The question of why simultaneous interpreting merits our interest needs to be asked a priori because after sixty years of providing a vital daily service to the international community, the activity remains an arcane field of study. This status of the discipline is probably due in equal parts to the occult, not-quite-respectable odour of translation generally, and to the extreme difficulty of capturing SI for research. Translation is often regarded, particularly by monolingual speakers of

more »

... nt languages, as an irritating necessary evil. When sacred texts have had to be translated to evangelical ends, the translations have never quite acquired the status of holy writ; translations of legal documents gain equal status with the original when multiple official languages are constitutionally recognised, but then they are explicitly not to be considered translations. In the folk view of language as encoding thoughts, or as inextricably bound to them, a translation can never be more than an approximation, a pale substitute for the original. Public figures often seem impatient or resentful of the need to go through interpretation. The prospect of imposing an international language is seen by many now as an historic opportunity far outweighing the sad loss of all the other languages of humanity. Philosophical and linguistic inquiry into translation has contributed to this view in concluding that translation can never be perfect (Quine 1960), but only good enough for practical purposes (Keenan 1978). If the mainstream linguistic and cognitive sciences have not looked to translation or interpretation as valid objects of study, it is probably due to a belief that we can learn about language and thought only by studying the 'pure' products of spontaneous sovereign linguistic creation from spontaneous thought. The analyses and conclusions of authors like Quine and Keenan are valid, but the mistaken inference that translations are therefore products of linguistic communication which are less pure than spontaneous production, or epistemologically inferior to it, reflects an obsolete logical-semantics paradigm which sees language as a code capable of perfectly expressing thought if only it is perfectly used. In a modern, code-plus-inference model of linguistic communication, in contrast, ordinary spontaneous (or even carefully composed) linguistic productions do not perfectly express thoughts and communicative intentions, but merely offer sophisticated evidence for inferring them. In this paradigm, sovereign utterances, as best attempts to represent and communicate 'original' thoughts, enjoy no special status over 'translations', which are best attempts to communicate a thought originating in the act of understanding someone else's utterance; the resulting text or discourse is no less capable of enriching and fertilising a target culture, or stimulating associations in an audience, as the 'original' in its own domain. Because of their 'necessary evil' status, translating and interpreting attract more pedagogical than scientific interest, since translators must continue to be trained for the foreseeable future. To the extent that interpreting has been studied, the emphasis has also been on the help that cognitive modelling might provide to training. However, quite soon after the emergence of SI, some trainer-researchers who examined recordings and transcripts more closely came to believe that this process of transfer from language to thought to language under special constraints might offer new insights into the relationship between language and mind; and interpreting has also periodically aroused the curiosity of some linguists and psychologists. The question of 'what is so interesting' about SI therefore goes beyond the 70