True, I was rather pointing out that being able to parse symbolic language deterministically doesn't imply that we could then "reason" deterministically in general; the reasoning would still need to involve some level of stochasticism. Whether or not that's a problem in practice depends on specifics.