Drawing on the resources of reception theory, binding theory and text technology, I present a formal grammar especially designed to identify expressions co-referent with the German reflexive sich. I take co-reference to be an empirical relation determined by the use of native
speakers. The aim of my paper, now, is to investigate in how far the empirical performance of native speakers can be simulated by a formal grammar. According to my formal grammar, sich may occur in twelve different syntactical schemata. I investigate the possibilities of determining
the expressions co-referring with sich in the two most frequent schemata. I find that 45 formal rules are needed to do the job. Of these, 39 rules are grammatically decidable in the sense that they identify for any given grammatically analysed construction at most one expression as
the co-referring one. For the remaining six rules, I assume without investigating the matter other knowledge components than grammar to enable native speakers to disambiguate the relation between the reflexive and its co-referring expressions. I end with a comment on the nature of the grammar
Document Type: Research Article
Publication date: August 1, 2008
More about this publication?
"Linguistische Berichte" is open-minded regarding linguistic subjects and methods, but attaches great importance to theoretical reasoning and empirical validation. Since its foundation in 1969 by Peter Hartmann and Arnim von Stechow, the journal is an academic forum where serious views within linguistics and similar fields of studies (psycholinguistics, sociolinguistics, language acquisition and teaching, literary studies, philosophy, and computer linguistics) are discussed equally. The purpose is to provide an objective and critical documentation of linguistic developments. Besides, the forum is open for each level of scientific work, i.e. for everyone who is dealing with linguistic-based work (undergraduates, assistants, and professors). New research results and linguistic developments are published quickly and comprehensively, and contributions are normally published within one year after acceptance by the editors.