Machine Translation as a complex system, and the phenomenon of Esperanto

The history of machine translation and the history of Esperanto have long been connected, as they are two different ways to deal with the same problem: the problem of communication across language barriers. Language can be considered a Complex Adaptive System (CAS), and machine translation too. In fact, there are multiple agents (both natural and artificial) involved, interacting with one another and committed to achieve a common goal, i.e., the machine translation task. The main characteristics of language as a CAS are also shared in machine translation, especially if we consider the example-based, statistical approach, which is nowadays paradigmatic and unavoidable. In fact, control is distributed, there is no ideal representing agent (intrinsic diversity), there are perpetual dynamics in performance, adapted through amplification and competition of new examples from the crowd of users. On the other hand, Esperanto, being a living language, can be considered a CAS, but of a special kind, because its intrinsic regularity in structure simplifies the task of machine translation, at least up to a certain level. This paper reviews how Esperanto has enhanced the development of human-machine communication in general and within machine translation in particular, tracing some prospects for further development of machine translation, where Esperanto could play a key role