Proceedings of the Tenth ESSLLI Student Session

This paper is about the development of semi-automatic methods to disambiguate genitive attributes of German deverbal nouns. Human listeners use structural and semantic clues to disambiguate these genitives — tools for automatic semantic text analysis, information extraction and information retrieval need to mimic this process by extracting clues from corpus data. I present morphosyntactic and lexical-semantic features that can be used as clues in semi-automatic disambiguation and first results from the analysis of a large data sample.1 1 Nominalizations and genitive attributes In my Ph.D. dissertation project I develop methods for semi-automatic text analysis to disambiguate genitive attributes of German deverbal nouns, especially those ending in -ung, like Verfolgung “persecution”. The suffix -ung is used to derive feminine nouns (mostly) from transitive verbs. It shares an etymological root with English -ing but forms full nouns. The meaning of -ung is comparable to that of the Latin-based -(a)tion. As with many relational nouns — e.g. picture in Picasso’s picture —, genitive attributes of -ung nominals can be subject (agent-related, 1) or object (theme-related, 2) genitives: (1) a. In in den the Dörfern villages leben live rund about 45000 45000 Menschen, people die who vor before der the-dat Verfolgung persecution der Armee the-gen army geflohen fled waren. were I would like to thank Hans Kamp and Ulrich Heid for helpful comments and discussions and two anonymous reviewers for their suggestions. Proceedings of the Tenth ESSLLI Student Session Judit Gervain (editor) Chapter 1, Copyright c © 2005, Nadine Aldinger 1 “In the villages live about 45000 people who had fled the persecution of the army.” b. Ihr she-dat missfällt, disapproves dass that die the Verfolgung persecution des Geschlechts the-gen gender in in Deutschland Germany als as Asylgrund asylum-reason nicht not anerkannt accepted wird. is “She disapproves of the fact that persecution of gender is not accepted as a reason for asylum in Germany.” (2) a. Dort There ist is die the Radioaktivität radioactivity laut according-to Messungen measurements der Organisation the-gen organization hundertmal a-hundred-times höher higher als than normal. normal “According to measurings of/by the organization, the radioactivity level is a hundred times higher there.” b. Der the Satellit satellite dient serves der the-dat Messung measurement von Verschiebungen of shifts in in der the Erdkruste. earth’s-crust “The satellite serves to measure shifts in the earth’s crust.” In (1a), the genitive der Armee “of the army” denotes the role-player who persecutes (the people); in (1b), the genitive des Geschlechts “of gender” denotes the role-player who is persecuted. At the structural surface, both interpretations are possible in both cases. The interpretations are parallel in (2a) and (2b): the genitive denotes the measurer in (2a), the thing measured in (2b). Human listeners use structural and semantic clues as well as world knowledge to disambiguate genitive attributes — e.g. in (1a) and (1b) the knowledge that armies are military organizations and therefore persecute others more often than they are persecuted. Automatic text analyzers also need to disambiguate genitive attributes, in order to provide input for information extraction and information retrieval. But they do not have a very elaborate “understanding” of the world yet; in particular, they do not possess the world knowledge that humans have at their hands to determine the interpretation intended. 2I do not distinguish between “genuine” subject genitive and author’s genitive, as Ehrich and Rapp (2000) do; for this project, the important matter is the semantic relatedness of the genitive attribute to the agent/subject/causer of the event/state/process characterized by the nominal’s base verb. 3There are also non-thematic genitives, e.g. temporal (die Lieferungen dieses Monats “the deliveries of this month”) and modal genitives (die Lieferungen dieser Art “the deliveries of this kind”). They are not restricted to nominalizations and therefore will not be treated further in this paper, although they must of course be detected and filtered out in the disambiguation process. See section 4.4 for a list.