This is a plain-text version of the mail just sent. Sorry for any confusion caused.  



Our discussion lately has maintained that grammar is a model of the structure of the language.  This model may be formal or informal.  If it is informal, then it tends to be subjective and may change from time to time and place to place.  When it is formal such changes through time and place are seen as objective and rigid differences in particular models of Grammar.  The study of structural models is a branch of mathematics.  One desideratum of mathematics is that as a tool of science it be formal and subject to the construction of a proof.  Proofs may contain errors but it is frustrating for a scientist to speak of an "error" without a formal model to relate it to.  I have seen the "had-for-did" error described as an optional "back-shifting."  This is related to the fact that sometimes quoted material may be direct and sometimes indirect.  An "error" may come in maintaining the tense of the material quoted: "You did not clean your room!"  Since the mother is being quoted indirectly, in some models the tense may be back-shifted to agree with the past tense of "complained."  This, then, is taken as the origin of the "past perfect."  Other models may make it correspond to any past-in-the-past tense, and some might allow only the present perfect to have a past perfect.  Some models might allow any of these descriptions.  However, it seems that only when it is to be made relative to a formal model can there be an error one way or the other. 



Bruce