Varieties of English defined by users (as dialects, sociolects and chronolects) and by uses (according to medium, formality, text type, etc.) have been the object of description in various places, and have of course functioned conspicuously in this journal which has the topic as part of its title. By contrast, the questions of what can be considered English, and how its outer boundaries are defined, have been asked less frequently, and not in any comprehensive way. (In EWW I have followed a pragmatic editorial course in admitting varieties which have some linguistic relationship with English and are in a contact situation/coexistence with English in the speech community discussed.) My paper looks at a few 'problem cases' among utterances, in particular at various forms of broken English and linguistic experiments, at language mix and code-switching and then turns to linguistic systems, with semi-languages, pidgins, Creoles, cants and mixed languages singled out for detailed discussion. A classification of the varieties treated obviously depends on the degree of their divergence from English, their functional range and standardization, users' attitudes and the ways how the language is acquired -four factors which can have different weight for the classification in the individual case.