---
title: "Modeling Construction Grammar's Way into NLP — Lorenzi 2023"
description: "Authors from the FrameNet Brasil laboratory examined whether schematic clausal constructions and their elements… 23 references, cited by 1 paper here."
authors: ["Arthur Lorenzi", "Vânia Gomes de Almeida", "Ely Edison Matos", "Tiago Timponi Torrent"]
year: 2023
venue: "Proceedings of the First International Workshop on Construction Grammars and NLP (CxGs+NLP, GURT/SyntaxFest 2023)"
source: "https://aclanthology.org/2023.cxgsnlp-1.11/"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/lorenzi-2023-modeling-construction-grammar-s-way/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/lorenzi-2023-modeling-construction-grammar-s-way/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2023](https://md.research.elamur.ai/free/en/papers/2023/page/8/) / Modeling Construction Grammar's Way into NLP: Insights from negative results in automatically identifying schematic clausal constructions in Brazilian Portuguese

# Modeling Construction Grammar's Way into NLP: Insights from negative results in automatically identifying schematic clausal constructions in Brazilian Portuguese

**Authors:** Arthur Lorenzi, Vânia Gomes de Almeida, Ely Edison Matos and Tiago Timponi Torrent · **Year:** 2023 · **Venue:** Proceedings of the First International Workshop on Construction Grammars and NLP (CxGs+NLP, GURT/SyntaxFest 2023) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2023.cxgsnlp-1.11/) · [PDF](https://aclanthology.org/2023.cxgsnlp-1.11.pdf)

**Citations:** 1 in our library, of the 140,884 papers whose bibliographies we read

## Summary
Authors from the FrameNet Brasil laboratory examined whether schematic clausal constructions and their elements (CEs) can be automatically identified in sentences of Brazilian Portuguese. The difficulty is that such constructions have no characteristic lexical fillers. Moreover, different constructions often share the same syntactic structure, such as the intransitive and the ergative. To obtain data, the researchers took the test portion of the UD Portuguese GSD treebank and manually annotated it in the FN-Br WebTool with constructions from the FrameNet Brasil Constructicon. They then tested whether models based on Multilingual BERT and LSTM benefit from additional information: UD dependencies and frames selected by two FN-Br graph traversal algorithms. The result is negative. The best average performance came from a compact LSTM with a CRF layer that received neither UD nor frames. The work is valuable for its honest analysis of failure. The authors show that the linearization of tree and graph structures, the small amount of data, and the lack of post-processing hinder the use of constructicons in NLP.

## Key points
- The dataset contains 673 sentences from the test portion of UD pt-br GSD (1200 sentences). 24 CEs from 11 argument structure constructions were annotated. The full GSD comprises 12,019 sentences and 297,045 tokens.
- The models were evaluated by macro-F1, averaged over 10 independent training runs. The highest average score and the smallest variance were achieved by LSTM+CRF without frames or UD.
- For the CE "SUBJECT" of the Attributive_nominal_predicative construction, the model with CRF achieved an F1 of 0.628, while the variant with BFS token ordering achieved only 0.388. Thus, positional information is important, and UD does not replace it.
- Qualitative analysis of the best Transformer model revealed the following errors. In 26% of sentences only the heads of CEs were marked, and in 13% the spans were discontinuous. In 8.7% of cases the model overgeneralizes based on part of speech, and in another 8.7% it incorrectly annotates conjuncts via the conj relation.
- The authors propose several directions for future work: architectures that operate directly on trees and graphs, a larger and more diverse dataset, and dependency-tree-based post-processing that would expand spans from already identified heads.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

Fields: [Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/)

## Cite

Cite as: Arthur Lorenzi, Vânia Gomes de Almeida, Ely Edison Matos and Tiago Timponi Torrent. 2023. Modeling Construction Grammar's Way into NLP: Insights from negative results in automatically identifying schematic clausal constructions in Brazilian Portuguese. Proceedings of the First International Workshop on Construction Grammars and NLP (CxGs+NLP, GURT/SyntaxFest 2023).

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

```bibtex
@inproceedings{lorenzi2023modeling,
  title = {{Modeling Construction Grammar's Way into NLP: Insights from negative results in automatically identifying schematic clausal constructions in Brazilian Portuguese}},
  author = {Arthur Lorenzi, Vânia Gomes {de Almeida}, Ely Edison Matos and Tiago Timponi Torrent},
  booktitle = {Proceedings of the First International Workshop on Construction Grammars and NLP (CxGs+NLP, GURT/SyntaxFest 2023)},
  year = {2023},
  url = {https://aclanthology.org/2023.cxgsnlp-1.11/}
}
```

- [← previous Introducing DictaLM — A Large Generative Language Model for Modern Hebrew](https://md.research.elamur.ai/free/en/papers/shmidman-2023-introducing-dictalm-large-generative-language/)
- [next → Morphosyntactic aspects of agrammatism in Palestinian Arabic: Findings from a Semitic language](https://md.research.elamur.ai/free/en/papers/adam-2023-morphosyntactic-aspects-agrammatism-palestinian-arabic/)
- [↑ back to papers of 2023](https://md.research.elamur.ai/free/en/papers/2023/page/8/)

## References

23 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Vânia Almeida and Tiago Torrent. 2021. Construções de estrutura argumental com argumento preposicionado: uma modelagem linguístico-computacional na FrameNet Brasil. In Anais do XIII Simpósio Brasileiro de Tecnologia da Informação e da Linguagem Humana, pages 353–362, Porto Alegre, RS, Brasil. SBC. [find](https://scholar.google.com/scholar?q=V%C3%A2nia+Almeida+and+Tiago+Torrent%2E+2021%2E+Constru%C3%A7%C3%B5es+de+estrutura+argumental+com+argumento+preposicionado%3A+uma+modelagem+lingu%C3%ADstico%2Dcomputacional+na+FrameNet+Brasil%2E+In+Anais+do+XIII+Simp%C3%B3sio+Brasileir)
2. Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory. Neural Computation, 9(8):1735– 1780. [find](https://scholar.google.com/scholar?q=Sepp+Hochreiter+and+J%C3%BCrgen+Schmidhuber%2E+1997%2E+Long+short%2Dterm+memory%2E+Neural+Computation%2C+9%288%29%3A1735%E2%80%93+1780%2E)
3. Vânia Gomes de Almeida. 2022. Modelagem e Identificação Automática de Construções de Estrutura Argumental: uma proposta para o Constructicon da FrameNet Brasil. Ph.D. thesis, Graduate Program in Linguistics, Federal University of Juiz de Fora, Brazil. [find](https://scholar.google.com/scholar?q=V%C3%A2nia+Gomes+de+Almeida%2E+2022%2E+Modelagem+e+Identifica%C3%A7%C3%A3o+Autom%C3%A1tica+de+Constru%C3%A7%C3%B5es+de+Estrutura+Argumental%3A+uma+proposta+para+o+Constructicon+da+FrameNet+Brasil%2E+Ph%2ED%2E+thesis%2C+Graduate+Program+in+Lingu)
4. Jing Li, Aixin Sun, Jianglei Han, and Chenliang Li. 2020. A survey on deep learning for named entity recognition. IEEE Transactions on Knowledge and Data Engineering, 34(1):50–70. [find](https://scholar.google.com/scholar?q=Jing+Li%2C+Aixin+Sun%2C+Jianglei+Han%2C+and+Chenliang+Li%2E+2020%2E+A+survey+on+deep+learning+for+named+entity+recognition%2E+IEEE+Transactions+on+Knowledge+and+Data+Engineering%2C+34%281%29%3A50%E2%80%9370%2E)
5. John Edward Bryant. 2008. Best-Fit Constructional Analysis. Ph.D. thesis, EECS Department, University of California, Berkeley. [find](https://scholar.google.com/scholar?q=John+Edward+Bryant%2E+2008%2E+Best%2DFit+Constructional+Analysis%2E+Ph%2ED%2E+thesis%2C+EECS+Department%2C+University+of+California%2C+Berkeley%2E)
6. Alebachew Chiche and Betselot Yitagesu. 2022. Part of speech tagging: a systematic review of deep learning and machine learning approaches. Journal of Big Data, 9(1):1–25. [find](https://scholar.google.com/scholar?q=Alebachew+Chiche+and+Betselot+Yitagesu%2E+2022%2E+Part+of+speech+tagging%3A+a+systematic+review+of+deep+learning+and+machine+learning+approaches%2E+Journal+of+Big+Data%2C+9%281%29%3A1%E2%80%9325%2E)
7. Alexandre Diniz da Costa, Mateus Coutinho Marim, Ely Matos, and Tiago Timponi Torrent. 2022. Domain adaptation in neural machine translation using a qualia-enriched FrameNet. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 1–12, Marseille, France. European Langua… [find](https://scholar.google.com/scholar?q=Alexandre+Diniz+da+Costa%2C+Mateus+Coutinho+Marim%2C+Ely+Matos%2C+and+Tiago+Timponi+Torrent%2E+2022%2E+Domain+adaptation+in+neural+machine+translation+using+a+qualia%2Denriched+FrameNet%2E+In+Proceedings+of+the+Thi)
8. Alexandre Diniz da Costa, Vânia Almeida, Ludmila Lage, Gustavo Barbosa, Natália Marção, Vanessa Paiva, Ely da Silva Matos, and Tiago Torrent. 2018. Representação computacional das construções de sujeito-predicado do português do brasil. Revista LinguíStica, 14(1):149–178. [find](https://scholar.google.com/scholar?q=Alexandre+Diniz+da+Costa%2C+V%C3%A2nia+Almeida%2C+Ludmila+Lage%2C+Gustavo+Barbosa%2C+Nat%C3%A1lia+Mar%C3%A7%C3%A3o%2C+Vanessa+Paiva%2C+Ely+da+Silva+Matos%2C+and+Tiago+Torrent%2E+2018%2E+Representa%C3%A7%C3%A3o+computacional+das+constru%C3%A7%C3%B5es+de+sujei)
9. Marie-Catherine de Marneffe, Christopher D. Manning, Joakim Nivre, and Daniel Zeman. 2021. Universal Dependencies. Computational Linguistics, 47(2):255–308. [our card](https://md.research.elamur.ai/free/en/papers/marneffe-2021-universal-dependencies-2/)
10. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,… [our card](https://md.research.elamur.ai/free/en/papers/devlin-2019-bert-pre-training-deep-bidirectional/)
11. Charles J. Fillmore. 1968. The case for case. In Emmon Bach and Robert T. Harms, editors, Universals in Linguistic Theory, pages 0–88. Holt, Rinehart and Winston, New York. [find](https://scholar.google.com/scholar?q=Charles+J%2E+Fillmore%2E+1968%2E+The+case+for+case%2E+In+Emmon+Bach+and+Robert+T%2E+Harms%2C+editors%2C+Universals+in+Linguistic+Theory%2C+pages+0%E2%80%9388%2E+Holt%2C+Rinehart+and+Winston%2C+New+York%2E)
12. Charles J. Fillmore. 2008. Border conflicts: FrameNet meets Construction Grammar. In Proceedings of the XIII EURALEX International Congress, pages 49–68, Barcelona. Universitat Pompeu Fabra, Universitat Pompeu Fabra. [find](https://scholar.google.com/scholar?q=Charles+J%2E+Fillmore%2E+2008%2E+Border+conflicts%3A+FrameNet+meets+Construction+Grammar%2E+In+Proceedings+of+the+XIII+EURALEX+International+Congress%2C+pages+49%E2%80%9368%2C+Barcelona%2E+Universitat+Pompeu+Fabra%2C+Universit)
13. Adele E. Goldberg. 2013. Constructionist approaches. In Thomas Hoffman and Graeme Trousadale, editors, The Oxford Handbook of Construction Grammar. Oxford University Press. [find](https://scholar.google.com/scholar?q=Adele+E%2E+Goldberg%2E+2013%2E+Constructionist+approaches%2E+In+Thomas+Hoffman+and+Graeme+Trousadale%2C+editors%2C+The+Oxford+Handbook+of+Construction+Grammar%2E+Oxford+University+Press%2E)
14. Rui Liu, Junjie Hu, Wei Wei, Zi Yang, and Eric Nyberg. 2017. Structural embedding of syntactic trees for machine comprehension. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 815–824, Copenhagen, Denmark. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Rui+Liu%2C+Junjie+Hu%2C+Wei+Wei%2C+Zi+Yang%2C+and+Eric+Nyberg%2E+2017%2E+Structural+embedding+of+syntactic+trees+for+machine+comprehension%2E+In+Proceedings+of+the+2017+Conference+on+Empirical+Methods+in+Natural+La)
15. Benjamin Lyngfelt, Lars Borin, Markus Forsberg, Julia Prentice, Rudolf Rydstedt, Emma Sköldberg, and Sofia Tingsell. 2012. Adding a constructicon to the Swedish resource network of Språkbanken. In Proceedings of KONVENS 2012, pages 452–461. [find](https://scholar.google.com/scholar?q=Benjamin+Lyngfelt%2C+Lars+Borin%2C+Markus+Forsberg%2C+Julia+Prentice%2C+Rudolf+Rydstedt%2C+Emma+Sk%C3%B6ldberg%2C+and+Sofia+Tingsell%2E+2012%2E+Adding+a+constructicon+to+the+Swedish+resource+network+of+Spr%C3%A5kbanken%2E+In+Pro)
16. ÖGAI. LexSem 2012 workshop. [find](https://scholar.google.com/scholar?q=%C3%96GAI%2E+LexSem+2012+workshop%2E)
17. Benjamin Lyngfelt, Lars Borin, Kyoko Ohara, and Tiago Timponi Torrent. 2018. Constructicography: Constructicon development across languages. John Benjamins, Amsterdam. [find](https://scholar.google.com/scholar?q=Benjamin+Lyngfelt%2C+Lars+Borin%2C+Kyoko+Ohara%2C+and+Tiago+Timponi+Torrent%2E+2018%2E+Constructicography%3A+Constructicon+development+across+languages%2E+John+Benjamins%2C+Amsterdam%2E)
18. Ely Matos, Tiago Torrent, Vânia Almeida, Adrieli Laviola, Ludmila Lage, Natália Marção, and Tatiane Tavares. 2017. Constructional Analysis Using Constrained Spreading Activation in a FrameNet-Based Structured Connectionist Model. In The AAAI 2017 Spring Symposium on Computational Construction Gramma… [find](https://scholar.google.com/scholar?q=Ely+Matos%2C+Tiago+Torrent%2C+V%C3%A2nia+Almeida%2C+Adrieli+Laviola%2C+Ludmila+Lage%2C+Nat%C3%A1lia+Mar%C3%A7%C3%A3o%2C+and+Tatiane+Tavares%2E+2017%2E+Constructional+Analysis+Using+Constrained+Spreading+Activation+in+a+FrameNet%2DBased+St)
19. Ely Edison da Silva Matos and Maria Margarida Martins Salomão. 2014. Ludi: um framework para desambiguação lexical com base no enriquecimento da semântica de frames. Revista Linguística, 12(1). [find](https://scholar.google.com/scholar?q=Ely+Edison+da+Silva+Matos+and+Maria+Margarida+Martins+Salom%C3%A3o%2E+2014%2E+Ludi%3A+um+framework+para+desambigua%C3%A7%C3%A3o+lexical+com+base+no+enriquecimento+da+sem%C3%A2ntica+de+frames%2E+Revista+Lingu%C3%ADstica%2C+12%281%29%2E)
20. Ryan McDonald, Joakim Nivre, Yvonne Quirmbach-Brundage, Yoav Goldberg, Dipanjan Das, Kuzman Ganchev, Keith Hall, Slav Petrov, Hao Zhang, Oscar Täckström, et al. 2013. Universal dependency annotation for multilingual parsing. In Proceedings of the 51st Annual Meeting of the Association for Computatio… [our card](https://md.research.elamur.ai/free/en/papers/mcdonald-2013-universal-dependency-annotation-multilingual-parsing/)
21. Kyoko Hirose Ohara. 2014. Relating frames and constructions in Japanese FrameNet. In Proceedings of the 9th International Conference on Language Resources and Evaluation, LREC 2014, Proceedings of the 9th International Conference on Language Resources and Evaluation, LREC 2014, pages 2474– 2477. Eur… [find](https://scholar.google.com/scholar?q=Kyoko+Hirose+Ohara%2E+2014%2E+Relating+frames+and+constructions+in+Japanese+FrameNet%2E+In+Proceedings+of+the+9th+International+Conference+on+Language+Resources+and+Evaluation%2C+LREC+2014%2C+Proceedings+of+the)
22. Copyright: Copyright 2017 Elsevier B.V., All rights reserved.; 9th International Conference on Language Resources and Evaluation, LREC 2014 ; Conference date: 26-05-2014 Through 31-05-2014. [find](https://scholar.google.com/scholar?q=Copyright%3A+Copyright+2017+Elsevier+B%2EV%2E%2C+All+rights+reserved%2E%3B+9th+International+Conference+on+Language+Resources+and+Evaluation%2C+LREC+2014+%3B+Conference+date%3A+26%2D05%2D2014+Through+31%2D05%2D2014%2E)
23. Chi Sun, Xipeng Qiu, Yige Xu, and Xuanjing Huang. 2019. How to fine-tune bert for text classification? In China national conference on Chinese computational linguistics, pages 194–206. Springer. [our card](https://md.research.elamur.ai/free/en/papers/sun-2019-fine-tune-bert-text-classification/)