---
title: "An Activation-based Sentence Processing Model of English (2006)"
description: "They propose a working-memory model of incremental parsing. 10 references. Among them: 0."
authors: ["Kei Takahashi", "Kiyoshi Ishikawa", "Kei Yoshimoto"]
year: 2006
venue: "Proceedings of the 20th Pacific Asia Conference on Language, Information and Computation"
source: "https://aclanthology.org/Y06-1040/"
doi: "http://hdl.handle.net/2065/29018"
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/takahashi-2006-activation-based-sentence-processing-model/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/takahashi-2006-activation-based-sentence-processing-model/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2006](https://md.research.elamur.ai/free/en/papers/2006/page/2/) / An Activation-based Sentence Processing Model of English

# An Activation-based Sentence Processing Model of English

**Authors:** Kei Takahashi, Kiyoshi Ishikawa and [Kei Yoshimoto](https://md.research.elamur.ai/free/en/authors/kei-yoshimoto/) · **Year:** 2006 · **Venue:** [Proceedings of the 20th Pacific Asia Conference on Language, Information and Computation](https://md.research.elamur.ai/free/en/venues/paclic/) ([PACLIC 2006](https://md.research.elamur.ai/free/en/venues/paclic/2006/)) · **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/Y06-1040/) · [PDF](https://aclanthology.org/Y06-1040.pdf) · [DOI http://hdl.handle.net/2065/29018](https://doi.org/http://hdl.handle.net/2065/29018) · [OpenAlex](https://openalex.org/W2404115783)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
The paper looks at English sentences where acceptability depends on word order and not only on hierarchical structure. One example is a that-clause after "about": it is fine when topicalized, coordinated second, or separated from the preposition by a pause or by an inserted phrase such as "by the way", but bad when it directly follows the preposition. The authors argue that grammar-based accounts, including LFG's grammatical-function solution, cannot explain these contrasts. They propose a working-memory model of incremental parsing. A head predicts its complement's category ahead of time, and each candidate category carries a numeric activation value. These values are lowered in three ways: fully once the predicate-argument structure is built, gradually as processing time passes, and by a further amount when overt material intervenes. A well-formedness score, computed from activation clashes and multiplied upward to the S node, gives graded acceptability. The model accounts for topicalization, complement coordination, insertion and Right Node Raising in one framework. It also suggests that garden-path cost could be measured as the activation gap between competing parses. The work is theoretical, with no implementation and no corpus.

## Key points
- Core data: the topicalization contrast from Kaplan and Bresnan (1982), coordination where only the order of conjuncts differs, Right Node Raising with "agreed with", and insertion of a pause or "by the way".
- The authors state a "Linear Order Effect": a head's syntactic demands apply only while the argument is close enough in linear order. Semantic constraints, in contrast, stay fully in force.
- Activation values for the competing category labels of one node sum to a constant T with 0 ≤ T ≤ 1. When the predicted label is confirmed, it absorbs the activation of the rival labels, e.g. PP and CP at 0.5 each become PP at 1.0.
- There are three deactivation functions:
  - f sets activation to 0 once the semantic content is built.
  - g subtracts a per processing cycle, with t/c cycles in total.
  - h subtracts b when an overt phrase intervenes.
- Node well-formedness scores are multiplied and passed up to S, giving graded acceptability (e.g. 0.7). The coordination analysis relies on the incremental conjunct-attachment findings of Sturt and Lombardo (2005).

## Interesting ideas
- Store a probability-like budget over a node's possible categories, and move the losers' mass to the winner on confirmation, so ambiguity and its resolution become one explicit number.
- Selection constraints weaken with linear distance and intervening material, while semantic constraints persist.
- Once predicate-argument content is assembled, the syntactic details can be dropped and only the meaning kept.
- Acceptability is computed as a product of local well-formedness scores passed up to the root, giving graded rather than binary judgments.
- Garden-path difficulty is measured as the gap in activation between the chosen parse and its rivals.

## For Mova

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

- **I1** (tree): The parser could record, for each predicted dependent, an activation distribution over expected categories, so traces show why a reading won.
- **I2** (tree): The validator of UD trees could soften category-selection checks (e.g. a clause under a preposition) when a pause, comma or adjunct separates head and dependent, instead of flagging them as errors.
- **I3** (frames): Frame building could mark syntactic features as consumed after a role is filled, so later reasoning works only with frames and stays explainable.
- **I4** (learn): Silver-data filtering could score sentences by multiplying local clash penalties, keeping borderline order-dependent constructions with a confidence rather than discarding them.
- **I5** (learn): Evaluation could use this gap as a difficulty signal, picking hard sentences for LLM-teacher review or as negative controls.

*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 Edinburgh](https://md.research.elamur.ai/free/en/universities/university-of-edinburgh/) — University, Edinburgh, United Kingdom · 381 works · 80,747 citations

📚 Shelves (6): [🎭 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 · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 1,649 works · [💭 Developmental and Educational Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/developmental-and-educational-psychology/) · 523 works · [🧠 Neuroscience](https://md.research.elamur.ai/free/en/shelves/neuroscience/) · 186 works · [🧠 Cognitive Neuroscience](https://md.research.elamur.ai/free/en/shelves/neuroscience/cognitive-neuroscience/) · 185 works

🧭 Topics (2): [English](https://md.research.elamur.ai/free/en/topics/english/) · 2,379 works · [🧒 Psycholinguistics & language acquisition](https://md.research.elamur.ai/free/en/topics/psycholinguistics/) · 108 works

## Cite

Cite as: Kei Takahashi, Kiyoshi Ishikawa and Kei Yoshimoto. 2006. An Activation-based Sentence Processing Model of English. Proceedings of the 20th Pacific Asia Conference on Language, Information and Computation.

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

```bibtex
@inproceedings{takahashi2006activation,
  title = {{An Activation-based Sentence Processing Model of English}},
  author = {Kei Takahashi, Kiyoshi Ishikawa and Kei Yoshimoto},
  booktitle = {Proceedings of the 20th Pacific Asia Conference on Language, Information and Computation},
  year = {2006},
  doi = {http://hdl.handle.net/2065/29018},
  url = {https://doi.org/http://hdl.handle.net/2065/29018}
}
```

- [← previous Va'yehi as a Discourse Marker in Kings](https://md.research.elamur.ai/free/en/papers/li-2006-va-yehi-discourse-marker-kings/)
- [next → Constructing an English Valency Lexicon](https://md.research.elamur.ai/free/en/papers/semecky-2006-constructing-english-valency-lexicon/)
- [↑ back to papers of 2006](https://md.research.elamur.ai/free/en/papers/2006/page/2/)

## References

10 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Bresnan, Joan. 2000. Lexical-Functional Syntax . Oxford: Blackwell. [find](https://scholar.google.com/scholar?q=Bresnan%2C+Joan%2E+2000%2E+Lexical%2DFunctional+Syntax+%2E+Oxford%3A+Blackwell%2E)
2. Falk, Yehuda N. 2001. Lexical-Functional Grammar: An Introduction to Parallel Constraint-Based Syntax . Stanford: CSLI Publications. [find](https://scholar.google.com/scholar?q=Falk%2C+Yehuda+N%2E+2001%2E+Lexical%2DFunctional+Grammar%3A+An+Introduction+to+Parallel+Constraint%2DBased+Syntax+%2E+Stanford%3A+CSLI+Publications%2E)
3. Hawkins, John. 1994. Performance Theory of Order and Constotuency. Cambridge: Cambridge University Press. [find](https://scholar.google.com/scholar?q=Hawkins%2C+John%2E+1994%2E+Performance+Theory+of+Order+and+Constotuency%2E+Cambridge%3A+Cambridge+University+Press%2E)
4. Kaplan, Ronald M., and Joan Bresnan. 1982. Lexical-Functional Grammar: A Formal System for Grammatical Representation. The Mental Representation of Grammatical Relations, ed. Joan Bresnan, 173–281. Cambridge, MA: MIT Press. [find](https://scholar.google.com/scholar?q=Kaplan%2C+Ronald+M%2E%2C+and+Joan+Bresnan%2E+1982%2E+Lexical%2DFunctional+Grammar%3A+A+Formal+System+for+Grammatical+Representation%2E+The+Mental+Representation+of+Grammatical+Relations%2C+ed%2E+Joan+Bresnan%2C+173%E2%80%93281%2E+Ca)
5. Kaplan, Ronald M., and Annie Zaenan 1989. Long-Distance Dependencies, Constituent Structure, and Functional Uncertainty. Alternative Conception of Phrase Structure, eds. Mark R. Bartin and Anthony S. Kroch, 17–42, Chicago: University of Chicago Press. [find](https://scholar.google.com/scholar?q=Kaplan%2C+Ronald+M%2E%2C+and+Annie+Zaenan+1989%2E+Long%2DDistance+Dependencies%2C+Constituent+Structure%2C+and+Functional+Uncertainty%2E+Alternative+Conception+of+Phrase+Structure%2C+eds%2E+Mark+R%2E+Bartin+and+Anthony+S%2E+)
6. Moosally, Michelle J. 1999. Subject and Object Coordination in Ndebele: An HPSG Analysis. WCCFL 18, eds. S.Bird, A Camnie, J. Haugen, P. Norquest, 379–392. MA: Cascadilla Press. [find](https://scholar.google.com/scholar?q=Moosally%2C+Michelle+J%2E+1999%2E+Subject+and+Object+Coordination+in+Ndebele%3A+An+HPSG+Analysis%2E+WCCFL+18%2C+eds%2E+S%2EBird%2C+A+Camnie%2C+J%2E+Haugen%2C+P%2E+Norquest%2C+379%E2%80%93392%2E+MA%3A+Cascadilla+Press%2E)
7. Sadock, Jerrold M. 1998. Grammatical Tension. CLS 34: The Panels, 179–198. [find](https://scholar.google.com/scholar?q=Sadock%2C+Jerrold+M%2E+1998%2E+Grammatical+Tension%2E+CLS+34%3A+The+Panels%2C+179%E2%80%93198%2E)
8. Sturt, Patrick, and Vincenzo Lombardo. 2005. Processing Coordinated Structures: Incrementality and Connectedness. Cognitive Science 29, 291–305. [find](https://scholar.google.com/scholar?q=Sturt%2C+Patrick%2C+and+Vincenzo+Lombardo%2E+2005%2E+Processing+Coordinated+Structures%3A+Incrementality+and+Connectedness%2E+Cognitive+Science+29%2C+291%E2%80%93305%2E)
9. Takahashi, Kei, and Kiyoshi Ishikawa. 2004. An Adjacecy Constraint on Argument Selection. Language, Information and Computation: Proceedings of the 18th Pacific Asia Conference. eds. Hiroshi Masuichi, Tomoko Ohkuma, Kiyoshi Ishikawa, Yasunari Harada, and Kei Yoshimoto. 23–34. Tokyo: The Logico-Linguistic Society of Japan. [find](https://scholar.google.com/scholar?q=Takahashi%2C+Kei%2C+and+Kiyoshi+Ishikawa%2E+2004%2E+An+Adjacecy+Constraint+on+Argument+Selection%2E+Language%2C+Information+and+Computation%3A+Proceedings+of+the+18th+Pacific+Asia+Conference%2E+eds%2E+Hiroshi+Masuichi%2C)
10. Takahashi, Kei, and Kei Yoshimoto. 2006. A Memory-based Sentence Processing Model. The fifth International Workshop on Evolutionary Cognitive Sciences Human Sentence Processing and Production (Co-sponsored by The Technical Group of Thought and Language (The Institute of Electronics, Information and Communication Engineers) ). IEICE Technical Report. TL-2006-11, 25–31. In the present model, the number of the intervening overt expressions has no effect; all that counts is whether some overt lingusitic expression or other intervenes. This is against our intuition, and we intend to remedy this defect in the future development of our model. [find](https://scholar.google.com/scholar?q=Takahashi%2C+Kei%2C+and+Kei+Yoshimoto%2E+2006%2E+A+Memory%2Dbased+Sentence+Processing+Model%2E+The+fifth+International+Workshop+on+Evolutionary+Cognitive+Sciences+Human+Sentence+Processing+and+Production+%28Co%2Dspon)

*Updated 2 November 2026*