Syntax and Semantics of Adpositions and Case Marking. Description of Prepositions in English and Adpositions and Case Marking in Finnish within the Framework of a Parallel Database

This document is a parallel research on prepositions in English and case marking and adpositions in Finnish (ADPOS-CASE relators) which consists of a three-fold system of locative and de- and adlocative relators and circumstantial relators bound with various syntactic positions. The three-fold system of relators is examined with respect to their occurrences in three-dimensional space based on the closure of space and distances between objects. The ADPOS-CASE relators in the parallel database are extracted with the help of rules which structurally are Regular Expressions, and ambiguous results are examined with the help of disambiguation rules. Various statistical methods are used in revealing differences between structural systems of languages. The statistical model Self-Organizing Maps is used in examining the organization of ADPOS-CASE relators and also clusters of relators in combination with various semantic groups of verbs. For evaluating the value of semantic categories used in the analysis of English and Finnish, descriptions of ADPOS-CASE relators are extended to concern some Indo-European and Finno-Ugric languages and a Malayo-Polynesian language. The parallel database used as research material consists of English and Finnish translations of the Bible in the machine-readable form. It is shown that it is possible to examine the distribution of the ADPOS-CASE relators with respect to semantic categories which are independent from the surface grammar. Disambiguation of the results extracted from the database completes information expressed with grammatical rules. In Finnish most disambiguation rules deal with morphological categories, but in English in which differentiation of lexical categories is weaker, disambiguation concerns, in addition to lexical, also phrasal units. A special value of statistical methods is shown in opening clusters of properties in investigating interaction between properties of languages.