Dennis King wrote:

> Question: is do-
> periphrasis (aka "irrelevant 'do'" -- ;-) ) something that Old Irish 
> and Brittonic both had as part of their common Celtic patrimony?  

I doubt this very much. Neither in Old Irish, nor in the old British 
Celtic languages is the DO-periphrasis more than an option. Given 
that both branches possess verbal nouns, it does not appear to be 
unnatural that the languages could develop constructions of the type 
"periphrastic verb + VN" as a variant beside the non-periphrastic 
construction, for example for such pragmatic reasons as morphological 
ease. Alternatively, the periphrastic construction may have arisen as 
a contact phenomenon among Celtic languages in the middle ages.

In any case, if periphrastic DO was inherited in Goidelic and 
British, it can only go back to the Insular Celtic node. There is no 
trace of such a construction in Continental Celtic languages. And, 
furthermore, in any case there's a decisive difference between the 
Celtic and the English periphrastic construction: whereas in the 
former it is optional, it has become an obligatory feature of modern 
English syntax.

David