🚀 go-pugleaf

My reference is Pierre Tre'pos' "Grammaire bretonne" (Imprimerie Simon,
Rennes, undated, probably ca 1975)

Page 187, Tre'pos discusses verbal particles. Of "a", paragraph
422, he writes:

Elle a la valeur du _pronom relatif_ A: qui, que et ne s'emploie
que lorsque le _sujet_ ou le _comple'ment d'objet_ direct 
_pre'ce`de_ le verbe: [his emphasis]

         tad-koz a bren eur pakad butun beb sul (eur pakad
         butun a bren tad-koz beb sul), grand-p`ere ache`te un
         paquet de tabac chaque dimanche.

In English, literally:

[Verbal particle A] is equivalent to the relative pronoun: who,
whom, which, and is used only when the _subject_ or the _direct
object_ complement _precedes_ the verb:
        
         tad-koz a bren eur pakad butun beb sul (eur pakad
         butun a bren tad-koz beb sul), granddad buys a packet
         of tobacco every Sunday.


Interlinear translation:

         tad    -  koz    a   bren   eur   pakad    butun
         father    old        buys    a    packet   tobacco

         beb   sul
         every Sunday


In the alternative construction, in parentheses, subject and
direct object have switched positions.  Disambiguation is
purely by meaning.

Tre'pos points out that, in the part on syntax, that the passive
is extremely frequent in Breton. There is no ambiguity in passive
sentences, the agent being marked by the preposition "gand". E.g.
(p.243, paragraph 575):

        drailhet eo bet   e    vraoig   gand     ar   paotr bihan
        broken   is been  his  toy      (agent)  the  boy   little.
        The little boy has broken his toy