Project Description

The project aims to compare argument structure in typologically different languages, and to develop tools for a quantitative analysis of the data regarding argument structure. This should lead to a better understanding of the relation between the verb and the other sentence components, while in the meantime enhancing available methods for language comparison. In the future, our results might also lead to practical improvements deriving from a deeper understanding of translatability of verbs and clauses.

We will use two typologically distant languages, i.e. Ancient Greek and Yucatec Maya, and will develop a tool that allows to query argument structure in annotated copora of texts in the two languages. The data will then be analyzed and compared through quantitative methodologies. In particular, statistical methods will be used to assess frequency and productivity of argument structures, as well as their developments or tendencies in their possible development.

The remarkable typological and sociolinguistic differences exhibited by the two languages will grant the results of our research a wider range of usability, and will allow the applications of methods and tools developed by our research to other languages.

Basic concepts in our understanding of language and language structure are those of text and usage. Following this view, linguistic structures emerge from usage, rather than existing before and independently of it. For this reason we make extensive use of large electronic corpora, whose size makes it possible to reach statistically significant results. Numerous and useful tools for corpus research have been developed over the last few decades, among which syntactic Treebanks, which, however, do not yet rely on wide corpora. For this reason we will need a preparatory stage during which we will build on existing corpora, in order to enlarge the usable database and smoothen down differences between corpora in the two languages. Query tools for argument structures exist and are implemented for several languages, including Indo-European and non-Indo-European, but never for head marking languages such as Yucatec. For this reason, a stage in our research will be devoted to the implementation of such tools. In addition, conversion tools will be developed to deal with diferent types of annotation, This will lead especially to an enlargenment of the Greek corpus, as corpora already exist based on different annotation standards (PerseusTreebank, Prague Treebank, Proiel, Exmaralda, Penn Treebank).

The envisaged results of the project are summarized as follows:

1. We will shed light on a number of empyrical issues regarding a specific area of language, i.e. argument structure and its development in connection with the realization of referential arguments.

2. Our research provides a contribution to the clarification of theoretical problems such as inclusion of verbs in semantic classes and their argument structure, and referent tracking through null vs. overt anaphora.

3. The project provides a durable contribution to the preparation and to the improvement of corpora, which will become available for futher research in the future. This part of the project will be developed in partnership with leading international institutions in the field of computational linguistics and annotation.