Grzegorz Jagodzinski wrote:
> Were they more frequent than "head" and "hawk"? And especially than "has"
> and "had"?
Fortunately, thanks to the enormous size of the ME corpus we can leave
the realm of speculation. It is possible to estimate these frequencies,
e.g. by counting quotations under the relevant entries in Kurath's
Middle English Dictionary. It turns out that <seven> and <heven>, while
less frequent than <hed> 'head', were several times more frequent than
<hauk> 'hawk'. (The frequency of <seven> would further rise, quite
dramatically, if I'd taken into account <seventh>, <seventi> and
compounds containing this numeral.) The bird names <raven> and <hauk>
'hawk' were in the same range, but both were more frequent than <lark(e>
'skylark' (< laverke).
As for <haven> 'have' (disyllabic till Late ME), it is noteworthy that
Mod.E <have> has kept its /v/ despite its top-range frequency. It seems
it wasn't an _intervocalic_ /v/ that was lost in forms of this verb
(except occasionally) but the preconsonantal fricative in <hafst>,
<hafth> and <hafde(n>. I absolutely agree that this particular cluster
simplification has to do with the frequency of the verb.
Piotr