---
title: "Identifying Discourse Markers in Spoken Dialog — Heeman 1998"
description: "The authors annotate discourse-marker uses in the Trains corpus with dedicated POS tags, so finding… 32 references, cited by 1 paper here."
authors: ["Peter A. Heeman", "Donna Byron", "James F. Allen"]
year: 1998
venue: "AAAI 1998 Spring Symposium on Applying Machine Learning to Discourse Processing"
source: "https://arxiv.org/abs/cmp-lg/9801002"
doi: "10.48550/arxiv.cmp-lg/9801002"
arxiv: "cmp-lg/9801002"
licence: "arXiv non-exclusive distribution licence (read only, no redistribution)"
licence_url: "http://arxiv.org/licenses/nonexclusive-distrib/1.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/heeman-1998-identifying-discourse-markers-spoken-dialog/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/heeman-1998-identifying-discourse-markers-spoken-dialog/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [1998](https://md.research.elamur.ai/free/en/papers/1998/) / Identifying Discourse Markers in Spoken Dialog

# Identifying Discourse Markers in Spoken Dialog

**Authors:** [Peter A. Heeman](https://md.research.elamur.ai/free/en/authors/peter-a-heeman/), Donna Byron and [James F. Allen](https://md.research.elamur.ai/free/en/authors/james-f-allen/) · **Year:** 1998 · **Venue:** AAAI 1998 Spring Symposium on Applying Machine Learning to Discourse Processing · **Licence:** [arXiv non-exclusive distribution licence (read only, no redistribution)](http://arxiv.org/licenses/nonexclusive-distrib/1.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/cmp-lg/9801002) · [PDF](https://arxiv.org/pdf/cmp-lg/9801002) · [DOI 10.48550/arxiv.cmp-lg/9801002](https://doi.org/10.48550/arxiv.cmp-lg/9801002) · [arXiv cmp-lg/9801002](https://arxiv.org/abs/cmp-lg/9801002) · [OpenAlex](https://openalex.org/W2124521363)

**Citations:** 32 citations (OpenAlex, 2026-10-14) · [1](https://md.research.elamur.ai/free/en/papers/heeman-1998-identifying-discourse-markers-spoken-dialog/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
In task-oriented spoken dialog, words like "so", "well", "and" or "now" often mark how an upcoming utterance connects to prior talk, yet dialog systems usually discard them as noise. They are also ambiguous: "so" can act sententially, and "well" can sit inside a speech repair. The authors annotate discourse-marker uses in the Trains corpus with dedicated POS tags, so finding markers becomes part of POS tagging. That tagging is built into a speech-recognition language model that searches for the best joint word-and-tag sequence. Probabilities are estimated with decision trees that query hierarchical word and tag clusters. Separating marker uses yields a small drop in word perplexity and in POS error. Jointly modeling utterance segmentation (boundary tones) and speech repairs improves marker identification further. A corpus analysis shows marker-initial turns rarely follow the first part of an adjacency pair. They appear mainly where the next move is not already predictable, which makes markers useful cues for dialog act prediction.

## Key points
- Trains corpus: 98 dialogs, 34 speakers, 58,298 words, 6163 turns and 8278 discourse markers, roughly six and a half hours of speech. 44.1% of non-acknowledgment turns begin with a marker.
- Four tags extend the Penn Treebank tagset:
  - AC for acknowledgments;
  - UH_D for discourse interjections;
  - CC_D for discourse conjuncts;
  - RB_D for discourse adverbials.
- Verbal markers such as "wait" stay tagged as verbs. Filled pauses get UH_FP and are not counted as markers.
- The language model factors Pr(W P) into a word part and a tag part over full histories, avoiding the usual simplified approximations. Both parts are estimated with decision trees grown and smoothed on heldout data.
- The decision-tree questions rely on Brown et al. clustering: a binary tree over POS tags, plus a separate word tree per tag. The induced tag tree places the marker tags near each other.
- Unlike Litman (1996), who classifies each candidate word independently, this approach picks the best overall interpretation of the signal, and marker detection falls out of it.

## Interesting ideas
- Encode a pragmatic function (discourse-marker use) as a separate POS tag, so an ordinary tagger resolves the ambiguity with no extra classifier.
- Build a word cluster tree inside each POS tag rather than one global tree, so ambiguous words like "loads" are clustered separately per sense and sparse data stays manageable.
- Turn hierarchical cluster paths into bit strings that a decision tree can query with simple yes/no questions, which keeps the model explainable.
- Discourse markers are frequent where the next move is open and rare right after a question that sets up a strong expectation, so marker presence carries information about the coming speech act.
- Solve marker detection jointly with segmentation and repair detection instead of in isolation, since each decision constrains the others.

## For Mova

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

- **I1** (tokens): add marker-specific tags or features for "so/well/now/anyway" in the en tagger, so later stages see the disambiguated function.
- **I2** (learn): when inducing lexical classes from silver data, cluster within each UPOS so features stay interpretable and sense-separated.
- **I3** (tree): offer cluster-path bits as transparent features to the perceptron parser, so each split can be inspected.
- **I4** (frames): in prag, let a leading marker ("well" for correction, "so" for conclusion) shift speech-act and relation guesses, and weigh it against open adjacency-pair obligations.
- **I5** (tree): treat dialog-initial "well/so" and repair "well" jointly with sentence segmentation and disfluency handling before attaching them as discourse dependents.

*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 Rochester](https://md.research.elamur.ai/free/en/universities/university-of-rochester/) — University, Rochester, United States · 36 works · 7,398 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

🧭 Topics (2): [🗨️ Dialogue systems](https://md.research.elamur.ai/free/en/topics/dialogue-systems/) · 624 works · [🧵 Discourse analysis](https://md.research.elamur.ai/free/en/topics/discourse-analysis/) · 466 works

## Cite

Cite as: Peter A. Heeman, Donna Byron and James F. Allen. 1998. Identifying Discourse Markers in Spoken Dialog. AAAI 1998 Spring Symposium on Applying Machine Learning to Discourse Processing.

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

```bibtex
@inproceedings{heeman1998identifying,
  title = {{Identifying Discourse Markers in Spoken Dialog}},
  author = {Peter A. Heeman, Donna Byron and James F. Allen},
  booktitle = {AAAI 1998 Spring Symposium on Applying Machine Learning to Discourse Processing},
  year = {1998},
  doi = {10.48550/arxiv.cmp-lg/9801002},
  eprint = {cmp-lg/9801002},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.cmp-lg/9801002}
}
```

- [← previous A Lexicalized Tree Adjoining Grammar for English](https://md.research.elamur.ai/free/en/papers/group-1998-lexicalized-tree-adjoining-grammar-english/)
- [next → Multilingual Robust Anaphora Resolution](https://md.research.elamur.ai/free/en/papers/mitkov-1998-multilingual-robust-anaphora-resolution/)
- [↑ back to papers of 1998](https://md.research.elamur.ai/free/en/papers/1998/)

## References

32 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Allen, J. F.; Schubert, L.; Ferguson, G.; Heeman, P.; Hwang, C.; Kato, T.; Light, M.; Martin, N.; Miller, B.; Poesio, M.; and Traum, D. 1995. The Trains project: A case study in building a conversational planning agent. Journal of Experimental and Theoretical AI 7:7–48. [↗](https://doi.org/10.1080/09528139508953799)
2. Bahl, L. R.; Brown, P. F.; deSouza, P. V.; and Mercer, R. L. 1989. A tree-based statistical language model for natural language speech recognition. IEEE Transactions on Acoustics, Speech, and Signal Processing 36(7):1001–1008. [↗](https://doi.org/10.1109/29.32278)
3. Black, E.; Jelinek, F.; Lafferty, J.; Mercer, R.; and Roukos, S. 1992. Decision tree models applied to the labeling of text with parts-of-speech. In Proceedings of the DARPA Speech and Natural Language Workshop, 117–121. Morgan Kaufman. [find](https://scholar.google.com/scholar?q=Black%2C+E%2E%3B+Jelinek%2C+F%2E%3B+Lafferty%2C+J%2E%3B+Mercer%2C+R%2E%3B+and+Roukos%2C+S%2E+1992%2E+Decision+tree+models+applied+to+the+labeling+of+text+with+parts%2Dof%2Dspeech%2E+In+Proceedings+of+the+DARPA+Speech+and+Natural+Languag)
4. Breiman, L.; Friedman, J. H.; Olshen, R. A.; and Stone, C. J. 1984. Classification and Regression Trees. Monterrey, CA: Wadsworth & Brooks. [find](https://scholar.google.com/scholar?q=Breiman%2C+L%2E%3B+Friedman%2C+J%2E+H%2E%3B+Olshen%2C+R%2E+A%2E%3B+and+Stone%2C+C%2E+J%2E+1984%2E+Classification+and+Regression+Trees%2E+Monterrey%2C+CA%3A+Wadsworth+%26+Brooks%2E)
5. Brown, G., and Yule, G. 1983. Discourse Analysis. Cambridge: Cambridge University Press. [find](https://scholar.google.com/scholar?q=Brown%2C+G%2E%2C+and+Yule%2C+G%2E+1983%2E+Discourse+Analysis%2E+Cambridge%3A+Cambridge+University+Press%2E)
6. Brown, P. F.; Della Pietra, V. J.; deSouza, P. V.; Lai, J. C.; and Mercer, R. L. 1992. Class-based ngram models of natural language. Computational Linguistics 18(4):467–479. [find](https://scholar.google.com/scholar?q=Brown%2C+P%2E+F%2E%3B+Della+Pietra%2C+V%2E+J%2E%3B+deSouza%2C+P%2E+V%2E%3B+Lai%2C+J%2E+C%2E%3B+and+Mercer%2C+R%2E+L%2E+1992%2E+Class%2Dbased+ngram+models+of+natural+language%2E+Computational+Linguistics+18%284%29%3A467%E2%80%93479%2E)
7. Byron, D. K., and Heeman, P. A. 1997a. Discourse marker use in task-oriented spoken dialog. In Proceedings of the 5 th European Conference on Speech Communication and Technology (Eurospeech). [find](https://scholar.google.com/scholar?q=Byron%2C+D%2E+K%2E%2C+and+Heeman%2C+P%2E+A%2E+1997a%2E+Discourse+marker+use+in+task%2Doriented+spoken+dialog%2E+In+Proceedings+of+the+5+th+European+Conference+on+Speech+Communication+and+Technology+%28Eurospeech%29%2E)
8. Byron, D. K., and Heeman, P. A. 1997b. Discourse marker use in task-oriented spoken dialog. Technical report, Department of Computer Science, University of Rochester, 664. [find](https://scholar.google.com/scholar?q=Byron%2C+D%2E+K%2E%2C+and+Heeman%2C+P%2E+A%2E+1997b%2E+Discourse+marker+use+in+task%2Doriented+spoken+dialog%2E+Technical+report%2C+Department+of+Computer+Science%2C+University+of+Rochester%2C+664%2E)
9. Charniak, E.; Hendrickson, C.; Jacobson, N.; and Perkowitz, M. 1993. Equations for part-of-speech tagging. In Proceedings of the National Conference on Artificial Intelligence (AAAI ’93). [find](https://scholar.google.com/scholar?q=Charniak%2C+E%2E%3B+Hendrickson%2C+C%2E%3B+Jacobson%2C+N%2E%3B+and+Perkowitz%2C+M%2E+1993%2E+Equations+for+part%2Dof%2Dspeech+tagging%2E+In+Proceedings+of+the+National+Conference+on+Artificial+Intelligence+%28AAAI+%E2%80%9993%29%2E)
10. Chu-Carroll, J. 1998. Statistical model for discourse act recognition in dialogue interactions. In Proceedings of the AAAI Workshop on Applying Machine Learning to Discourse Processing. [find](https://scholar.google.com/scholar?q=Chu%2DCarroll%2C+J%2E+1998%2E+Statistical+model+for+discourse+act+recognition+in+dialogue+interactions%2E+In+Proceedings+of+the+AAAI+Workshop+on+Applying+Machine+Learning+to+Discourse+Processing%2E)
11. Church, K. 1988. [A stochastic parts program and noun phrase parser for unrestricted text](https://md.research.elamur.ai/free/en/papers/church-1988-stochastic-parts-program-noun-phrase/). In Proceedings of the 2nd Conference on Applied Natural Language Processing, 136–143.
12. Cohen, R. 1984. A computational theory of the function of clue words in argument understanding. In Proceedings of the 10 th International Conference on Computational Linguistics (COLING), 251–255. [find](https://scholar.google.com/scholar?q=Cohen%2C+R%2E+1984%2E+A+computational+theory+of+the+function+of+clue+words+in+argument+understanding%2E+In+Proceedings+of+the+10+th+International+Conference+on+Computational+Linguistics+%28COLING%29%2C+251%E2%80%93255%2E)
13. Cohen, W. W. 1992. Compiling knowledge into an explicit bias. In Proceedings of the Ninth International Conference on Machine Learning. [find](https://scholar.google.com/scholar?q=Cohen%2C+W%2E+W%2E+1992%2E+Compiling+knowledge+into+an+explicit+bias%2E+In+Proceedings+of+the+Ninth+International+Conference+on+Machine+Learning%2E)
14. Cohen, W. W. 1993. Efficient pruning methods for separate-and-conquer rule learning systems. In Proceedings of the International Joint Conference on Artificial Intelligence (IJ-CAI ’93). [find](https://scholar.google.com/scholar?q=Cohen%2C+W%2E+W%2E+1993%2E+Efficient+pruning+methods+for+separate%2Dand%2Dconquer+rule+learning+systems%2E+In+Proceedings+of+the+International+Joint+Conference+on+Artificial+Intelligence+%28IJ%2DCAI+%E2%80%9993%29%2E)
15. Dahlbäck, N., and Jönsson, A. 1992. An empirically based computationally tractable dialogue model. In Program of the 14 th Annual Conference of the Cognitive Science Society. [find](https://scholar.google.com/scholar?q=Dahlba%CC%88ck%2C+N%2E%2C+and+Jo%CC%88nsson%2C+A%2E+1992%2E+An+empirically+based+computationally+tractable+dialogue+model%2E+In+Program+of+the+14+th+Annual+Conference+of+the+Cognitive+Science+Society%2E)
16. DeRose, S. J. 1988. [Grammatical category disambiguation by statistical optimization](https://md.research.elamur.ai/free/en/papers/derose-1988-grammatical-category-disambiguation-statistical-optimization/). Computational Linguistics 14(1):31–39. [↗](https://doi.org/10.5555/49084.49087)
17. Grosz, B. J., and Sidner, C. L. 1986. [Attention, intentions, and the structure of discourse](https://md.research.elamur.ai/free/en/papers/grosz-1986-attention-intentions-structure-discourse/). Computational Linguistics 12(3):175–204. [↗](https://doi.org/10.5555/12457.12458)
18. Heeman, P. A., and Allen, J. F. 1995. The Trains spoken dialog corpus. CD-ROM, Linguistics Data Consortium. [find](https://scholar.google.com/scholar?q=Heeman%2C+P%2E+A%2E%2C+and+Allen%2C+J%2E+F%2E+1995%2E+The+Trains+spoken+dialog+corpus%2E+CD%2DROM%2C+Linguistics+Data+Consortium%2E)
19. Heeman, P. A., and Allen, J. F. 1997a. Incorporating POS tagging into language modeling. In Proceedings of the 5 th European Conference on Speech Communication and Technology (Eurospeech). [find](https://scholar.google.com/scholar?q=Heeman%2C+P%2E+A%2E%2C+and+Allen%2C+J%2E+F%2E+1997a%2E+Incorporating+POS+tagging+into+language+modeling%2E+In+Proceedings+of+the+5+th+European+Conference+on+Speech+Communication+and+Technology+%28Eurospeech%29%2E)
20. Heeman, P. A., and Allen, J. F. 1997b. [Intonational boundaries, speech repairs, and discourse markers: Modeling spoken dialog](https://md.research.elamur.ai/free/en/papers/heeman-1997-intonational-boundaries-speech-repairs-discourse/). In Proceedings of the 35 th Annual Meeting of the Association for Computational Linguistics, 254– 261. [↗](https://doi.org/10.48550/arxiv.cmp-lg/9704008)
21. Heeman, P. A. 1997. Speech repairs, intonational boundaries and discourse markers: Modeling speakers’ utterances in spoken dialog. Technical report, Department of Computer Science, University of Rochester. Doctoral dissertation. [↗](https://doi.org/10.48550/arxiv.cmp-lg/9712009)
22. Hirschberg, J., and Litman, D. 1993. [Empirical studies on the disambiguation of cue phrases](https://md.research.elamur.ai/free/en/papers/hirschberg-1993-empirical-studies-disambiguation-cue-phrases/). Computational Linguistics 19(3):501–530.
23. Jelinek, F. 1985. Self-organized language modeling for speech recognition. Technical report, IBM T.J. Watson Research Center, Continuous Speech Recognition Group, Yorktown Heights, NY. [find](https://scholar.google.com/scholar?q=Jelinek%2C+F%2E+1985%2E+Self%2Dorganized+language+modeling+for+speech+recognition%2E+Technical+report%2C+IBM+T%2EJ%2E+Watson+Research+Center%2C+Continuous+Speech+Recognition+Group%2C+Yorktown+Heights%2C+NY%2E)
24. Litman, D. J. 1996. Cue phrase classification using machine learning. Journal of Artificial Intelligence Research 5:53–94. [↗](https://doi.org/10.1613/jair.327)
25. Magerman, D. M. 1995. Statistical decision trees for parsing. In Proceedings of the 33 th Annual Meeting of the Association for Computational Linguistics, 7–14. [find](https://scholar.google.com/scholar?q=Magerman%2C+D%2E+M%2E+1995%2E+Statistical+decision+trees+for+parsing%2E+In+Proceedings+of+the+33+th+Annual+Meeting+of+the+Association+for+Computational+Linguistics%2C+7%E2%80%9314%2E)
26. Marcus, M. P.; Santorini, B.; and Marcinkiewicz, M. A. 1993. [Building a large annotated corpus of english: The Penn Treebank](https://md.research.elamur.ai/free/en/papers/marcus-1993-building-large-annotated-corpus-english/). Computational Linguistics 19(2):313–330. [↗](https://doi.org/10.21236/ada273556)
27. Quinlan, J. R. 1993. [C4.5: Programs for Machine Learning](https://md.research.elamur.ai/free/en/papers/quinlan-1993-c45-programs-machine-learning/). San Mateo, California: Morgan Kaufman.
28. Reichman, R. 1985. Getting Computers to Talk Like You and Me: Discourse Context, Focus, and Semantics (An ATN Model). Cambridge, MA: MIT Press. [find](https://scholar.google.com/scholar?q=Reichman%2C+R%2E+1985%2E+Getting+Computers+to+Talk+Like+You+and+Me%3A+Discourse+Context%2C+Focus%2C+and+Semantics+%28An+ATN+Model%29%2E+Cambridge%2C+MA%3A+MIT+Press%2E)
29. Schegloff, E. A., and Sacks, H. 1973. Opening up closings. Semiotica 7:289–327. [↗](https://doi.org/10.1515/semi.1973.8.4.289)
30. Schiffrin, D. 1987. Discourse Markers. New York: Cambridge University Press. [find](https://scholar.google.com/scholar?q=Schiffrin%2C+D%2E+1987%2E+Discourse+Markers%2E+New+York%3A+Cambridge+University+Press%2E)
31. Stolcke, A.; Shriberg, E.; Bates, R.; Coccaro, N.; Jurafsky, D.; Martin, R.; Meteer, M.; Ries, K.; Taylor, P.; and Ess-Dykema, C. V. 1998. Dialog act modeling for conversational speech. In Proceedings of the AAAI Workshop on Applying Machine Learning to Discourse Processing. [find](https://scholar.google.com/scholar?q=Stolcke%2C+A%2E%3B+Shriberg%2C+E%2E%3B+Bates%2C+R%2E%3B+Coccaro%2C+N%2E%3B+Jurafsky%2C+D%2E%3B+Martin%2C+R%2E%3B+Meteer%2C+M%2E%3B+Ries%2C+K%2E%3B+Taylor%2C+P%2E%3B+and+Ess%2DDykema%2C+C%2E+V%2E+1998%2E+Dialog+act+modeling+for+conversational+speech%2E+In+Proceedings)
32. Taylor, P.; King, S.; Isard, S.; Wright, H.; and Kowtko, J. 1997. Using intonation to constrain language models in speech recognition. In Proceedings of the 5 th European Conference on Speech Communication and Technology (Eurospeech). [find](https://scholar.google.com/scholar?q=Taylor%2C+P%2E%3B+King%2C+S%2E%3B+Isard%2C+S%2E%3B+Wright%2C+H%2E%3B+and+Kowtko%2C+J%2E+1997%2E+Using+intonation+to+constrain+language+models+in+speech+recognition%2E+In+Proceedings+of+the+5+th+European+Conference+on+Speech+Communic)

*Updated 1 November 2026*