A Linguistic Discovery Program that Verbalizes its Discoveries.pdf

A Linguistic Discovery Program that Verbalizes its Discoveries Vladimir Pericliev* Max-Planck Institute for Evolutionary Anthropology, 04103 Leipzig, Germany pericliev@eva.mpg.de * Currently on leave from the Institute of Mathematics Informatics, 1113 Sofia, Bulgaria (peri@math.bas.bg). Abstract We describe a discovery program, called UNIVAUTO (UNIVersals AUthoringTOol), whose domain of application is the study of language universals, a classic trend in contemporary linguistics. Accepting as input information about languages, presented in terms of feature-values, the discoveries of another human agent arising from the same data, as well as some additional data, the program discovers the universals in the data, compares them with the discoveries of the human agent and, if appropriate, generates a report in English on its discoveries. Running UNIVAUTO on the data from the seminal paper of Greenberg (1966) on word order universals, the system has produced several linguistically valuable texts, two of which are published in a refereed linguistic journal. 1 Introduction Previous works in machine scientific discovery have mostly focussed on historical reconstruction (work culminating in the book by Langley et. al. 1987), but more recent efforts are directed towards designing programs that discover new scientific knowledge. Such systems operate in disciplines as diverse as mathematics, chemistry, astronomy, medicine or linguistics. The field is currently very active (for recent developments, cf. e.g. the special issues on discovery of the journals Artificial Intelligence, April 1997 or Foundations of Science 1999; the ECAI-98 Workshop on Discovery, The International Conferences on Discovery Science, Japan, 1998 and 1999). In this paper, we present UNIVAUTO (UNIVersals AUthoringTOol), a system whose domain of application is linguistics, and in particular, the study of language universals, a classic trend in contemporar