OK, time to admit I was wrong.
My last posting, on the borrowing of words, failed to get to the core of the problems of word borrowing.
The crux of the matter is publishing. As soon as a word comes into our language today it is written down, printed and seen by millions of people. That makes it very hard for the spelling to change. This is the reason we have curious spellings like niche in our language. Of course, this wasn't the situation in the past. After the vikings were expelled from England, it took hundreds of years for the words they left behind to appear in writing. As a result, the original foreign spelling was long forgotten and the words were written as they pronounced them. The pronunciation probably changed too, to better fit the cadences of English.
These problems with borrowing do not only occur in English; this is a current problem in Scots Gaelic and the strong viking influence in the Hebrides makes the two languages quite comparable.
The viking word for hall was halla (as in Valhalla). This was borrowed into Gaelic, but over time the pronunciation changed to better fit Gaelic and became talla. This is due to a process called delinition. (Under certain circumstances an initial T is lenited to the form TH and pronounced like the English H. As this sound cannot occur at the start of a true Gaelic word, this sound was retained for the lenited form and the root form of the word had the delinited T.)
Modern borrowings, however, are taken as close as they can be, eg bhuruca for verruca. This is inappropriate for two reasons:
1) It's a lenited form and should be delenited.
2) The stress in Gaelic words is on the first syllable, whereas the stress in the English verruca is on the second.
No-one wants to engage in artificial language change, quite understandably; languages can and should change of their own accord. However, when we put a newly borrowed term in writing, we are fixing its spelling before it's had time to "bed in" with its new host language. The spelling encodes a pronunciation which is therefore enforced by the use of the dictionary as an authority. By fixing things in an alien form, you force this new form to be accepted as part of the host language. As such, a written borrowing is a form of language change.
We are now at a stage where we understand language change enough to model the changes that have occurred in older borrowings. I concede that consciously applying these models is not a natural form of language, but I contest that it is a naturalistic form of change. If writing borrowed words down is a language change then it must be considered artificial, as language is first and foremost a spoken system.
In the absence of natural language change we must embrace the naturalistic.