Bryan Jenner (Vienna)
The original idea of the Common Core (Jenner 1989) was based on the belief that the only conceivable model of pronunciation for the learner was provided by some sort of native speaker, but that we should only be able to achieve that if we changed our priorities and refrained from spending all our time on details of vowels.
This idea was taken up and modified (e.g. by Jenkins 1996, 2000) on the basis that most learners did not want or need to interact with native speakers, and the important thing was that they should be able to communicate with and understand each other: Jenkins therefore proposed a reduced common core .
In practice, however, both of these notions are based on the idea that the learner can either imitate native-speaker norms or deviate from them, in certain approved ways. It seems to me that neither Jenkins nor I went far enough in the direction of finding a true phonology of International English, in that like most British phoneticians we have been pre-occupied with details of realisation, rather than with systems.
What is now needed is an attempt to determine a 'real' phonological common core underlying L2 (and possibly also L1) realizations, based on data from fluent EIL performers rather than learners. Such a model would be at a very abstract or 'reference' level, and would in some respects resemble a model for grammar. Actual pronunciation variants would then be seen simply as alternative exponents of this.
Such an ultimate phonology of English might contain, for example, a very small number of vowels (perhaps 6), with no central vowels at all; and few, if any, diphthongs. It would be rhotic in principle; and length contrasts would be largely redundant.