In Bialystok and K ellerm an (1987), it was proposed tha t any a ttem p t to p roduce a taxonom y o f com m unication strategies as cognitive processes should m eet th ree requirem ents: parsimony, generalizability and psychological plausibility. These th ree are con sidered desirable conditions for a p ro p er accoun t o f strategic behaviour. T h e first requ irem ent, parsimony, is tha t there should be as few discrete strategy types proposed to classify the data as possible. In the absence o f o th e r relevant criteria, O ccam ’s Razor should be applied in favour o f the m ost parsim onious taxonomy. T he second requ irem en t, generalizability, states th a t any proposed taxonom y should be equally applicable to all norm ally endow ed speakers perfo rm in g any task eng en d erin g strategic behaviour, irrespective o f w hether first o r second languages are involved, and in the la tte r case irrespective o f proficiency level o r the particu lar coupling o f the lea rn ers’ native and non-native languages. T he th ird and m ost im portan t requ irem ent, psychological plausibility, holds tha t any proposed taxonom y o f strategies should be in fo rm ed by what is curren tly known abou t language processing, cognition and problem -solving behaviour.