The Assessment of Semantic Cases Using English Positional, Prepositional and Adverbial Case Markers

Semantic analysis of text is often based on semantic roles drawn from a small, closed set. The set of roles describing relationships between a verb and its arguments within a clause are cases. This report assesses the cases used in the TANKA project (Text ANalysis for Knowledge Acquisition) according to the distribution of the words and syntactic constituents that mark them. The definitions of TANKA’s cases, the number of instances of each case in a test text and numerous English example sentences and counterexamples are also used in the assessment. Results show that TANKA’s cases are supported by the marker data. 1 Background The aim of the TANKA project is to acquire knowledge from unedited English technical text using a minimum of pre-coded semantic information. To realize this goal, semantic analysis in TANKA semi-automatically identifies relationships among the different elements of English sentences—clauses, verbs, verb arguments, noun phrases—found in parse trees produced by the DIPETT parser (Delisle 1994). The HAIKU semantic analyzer determines which pairs of elements are involved in relationships, based on the presence of particular words and syntactic constituents. There are three kinds of relationship in this framework: clause-level relationships (between two or more clauses); case relationships (between a verb and its arguments) and noun-modifier relationships (between a noun and its modifiers). A more detailed discussion of the semantic relationships in TANKA appears in Delisle et al. (1996). The work described in this report deals only with case relationships. A case is a generalization from a particular verb-argument pair to one of a finite set of semantic role labels. The cases used in TANKA are listed in section 3 and defined in Appendix III. A case marker is a lexical or syntactic element that indicates the presence of a case. Detailed discussions of case theory and case markers can be found in Fillmore (1968), Bruce (1975), Somers (1987) and Palmer (1990). The purpose of this report is to use the distribution of case markers among TANKA’s case set to support the inclusion of individual cases in the set. Instances of the cases in an English text will also be considered to determine if there is any empirical support for the inclusion of cases that are relatively weakly represented by the markers.