Colorless Green Recurrent Networks Dream Hierarchically. Gulordava, K., Bojanowski, P., Grave, E., Linzen, T., & Baroni, M. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers), pages 1195–1205, Stroudsburg, PA, USA, 2018. Association for Computational Linguistics. ![Colorless Green Recurrent Networks Dream Hierarchically [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Recurrent neural networks (RNNs) have achieved impressive results in a variety of linguistic processing tasks, suggesting that they can induce non-trivial properties of language. We investigate here to what extent RNNs learn to track abstract hierarchical syntactic structure. We test whether RNNs trained with a generic language modeling objective in four languages (Italian, English, Hebrew, Russian) can predict long-distance number agreement in various constructions. We include in our evaluation nonsensical sentences where RNNs cannot rely on semantic or lexical cues ("The colorless green ideas I ate with the chair sleep furiously"), and, for Italian, we compare model performance to human intuitions. Our language-model-trained RNNs make reliable predictions about long-distance agreement, and do not lag much behind human performance. We thus bring support to the hypothesis that RNNs are not just shallow-pattern extractors, but they also acquire deeper grammatical competence.
@inproceedings{Gulordava2018,
abstract = {Recurrent neural networks (RNNs) have achieved impressive results in a variety of linguistic processing tasks, suggesting that they can induce non-trivial properties of language. We investigate here to what extent RNNs learn to track abstract hierarchical syntactic structure. We test whether RNNs trained with a generic language modeling objective in four languages (Italian, English, Hebrew, Russian) can predict long-distance number agreement in various constructions. We include in our evaluation nonsensical sentences where RNNs cannot rely on semantic or lexical cues ("The colorless green ideas I ate with the chair sleep furiously"), and, for Italian, we compare model performance to human intuitions. Our language-model-trained RNNs make reliable predictions about long-distance agreement, and do not lag much behind human performance. We thus bring support to the hypothesis that RNNs are not just shallow-pattern extractors, but they also acquire deeper grammatical competence.},
address = {Stroudsburg, PA, USA},
archivePrefix = {arXiv},
arxivId = {1803.11138},
author = {Gulordava, Kristina and Bojanowski, Piotr and Grave, Edouard and Linzen, Tal and Baroni, Marco},
booktitle = {Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers)},
doi = {10.18653/v1/N18-1108},
eprint = {1803.11138},
file = {:Users/shanest/Documents/Library/Gulordava et al/Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics Human Language Technologie./Gulordava et al. - 2018 - Colorless Green Recurrent Networks Dream Hierarchically.pdf:pdf},
keywords = {method: cross-linguistic,method: nonsense,model,phenomenon: number agreement},
pages = {1195--1205},
publisher = {Association for Computational Linguistics},
title = {{Colorless Green Recurrent Networks Dream Hierarchically}},
url = {http://aclweb.org/anthology/N18-1108},
year = {2018}
}
Downloads: 0
{"_id":"dnDffNA8TiAnqtZB6","bibbaseid":"gulordava-bojanowski-grave-linzen-baroni-colorlessgreenrecurrentnetworksdreamhierarchically-2018","authorIDs":[],"author_short":["Gulordava, K.","Bojanowski, P.","Grave, E.","Linzen, T.","Baroni, M."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","abstract":"Recurrent neural networks (RNNs) have achieved impressive results in a variety of linguistic processing tasks, suggesting that they can induce non-trivial properties of language. We investigate here to what extent RNNs learn to track abstract hierarchical syntactic structure. We test whether RNNs trained with a generic language modeling objective in four languages (Italian, English, Hebrew, Russian) can predict long-distance number agreement in various constructions. We include in our evaluation nonsensical sentences where RNNs cannot rely on semantic or lexical cues (\"The colorless green ideas I ate with the chair sleep furiously\"), and, for Italian, we compare model performance to human intuitions. Our language-model-trained RNNs make reliable predictions about long-distance agreement, and do not lag much behind human performance. We thus bring support to the hypothesis that RNNs are not just shallow-pattern extractors, but they also acquire deeper grammatical competence.","address":"Stroudsburg, PA, USA","archiveprefix":"arXiv","arxivid":"1803.11138","author":[{"propositions":[],"lastnames":["Gulordava"],"firstnames":["Kristina"],"suffixes":[]},{"propositions":[],"lastnames":["Bojanowski"],"firstnames":["Piotr"],"suffixes":[]},{"propositions":[],"lastnames":["Grave"],"firstnames":["Edouard"],"suffixes":[]},{"propositions":[],"lastnames":["Linzen"],"firstnames":["Tal"],"suffixes":[]},{"propositions":[],"lastnames":["Baroni"],"firstnames":["Marco"],"suffixes":[]}],"booktitle":"Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers)","doi":"10.18653/v1/N18-1108","eprint":"1803.11138","file":":Users/shanest/Documents/Library/Gulordava et al/Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics Human Language Technologie./Gulordava et al. - 2018 - Colorless Green Recurrent Networks Dream Hierarchically.pdf:pdf","keywords":"method: cross-linguistic,method: nonsense,model,phenomenon: number agreement","pages":"1195–1205","publisher":"Association for Computational Linguistics","title":"Colorless Green Recurrent Networks Dream Hierarchically","url":"http://aclweb.org/anthology/N18-1108","year":"2018","bibtex":"@inproceedings{Gulordava2018,\nabstract = {Recurrent neural networks (RNNs) have achieved impressive results in a variety of linguistic processing tasks, suggesting that they can induce non-trivial properties of language. We investigate here to what extent RNNs learn to track abstract hierarchical syntactic structure. We test whether RNNs trained with a generic language modeling objective in four languages (Italian, English, Hebrew, Russian) can predict long-distance number agreement in various constructions. We include in our evaluation nonsensical sentences where RNNs cannot rely on semantic or lexical cues (\"The colorless green ideas I ate with the chair sleep furiously\"), and, for Italian, we compare model performance to human intuitions. Our language-model-trained RNNs make reliable predictions about long-distance agreement, and do not lag much behind human performance. We thus bring support to the hypothesis that RNNs are not just shallow-pattern extractors, but they also acquire deeper grammatical competence.},\naddress = {Stroudsburg, PA, USA},\narchivePrefix = {arXiv},\narxivId = {1803.11138},\nauthor = {Gulordava, Kristina and Bojanowski, Piotr and Grave, Edouard and Linzen, Tal and Baroni, Marco},\nbooktitle = {Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers)},\ndoi = {10.18653/v1/N18-1108},\neprint = {1803.11138},\nfile = {:Users/shanest/Documents/Library/Gulordava et al/Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics Human Language Technologie./Gulordava et al. - 2018 - Colorless Green Recurrent Networks Dream Hierarchically.pdf:pdf},\nkeywords = {method: cross-linguistic,method: nonsense,model,phenomenon: number agreement},\npages = {1195--1205},\npublisher = {Association for Computational Linguistics},\ntitle = {{Colorless Green Recurrent Networks Dream Hierarchically}},\nurl = {http://aclweb.org/anthology/N18-1108},\nyear = {2018}\n}\n","author_short":["Gulordava, K.","Bojanowski, P.","Grave, E.","Linzen, T.","Baroni, M."],"key":"Gulordava2018","id":"Gulordava2018","bibbaseid":"gulordava-bojanowski-grave-linzen-baroni-colorlessgreenrecurrentnetworksdreamhierarchically-2018","role":"author","urls":{"Paper":"http://aclweb.org/anthology/N18-1108"},"keyword":["method: cross-linguistic","method: nonsense","model","phenomenon: number agreement"],"metadata":{"authorlinks":{}},"downloads":0},"bibtype":"inproceedings","biburl":"https://www.shane.st/teaching/575/win20/MachineLearning-interpretability.bib","creationDate":"2020-01-05T04:04:02.901Z","downloads":0,"keywords":["method: cross-linguistic","method: nonsense","model","phenomenon: number agreement"],"search_terms":["colorless","green","recurrent","networks","dream","hierarchically","gulordava","bojanowski","grave","linzen","baroni"],"title":"Colorless Green Recurrent Networks Dream Hierarchically","year":2018,"dataSources":["okYcdTpf4JJ2zkj7A","znj7izS5PeehdLR3G"]}