Even on the basis of a documentation as rich and extensive, cataloguing the pronunciations of English in a global perspective seems a herculean task, due to several basic problems and pitfalls involved. For one thing, there is the immense amount of variability that can be observed: While the range of possible pronunciations is naturally constrained by the conditions and limitations imposed by articulatory space and organs, the amount of detail of sound realizations – idiosyncratic, phonologically conditioned or not, socially or regionally motivated – is extremely difficult to grasp and categorize. Essentially, a resultant problem of this is the difference in levels of details of phonetic descriptions from one study or description to another, also in this Handbook: it ranges from minute phonetic analyses with lots of diacritics to essentially broad phonemic categorizations. Secondarily, conventional descriptive models, most notably the structuralist idea of a phoneme system selected by any individual language (or variety?) from an infinite set of articulatory possibilities, fail in the absence of phonological analyses of practically all nonstandard varieties of English (the only attempt at a systematic analysis of the phonological system of a local dialect that I know of is McDavid 1985; Wells 1982 contains a few sections on regionally varying phonemic part-systems and many remarks on facets of the phonologies of many varieties). Essentially, this is the framework in which most descriptions operate; feature-based theories or other advances of phonological theory are therefore largely ignored here. It is clear that the phonemic load of individual phonemes (as determined by their frequency overall, or the number of minimal pairs that they enter) varies greatly, even in “Standard” varieties (for instance, in RP, /Ʒ/ is known to be relatively rare), and whether two phonetically observed sounds are to be credited the status of phonemes or not is a matter of more detailed analysis and argumentation in many instances (cf. Gleason 1970): take the fact that argumentation is required to underline the status of affricates as single phonemes in English, or the observation that /h/ and /ŋ/ always occur in complementary distribution. Similarly, certain sounds are assumed to have merged in certain varieties, but then some mergers have turned out to be near-mergers only (a concept which oscillates fuzzily between a phonological and a phonetic perspective). Some sounds are assumed to have “changed” in certain ways in certain varieties – but then some of these changes have been found to be a matter of lexical incidence, i.e. to affect some words in which the sound (i.e. “phoneme”) occurs but not others; so has the phonetic realization of the phoneme in question been changed, or has a phonemic split occurred? (Essentially, this relates to the fundamental distinction between phonetically gradual “neogrammarian change” and abrupt “lexical diffusion”, as discussed, for instance, by Labov 1981.) Thirdly, it seems equally difficult to tackle the most interesting question involved, that for the motivation behind accent differences. Do natural principles play a role, could it be the case that chain shifts or other phonemic rotations diffuse globally? Possibly so – but then, the distribution of vocalic space in many varieties, including RP, is anything but symmetric (or to be accounted for by a principle of an optimal distribution of the available vocalic space). Are sociopsychological motivations decisive, like a group’s desire to express their identity by some phonetic means loaded with symbolic meaning? Possibly so, but then, which variants are likely to be chosen for such functions, and why – or does such a selection simply occur haphazardly? Is all variability barely local? The set of possibly pertinent parameters seems endless.

What the above considerations are meant to imply is that any attempt at a bird’s eye view, as in this paper, unavoidably is bound to leak: Rich as the documentation of the accent variability of English in a global perspective is, it seems impossible to do more than touch upon a few generalizing tendencies and observations. For more details, and generalizations at different levels, the reader is referred back to the individual papers and the regional synopses. By necessity, the coverage of the material in what follows is selective, and abstracting from many other facts and observations which might be equally interesting but cannot be addressed here. What follows is a synopsis – it is neither a thorough documentation nor a systematic analysis.