Methodological background

To provide a uniform basis for the cataloguing of the global pronunciation variability, I devised a checklist of phonetic features that was to be specified for each of the varieties under investigation. Essentially, the checklist was meant to anticipate and provide a categorial framework for the major variants that I expected to come up, based upon my familiarity with the variation of English and a perusal of some pertinent publications. It is divided into four sections. The first one, with 121 items by far the most voluminous one, covers the phonetic realization of vowels, based upon Wells’ (1982) lexical sets. For each of the key words, both one or two “canonical” realizations (as usually found in the major reference accents of BrE and AmE) and a few possible types of articulatory modifications (backing / fronting, raising / lowering, monophthongization / diphthongization with offglides, rounding / unrounding), with sample phonetic symbols, were specified. Items numbered 122 through 136, probed into vocalic distributions, i.e. asked whether specific mergers or similar phenomena occurred in the respective regions. The items number 137 through 174, checked the phonetic realization and distributional facts concerning consonants, and five items numbered 175–179, asked for prosodic features and intonation contours. In each case, the informants (i.e. the authors of the respective articles) were asked to indicate whether the respective feature occurs normally and is widespread (to be symbolized by an “A”), occurs with restrictions (“B”), or does not normally occur (“C”). Admittedly, this is a fairly crude categorization. In particular, category “B” covers a variety of fairly distinct constellations, namely, as spelled out in the instructions, that the feature “occurs sometimes / occasionally”, that it is found “with some speakers / groups”, or that it is restricted “to some environments” – in other words, it encompasses restricted frequency but also the cases of external, social and internal, linguistic conditioning. Category C is of course also possibly open to interpretation, given that it is practically not possible to positively document that a certain phenomenon does not occur at all in a given region; but the possibility of idiosyncratic occurrences should be provided for by the description of the category as “not normally” occurring. A number of contributors left many cells blank, indicating that these are cases of non-occurrence, i.e. “C”. Occasionally, some authors felt a need to be more specific, and they suggested or generated intermediate categories like “BC”. In such cases, specific details or added comments (which also were provided in individual cases) were put aside and collected in a separate fi le; for the table and mapping procedure itself, the articles themselves were checked for more acccurate information (so that a clear categorization could be achieved), but usually “B” tended to be the catch-all category for such intermediate instances.

In general, however, the feature listing worked well, and some contributors stated that they found this preconceived categorization an interesting and useful tool for comparative analyses. Very rarely did a variant come up which could not be grasped by the suggested categories. Of course, the variants suggested are not mutually exclusive: several alternative pronunciations of a given key word may co-occur in a given region – typically one as the major one (“A”) and others as group-specific or environment-specific (“B”) variants. In that sense, the various groups of phenomena suggested for the same key word, belong together as possible variants of a variable.

The list of features itself, which encompasses a total of 179 items and is thus fairly long, is made fully accessible, as distributed to all article authors, in the Appendix to this paper. Regrettably, not all contributors responded, however. I would like to thank those who did, and also Raj Mesthrie and Kate Burridge, who practically produced almost all of the feature lists for Africa and Asia and the Pacific region themselves, based upon the articles. Similarly, a few of the lists for the Americas and the Caribbean (notably, the ones for the Urban South, Barbados, and Suriname) were compiled by me.

In the following discussion, vowels are classified into “short” and “long” ones. The quotation marks are meant to indicate that the labels are conventional categorizing devices rather than phonetically descriptive statements, given that the relationship between the phonological “length” and the physical duration of a sound is a highly complex and problematic one, and that lengthening and shortening processes are common in many varieties. Hence, “short” is meant to imply “classed as short in RP (as a primary reference accent) and short in the majority of (but not necessarily all) accents”, and vice versa for “long”.