Rob wrote:
>> *skurta- > *skort > sceort 'short'
>> *juka- > *jok > geoc 'yoke'
>
>
> But if all /sk/ > /S/, why would the <e> be needed?
Not all; there were a few lexical exceptions, e.g. <a:scian>, <a:scode>
etc. 'to ask' (beside metathesised <a:hs-, a:x- ~ a:cs->). The <e> is
frequently written even in unstressed syllables, e.g. <fisceas>,
<bisceop>, where a diphthong had no business to be; the spelling is
certainly diacritic at least in such cases.
> On the other hand, perhaps <cealf> also implies a diacritic, whereby
> the pronunciation would have been /c^ælf/.
Not in West Saxon. Pre-liquid breaking was earlier than palatal
diphthongisation (which is why we have WS ceorl, georn, etc. rather than
*cierl/*cyrl), and so it must be pre-liquid breaking that is responsible
for the short diphthong in <cealf>. In this lexical set palatal the
effect of palatal diphthongisation is phonologically invisible.
Piotr