--- title: "An ascription-based approach to Speech Acts — Lee 1996" description: "The authors build on ViewGen, a Quintus Prolog program by Ballim and Wilks that represents an agent's view of other… Cited by 1 paper here." authors: ["Mark Lee", "Yorick Wilks"] year: 1996 venue: "COLING 1996 Volume 2: The 16th International Conference on Computational Linguistics" source: "https://aclanthology.org/C96-2118/" licence: "CC BY-NC-SA 3.0 (ACL Anthology, before 2016)" licence_url: "https://creativecommons.org/licenses/by-nc-sa/3.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/lee-1996-ascription-based-approach-speech-acts/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/lee-1996-ascription-based-approach-speech-acts/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [1996](https://md.research.elamur.ai/free/en/papers/1996/) / An ascription-based approach to Speech Acts # An ascription-based approach to Speech Acts **Authors:** [Mark Lee](https://md.research.elamur.ai/free/en/authors/mark-lee/) and [Yorick Wilks](https://md.research.elamur.ai/free/en/authors/yorick-wilks/) · **Year:** 1996 · **Venue:** [COLING 1996 Volume 2: The 16th International Conference on Computational Linguistics](https://md.research.elamur.ai/free/en/venues/coling/) ([COLING 1996](https://md.research.elamur.ai/free/en/venues/coling/1996/)) · **Licence:** [CC BY-NC-SA 3.0 (ACL Anthology, before 2016)](https://creativecommons.org/licenses/by-nc-sa/3.0/) **Links:** [Publisher's page](https://aclanthology.org/C96-2118/) · [PDF](https://aclanthology.org/C96-2118.pdf) · [OpenAlex](https://openalex.org/W2023997760) **Citations:** 15 citations (OpenAlex, 2026-10-14) · [1](https://md.research.elamur.ai/free/en/papers/lee-1996-ascription-based-approach-speech-acts/cited-by/) in our library, of the 149,439 papers whose bibliographies we read ## Summary The paper argues that speech act processing in dialogue has relied on weak belief models. Earlier systems assumed fixed, precomputed stacks of nested beliefs, including "mutual belief", which in principle regresses forever. The authors build on ViewGen, a Quintus Prolog program by Ballim and Wilks that represents an agent's view of other agents' beliefs as nested environments. They extend it from beliefs to goals and intentions, with intentions stored as linked planning actions. Instead of mutual belief, ViewGen ascribes attitudes to others only when the dialogue needs them. It uses two rules: default ascription for ordinary beliefs about the world, and stereotypical ascription for class-typical knowledge and goals. These rules, plus a rule for accepting a trusted speaker's claim, are themselves written as planning operators ("mental acts"). This lets speech acts such as inform carry only minimal, solipsistic effects. Deeper nesting is built on demand, for example for non-conventional language. The result is a simpler speech act theory that keeps speaker and hearer belief sets separate. ## Key points - The work extends ViewGen (Ballim & Wilks 1991), implemented in Quintus Prolog, to model goals and intentions as well as beliefs. - A corpus study (Lee 1994) found that no dialogue exchange needed more than two levels of belief nesting. Wrong assumptions about common ground were fixed quickly by corrections and repetitions. Taylor and Carletta (1994) reported similar findings on the HCRC Map Task corpus. - Default ascription copies a System belief into another agent's environment unless there is contrary evidence. It applies to beliefs only, not to goals or intentions. - Stereotypical ascription assigns a class's typical attitudes to agents of that class. The paper's example ascribes doctor-style medical knowledge to John. - The rules are written as plan operators with preconditions and effects, including an Accept_belief operator. It requires that the hearer holds no contrary belief and believes the speaker is trustworthy. Planned future work: a robust dialogue parser and a link to the LaSie IE platform using a small medical corpus. ## Interesting ideas - Build nested "A thinks B thinks…" structures lazily, only when an utterance's interpretation actually needs them, rather than precomputing a fixed depth. - Treat belief ascription itself as an action with preconditions and effects, so belief updates are traceable steps like any other plan step. - Distinguish defaultable attitudes (world beliefs, assumed shared unless blocked) from non-defaultable ones (goals, intentions), which instead need stereotypes or plan recognition. - Gate the acceptance of reported information on the absence of a contrary belief plus trust in the source, so the hearer's beliefs and what was merely said stay apart. - Use stereotypes (role classes such as doctor) as bundles of typical knowledge and goals attached through an isa link, overridable by explicit evidence. ## For Mova How these ideas could serve Mova, our small open language model — its files, stages and tests: - **I1** (frames): The speech-act layer in prag could keep a flat speaker/hearer view by default and nest views only when irony, indirect questions or deception require it. - **I2** (frames): Every inferred belief attribution in prag could carry its rule name and the preconditions that licensed it, which keeps pragmatics explainable. - **I3** (world): Entity models in the world of a text could inherit general facts freely, but assign goals only from role stereotypes or observed actions, avoiding invented motives. - **I4** (answer): Fable and tale QA readers could separate "the fox said X" from "X is true in the story", answering correctly about deceitful characters. - **I5** (world): Role-based stereotype bundles derived from WordNet/FrameNet could fill in default knowledge for characters, with explicit text statements overriding them. *Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.* ## Institutions - [University of Sheffield](https://md.research.elamur.ai/free/en/universities/university-of-sheffield/) — University, Sheffield, United Kingdom · 117 works · 22,645 citations 📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works 🧭 Topics (1): [💬 Speech act theory](https://md.research.elamur.ai/free/en/topics/speech-act-theory/) · 229 works ## Cite Cite as: Mark Lee and Yorick Wilks. 1996. An ascription-based approach to Speech Acts. COLING 1996 Volume 2: The 16th International Conference on Computational Linguistics. The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{lee1996ascription, title = {{An ascription-based approach to Speech Acts}}, author = {Mark Lee and Yorick Wilks}, booktitle = {COLING 1996 Volume 2: The 16th International Conference on Computational Linguistics}, year = {1996}, doi = {10.3115/993268.993290}, url = {https://doi.org/10.3115/993268.993290} } ``` - [← previous GRICE INCORPORATED: Cooperativity in Spoken Dialogue](https://md.research.elamur.ai/free/en/papers/dybkjaer-1996-grice-incorporated-cooperativity-spoken-dialogue/) - [next → Markov random field based English Part-Of-Speech tagging system](https://md.research.elamur.ai/free/en/papers/jung-1996-markov-random-field-based-english/) - [↑ back to papers of 1996](https://md.research.elamur.ai/free/en/papers/1996/) *Updated 29 October 2026*