On Language Spaces, Scales and Cross-Lingual Transfer of UD Parsers

Tanja Samardžić; Ximena Gutierrez-Vasques; Rob van der Goot; Max Müller-Eberstein; Olga Pelloni; Barbara Plank

doi:10.18653/v1/2022.conll-1.18

On Language Spaces, Scales and Cross-Lingual Transfer of UD Parsers

Tanja Samardžić, Ximena Gutierrez-Vasques, Rob van der Goot, Max Müller-Eberstein, Olga Pelloni, Barbara Plank

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

Abstract

Cross-lingual transfer of parsing models has been shown to work well for several closely-related languages, but predicting the success in other cases remains hard. Our study is a comprehensive analysis of the impact of linguistic distance on the transfer of UD parsers. As an alternative to syntactic typological distances extracted from URIEL, we propose three text-based feature spaces and show that they can be more precise predictors, especially on a more local scale, when only shorter distances are taken into account. Our analyses also reveal that the good coverage in typological databases is not among the factors that explain good transfer.

Original language	English
Title of host publication	Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL)
Number of pages	16
Volume	26
Publisher	Association for Computational Linguistics
Publication date	Dec 2023
Pages	266-281
DOIs	https://doi.org/10.18653/v1/2022.conll-1.18
Publication status	Published - Dec 2023
Event	The SIGNLL Conference on Computational Natural Language Learning - Abu Dhabi, United Arab Emirates Duration: 7 Dec 2022 → 8 Dec 2022 Conference number: 26 https://conll.org/

Conference

Conference	The SIGNLL Conference on Computational Natural Language Learning
Number	26
Country/Territory	United Arab Emirates
City	Abu Dhabi
Period	07/12/2022 → 08/12/2022
Internet address	https://conll.org/

Keywords

Natural Language Processing

Access to Document

10.18653/v1/2022.conll-1.18

2022.conll-1.18Final published version, 355 KBLicence: CC BY

https://aclanthology.org/2022.conll-1.18.pdfLicence: CC BY

Cite this

Samardžić, T., Gutierrez-Vasques, X., van der Goot, R., Müller-Eberstein, M., Pelloni, O., & Plank, B. (2023). On Language Spaces, Scales and Cross-Lingual Transfer of UD Parsers. In Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL) (Vol. 26, pp. 266-281). Association for Computational Linguistics. https://doi.org/10.18653/v1/2022.conll-1.18

@inproceedings{4f2e68b592414e0a9003fc3d742d781f,

title = "On Language Spaces, Scales and Cross-Lingual Transfer of UD Parsers",

abstract = "Cross-lingual transfer of parsing models has been shown to work well for several closely-related languages, but predicting the success in other cases remains hard. Our study is a comprehensive analysis of the impact of linguistic distance on the transfer of UD parsers. As an alternative to syntactic typological distances extracted from URIEL, we propose three text-based feature spaces and show that they can be more precise predictors, especially on a more local scale, when only shorter distances are taken into account. Our analyses also reveal that the good coverage in typological databases is not among the factors that explain good transfer.",

keywords = "Natural Language Processing, Natural Language Processing",

author = "Tanja Samard{\v z}i{\'c} and Ximena Gutierrez-Vasques and {van der Goot}, Rob and Max M{\"u}ller-Eberstein and Olga Pelloni and Barbara Plank",

year = "2023",

month = dec,

doi = "10.18653/v1/2022.conll-1.18",

language = "English",

volume = "26",

pages = "266--281",

booktitle = "Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL)",

publisher = "Association for Computational Linguistics",

address = "United States",

note = "The SIGNLL Conference on Computational Natural Language Learning, CoNLL ; Conference date: 07-12-2022 Through 08-12-2022",

url = "https://conll.org/",

}

Samardžić, T, Gutierrez-Vasques, X, van der Goot, R , Müller-Eberstein, M, Pelloni, O & Plank, B 2023, On Language Spaces, Scales and Cross-Lingual Transfer of UD Parsers. in Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL). vol. 26, Association for Computational Linguistics, pp. 266-281, The SIGNLL Conference on Computational Natural Language Learning, Abu Dhabi, United Arab Emirates, 07/12/2022. https://doi.org/10.18653/v1/2022.conll-1.18

On Language Spaces, Scales and Cross-Lingual Transfer of UD Parsers. / Samardžić, Tanja; Gutierrez-Vasques, Ximena; van der Goot, Rob et al.
Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL). Vol. 26 Association for Computational Linguistics, 2023. p. 266-281.

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

TY - GEN

T1 - On Language Spaces, Scales and Cross-Lingual Transfer of UD Parsers

AU - Samardžić, Tanja

AU - Gutierrez-Vasques, Ximena

AU - van der Goot, Rob

AU - Müller-Eberstein, Max

AU - Pelloni, Olga

AU - Plank, Barbara

N1 - Conference code: 26

PY - 2023/12

Y1 - 2023/12

N2 - Cross-lingual transfer of parsing models has been shown to work well for several closely-related languages, but predicting the success in other cases remains hard. Our study is a comprehensive analysis of the impact of linguistic distance on the transfer of UD parsers. As an alternative to syntactic typological distances extracted from URIEL, we propose three text-based feature spaces and show that they can be more precise predictors, especially on a more local scale, when only shorter distances are taken into account. Our analyses also reveal that the good coverage in typological databases is not among the factors that explain good transfer.

AB - Cross-lingual transfer of parsing models has been shown to work well for several closely-related languages, but predicting the success in other cases remains hard. Our study is a comprehensive analysis of the impact of linguistic distance on the transfer of UD parsers. As an alternative to syntactic typological distances extracted from URIEL, we propose three text-based feature spaces and show that they can be more precise predictors, especially on a more local scale, when only shorter distances are taken into account. Our analyses also reveal that the good coverage in typological databases is not among the factors that explain good transfer.

KW - Natural Language Processing

U2 - 10.18653/v1/2022.conll-1.18

DO - 10.18653/v1/2022.conll-1.18

M3 - Article in proceedings

VL - 26

SP - 266

EP - 281

BT - Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL)

PB - Association for Computational Linguistics

T2 - The SIGNLL Conference on Computational Natural Language Learning

Y2 - 7 December 2022 through 8 December 2022

ER -

On Language Spaces, Scales and Cross-Lingual Transfer of UD Parsers

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this