Vol. 7:2 (2021) ► pp.259–274
fsca
French syntactic complexity analyzer [*] *
This article reports on an open-source R package for the extraction of syntactic units from dependency-parsed French texts. To evaluate the reliability of the package, syntactic units were extracted from a corpus of L2 French and were compared to units extracted manually from the same corpus. The f-score of the extracted units ranged from 0.53–0.97. Although units were not always identical between the two methods, manual and automatically-derived syntactic complexity measures were strongly and significantly correlated (ρ = 0.62–0.97, p < 0.001), suggesting that this package may be a suitable replacement for manual annotation in some cases where manual annotation is not possible but that care should be used in interpreting the measures based on these units.
Article outline
- 1.Introduction
- 2.Methodology
- 2.1Manual annotation
- 2.2Automatic extraction of syntactic units
- 3.Results
- 3.1Precision and recall of automatically identified units
- 3.2Correlation between manual and automatic methods
- 3.3Sources of error
- 4.Discussion and conclusion
- Disclosures
- Acknowledgements
- Notes
-
References
https://doi.org/10.1075/ijlcr.20018.van