--- title: "On the Derivation of a Conversational Maxim — Hofmann 1980" description: "Hofmann asks whether Grice's maxim of relation (\"be relevant\") is a basic rule that speakers follow, or whether it falls out of… 1 reference." authors: ["Th. R. Hofmann"] year: 1980 venue: "COLING 1980 Volume 1: The 8th International Conference on Computational Linguistics" source: "https://aclanthology.org/C80-1034/" 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/hofmann-1980-derivation-conversational-maxim/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/hofmann-1980-derivation-conversational-maxim/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [1980](https://md.research.elamur.ai/free/en/papers/1980/) / On the Derivation of a Conversational Maxim # On the Derivation of a Conversational Maxim **Authors:** Th. R. Hofmann · **Year:** 1980 · **Venue:** [COLING 1980 Volume 1: The 8th International Conference on Computational Linguistics](https://md.research.elamur.ai/free/en/venues/coling/) ([COLING 1980](https://md.research.elamur.ai/free/en/venues/coling/1980/)) · **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/C80-1034/) · [PDF](https://aclanthology.org/C80-1034.pdf) · [OpenAlex](https://openalex.org/W1971208965) **Citations:** 0 citations (OpenAlex, 2026-10-14) ## Summary Hofmann asks whether Grice's maxim of relation ("be relevant") is a basic rule that speakers follow, or whether it falls out of something deeper. He first narrows the maxim to a connectedness principle: each sentence should share semantic content with the context just before it, and abrupt jumps in subject need an explicit signal. He then derives this from PIM (principe d'interprétation minimale), a rule from his earlier thesis. Under PIM, a hearer resolves word senses, constructions and pronouns by picking the reading that adds the least new material to the prior semantic context. If hearers interpret this way, an unmarked topic jump will be misread. Speakers must therefore keep overlap or flag the break. PIM also covers cases the maxim cannot, such as pronoun resolution driven by verb meaning, so the derivation works in one direction only. The paper is theoretical and uses no corpus or experiments. Hofmann concludes that semantic processes are best formalized in the interpretive direction, that generation is trial and error, and that "discourse topic" may be no more than the overlap between successive sentences. ## Key points - The paper reformulates Grice's four maxim groups (quantity, quality, relation, manner). It reduces relation to one claim: rapid shifts of subject must be overtly marked, while gradual drift is harmless. - The paper sorts topic-change markers into two kinds: - Temporary shifters ("by the way", "incidentally") mark a segment that need not overlap and can be dropped from later context. - Complete shifters ("speaking of …", "to change the topic") tell the hearer to seek no overlap at all. - PIM disambiguation examples: - "flying planes can be dangerous" is disambiguated by the preceding context. - "it is light" means weight or brightness depending on the prior question. - In "she misunderstood it", "she" resolves to the hearer of a telling event, because "misunderstand" overlaps with "tell". Changing the verb (come, say softly) shifts the referent. - The other Gricean maxims get other explanations: - "Be as informative as required" is traced to the social rituals behind performatives such as questions. A curt answer like "Out" still completes the ritual. - "Don't be over-informative" is better captured by McCawley's least-effort principle. - Hofmann claims fine generation (high rhetoric) is reached by only about 5% of people. Interpretation, by contrast, is fast and deterministic for good texts, which is why he argues it is easier to formalize. ## Interesting ideas - Resolve ambiguity by choosing the reading that maximally overlaps with the already-built semantic context, i.e. adds the least new information; one explicit criterion covers word senses, attachments and pronouns. - A pronoun antecedent can be chosen by semantic links between verbs (misunderstand ↔ tell → the addressee of the telling) rather than by recency or subjecthood. - Discourse markers act as instructions to the interpreter: "by the way" opens a disposable side segment, and "speaking of" resets the overlap search. - Generation can be done as compose, interpret in context, then revise, using the interpreter as the judge of what a sentence will mean. - Text quality is defined operationally: good text leaves no unresolved ambiguity under PIM, and poorly written text is where PIM fails. ## For Mova How these ideas could serve Mova, our small open language model — its files, stages and tests: - **I1** (frames): coreference and sense selection in prag could score candidates by how many semantic components they share with the existing frames, which gives an explainable reason for each choice. - **I2** (world): store lexical decompositions such as "tell = cause to understand by speaking" in the verb-class knowledge so the world builder can bind pronouns through shared meaning components. - **I3** (frames): treat these markers as context-control speech acts that open a temporary sub-context or start a fresh one. This would stop spurious coreference links across the boundary. - **I4** (back): run each generated sentence back through Mova's own text → graph reader. Accept it only if its minimal-overlap reading matches the intended graph, which gives the compressor a built-in round-trip check. - **I5** (learn): use sentences where minimal-overlap scoring finds no unique winner as an evaluation flag or negative control. Disagreement there reflects genuinely underdetermined input rather than parser error. *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 Also on the byline: [Shimane University](https://ror.org/01jaaym28). 📚 Shelves (3): [🎭 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 · [🤔 Philosophy](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/philosophy/) · 287 works 🧭 Topics (2): [🗨️ Dialogue systems](https://md.research.elamur.ai/free/en/topics/dialogue-systems/) · 654 works · [😉 Implicature](https://md.research.elamur.ai/free/en/topics/implicature/) · 138 works ## Cite Cite as: Th. R. Hofmann. 1980. On the Derivation of a Conversational Maxim. COLING 1980 Volume 1: The 8th International Conference on Computational Linguistics. The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{hofmann1980derivation, title = {{On the Derivation of a Conversational Maxim}}, author = {Th. R. Hofmann}, booktitle = {COLING 1980 Volume 1: The 8th International Conference on Computational Linguistics}, year = {1980}, doi = {10.3115/990174.990215}, url = {https://doi.org/10.3115/990174.990215} } ``` - [← previous On FROFF: A Text Processing System for English Texts and Figures](https://md.research.elamur.ai/free/en/papers/abe-1980-froff-text-processing-system-english/) - [↑ back to papers of 1980](https://md.research.elamur.ai/free/en/papers/1980/) ## References 1 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. i. Grice, H.P. 'Logic & conversation' (1975) in Cole & Morgan (Ed) Syntax & Semantics 3: S~eech Acts, Academic Press (N.Y.). & (1967) part of William James Lectures. 2. Hofmann, Th.R. Description s~mantique & dynamique du discours (1978)th~se pour la Sorbonne (Paris)j & G.E.T.A (Grenoble). 3. --- 'What some semantic theories can't do' (1979) C.L.S. Parasession on the elements, Univ Chicago (Chicago). 4. --- 'How to ignore pragmatics & get away with it, almost' (1980) Canadian Semiotics Assn --~239-- [find](https://scholar.google.com/scholar?q=i%2E+Grice%2C+H%2EP%2E+%27Logic+%26+conversation%27+%281975%29+in+Cole+%26+Morgan+%28Ed%29+Syntax+%26+Semantics+3%3A+S%7Eeech+Acts%2C+Academic+Press+%28N%2EY%2E%29%2E+%26+%281967%29+part+of+William+James+Lectures%2E+2%2E+Hofmann%2C+Th%2ER%2E+Description+s%7Ema) *Updated 29 October 2026*