Discourse Deixis and Null Anaphora in German

The main aim of this thesis is to provide insight into the interaction of the syntactic and pragmatic properties of German, particularly with respect to the issue of congurationality. This language is particularly di cult to classify as it displays both subject-object asymmetries (a feature of \con gurational" languages), but also has a topic position (a feature of \discourse-con gurational" languages). In order to avoid the di culties associated with subtle acceptability judgements from informants, the study presented here is based on a frequency analysis of word order variation in spoken language corpora. In the rst part, I concentrate on the initial position in German main clauses, which is traditionally referred to as the topic position, and using a task-oriented corpus provide the statistics for the following: The frequency of the di erent grammatical functions in initial position, in order to determine the relative frequency of the canonical SVO word order. The frequency of the di erent NP-forms (null anaphora, pronouns, demonstratives, de nite and inde nite NPs) in initial position, which are each associated with a particular cognitive status, thus giving insight into the precise nature of the initial position. These analyses show that the canonical word order occurs in less than 50% of the utterances and that the initial position is most frequently lled with null topics, pronouns and demonstratives, i.e. NP-forms associated with the most salient, \topiclike" entities in the discourse model. Furthermore, the placing of certain adverbials in initial 1 position is frequently employed to signal the beginning of a new subsection (\transaction") of the dialogue. These results indicate that German is to be placed at the discourse-con gurational end of the con gurationality scale. The second part of the thesis examines a speci c aspect of the initial position in greater detail, namely the observed correlation between initial position, null topics and discourse-deictic reference to events and propositions. A study of the referents of null topics in the corpus shows that null topics are mainly discourse-deictic, ie have no NPantecedents but refer to preceding sections of the text. This con icts with the standard assumption that discourse-deictic reference involves topic shift rather than reference to a continuing topic. In addition, a study of discourse-deictic reference in general in the corpus shows that there is a strong preference for establishing it in initial position, regardless of whether this is done by null anaphora, pronouns or demonstratives. The results also show that null topics and demonstratives are used far more frequently for establishing discourse-deictic reference than pronouns. This argues against traditional hierarchies of NP-forms and cognitive status which group null anaphora and pronouns together. A comparative study of discourse deixis in an English corpus of similar sort shows that in this con gurational language the frequency of anaphora forms is in line with the predictions made by the hierarchies. A solution to the problems resulting from these analyses is presented in the nal part. Here, it is proposed that discourse-deictic reference does not necessarily imply topic shift. Instead, reference to events and propositions expressed by the previous utterance can be the default centre of attention. Compatibility with verbal subcategorisation restrictions assures that the anaphor is interpreted correctly as referring to either an NP or a that-clause. A comparison with a non-task-oriented corpus shows that in text-types revolving around the completion of a task, the expected negotiations allow propositions and events referred to by whole utterances to be the centre of attention and reference to them is treated as a continuing topic. In non-task-oriented corpora this is not the case and discourse-deictic reference, whilst being in topic position, is established by demonstratives, indicating topic shift. The general conclusion to be drawn from this is 2 that the non-linguistic context, ie text genre and situation, can in uence the choice of NP-form. Finally, the lack of pronouns for this type of reference is explained by the syntactic restrictions which disallow the occurrence of unstressed object pronouns in initial position in German main clauses. This indicates that given the choice between expressing pragmatic functions by word order or by NP-form, English speakers choose NP-form, whereas German speakers choose word order and adjust the NP-form accordingly { further evidence for the discourse-con gurational nature of German. 3 Declaration This thesis has been composed by myself and it has not been submitted in any previous application for a degree. The work reported within was executed by myself, unless otherwise stated. November 1998 4 Acknowledgements First and foremost I wish to thank my supervisor Ronnie Cann for his advice and support and for the encouragement he gave me throughout the years it took to complete this thesis. I am greatly indebted to him for this. A big thank you also to Caroline Heycock, who took a lot of time to read and correct the drafts, and gave me many useful comments on the work presented here. I am indebted to Massimo Poesio, who despite being under no o cial obligation to do so, took a great deal of time to discuss my work and to read the whole of the nal draft. Ellen Prince and Enric Vallduv 's course on the Discourse-Syntax Interface taught at the LSA Summer Institute 1997, aside from being very enjoyable, also gave me inspiration for much of the work presented here. Any faults in this thesis are, of course, my own. The sta and students of the Department of Linguistics here at Edinburgh have provided me with a friendly and motivating atmosphere in which to complete my work. In particular I'd like to thank Ethel Jack, Bob Ladd, Robin Lickley, Diane Nelson, Louise Kelly, and Theodora Alexopoulou for being good friends and colleagues. Also, a big thanks to my wonderful o cemates Max Louwerse and Dave Patterson { Max, for the discussions about causality which made my own thesis topic seem crystal clear, and Dave, whose incessant rambling about rowing scores made just about every aspect of linguistics appear quite fascinating by comparison. On a personal note, I'd like to thank the following: Paul Taylor for being my best friend throughout and for the inspirational pep talks 5 containing many a four-letter word. Couldn't have done it without you! Celia for being a great pal and for the wonderful month spent in the Alps (involving climbing Mont Blanc on a diet of lukewarm pot noodles and glacier water). Also my other climbing pals Chris, Jacqui and Paul Schweizer. There's nothing like hanging of a cli face to make writing a PhD seem like a doddle. My fantastic atmates Celia, Ant and Julian and our Morningside neighbours Tim and Conor for the much needed tequila nights. Thanks are due also to Sneeky Pete and Jose Cuervo. Finally, I'd like to thank my parents Hartwig and Brigitte Eckert for being helpful and encouraging in every way. I promise from now on to pay attention to the content of what they are saying rather than to the constituents they put in their Vorfeld. This thesis is dedicated to them. 6 You get to know how people put things, the way they use words, just where the stresses fall...I couldn't help thinking that if Rick said 'This is good news,' he probably meant 'This is good news,' contrasting it with something else, that wasn't. Or maybe I was just being paranoid, as usual. Iain Banks, Espedair Street. 7