The Treatment of Information Structure in Machine Translation Patrizia Paggio September 1996 Contents Acknowledgements i 1. Introduction 1 1.1 The aims of the research 2 1.2 The data 3 1.3 Structure of this thesis 3 2 Information structure and Machine Translation 4 2.1 PC-based translation 6 2.2 Linguistically sophisticated models 7 2.3 Research-oriented MT 10 2.4 The ALEP system 11 2.5 Summary 12 3 Some previous approaches to information structure 13 3.1 The Prague model: the topic-focus articulation 13 3.2 The theme in Halliday's approach 16 3.3 Focus in generative grammar 19 3.3.1 Chomsky: focus and deep structure 19 3.3.2 Rochemont and Culicover: focus marking constructions 20 3.4 Information focus in artificial intelligence 26 3.5 Discourse analysis 28 3.6 Summary 30 4 Information Structure Revisited 32 4.1 Topicality 33 4.1.1 New, given and resumed topic: three different pragmatic functions 33 4.1.2 The primary topic 35 4.1.3 Contrastive topics 37 4.2 Focality 39 4.2.1 The pragmatic functions of focus 39 4.2.2 A semantic approach to focus 41 4.2.3 Focus and Presupposition 46 4.2.4 Multiple and listed foci 47 4.3 Proposal for an HPSG-like representation of information structure 47 4.4 Summary 50 5 The syntactic realisation of information structure in Danish 52 5.1 Word order: theme and end-focus 52 5.2 Heuristics for the Computation of Information Structure 55 5.2.1 Position and Definiteness 56 5.2.2 Semantic Classifications 57 5.2.3 Systemic Ordering 58 5.2.4 Heuristics for Danish 63 5.3 Focussing and topicalising constructions 67 5.3.1 Clefts 68 5.3.2 Existential sentences 86 5.3.3 Topicalisation 99 5.3.4 Extraposition 108 5.4 Summary 118 6 Automatic analysis, transfer and generation 120 6.1 Implementing an MT system in ALEP 120 6.1.1 The formalism 120 6.1.2 System architecture 127 6.2 Implementing focus heuristics 129 6.2.1 Fronted adverbials 130 6.2.2 Assignment of INFOSTR values to subject and verb 132 6.2.3 Assignment of INFOSTR values to verbal complements and adjuncts 133 6.3 Implementing focussing constructions 141 6.3.1 Clefts 141 6.3.2 Existential constructions 148 6.4 Summary 152 7 Conclusions 153 Bibliography 155 Dansk resume 158 Appendix I: type declarations and coverage description 160 Appendix II: test suites and results 177