formulaic sequences

*** Apologies for multiple postings! ***

Dear list members,

As a part of my M.A. thesis, I am investigating multi-word units 
(otherways called recurrent sequences, word bundles, etc.) in non-native 
English speech.

I remember having read somewhere that if such a string of words contains 
repeats (of the same words) and/or hesitation (like erm, mm, uhu) within 
it, it cannot be treated as a formula, for the very inclusion of 
repetition or hesitation proves the sequence was not processed as a 
whole but rather as a set of words.

Unfortunately, I do not remember from which scolar the idea comes and I 
need it for justification of my very study.

I'll appreciate any hints or direct reference to the source.

Best regards,

Joanna Jendryczka
------------------
------------------
M.A. English Linguistics student
A.Mickiewicz University Poznan, Poland