---
title: "Utilization of Multiple Language Resources for Robust"
description: "The paper targets the automatic labelling of every predicator (verbs and nominals) in real English prose… 9 references, cited by 1 paper here."
authors: ["Alexis Palmer", "Jonas Kuhn", "Carlota Smith"]
year: 2004
venue: "Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC'04)"
source: "https://aclanthology.org/L04-1232/"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/palmer-2004-utilization-multiple-language-resources-robust/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/palmer-2004-utilization-multiple-language-resources-robust/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2004](https://md.research.elamur.ai/free/en/papers/2004/page/2/) / Utilization of Multiple Language Resources for Robust Grammar-Based Tense and Aspect Classification

# Utilization of Multiple Language Resources for Robust Grammar-Based Tense and Aspect Classification

**Authors:** [Alexis Palmer](https://md.research.elamur.ai/free/en/authors/alexis-palmer/), [Jonas Kuhn](https://md.research.elamur.ai/free/en/authors/jonas-kuhn/) and Carlota Smith · **Year:** 2004 · **Venue:** [Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC'04)](https://md.research.elamur.ai/free/en/venues/lrec/) ([LREC 2004](https://md.research.elamur.ai/free/en/venues/lrec/2004/)) · **Publisher:** [European Language Resources Association (ELRA)](https://md.research.elamur.ai/free/en/publishers/european-language-resources-association-elra/) · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/L04-1232/) · [PDF](https://aclanthology.org/L04-1232.pdf) · [OpenAlex](https://openalex.org/W2402514949)

**Citations:** 2 citations (OpenAlex, 2026-10-14) · 1 in our library, of the 149,439 papers whose bibliographies we read

## Summary
The paper targets the automatic labelling of every predicator (verbs and nominals) in real English prose with a "situation entity" type, following Smith's theory of Modes of Discourse. Three coarse classes are used: eventualities (specific events and states), generalizing statives (generics, habituals) and abstract entities (facts, propositions). Texts are parsed with PARC's broad-coverage LFG grammar under XLE, with statistical disambiguation and a robustness fallback, so every sentence yields an f-structure. These structures are converted to flat Prolog terms. They are then enriched with lexical features: aspectual classes mined from Dorr's LCS verb database, plus lists of factive and propositional predicates. Ranked linguistic tests, written as term-rewriting transfer rules originally built for machine translation, assign the labels. A version using only syntax was compared with a lexically enriched one on three National Geographic texts annotated by humans. Adding lexical knowledge raised recall but reduced precision, because the richer rules tended to overgenerate. The work matters as an early, inspectable pipeline that links deep parsing to discourse semantics.

## Key points
- The core resources are the ParGram English LFG grammar, the XLE parser, the stochastic disambiguation of Riezler et al. (2002) and PARC's transfer system (Frank, 1999).
- Pattern searches over Dorr's LCS database (4269 verbs, 9806 readings) yielded the following verb classes:
  - 219 verbs purely stative and 310 ambiguous between state and event;
  - 1399 verbs purely telic and 1030 ambiguous between telic and atelic.
- Seventeen linguistic tests were ranked by how reliably each signals a class, and a stronger test overrides a weaker one when they conflict. One example is a bare plural subject signalling a generic.
- The evaluation set was three National Geographic excerpts of roughly 800–1000 words each, with 52, 85 and 59 gold situation entities.
- Comparing "nolex" with "withlex", the lexical augmentation improved recall but lowered precision. The authors present all results as preliminary.

## Interesting ideas
- Tests run in order of strength, and a "not-yet-marked" flag is consumed once a predicate is labelled, so weaker heuristics only fire where stronger ones stayed silent. This gives deterministic, explainable conflict resolution.
- The parse is flattened into an unordered set of local facts, so rules can match features by their neighbourhood rather than by exact tree paths. This makes rules robust to structural variation.
- Lexical knowledge is compiled once from external databases into reusable features (e.g., stative, telic, factive) that are attached to predicates only when they appear. New resources can be plugged in without touching the rules.
- Composite clues are built in stages: one rule marks an intermediate property (a bare plural NP), and a later rule combines it with a grammatical role (subject) to decide genericity.
- The ablation runs the same rule set with and without the lexical resource and reports precision and recall separately. This revealed that the extra knowledge raises coverage while causing overgeneration.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (frames): Prag's frame and speech-act rules could carry an explicit strength rank and a consumable "unresolved" marker, so every label traces back to the single rule that decided it.
- **I2** (tree): UD trees could be exposed to downstream rules as flat triples (head, relation, features), so that rules tolerate dialect or attachment differences between parsers.
- **I3** (world): verb aspect and factivity could be precompiled from VerbNet/WordNet into the knowledge layer and attached as predicate features. This helps decide event order and which embedded clauses count as facts versus beliefs.
- **I4** (world): generic-versus-particular detection ("Lions sleep" vs "the lion slept") could use the same two-step marking. This would stop generic statements from being stored as concrete events in the text world.
- **I5** (learn): each new knowledge source or repair rule could be evaluated as an on/off ablation with precision and recall reported separately, to catch overgeneration hidden behind recall gains.

*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

- [The University of Texas at Austin](https://md.research.elamur.ai/free/en/universities/the-university-of-texas-at-austin/) — University, Austin, United States · 109 works · 56,942 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

## Cite

Cite as: Alexis Palmer, Jonas Kuhn and Carlota Smith. 2004. Utilization of Multiple Language Resources for Robust Grammar-Based Tense and Aspect Classification. Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC'04).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{palmer2004utilization,
  title = {{Utilization of Multiple Language Resources for Robust Grammar-Based Tense and Aspect Classification}},
  author = {Alexis Palmer, Jonas Kuhn and Carlota Smith},
  booktitle = {Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC'04)},
  year = {2004},
  doi = {10.63317/2mr9byhgyv2q},
  url = {https://doi.org/10.63317/2mr9byhgyv2q}
}
```

- [← previous Fractal geometry of literature: first attempt to Shakespeare's works](https://md.research.elamur.ai/free/en/papers/eftekhari-2004-fractal-geometry-literature-first-attempt/)
- [next → Three English Learner Assistance Systems Using Automatic Paraphrasing Techniques](https://md.research.elamur.ai/free/en/papers/murata-2004-three-english-learner-assistance-systems/)
- [↑ back to papers of 2004](https://md.research.elamur.ai/free/en/papers/2004/page/2/)

## References

9 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Carlson, G.N. and F.J. Pelletier (eds.), 1995. The Generic Book. Chicago: University of Chicago Press. [find](https://scholar.google.com/scholar?q=Carlson%2C+G%2EN%2E+and+F%2EJ%2E+Pelletier+%28eds%2E%29%2C+1995%2E+The+Generic+Book%2E+Chicago%3A+University+of+Chicago+Press%2E)
2. Dorr, Bonnie and Mari Olsen, 1997. Deriving verbal and compositional lexical aspect for NLP applications. In Proceedings of ACL35. [find](https://scholar.google.com/scholar?q=Dorr%2C+Bonnie+and+Mari+Olsen%2C+1997%2E+Deriving+verbal+and+compositional+lexical+aspect+for+NLP+applications%2E+In+Proceedings+of+ACL35%2E)
3. Dorr, Bonnie J., 2001. LCS verb database. University of Maryland. www.umiacs.umd.edu/˜bonnie/LCS Database Documentation.html. [find](https://scholar.google.com/scholar?q=Dorr%2C+Bonnie+J%2E%2C+2001%2E+LCS+verb+database%2E+University+of+Maryland%2E+www%2Eumiacs%2Eumd%2Eedu%2F%CB%9Cbonnie%2FLCS+Database+Documentation%2Ehtml%2E)
4. Frank, Anette, 1999. From parallel grammar development towards machine translation. a project overview. In Proceedings of Machine Translation Summit VII. [find](https://scholar.google.com/scholar?q=Frank%2C+Anette%2C+1999%2E+From+parallel+grammar+development+towards+machine+translation%2E+a+project+overview%2E+In+Proceedings+of+Machine+Translation+Summit+VII%2E)
5. Passonneau, Rebecca and Diane Litman, 1997. [Discourse segmentation by human and automated means](https://md.research.elamur.ai/free/en/papers/passonneau-1997-discourse-segmentation-human-automated-means/). Computational Linguistics, 23(1):103–139. Special Issue on Empirical Studies in Discourse Interpretation and Generation. [↗](https://doi.org/10.5555/972684.972689)
6. Peterson, Philip (ed.), 1997. Fact Proposition Event. Kluwer. [↗](https://doi.org/10.1007/978-94-015-8959-8)
7. Riezler, Stefan, Dick Crouch, Ron Kaplan, Tracy King, John Maxwell, and Mark Johnson, 2002. [Parsing the Wall Street Journal using a Lexical-Functional Grammar and discriminative estimation techniques](https://md.research.elamur.ai/free/en/papers/riezler-2002-parsing-wall-street-journal-lexical/). In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics (ACL’02), Pennsylvania, Philadelphia.
8. Smith, Carlota, 2003. Modes of Discourse. Cambridge University Press. [↗](https://doi.org/10.1017/cbo9780511615108)
9. Whitelock, P., 1992. [Shake-and-bake translation](https://md.research.elamur.ai/free/en/papers/whitelock-1992-shake-bake-translation/). In Proceedings of COLING-92. Nantes.

*Updated 1 November 2026*