The surname <Knivett> (<Knyvett>, <Knevet>, etc.) is usually
explained as in the following entry from Reaney & Wilson, A
Dictionary of English Surnames, OUP, 1995; angle brackets
for italics in the original):
Knivett, Knyvett, Knevet, Knevit, de Knevett, Nevet,
Nevett, Nevitt: Leuricus, Nicholas <Cnivet> 1087-97
Crispin (Mx), 1185 Templars (Herts); William, Osbert
<Knivet> Hy 2 DC (L), 1199 AssSt; Ernald <Knikt> 1275 RH
(Nf); Thomas <Knifet> ib. (L); Walter <le Knift> 1279 RH
ib. (O); John <Knyft>, <Knyvet> 1311, 1337 ColchCt. A
Norman pronunciation of <Knight>, owing to the French
difficulty with the <h> of <cniht>. Matthew <de Knyvet>
1273 RH (Nt) must be identical with Matthew <de Knyveton_
above [in the previous entry -BMS] and William <de Knyvet>
1327 SRDb must similarly be for <Knyvet'>, i.e.,
<Kniveton>. Alexander <de Knyft> 1279 RH (O) is an error
for <le Knyft> (cf. Walter <le Knift> above). There is no
evidence for a place <Knevet> and the modern <de Knevett>
is either a perpetuation of one of these errors or due to
a late prefixing of an unetymological <de>.
The previous entry is for the surname <Kniveton>, <Knifton>,
which is locative, from Kniveton in Derbyshire.
The AFr problems with OE /x/ are real enough: spellings show
that it could be replaced with /k/ or /s/ or dropped
altogether (e.g., <cnict>, <cnist>, <kneit>). The initial
/kn-/ was also a bit of a problem, sometimes solved by
epenthesis (<chenict>, where <ch> = /k/) and sometimes by
metathesis (<cincht>).
Now while a substitution of /f/ for /x/ is certainly not
foreign to English (<laugh>, <enough>), I have not seen an
undoubted example of the word with <u/v> or <f>, or for that
matter with <-et>. Nor have I noticed similar treatments of
other words of the same shape, like <night> and <right>.
Also, it's notable that all of the early citations that I
can find have <u/v>, not <f>, and final <-et>.
The earliest is the first mentioned by Reaney & Wilson:
Leuricus Cnivet witnessed the grant by Geoffrey de
Mandeville (<Goffridus de Magna Villa>) to St Peter of
Westminster of a small manor 'in presentia Gisleberti
abbatis et monachorum et multorum militum meorum et suorum',
presumably as one of the knights.
<
http://www.british-history.ac.uk/report.aspx?compid=63959>
<Leuricus> is OE <Le:ofri:c>, and I'm strongly tempted to
see his byname as a very early instance of the Anglo-French
diminutive <cnivet> from OE <cni:f> rather than as an AFr
distortion of OE <cniht>.
I've two questions. First, have I overlooked some obvious
problem with the idea? Secondly, has anyone seen <-ivet> or
the like as an AFr treatment of OE <-iht>?
Brian