Abstract

The original version of this talk was prompted by an invitation to the conference with the same title organised to celebrate the career of the phonetician John Esling. In it, I consider whether in reality there is an inventory of sounds, along the lines of the IPA framework, from which languages can select to make their phonemes and ‘extrinsic’ allophones, or whether, given the multiple degrees of freedom inherent in the vocal tract, sounds are infinitely variable across languages, and phonetic categories exist merely by dint of analysts imposing them on a continuously variable phonetic space. In answering this question we need to consider the role of ‘articulatory settings’, long term configurational trends which may characterise speech in a given language, or indeed a given speaker. Can we abstract away from variation in the realisation of a putative ‘phonetic category’ by attributing it to a language-specific vocal ‘gait’ which imposes itself on the individual steps – i.e. sounds? Ultimately, I conclude that ‘building site’ metaphors may be inappropriate. Instead, an imp®udent analogy will be proposed with a model in another area of science.