abstracts from the 4th Workshop on Discourse Structure
Austin, Texas :: March 3-5, 2006
Craige Roberts
‘Only’: Presupposition and Implicature
The meaning of the English adverb only has been the subject of intense debate; in particular, regarding the status of its prejacent. It has been argued that the latter is either a presupposition, a conversational implicature, or an entailment. I argue that it is best characterized as a non-presuppositional conventional implicature. But in several important respects, the prejacent of only is also unlike the conventional implicatures discussed by Potts. The detailed consideration of the meaning of only sheds light on the relationship between all these various types of meanings, contributing to the recent reconsideration in the literature of the nature of presuppositions and implicatures, both conversational and conventional (Abbott, Abusch, Bach, Bezuidenhout, von Fintel, Horn, Potts, Schwenter, Simons).
Patrick Caudal
Aspectual classes of tenses and the semantics/pragmatics interface – a SDRT investigation
The ultimate goal of the research project I'm presenting here is to formulate an SDRT account of the function of tenses as discourse-structuring items. At least since Weinrich (1964), it is almost common wisdom that tenses are a key factor w.r.t. discourse structure. The idea that the aspectual contribution of tenses helps determining the temporal ordering and overall structure of discourse has been steadily gaining ground in the litterature since the early 1990's – a development in which the SDRT framework has played a pioneering role, cf. e.g. Asher & Lascarides (1993).
The hypothesis providing the theoretical backbone of the present paper is the following : tenses possess an illocutionary force which plays a fundamental role at the semantics/pragmatics interface, and which is somehow related to their aspectuo-temporal content. This fact is reflected by their being associated with a particular subset of discourse relations, and even sometimes implicature-like entailments which in turn can trigger certain specific interpretative effects (possibly causing/enabling additional discourse relations to be established).
Capitalizing on a series of papers dedicated to the study of tenses at the semantics/pragmatics interface across a handful of Romance and Germanic languages (i.e., French, English and Allemanic; cf. Caudal & Roussarie (2004, 2005), Caudal & Schaden (2005), Caudal & Vetters (2003, 2005), etc.), I intend to sketch here a tentative classification of the aspectuo-temporal contribution of tenses at the semantics/pragmatics interface, and to propose distinct formal options to account for them within the SDRT framework – in short, I will put forth some sort of formal “toolbox” with cross-linguistic and historical purpose. This proposal is in line with the grammaticalization hypothesis (see e.g., Bybee et al. (2004), or Dahl (2005)) insofar as it draws on regular cross-linguistic and historical patterns about the semantics and pragmatics of tenses. This “toolbox” will comprise a number different mechanisms (operating mostly at the semantics/pragmatics interface) that can be helpful when trying to model the interpretation of tenses in discourse.
Chris Barker
Grammaticized metadiscourse
Traditional formal models of discourse update rigidly separate the informational payload (what a sentence is about) from its linguistic side effects. For instance, a statement such as "A man walked in" describes a certain class of worlds. As a side effect, the use of the indefinite creates a new discourse referent that can serve to anchor subsequent definite reference. Traditionally, side effects are handled through a formal mechanism that is distinct and separate from the information update corresponding to the descriptive content. I argue that the division between content and side effects does not stand up to scrutiny. Building on an idea of Stalnaker, I show how informational update and side effects can be combined within a single update mechanism. Based on two detailed case studies, both involving absolute uses of gradable adjectives, I go on to suggest that some expression types have no update effect apart from their side effects. I conclude with a tentative connection with performatives.
Chiyo Nishida
Expressing Subjectivity: A Case Study on the Spanish Clitic ‘se’and the Japanese Auxiliary Verb ‘shimau’
In this presentation, we look at the Japanese auxiliary verb –shimau, as in 1a and 2a as well as a certain use of the Spanish reflexive clitic se, as in 3a and 4a, as markers of subjectivity (Smith 2003). Each example is contrasted with a sentence without the corresponding morpheme.
Japanese
1a. Taroo ga wain o nondeshimatta.
Taroo nom wine acc drink-TE shimau-past
‘Taroo drank up the wine/ Taroo drank wine/the wine’
b. Taroo ga wain o nonda
Taroo nom wine acc drink-past
‘Taroo drank wine/the wine’
2a. Kago kara tori ga deteshimatta.
cage from bird nom get.out-TE shimau-past
‘The bird/the birds got away from the cage’
b. Kago kara tori ga deta
cage from bird nom get.out-past
‘A bird/birds/the bird/the birds came out of the cage’
Spanish
3a. Tomás SE tomó el vino/*vino.
Tom se-3 drank-3sg the the wine/*wine
‘Tom drank up the wine/*wine’
b. Tomás tomó el vino/vino
Tom drank-3sg the wine/wine
‘Tom drank the wine/wine’
4a. SE salió el pájaro de la jaula.
se-3 got.out-3sg the bird from the cage
‘The bird got away from the cage’
b. El pájaro salió de la jaula.
the bird got.out-3sg from the cage.
‘The bird came out of the cage’
The Japanese auxiliary verb shimau and the Spanish reflexive clitic se, while they retain the primary role assigned to them in their respective languages, i.e., shimau as a lexical verb to mean ‘put.away’ or ‘close.up’ and se as an anaphor, have been grammaticalized to assume multi-functions in a very similar fashion. With respect to shimau, it has been widely accepted that, when it’s used as a part of a compound verb, it marks not only the completive aspect, but also a certain stance that the speaker takes toward the event denoted by its argument predicate (Iwasaki 2003, Kayama 2001, Soga, 2000, among others). The Spanish clitic se has also been said to mark the completive aspect (Nishida 1994, Zagona 1996, Sanz 2000, McCready and Nishida 2006, among others) as well as to convey a certain pragmatic meaning (Maldonado 1999).
What I hope to achieve in this presentation is two-fold: a) To provide a detailed account of the data involved of the behaviors of the two morphemes and to show how they mark subjectivity (or a particular way of conceptualizing an event), and b) to show how the two morphemes differ in their distributions as such markers. The ultimate goal of this presentation, however, is to demonstrate that language tends to mark what we can call “personal (or private)” discourse, as shown in 6a, in contrast to the neutral (or public) counterpart, as shown in 6b, by way of grammatical means, such as shimau or se.
6a. (Personal/Private)
Jpn:
Watashi wa Kinoo ojiichan ga nakunatteSHIMATta.
I top yesterday grandpa nom die TE shimau-past
Spn:
Ayer SE me murió mi abuelito.
yesterday se-3 CL-1sg died-3sg my grandpa ‘Yesterday my grandpa died on me’
b. (Neutral/Public)
Jpn:
Sakujitu jidoosha jiko de gomei no mono ga
yesterday car accident Loc five gen people nom
nakunatta.
die-past
Spn:
Ayer murieron cinco personas en un accidente
yesterday died-3pl five people in an accident
de coche.
of car
‘Yesterday five people died in a car accident’
Carlota Smith
Discourse Modes across languages
I consider the linguistic correlates for semantic and conceptual categories in languages of very different structures. First as background, I introduce the notion of discourse modes, the classes of situation entity, and the relevant linguistic correlates. Here I attempt to extend the analysis of English to Mandarin Chinese, Navajo, and French, languages of different families. I'm interested in whether there is linguistic evidence for Discourse Modes, and for the proposed classes of entities, in these languages. I will concentrate on astract entities (Facts and Propositions) and on the argument mode, which depends on clausal structures that vary dramatically across languages.
Paul Portner
The nature of imperative force in imperative sub-types
I am going to discuss the ways in which imperatives, and various sub-types of imperatives, are interpreted in root and embedded contexts. A key piece of data concerns the “psycho boss”. You’re being sent out of town to a meeting, and your boss says:
(1) a. Be there at least two hours early.
b. Then, have a bite to eat.
By itself, (1a) can be taken as an order, and if your boss is that kind, you might well understand it that way. However, if you do, (1b) is bizarre. You have to take it as an order too, and your boss must be psycho to order you around at that level of detail. Of course it makes sense to interpret (1b) as a suggestion, but then you have to interpret (1a) as a suggestion too. Why can’t (1a) be understood as an order and (1b) as a suggestion? You might think it’s just impossible to switch from being the ordering–boss to being the suggesting–boss so abruptly, but a parallel sequence with overt operators wouldn’t be odd in the same way as (1):
(2) a. You must be there at least two hours early.
b. Then, I suggest you have a bite to eat.
I am going to propose an extension of the model of discourse semantics for imperatives given in Portner (2004) which accounts for these facts, as well as for the interpretation of imperatives embedded under verbs like order in those languages that allow embedded imperatives. The analysis emphasizes the parallels between the semantics of imperatives and the semantics of modals (particularly as in Kratzer (1981)) while avoiding (what I take to be the mistake of ) saying that imperatives contain convert modals.
References
Kratzer, Angelika. 1981. The notional category of modality. In Words, worlds, and contexts , ed. Hans-Jurgen Eikmeyer and Hannes Rieser, 38–74. Berlin: de Gruyter.
Portner, Paul. 2004. The semantics of imperatives within a theory of clause types. In Proceedings of semantics and linguistic theory 14 , ed. Kazuha Watanabe and Robert B. Young. Cornell University Linguistics Department: CLC Publications. URL
http://semanticsarchive.net/Archive/mJlZGQ4N/
Jason Baldridge
Strategies and Opportunities for Computational Discourse Structure Analysis
The automatic determination of discourse structure is a computationally challenging task. For one thing, it is usually assumed that discourse segments are related to one another in a hierarchical fashion, and there are strong arguments that quite general acyclic graphs are representationally necessary in order to adequately capture the rhetorical connections of discourse segments within a text or dialog. This leads to an explosion of alternative potential analyses that is difficult to reign in even with very sophisticated machine learning models. Another challenge is that there are many sources of information that go into the determination of rhetorical relationships -- this information can be difficult to manage and make use of in both symbolic and statistically based systems for determining discourse structure. Finally, computing discourse structure and other discourse-level relationships, such as coreference, are co-dependent tasks -- determining one before the other is likely to yield sub-optimal results, yet determining them jointly is likely to be intractable if done naively.
In this talk, I will discuss a statistical parsing approach to building discourse structures of Segmented Discourse Representation Theory for appointment scheduling dialogs, and highlight some of its adequacies and inadequacies for the task. I will then present a new project to build such discourse structures in conjunction with reference resolution for news-wire texts. In that context, I will discuss how recent developments in machine learning and discourse annotation have the potential to help yield more accurate and representationally adequate discourse structure analyzers.
return to proceedings