Natural languages (excluding deliberately-crafted artifacts such as Esperanto) undergo constant evolution in words, meanings and structures which may be analogous, in many particulars, to the processes described by biological evolution. The OED, cited in my previous post, gathers up something like the 'fossil record' of a language, the earlier variants of a word's orthography and semantics. From this record, it is clear that the meanings of words vary over time, often gradually, generally by small, incremental changes which are favoured by selective pressures.
Endless examples abound. To pick a personal hobby-horse, disinterested previously meant--and in the British Isles still primary means--"impartial, fair-minded, not having a personal stake in the outcome." In the United States, a colloquial usage arose whereby it was used as a synonym for "uninterested," which colloquial usage is becoming more widespread and it seems likely the use of 'disinterested' to mean 'impartial' will become extinct. This is descent with modification--but it is not random. It is not the case that on one particular date, the meaning of the word 'disinterested' suddenly mutated, far less 'randomly' mutated into a wholly unrelated meaning such as 'purple' or 'duct tape.' No one decided to change it, it changed through its changed usage being accepted by a growing number of speakers of English. It was 'selected,' in effect, without anyone doing the conscious selection.
That there are 'selective' processes on words does not mean that all words change meanings at the same rate. The verb "to be" has been under no evolutionary pressure to change in a thousand years--until your Mr. Clinton tried (and mercifully failed!) to challenge the meaning of the word "is"!
Linguistics, like biology, has nothing to say on abiogenesis, simply because we have no data at all on how language ever got started. But once started, the evolution of existing languages from earlier ones can be documented.
The analogy can be extended, but cutting to the chase: It would be easy to suppose, on casual examination, that a language--with its extensive vocabulary, shades of semantic nuance, extended rules of grammar--was too 'irreducibly complex' to be anything other than a designed artifact. But this is demonstrably false!
I'm wandering too far away from the topic here--I'll shut up now!
Technology: Something gets developed, then numerous improvements are added on, etc., often with various categories of devices that go on to serve different purposes. Example: steam engines, electric motors.
Free enterprise: same as above.
Religion: consider the current state of Protestantism in the US.
All of the foregoing are examples of unguided descent with modification. "Unguided" in the sense that although each little step was someone's plan, there was never any grand plan to take things to their present state.
Well, I think that linguistics as you have described it here is a poor analogy. Natural selection does not have a goal(supposedly). Language has a goal, namely communication.