This book presents a new theory of grammatical categories - the Universal Spine Hypothesis - and reinforces generative notions of Universal Grammar while accommodating insights from linguistic typology.
Date: Fri, 11 Apr 2003 12:42:11 +0200 From: Thora Tenbrink <tenbrink@Informatik.Uni-Bremen.DE> Subject: Semantic and Pragmatic Issues in Discourse and Dialogue
Bras, Myriam, and Laure Vieu (2001) Semantic and Pragmatic Issues in Discourse and Dialogue: Experimenting With Current Dynamic Theories. Elsevier Science Ltd, x+250pp, hardback ISBN 0-08-043943-8, $86.50.
Thora Tenbrink, University of Bremen, Germany
This book contains a collection of eight carefully revised and extended papers, most of which were originally presented at a workshop on "Theoretical Bases for Semantics and Pragmatics in NLP: The Expression of Time, Space and Movement in Lexicon, Discourse and Dialogue"; 6th TALN conference, Corsica, July 1999.
In the first paper ("She's Character"), Paul Dekker deals with the semantics of pronouns. In previous accounts in logic and formal semantics, pronouns were treated like variables: functional on the context, and void of content. Dekker points out that pronouns carry two kinds of indexical presuppositions that need to be taken into account when specifying their character: that of utterance (a pronoun has no 'character' independent of its actually being employed) and that of the presence of an individual that it points to (such as, in the case of 'she', a female). Both of these presuppositions characteristic of pronouns are captured in a formal specification. Highlighting how such presupposed presence is dependent on a speaker's intentions and on the context of utterance, Dekker now introduces the notion of 'presence in intentional space'. Many intuitive examples are introduced to illustrate this concept: for instance, it is possible to refer to a non-present (and discourse-new) woman as 'she' if both interlocutors at this moment see her husband enter the bar. In other words, for a pronoun to be employed in a discourse context it is sufficient that the interlocutor can identify the intended referent on the basis of what is said about the referent (as in 'She is very sick' if the husband looks downcast), of the interlocutors' background knowledge, etc. Note that this concept of 'presence in intentional space', which Dekker proposes to capture formally in a many-sorted modal logic, embraces both anaphoric and demonstrative uses of pronouns, and it applies independently of epistemic complications.
The second paper, "Exhaustivity and Specificity: A Parallelism between Answers and Pronouns" by Robert van Rooy, also deals with pronouns; more precisely, it addresses the distinction between their referential vs. descriptive uses. This distinction, re-analysed as differentiating between exhaustivity and specificity, is shown to reflect a more general phenomenon than previously assumed: very similar effects can be detected in the uses of wh-questions, in the classical distinction between referentially and attributively used definite descriptions, and in reference to specific events vs. event-types. This parallelism is extended to account for functional dependencies, and worked out in detail in a dynamic logic framework that integrates choice functions in addition to worlds and assignment functions.
In the third paper, "Presupposition Computation and Presupposition Justification: One Aspect of the Interpretation of Multi-Sentence Discourse", Hans Kamp presents an in-depth DRT (Discourse Representation Theory) analysis of the following mini-discourse: "I gave the workers a generous tip. One thanked me. The other one left without saying a word." The analysis highlights the specific mechanisms of presupposition justification, including the specific contributions of expressions like "other" or "one", that enable the reader to infer (after processing the whole mini-discourse) that the number of workers in the first sentence must be two. This procedure gives insight into the more general processes underlying the interpretation of discourse in those cases where presuppositions are not justifiable directly by the previous context, while at the same time offering a comprehensible introduction to the formalisms adopted in DRT (starting with the syntax-semantics interface). On the basis of his analysis, Kamp questions the dichotomy in the literature between presuppositions that either do or do not permit the accommodation of discourse referents, and points to the need for a finer-grained classification of presupposition types. He concludes by considering the complexities involved in accounting for the ambiguities arising in natural discourse in a working implementation, a task that needs to be approached using Underspecified DRT.
The next paper, "Presupposition Triggered by Temporal Connectives" by Frank Schilder, deals with the presuppositional effects associated not with a specific discourse as in the previous paper, but with specific expressions, namely, 'before' and 'after'. By analysing longer stretches of discourse in which these connectives occur, using the DRT framework, Schilder works out the regularities of presuppositions to be found both within and beyond the sentence, in contrast to previous accounts that concentrated on the temporal semantics of the connectives and on the relations between main and subordinate clause within a sentence. Concentrating on sentences reflecting the natural order of events, he points to several distinct kinds of sentential relations and discourse linking relations established by 'before' and 'after', such as a 'termination relation' (established as a discourse linking relation by 'after', but as a sentential relation by 'before'). The analysis also shows that a causal relation cannot be established by 'before'. Moreover, while 'after' can establish a causal relation between the two events described within the sentence, no causality can be inferred between a preposed after-clause and the preceding discourse. The refined discourse semantics of the two connectives is formally specified within DRT, incorporating and extending results taken from Asher and Lascarides (1998).
In the following paper, "French Adverbial Puis between Temporal Structure and Discourse Structure", Myriam Bras, Anne Le Draoulec, and Laure Vieu also present a detailed analysis of a specific expression ('puis') in its relation to the wider discourse context. The authors make extensive use of natural language examples drawn from a corpus of French literature (most of which they do not translate, which may hamper comprehension for English speaking readers). Their central finding, which they work out in detail in the framework of SDRT (Segmented DRT), is that 'puis' is a marker of the discourse relation Narration, instead of directly contributing to the semantics of the clause. This implies that 'puis' does not by itself introduce a temporal referent or relation, but that the relation of temporal succession is implied by the underlying discourse relation that is marked by the adverb. This proposal, which is well justified by the examples presented, mirrors findings on other discourse markers (in French: donc as a marker of Result, mais as one of Contrast) and opens up research questions with regard to related expressions in other languages, such as "then" in English and "dann" in German.
Ana Teresa Alves and Isabel Gómez Txurruka analyse "The Meaning of Same in Anaphoric Temporal Adverbials". Their focus is on those occurrences of 'same' in which a discourse relation is canceled that would be inferred in the absence of 'same'. The authors introduce the concept of Unexpected Identity (UI) to account for the discourse effects triggered by 'same', and formalise the refined semantics of 'same' in the framework of SDRT. UI indicates that a particular identity of an entity with a previously mentioned one is not logically implied by the discourse. This constraint is shown to capture most of the interaction effects between 'same' and diverse kinds of discourse relations that are put to the test.
The next contribution, "Spatial Inferences in a Localization Dialogue" by Peter Krause, Uwe Reyle, and Michael Schiehlen, highlights the many instances of inferences and presuppositions that occur in a natural dialogue dealing with spatial surroundings. The authors present a fairly thorough DRT formalisation of a localisation dialogue consisting of 10 turns, which is a formidable task in itself. Central to their analysis are the mereotopological relations originally proposed by Asher and Sablayrolles, 1995, which are extended. Furthermore, the authors present axioms that build the basis for accounting for the inferences associated with the interpretation of the lexical items in the dialogue, and they formally specify the most important lexical entries. Having accomplished this basis for analysis, they turn to a step-by-step DRT analysis of the dialogue, specifying in detail the implicit presuppositions and their justification. This analysis exemplifies the complexity of such an approach, leaving many issues and details open for further clarification, such as the decision about which information to include in the Common Ground before the dialogue has started. Another problem is posed by the fact that not all presuppositions that need to be accommodated, and are then included in the formalisation of the 'Common Ground', are indeed understood and/or accepted by the interlocutor, which makes the information one-sided (rather than, in the original sense, Common Ground). This differentiation is not clearly represented in the formalisation. Altogether, the integration of the diverse kinds of formal and cognitive problems that are merged in a localisation dialogue constitute a major challenge for research. The present article is certainly a good start in this direction.
The last paper, "Cooperativity in Dialogue" by Nicholas Asher, Joan Busquets, and Anne Le Draoulec, uses the SDRT framework to analyse another localisation dialogue consisting of 23 turns. To begin with, the main features of SDRT are introduced, and its relationship to DRT is outlined. Then, the more recent extensions to the specific requirements in dialogue are described, and the authors discuss the relationship of the discourse relations to cognitive modeling. Their main focus of analysis of the present dialogue is the question how the Gricean principle of Cooperativity is manifested in the speakers' contributions, also in cases of disagreement or indirect answers. Thus, discourse relations are analysed in terms of speakers' goals. One result is that there must be a further constraint pulling in a different direction than Cooperativity: a need to 'save face', i.e., to explain own actions to justify them. Many utterances cannot be interpreted taking Cooperativity as the only driving force in dialogue. Furthermore, the authors deal in some detail with the effects of the discourse relation Correction, suggest some changes to the SDRT framework, and open up issues for further research.
The book is carefully edited and the papers are sorted in a reasonable fashion, starting with traditional theoretical approaches, leading over to analyses of single lexemes occurring in natural language samples, and ending with articles that provide a good insight into the current status of understanding the intricate phenomena contained in longer stretches of dialogue. Some errata remain, e.g., a doubling of "the" on p11 as well as p236, and some confusions of letters ("ellispses" on p.132, etc.). But readability is enhanced by the clear and concise layout plus editorial decisions such as using footnotes rather than endnotes.
Although the book comprises only eight contributions, it still covers a wide range of issues related to the issue of formally capturing the interface between semantics and pragmatics: some contributions deal with monologue, others with dialogue; some are fairly readable and accessible to newcomers to the field or readers more familiar with less formal (but related) research directions, others are strictly formal and presuppose extensive previous knowledge of the approach adopted in the paper; some deal with one single lexical item and its occurrence in diverse kinds of discourse, while others concentrate on the diverse discourse phenomena occurring in one stretch of discourse, such as a localisation dialogue. Readers are presented with spatial and temporal phenomena as well as issues concerning presupposition and anaphora, discourse relations, and their treatment in different dynamic semantic frameworks such as DPL (Dynamic Predicate Logic), DRT and SDRT. All these issues reflect current moves towards integrating pragmatic phenomena into formal semantic accounts of natural discourse, as they are researched in diverse (almost entirely European) projects (reflected in the range of countries represented by the authors in this volume: France, the Netherlands, Germany, Spain, Portugal, and the US). Here, also, the contributions vary between innovative approaches, indicating many open issues for further research, and in-depth analyses of specific phenomena building on previous work.
Some of the analyses (specifically Bras et al., Krause et al., and Asher et al.) are based on natural language data, which can be considered as a special virtue. In Spenader (2002) some of the complications of such an approach are exemplified: in naturally occurring language relations and categories tend to be much less clear- cut than in made-up examples, which are invented precisely for the reason of exemplifying a concept. Therefore, while the respective contributions generally succeed in outlining the relevant contrasts, they also leave space for further analysis, often indicated by hints at intuitions of the authors that could be worth more precise consideration.
Concerning the range of issues covered, the book offers a good overview of the state of the art with regard to dynamic theories. However, concerning the potential audience some cautious remarks are expedient. Since some of the articles provide accessible introductions to all relevant concepts employed, readers from other fields, especially those concerned with pragmatic issues in discourse, will benefit from the analyses. Other contributions, however, have obviously not been written with a broader audience in mind. One particular example is the contribution by Krause et al., which - because of its relation to spatial phenomena and its treatment of "common ground" - has a potential readership within the communities of cognitive science as well as conversation analysis (or other areas dealing with the analysis of dialogue). However, since relevant concepts are presupposed in this article and innovative trains of thought are mostly not covered in detail other than by presenting formulae, much potentially interesting material will only be accessible to readers who are completely at ease with DRT. In contrast, Kamp manages to provide a brief (but effectual) introduction to his approach, while at the same time extending it with new and challenging issues not covered before. His paper is a fine exemplification of how the diverse elements of (even a mini-)discourse interact to create meaning, indicating just why it is so difficult to grasp the subtleties of the semantics and pragmatics of single terms like 'same', 'before', etc., covered in other papers in the book (not to mention multi-sentential discourse). Thus, Kamp's contribution provides useful background knowledge to some of the other articles - a fact which is not transparent beforehand since the two preceding papers in the book deal with entirely different (fairly theoretical) issues.
While, in the case of DRT, the contribution containing background knowledge for the understanding of DRT is placed before other articles in the book building on such knowledge, the case of SDRT is not so lucky. The first article in the book dealing with SDRT (Bras et al.) is fairly self-contained in that only those discourse relations are explained which are relevant to the paper (here, the difference between the concepts Occasion and Topic Contingency should be made explicit, since the intuition of 'belonging to the same story' is used to explain both (p119f.)). The next article (Alves & Gómez Txurruka) refers to a considerable variety of discourse relations most of which are neither explained in intuitive terms nor defined in any detail (e.g., Generalization on p155), although a more general introduction to the SDRT framework is provided (but only later in the text - reflecting a peculiarity with regard to the text structure of this particular contribution, which also provides the relation to previous approaches only at the end, rather than as an introduction, as could be expected. A further peculiarity is that they refer to chapter sections by using the paragraph symbol §). Moreover, although the authors state that this phenomenon is crucial to the understanding of the effects triggered by 'same', it is left open why in some cases the discourse relation Result seems to entail temporal abutment (p160), while in others it does not (p164). - Taken together, the articles dealing with SDRT provide a relatively broad picture of the applications of SDRT, even though every single contribution only represents a portion of it.
Naturally, however, the major part of the expected audience will be interested in the individual publications precisely because of their previous knowledge of the respective formal approaches. Therefore it is worth mentioning that, throughout, the authors have managed to work out convincingly their particular contributions to the research in each addressed field based on the previous state of the art, regardless of how intensively such previous work is presented in the respective article as relevant background knowledge.
The book is highly recommended to researchers dealing with any of the fields outlined above. Many central aspects of the discourse phenomenon of presupposition, for instance, are addressed from various directions that need to be taken into account in further research independent of the approach taken. Further major insights condensed from several contributions (and equally transcending theoretical stance) are the facts that temporal and spatial aspects are intricately entwined with other discourse issues in more respects than previously assumed, and that speakers' intentions play a major role that needs to be accounted for in the analysis. Finally, researchers working on the further development of both DRT and SDRT will benefit from the great diversity of insights (concerning solutions as well as open questions) indicated in the majority of contributions in this book.
Asher, N. and Sablayrolles, P. 1995. A Typology and Discourse Semantics for Motion Verbs and Spatial PP in French. Journal of Semantics, 12:163-209.
Asher, N. and Lascarides, A. 1998. Bridging. Journal of Semantics, 15(1), 83-113.
Spenader, Jennifer. 2002. Presupposed Propositions in a Corpus of Dialogue. In: van Deemter, Kees and Rodger Kibble. 2002. Information Sharing: Reference and Presupposition in Language Generation and Interpretation. Stanford: CSLI Publications.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Thora Tenbrink is a research assistant in the newly established DFG Collaborative Research Center SFB/TR8 "Spatial Cognition: Reasoning, Action, Interaction" (Bremen & Freiburg, Germany). Her research interests focus on the fields of discourse analysis and text linguistics; previous work has dealt with discourse relations and information structure, presuppositions and non-temporal implications of temporal terms, especially 'before', 'after', and 'then'. Her dissertation project deals with discoursal applicability conditions and features of spatial (and temporal where applicable) expressions in human-robot interaction.