Sorry for the flippant quick answer above, hoping to get more back on topic with this post.
It sounds like you are referring to the vocabulary of a language, is that correct? Which concepts are expressed at all, or which are expressed by separate words vs. grouped together with similar concepts? Things like the fact that Latin had, and Arabic has, one word for maternal uncle (avunculus, khaal) and a different one for paternal uncle (patruus, 'amm) because, in a strictly patrilineal society where your mother's relatives aren't properly considered family, the two concepts are different enough to warrant a terminological distinction, while in English, Spanish and Inuit the two are collapsed?
As the example already shows (Latin splitting, modern Romance languages lumping), this kind of thing is neither well-correlated with genealogical families nor stable on timescales as centuries or (low single digit) millennia.
I'm making the claim that the words which exist in a language come into common usage because they're relevant to the lifestyle of those speaking the language.
Sure, but how is that relevant to Khoisan languages being any more representative of what languages people may have spoken in deep prehistory than other extant languages?
Therefore, if hunter-gatherers have been living essentially the same static lifestyle for [x] period of time, then the evolution of the language is very slow, and it's likely that the modern one resembles the original one more closely than if more social evolution had occurred. I don't particularly care about structure, I'm discussing what's been symbolized.
That if clause, at least when applied to Khoisan languages, seems to contain several unspoken assumptions that are either demonstrably false or not known to be true. It is not only false but logically impossible that the San have been living the same static lifestyle for anywhere near the 50k years you were talking about - if nothing else, because the climate in the larger Kalahari region has been constantly changing becoming dryer or wetter and thus more or less abundant on timescales much shorter. As of today, in Southern Africa and elsewhere, hunter-gatherers are largely confined to marginal lands, but this wasn't always so, and the hospitability of a land directly affects group size, the viability of amassing possessions and the importance of clear inheritance rules and a host of other variables known to correlate with social structure. This is not merely theoretical, there are actual ethnographic records (e.g. from the Northwest USA and British Columbia Pacific Coast macroregion in the 19th century) showing that hunter-gatherers too will form complex stratified societies where the environment supports it. Assuming that a hunter-gatherer group living on what's now a scarce desert area will have "essentially the same static lifestyle" as their ancestors when the area was lush grassland a mere 2000 years ago is nonsense even in the absence of migration and contact (again not true, more on that below).
Even letting that fly, the timescale's just wrong. No serious linguistic reconstruction can go back more than, maybe, with a lot of conjecturing, 3000 years before the attested language stages under the best of circumstances - because languages change, period. Not "if there is a radical shift in lifestyle", not "when there is extensive contact with outsiders", it's just a thing language does
Yes social changes have occurred in that time, but I think it'd still be true that the language San Bushmen were speaking in the 19th century would be a good indicator of what they were speaking many thousands of years ago (in prehistory), if not an exact replica.
First, you were talking about Khoisan as a whole, not specifically about the San (Bushmen). The Khoisan are not one tribe speaking one language, they are a group of diverse people with diverse lifestyles, speaking a set of vaguely similar languages falling into three well-established families some of which may not even actually be related - and most of them were pastoralists herding sheep and cattle (both originally domesticated in Eurasia millennia earlier) when they were first encountered by Europeans, or for that matter by the Bantu. Even the "San" are not a monolithic group with a common language. A number of languages of hunter-gatherers classified as San by ethnography are actually closer to the languages of Khoekhoe pastoralists than to those of other San groups - and indeed some groups may be descended from pastoralists who gave up lifestock breeding when the climate deteriorated.
The history of Southern Africa in the two thousand years before the Europeans appeared on the scene is *much* more nuanced than you appear to believe. From my reading, the most plausible scenario is that a group of (probably, at least in a loose sense) Khoisan speakers in what's today Zambia or even Southern Tanzania picked up pastoralism from neighbouring Bantu (or possibly Kushitic) people around the year 500 BC or earlier and ran this new technology all the way to the Cape in quick succession, up to a millennium before Bantu farmers colonized Eastern South Africa. What appears less clear is if these people largely supplanted the previous inhabitants or whether it was more a case of cultural diffusion, where local foragers adopted the practice from their neighbours. If it is the former, most modern Khoisan as far west as Namibia actually trace most of their ancestry to this group. The jury, I gather, is still out on which is more plausible, though some researchers have argued that the existence of cognate terms for "cattle" and "sheep" among many Khoisan languages suggests the former, though there are other approaches (a fine example of language giving clues to prehistory, by the way, but not the way you seem to imagine). Interestingly, cognate forms for lifestock also appear in hunter-gatherer languages, suggesting at least that there ancestors knew about/where in contact with pastoralists prior to language diversification.
Here's some of my reading:
Southern African ancient genomes estimate modern human divergence to 350,000 to 260,000 years ago - don't let the title fool you, while they do report a genetic component that appears to be unique to Southern African Khoisan people and splitting from the rest at a very early age, they also estimate that "all modern-day Khoe-San groups have been influenced by 9 to 30% genetic admixture from East Africans/Eurasians" deriving from this pastoralist expanstion and/or the later Bantu (farmer) expansion.
Fine-Scale Human Population Structure in Southern Africa Reflects Ecogeographic Boundaries - suggesting neither language affiliation nor mode of subsistence are all that well correlated with genomic data, suggesting that both spread predominantly through diffusion rather than supplanting previous populations.
The early livestock-raisers of southern Africa - heavy on comparative linguistic data and discussing possible implication as to whether cattle and sheep arrived as one complex or in two waves (sheep first).
Tracing Pastoralist Migrations to Southern Africa with Lactase Persistence Alleles - providing data for positive selection for lactase persistence among Khoisan groups, by quantifying the overabundance of the corresponding (East African derived) alleles relative to the overall share of East African ancestry. An interesting marginal finding of theirs is evidence for selection among to
forage groups which can be construed as an indication that they've shifted to foraging from an earlier pastoralist lifestyle.
Was there an interchange between Cushitic pastoralists and Khoisan speakers in the prehistory of Southern Africa and how can this be detected? - Slightly different topic, but what I found interesting was the observation that South African Bantu languages (such as Zulu and Xhosa) actually have Khoisan loans for sheep and cattle, suggesting that the earliest Bantu cultivators reaching the region only brought goats and were familiarized with cattle and sheep by their non-Bantu neighbours.
Standard Average European, Standard Average European describe several linguistic features shared by many European languages and rare outside of Europe. They use lots of linguistic jargon, though they do have some examples.
The sharing of words for goat, horse/donkey, milk suggests to some that Basque-Burushaski broke up after the development of animal husbandry; but on this topic I think the evidence for P-I-E is more impressive: About ten P-I-E words have been reconstructed just related to horses.