I certainly agree with the first line, but I doubt that anyone has yet reached "a pretty good estimate". The Human Speechome Project is an ambitious attempt, but the papers that have been published so far show that the amount of variation and the number of factors to be considered is enormous. See the following list:
As just one example of the kinds of issues involved, people once claimed that exposing infants to TV would increase their exposure to words and improve their rate of language learning.
Unfortunately, more recent studies have showed the opposite: an infant's rate of learning language *decreases* with the amount of exposure to TV. Far more important than passive exposure is the emotional involvement of the child with caregivers and siblings in actively using language.
Trying to count exposure to words is difficult, but the quality of emotional involvement is far more important and much more difficult to estimate.