Very thoroughly indeed. The scenario
was that "of course" hobbits couldn't have spoken English (the story took
place far in the past); rather, they spoke their own language, called Westron
(but often referred to as the Common Speech). Tolkien "translated" this
language into English, which included "rendering" all the Common Speech
place-names into the equivalent English place-names. The object of the
exercise was to produce the following effect: names in the Common Speech
(which were familiar to the hobbits) were "rendered" into English (in which
form they would be familiar to us, the English-speaking
readers); names in other languages (usually Sindarin)
were "left alone", and thus were equally unfamiliar to the hobbits and
to us. Since the story was told largely from the hobbits' point of view,
that we should share their linguistic experience is a desirable result
(especially for Tolkien, who was unusually sensitive to such matters).
In portraying the linguistic landscape
of Middle-earth he carried this procedure much further. The main example
was his "substitution" of Anglo-Saxon for Rohirric. The "rationale" was
that the hobbits' dialect of Westron was distantly related to Rohirric;
therefore, when hobbits heard Rohirric they recognized many words but the
language nevertheless remained just beyond understanding (RK, 65 (V,3)).
Thus, Tolkien attempted to further "duplicate" hobbit linguistic perceptions
by "substituting" that language of our world (Anglo-Saxon) which has (more-or-less)
the same relation to English that Rohirric
had to the hobbit version of Westron.
There were many other nuances in the intricate and subtle linguistic web he devised (always, he carefully explained, in the interests of "reproducing" the linguistic map of Middle-earth in a way that could be easily assimilated by modern English-speaking readers). Thus:
a) Archaic English roots were used in those Common Speech place-names which were given long before the time of the story (e.g. Tindrock, Derndingle; see Guide).
b) Some of the Stoors (who later settled in Buckland and the Marish) dwelt in Dunland at one time (Tale of Years, entries for TA 1150 and 1630 (RK, App B)); the men of Bree also came from that region originally (RK, 408 (App F, I, "Of Men", "Of Hobbits")). "Since the survival of traces of the older language of the Stoors and the Bree-men resembled the survival of Celtic elements in England" (RK, 414 (App F, II)), the place-names in Bree were Celtic in origin (Bree, Archet, Chetwood) (see also Guide). Similarly, the names of the Buckland hobbits were Welsh (e.g. Madoc, Berilac).
c) Among hobbits some of the older Fallohide families
liked to give themselves high-sounding names from the legendary past (an
example of hobbit humor). Tolkien "represented" such names by names of
Frankish or Gothic origin (Isengrim, Rudigar,
Fredegar, Peregrin). These matters and much else is explained
in detail in Appendix F.