that having computers w/ AI - computers that cd really understand & really communicate w/ humans - is just a matter of time.