Workshop “Contrasting Underspecification and Overspecification of Discourse relations”

Anton Benz (SiGames, ZAS), Mingya Liu (SPOCC, Osnabrück), Torgrim Solstad (ZAS Berlin)

Time and venue:
two days in May 2019 at ZAS in Berlin

This workshop is a follow-up of the “Implicit and explicit marking of discourse relations” on May 24-25, 2018 at Osnabrück University with a related, but different focus.
Most of the literature and tools on discourse relations (DRs) focus on information-exchanging situations, where speaker and hearer share communicative goals. Therefore, the identification of a DR in the absence of explicit marking presupposes that there is one DR intended by the cooperative speaker. One possible reason for the lack of marking is to obey the economy principle, e.g. when discourse context makes it clear which DR is meant. However, the assumption of a specific, implicit DR is questionable. For example, when two events follow each other temporally, speakers may still be uncertain whether one is the cause for the other, and thus, opt for not using because to avoid over-commitment. Another point in case would be strategic situations in which speakers would choose to remain vague by avoiding the explicit marking of DRs or by using markers that are notoriously underspecified, such as and. This is important for DR annotations since the information of the speaker’s epistemic state is hard to detect from written texts. On the other hand, DRs may also be multiply marked (see Das & Taboada 2017). Such overspecification also challenges the economy principle, and the rationality behind it needs further investigation. The workshop aims at identifying factors that contribute to the decision of violating economy principles in the marking of DRs, in particular over- vs. underspecification. We will focus on various kinds of DRs that allow both explicit and implicit marking such as conditionals and causals. We particularly encourage contributions that take a production perspective in cooperative vs. strategic situations to shed light on the gaps and overlaps between production and comprehension. Below are listed a few dimensions which may prove important for the decision to underspecify, specify, or over-specify a DR:

  • Linguistic complexity of DRs
  • Predictability of a DR from context
  • Cognitive costs/resources for inferring a DR
  • Pragmatic constraints (e.g. avoidance of ambiguity or avoidance of falsity)
  • Strategic communication: the strategic speaker may opt for implicit DR marking to foster e.g. plausible deniability
  • Availability of fast and automatic mechanisms generating DR predictions
  • Interlocutors’ familiarity with the current (local) discourse topic

Invited speakers: t.b.a