Natural Language Processing

Natural Language Processing (NLP) – or Computational Linguistics – is a major field of computer-related research. And of course language has long been considered central to the study of Philosophy. So it is not surprising that an increasing amount of contemporary work on the Philosophy of Language is influenced by computational insights.

As in many other areas of cognitive activity (e.g. identifying objects within a visual scene), the attempt to reproduce human performance has revealed the complexity which underlies behaviour that we often take for granted. Early attempts to automate translation, for example, proved hopeless, showing that deep understanding of linguistic structure and processes would be required to make it feasible. A similar moral can be suggested in the case of philosophical discussions of "semantics", such as the classic problems of referring expressions and indirect contexts (e.g. Russell's Theory of Definite Descriptions, and theories of belief ascription). Many discussions of these problems in the philosophical literature can seem crude and naïve to those accustomed to the difficulties of practical AI. Appeal to "Gricean principles" or pragmatic domains of "salience" are typically too vague to generate any testable predictions, sometimes leaving the suspicion that a theory has been saved from refutation only by the expedient of losing its content. Solid progress in these sorts of areas seems most likely to come from concrete implementations of the relevant theories, proving by example that a theory can work and generate definite, testable predictions.

Learning about Natural Language Processing

Some initial insight into NLP – including simple syntactic processing and grammatical transformations – can be acquired by playing with a system called a "chatbot". The Elizabeth educational chatbot was designed for this purpose, and is hosted on this site. The system has sufficient power to permit the handling of complex grammatical transformations and resolution theorem-proving, and is packaged with self-teaching materials designed for the non-specialist.