Intentions, Information, and Inference: Two Rhetorical Questions
What it is about
A short position paper on discourse structure. Moore and Pollack argued that hearers can infer intentional relations from informational ones and back. Oberlander takes that inference model for granted and asks two questions that came up while modelling it in SDRT and DICE. First: when is an intentional relation not an intentional relation? His answer is that a pair of segments can carry two intentional relations at once (for example Narration and Result between two sentences about Betjeman and a best-seller). Second: when is an informational relation not an informational one? His answer is that "supports" is not a fixed domain relation: whether A supports B or B supports A depends on each speaker's network of prior beliefs, so both parties can be right. No quantitative data are reported; the argument is by worked examples.
Ideas
- I1. 36. Discourse relations and coherence: allow a pair of discourse segments to carry a set of relations, not one. In the world's discourse store (
world/src/narr.rs), keep a BTreeSet of labels per pair and measure how often the school texts need more than one label. - I2. Mark "supports"-type relations as belief-relative in
prag/src/infer.rs: when the direction of an evidential link depends on what the world already holds, record the dependency instead of fixing a direction. Measure the share of such links in a held-out set of school lessons. - I3. A small negative control: swap the two segments of a Result pair and check that the system reports the flip only for belief-relative relations, not for causes.
For Mova
The chapter is 36. Discourse relations and coherence, with a link to 33. Implicature and presupposition for the inference step. The code is world/src/narr.rs (narrative relations) and prag/src/infer.rs. The measure: count segment pairs with more than one relation, and the number of relations whose direction changes when the world's prior facts change; run twice to check determinism. The paper gives no numbers, so every threshold must be set on Mova's own data.
Cite
@inproceedings{oberlander1993intentions,
title = {{Intentions, Information, and Inference: Two Rhetorical Questions}},
author = {Oberlander, Jon},
booktitle = {Intentionality and Structure in Discourse Relations},
year = {1993},
url = {https://aclanthology.org/W93-0226/}
}References
- A s h e r , N. [1993] Reference to A b.stract Objects in Engli.~h: A Philosophi,'al ,¢h'manties for Natural Language Metaphysics. Dordrecht: Kluwer Aca.demic Publishers. find
- A s h e r , N. a n d M o r r e a u , M. [1991] ConmLon Sense Entaihnent: A Modal Theory of Non-monotonic Reasoning. In P~w:eedings of the 12th lnternatioual Joint C'onJ~.:rence on Artificial Intelligence. find
- Grosz, B. a n d S i d n e r , C. [1986] Attention, Intentions, and the Structure of Discourse. Computational Linguistics, 12, 175-2()4. find
- K n o t t , A. a n d Dale, It. [1992] Using Linguistic Phenomena to Motiva.te a Set of Rhetorical Relations. HCRC/ltP-39, Huma.n C, ommunication Research Centre, December, 1992. find
- Lascarides, A. a n d A s h e r , N. [1991] Discourse Relations and Defea.sible Knowledge. In P1o-ceedings of the 29th Annual Meeting of A.~.sociatio~ for Comp'utatio'ual Linguistics, 1)p55--63. find
- Lasearides, A. a n d O b e r l a n d e r , J. [1993] Temporal Connectiw,s in a Discourse Context. In Proceedings of the 6th Confe'rcncc of the E'm'opcau Chapter of the A.~.sociulion for C'omlmlational Linguistics, Utrecht, Netherlands, April 2 lst--23rd, 1993, pl)2(i0-2(i,~. find
- M o o r e , J. a n d Pollack, M. [1992] A Problem tor RST: The Need ['or Multi-Level Disc(mrse Analysis. Comlmtational Linguistics , 18,537-544. find
- O b e r l a n d e r , J. [1993] Beliefs a.nd Intentions iu the Abductive Gellel'~l.tion of Discourse. To appear in Proceedings of the Thild h~te'rnational Colloquium o'n Cognitive .%icuce, DonostbL-Sa,n Sebastian, May 4th-8th, 1993. find
This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/oberlander-1993-intentions-information-inference-two-rhetorical/