Machine Learning Models of Universal Grammar Parameter Dependencies

The use of parameters in the descrip- tion of natural language syntax has to balance between the need to discrim- inate among (sometimes subtly dier- ent) languages, which can be seen as a cross-linguistic version of Chomsky's (1964) descriptive adequacy, and the complexity of the acquisition task that a large number of parameters would imply, which is a problem for explana- tory adequacy. Here we present a novel approach in which a machine learning algorithm is used to nd dependencies in a table of parameters. The result is a dependency graph in which some of the parameters can be fully predicted from others. These empirical ndings can be then subjected to linguistic analy- sis, which may either refute them by providing typological counter-examples of languages not included in the origi- nal dataset, dismiss them on theoret- ical grounds, or uphold them as ten- tative empirical laws worth of further study.