Multiple Admissibility in Language Learning: Judging Grammaticality Using Unlabeled Data

Forskningsoutput: Kapitel i bok/rapport/konferenshandlingKonferensbidragVetenskapligPeer review

Sammanfattning

We present our work on the problem of detection Multiple Admissibility (MA) in language learning. Multiple Admissibility occurs when more than one grammatical form of a word fits syntactically and semantically in a given context. In second-language education—in particular, in intelligent tutoring systems/computer-aided language learning (ITS/CALL), systems generate exercises automatically. MA implies that multiple alternative answers are possible. We treat the problem as a grammaticality judgement task. We train a neural network with an objective to label sentences as grammatical or ungrammatical, using a "simulated learner corpus": a dataset with correct text and with artificial errors, generated automatically. While MA occurs commonly in many languages, this paper focuses on learning Russian. We present a detailed classification of the types of constructions in Russian, in which MA is possible, and evaluate the model using a test set built from answers provided by users of the Revita language learning system.
Originalspråkengelska
Titel på gästpublikationThe 7th Workshop on Balto-Slavic Natural Language Processing : Proceedings of the Workshop
Antal sidor11
UtgivningsortFlorence
FörlagThe Association for Computational Linguistics
Utgivningsdatumaug 2019
Sidor12-22
ISBN (tryckt) 978-1-950737-41-3
StatusPublicerad - aug 2019
MoE-publikationstypA4 Artikel i en konferenspublikation
EvenemangThe 7th Workshop on
Balto-Slavic Natural Language Processing
: BSNLP
- , Italien
Varaktighet: 2 aug 20192 aug 2019
http://bsnlp.cs.helsinki.fi

Vetenskapsgrenar

  • 113 Data- och informationsvetenskap

Citera det här

Katinskaia, A., Ivanova, S., & Yangarber, R. (2019). Multiple Admissibility in Language Learning: Judging Grammaticality Using Unlabeled Data. I The 7th Workshop on Balto-Slavic Natural Language Processing: Proceedings of the Workshop (s. 12-22). Florence: The Association for Computational Linguistics.
Katinskaia, Anisia ; Ivanova, Sardana ; Yangarber, Roman. / Multiple Admissibility in Language Learning : Judging Grammaticality Using Unlabeled Data. The 7th Workshop on Balto-Slavic Natural Language Processing: Proceedings of the Workshop. Florence : The Association for Computational Linguistics, 2019. s. 12-22
@inproceedings{e390c6a25b5a4a69a5aa678c2f9a0337,
title = "Multiple Admissibility in Language Learning: Judging Grammaticality Using Unlabeled Data",
abstract = "We present our work on the problem of detection Multiple Admissibility (MA) in language learning. Multiple Admissibility occurs when more than one grammatical form of a word fits syntactically and semantically in a given context. In second-language education—in particular, in intelligent tutoring systems/computer-aided language learning (ITS/CALL), systems generate exercises automatically. MA implies that multiple alternative answers are possible. We treat the problem as a grammaticality judgement task. We train a neural network with an objective to label sentences as grammatical or ungrammatical, using a {"}simulated learner corpus{"}: a dataset with correct text and with artificial errors, generated automatically. While MA occurs commonly in many languages, this paper focuses on learning Russian. We present a detailed classification of the types of constructions in Russian, in which MA is possible, and evaluate the model using a test set built from answers provided by users of the Revita language learning system.",
keywords = "113 Computer and information sciences",
author = "Anisia Katinskaia and Sardana Ivanova and Roman Yangarber",
year = "2019",
month = "8",
language = "English",
isbn = "978-1-950737-41-3",
pages = "12--22",
booktitle = "The 7th Workshop on Balto-Slavic Natural Language Processing",
publisher = "The Association for Computational Linguistics",
address = "United States",

}

Katinskaia, A, Ivanova, S & Yangarber, R 2019, Multiple Admissibility in Language Learning: Judging Grammaticality Using Unlabeled Data. i The 7th Workshop on Balto-Slavic Natural Language Processing: Proceedings of the Workshop. The Association for Computational Linguistics, Florence, s. 12-22, The 7th Workshop on
Balto-Slavic Natural Language Processing
, Italien, 02/08/2019.

Multiple Admissibility in Language Learning : Judging Grammaticality Using Unlabeled Data. / Katinskaia, Anisia; Ivanova, Sardana; Yangarber, Roman.

The 7th Workshop on Balto-Slavic Natural Language Processing: Proceedings of the Workshop. Florence : The Association for Computational Linguistics, 2019. s. 12-22.

Forskningsoutput: Kapitel i bok/rapport/konferenshandlingKonferensbidragVetenskapligPeer review

TY - GEN

T1 - Multiple Admissibility in Language Learning

T2 - Judging Grammaticality Using Unlabeled Data

AU - Katinskaia, Anisia

AU - Ivanova, Sardana

AU - Yangarber, Roman

PY - 2019/8

Y1 - 2019/8

N2 - We present our work on the problem of detection Multiple Admissibility (MA) in language learning. Multiple Admissibility occurs when more than one grammatical form of a word fits syntactically and semantically in a given context. In second-language education—in particular, in intelligent tutoring systems/computer-aided language learning (ITS/CALL), systems generate exercises automatically. MA implies that multiple alternative answers are possible. We treat the problem as a grammaticality judgement task. We train a neural network with an objective to label sentences as grammatical or ungrammatical, using a "simulated learner corpus": a dataset with correct text and with artificial errors, generated automatically. While MA occurs commonly in many languages, this paper focuses on learning Russian. We present a detailed classification of the types of constructions in Russian, in which MA is possible, and evaluate the model using a test set built from answers provided by users of the Revita language learning system.

AB - We present our work on the problem of detection Multiple Admissibility (MA) in language learning. Multiple Admissibility occurs when more than one grammatical form of a word fits syntactically and semantically in a given context. In second-language education—in particular, in intelligent tutoring systems/computer-aided language learning (ITS/CALL), systems generate exercises automatically. MA implies that multiple alternative answers are possible. We treat the problem as a grammaticality judgement task. We train a neural network with an objective to label sentences as grammatical or ungrammatical, using a "simulated learner corpus": a dataset with correct text and with artificial errors, generated automatically. While MA occurs commonly in many languages, this paper focuses on learning Russian. We present a detailed classification of the types of constructions in Russian, in which MA is possible, and evaluate the model using a test set built from answers provided by users of the Revita language learning system.

KW - 113 Computer and information sciences

M3 - Conference contribution

SN - 978-1-950737-41-3

SP - 12

EP - 22

BT - The 7th Workshop on Balto-Slavic Natural Language Processing

PB - The Association for Computational Linguistics

CY - Florence

ER -

Katinskaia A, Ivanova S, Yangarber R. Multiple Admissibility in Language Learning: Judging Grammaticality Using Unlabeled Data. I The 7th Workshop on Balto-Slavic Natural Language Processing: Proceedings of the Workshop. Florence: The Association for Computational Linguistics. 2019. s. 12-22