| { |
| "paper_id": "2022", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T15:21:07.589021Z" |
| }, |
| "title": "Tweaking UD annotations to investigate the placement of determiners, quantifiers and numerals in the noun phrase", |
| "authors": [ |
| { |
| "first": "Luigi", |
| "middle": [], |
| "last": "Talamo", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Saarland University", |
| "location": {} |
| }, |
| "email": "luigi.talamo@uni-saarland.de" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "We describe a methodology to extract with finer accuracy word order patterns from texts automatically annotated with Universal Dependency (UD) trained parsers. We use the methodology to quantify the word order entropy of determiners, quantifiers and numerals in ten Indo-European languages, using UDparsed texts from a parallel corpus of prosaic texts. Our results suggest that the combinations of different UD annotation layers, such as UD Relations, Universal Parts of Speech and lemma, and the introduction of languagespecific lists of closed-category lemmata has the twofold effect of improving the quality of analysis and unveiling hidden areas of variability in word order patterns.", |
| "pdf_parse": { |
| "paper_id": "2022", |
| "_pdf_hash": "", |
| "abstract": [ |
| { |
| "text": "We describe a methodology to extract with finer accuracy word order patterns from texts automatically annotated with Universal Dependency (UD) trained parsers. We use the methodology to quantify the word order entropy of determiners, quantifiers and numerals in ten Indo-European languages, using UDparsed texts from a parallel corpus of prosaic texts. Our results suggest that the combinations of different UD annotation layers, such as UD Relations, Universal Parts of Speech and lemma, and the introduction of languagespecific lists of closed-category lemmata has the twofold effect of improving the quality of analysis and unveiling hidden areas of variability in word order patterns.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Abstract", |
| "sec_num": null |
| } |
| ], |
| "body_text": [ |
| { |
| "text": "Most of the work on word order variation using Universal Dependencies (UD: de Marneffe et al., 2021) is based on curated dependency treebanks, with only a few works using dependency corpora derived from raw texts. Although the accuracy rate of NLP systems trained on UD models is reportedly very high (Haji\u010d and Zeman, 2017; Zeman and Haji\u010d, 2018; Straka et al., 2019; Qi et al., 2020) , a certain level of noise i.e., erroneous annotations is in fact present when working with automatically annotated texts (Levshina et al. , to appear; Talamo and Verkerk, to appear); furthermore, different layers of UD annotations such as Universal Parts of Speech (UPOS) and UD Relations are not always used consistently across languages, often resulting in the cross-linguistic comparison of different categories.", |
| "cite_spans": [ |
| { |
| "start": 301, |
| "end": 324, |
| "text": "(Haji\u010d and Zeman, 2017;", |
| "ref_id": "BIBREF7" |
| }, |
| { |
| "start": 325, |
| "end": 347, |
| "text": "Zeman and Haji\u010d, 2018;", |
| "ref_id": "BIBREF19" |
| }, |
| { |
| "start": 348, |
| "end": 368, |
| "text": "Straka et al., 2019;", |
| "ref_id": "BIBREF16" |
| }, |
| { |
| "start": 369, |
| "end": 385, |
| "text": "Qi et al., 2020)", |
| "ref_id": "BIBREF14" |
| }, |
| { |
| "start": 508, |
| "end": 524, |
| "text": "(Levshina et al.", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "We discuss a methodology to tweak the UD annotations in order to achieve a better representation of word order entropy; the methodology is exemplified on three categories that are particularly difficult to analyze with automatic methods and from a cross-linguistic perspective. Determiners, quantifiers and numerals are often treated in descriptive grammars as heterogeneous categories; the lexical category of determiners includes articles and demonstratives, while the category of quantifiers often includes elements of other closed categories, such as pronouns and gradation markers, and sometimes members of open categories, such as adjectives and adverbs; finally, numerals are often not restricted to cardinal, ordinal and distributive numbers, but overlap with quantifiers.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "This heterogeneity is reflected by the UD implementation of these categories, both at the Relation and the UPOS annotation layer. Numerals are treated as a separate category and represented at the syntactic level by the nummod UD Relation and at word category level by the NUM UPOS; by contrast, the UD framework conflates articles, demonstratives and quantifiers into one UPOS tag (DET) and into one UD Relation (det), resulting in the 'Determiners & Quantifiers' macro-category. At the language-specific level several individual POS tags and UD Relation subtypes are used; for instance, in Slavic languages quantifiers get two specific subtypes, det:numgov and det:numposs, and morpho-syntactic features of numerals can be specified using additional UD Relation subtypes.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "Our methodology combines two layers of UD annotations, UPOS and UD Relations, with manuallycompiled and language-specific lists of lemmata. We test the methodology against a parallel corpus of fiction texts and their translations in 10 Indo-European languages; given their particular genre, these texts are quite challenging for parsers that are mostly trained on non-fiction data such as Wiki and News. Following previous studies, we employ here Shannon's entropy as a metric for word order variation.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "Since its inception in 2015 (Nivre et al., 2015) , UD has been widely used in corpus-based studies on word order variability. However, as earlier mentioned, corpora used in previous studies are \"dependency corpora of the HamleDT 2.0 and Universal Dependencies 1.00\" (Futrell et al., 2015) , \"the Universal Dependencies Treebank version 2.2\" (Naranjo and Becker, 2018) , \"a selection of 55 treebanks from Universal Dependencies v2.4\" (Yu et al., 2019) , \"Surface-Syntactic Universal Dependencies (SUD) [treebanks]\" (Gerdes et al., 2019) and the \"Universal Dependencies project, release 2.1\" (Futrell et al., 2020) . UD Treebanks can be considered de-facto gold standards, as large parts of them are manually compiled or at least semiautomatically checked for wrong annotations, allowing scholars to work with high quality of data. However, as UD Treebanks wildly vary across languages with respect to size and text genres (Levshina et al., to appear), results from most of the previous works are biased against these factors. Exceptions are represented by works using the LISCA parse assessment algorithm (Dell'Orletta et al., 2013), whose models have been trained on UD-parsed Wikipedia corpora (Alzetta et al., 2018) and tested on the so-called 'reference corpora', which consist \"of a monolingual corpus of texts from the news and Wikipedia domains [...] morpho-syntactically annotated and dependency parsed by the UDPipe pipeline trained on the Universal Dependency treebanks, version 2.2\" (Alzetta et al., 2020) ; automatically annotated texts are also partially employed in Levshina (2019) , who uses eleven UD-parsed corpora from the Leipzig Corpora Collection for one of her case-studies on word order entropy. Finally, Talamo and Verkerk (to appear) is a study on word order variation in the nominal phrase and is entirely based on parallel texts that are parsed by the UDPipe pipeline trained on UD treebanks v.~2.5. To the best of our knowledge, Levshina (2019) and Talamo and Verkerk (to appear) are the only studies on word order variation combining UD Relations with other annotation layers; although her methodology is not fully disclosed, Levshina (2019) applies the UPOS annotation layer to the head in her first case study, where word order variability is taken from a syntactic perspective, and the wordform annotation layer to the dependent in her second case-study, where word order variability is investigated with respect to the lexically specific level; Talamo and Verkerk (to appear) take a step further and operationalize this methodology by introducing several combinations of UD Relation and UPOS annotation layer to restrict either the head, the dependent or both, and introducing language-specific list of lemmata to match modifiers at the lexical level.", |
| "cite_spans": [ |
| { |
| "start": 28, |
| "end": 48, |
| "text": "(Nivre et al., 2015)", |
| "ref_id": null |
| }, |
| { |
| "start": 266, |
| "end": 288, |
| "text": "(Futrell et al., 2015)", |
| "ref_id": "BIBREF5" |
| }, |
| { |
| "start": 341, |
| "end": 367, |
| "text": "(Naranjo and Becker, 2018)", |
| "ref_id": "BIBREF12" |
| }, |
| { |
| "start": 433, |
| "end": 450, |
| "text": "(Yu et al., 2019)", |
| "ref_id": "BIBREF18" |
| }, |
| { |
| "start": 514, |
| "end": 535, |
| "text": "(Gerdes et al., 2019)", |
| "ref_id": "BIBREF6" |
| }, |
| { |
| "start": 590, |
| "end": 612, |
| "text": "(Futrell et al., 2020)", |
| "ref_id": "BIBREF4" |
| }, |
| { |
| "start": 1195, |
| "end": 1217, |
| "text": "(Alzetta et al., 2018)", |
| "ref_id": "BIBREF1" |
| }, |
| { |
| "start": 1493, |
| "end": 1515, |
| "text": "(Alzetta et al., 2020)", |
| "ref_id": null |
| }, |
| { |
| "start": 1579, |
| "end": 1594, |
| "text": "Levshina (2019)", |
| "ref_id": "BIBREF10" |
| }, |
| { |
| "start": 1956, |
| "end": 1971, |
| "text": "Levshina (2019)", |
| "ref_id": "BIBREF10" |
| }, |
| { |
| "start": 1976, |
| "end": 2006, |
| "text": "Talamo and Verkerk (to appear)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Related work", |
| "sec_num": "2" |
| }, |
| { |
| "text": "3 Data and Methods", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Related work", |
| "sec_num": "2" |
| }, |
| { |
| "text": "We use the Parallel Corpus of Indo-European Prose and more (CIEP+: Talamo and Verkerk, to appear), which features prosaic texts and their translations in more than 30 languages; we select a sample of 10 Indo-European languages, belonging to the following branches: Balto-Slavic (Lithuanian, Polish), Celtic (Irish), Germanic (Danish, Dutch and German), Greek (Modern Greek), Romance (French, Portuguese and Spanish). All languages feature 18 books (approximately 120K of sentences for each language), with the exception of Irish that features 5 books (approximately 13K of sentences).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Corpus", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "The corpus has been parsed using Stanford Stanza 1 with pre-trained UD Models 2 for the 10 languages. The resulting CoNLL-U files are processed with a Python script using the pyconll library 3 . The script extracts the occurrences of the specific UD Relations (see below) and determines the relative position of head and dependent; for each occurrence, we collect the following annotation fields for both the head and the dependent: UD Relations, UPOS tag and lemma.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Corpus", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "Scripts and dataset, with the exception of the parsed corpus containing copyrighted texts, are available in the Supplementary Material 4 .", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Corpus", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "Working with a dependency grammar, the most important annotation layer is represented by the UD Relations, which identifies the head and the dependent within the phrase; for our case study, we deal with various dependents (Table 1 ) and one type of head, the noun. This basic layer of annotation is then combined with other layers of annotation. By formulating the categories in Table 1 and Universal Dependencies' in the Supplementary Material).", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 222, |
| "end": 230, |
| "text": "(Table 1", |
| "ref_id": "TABREF1" |
| }, |
| { |
| "start": 379, |
| "end": 386, |
| "text": "Table 1", |
| "ref_id": "TABREF1" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Tweaking the UD annotations", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "Elaborating on Talamo and Verkerk (to appear), we propose three combinations, with each combination building on top of the previous one. The first combination, rel, uses the specific UD relation for the dependent, thus corresponding to most of the approaches taken in previous works; the second combination, rel+upos, adds the UPOS layers, using specific UPOS tags for the nominal head, and specific and non-specific UPOS tags for the dependent; the third combination, rel+pos+lemma, introduces language-specific list of lemmata for the dependent.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Tweaking the UD annotations", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "With 'specific' UPOS tags we refer to the values that are described by the UD Annotation Guidelines 5 as relevant for the investigated categories; we additionally add non-specific UPOS tags, which are based on the consultation of descriptive grammars and on the comparison of UD-parsed texts across the ten languages of the sample. See Table 1 for a detailed list of these values. Finally, language-specific lists of lemmata are aimed to capture the three components of the 'Determiners & Quantifiers' macro-category, namely, articles, demonstratives and quantifiers; we use these lists of lemmata, which are compiled using descriptive grammars and with the aid of native speakers, in intersective queries (positive match) on the lemma field for articles, demonstratives and quantifiers, and in non-intersective queries (negative match) on the lemma field for numerals.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Tweaking the UD annotations", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "Word order variability is assessed using Shannon's entropy:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Metrics", |
| "sec_num": "3.3" |
| }, |
| { |
| "text": "H(X) = \u2212 n i=1 P (x i ) log 2 P (x i )", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Metrics", |
| "sec_num": "3.3" |
| }, |
| { |
| "text": "where the upper bound of summation, n, is set to 2, indicating that there are two possible word order patterns i.e., the modifier is either prenominal or postnominal, and P represents the probability of the two order patterns; the resulting value of entropy is given in bits and ranges from 0 (only one of the two word order patterns is attested) to 1 (both word order patterns are attested with the same probability). For instance, we find in the Danish part of CIEP+ 5089 occurences of postnominal nummod (and subtypes) and 1392 of prenominal nummod (and subtypes), with a resulting entropy of .75.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Metrics", |
| "sec_num": "3.3" |
| }, |
| { |
| "text": "As shown in the left panel of Figure 1 , entropy values captured by the second combination, rel+upos, are significantly lower than values captured by the rel combination. This is particularly clear for numerals, which display a substantial drop in entropy in half of the languages; furthermore, in some languages the entropy of numerals is reduced to near-zero values (German: .02, French and Irish: .04, Polish: .06). As for determiners & quantifiers, the introduction of the UPOS layer is overall less significant; a significant reduction of entropy is observed only for three languages (Greek, Lithuanian and Spanish). This is partly due to the low value of entropy already captured by the rel combination, but it also reflects the biunivocal relation between the DET UPOS tag det and the UD Relation and its subtypes; by contrast, the nummod UD relation and its subtypes are not in biunivocal relation with the NUM UPOS tag, since, as already mentioned above, this UD Relation is often used with quantifiers as well. The third combination, which adds language-specific lists of lemmata to the UPOS and UD Relation layers and is plotted in the right panel of Figure 1 , allows us to zoom in on the entropy of determiners & quantifiers, disentangling the category into articles, demonstratives and quantifiers; although the entropy of determiners & quantifiers is extremely low for all languages of the sample, two languages have moderate values of entropy for demonstratives or quantifiers. More specifically, Irish has the highest value of entropy for quantifiers (.71), while Greek the highest value for demonstratives (.59). According to Stenson (2020, 189-192) , the position of quantifiers in Irish is lexically determined; most Irish quantifiers precede the noun, while few follow it; postnominal quantifiers include the high-frequency lexeme uilig 'all', which explains the high entropy of Irish quantifiers. As for Greek, the high value of entropy for demonstratives can be accounted on a pragmatic and semantic basis, as postnominal demonstratives have an emphatic reading (Lascaratou, 1998, 164) . Furthermore, low-to-moderate values of entropy are observed for quantifiers in Portuguese (.30) and Spanish (.10) and for demonstratives in Portuguese (.15). As for numerals, we use language-specific lists of quantifiers as negative matches against the lemma field; this approach is however of little use, as entropy values captured by the third combination are the same of the second combination. Thanks to the introduction of language-specific list of lemmata, the third combination is suitable for closed categories such as articles, demonstratives and quantifiers, while the second combination is already effective for capturing open categories such as numerals.", |
| "cite_spans": [ |
| { |
| "start": 1644, |
| "end": 1667, |
| "text": "Stenson (2020, 189-192)", |
| "ref_id": null |
| }, |
| { |
| "start": 2085, |
| "end": 2108, |
| "text": "(Lascaratou, 1998, 164)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 30, |
| "end": 38, |
| "text": "Figure 1", |
| "ref_id": "FIGREF0" |
| }, |
| { |
| "start": 1162, |
| "end": 1170, |
| "text": "Figure 1", |
| "ref_id": "FIGREF0" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Results", |
| "sec_num": "4" |
| }, |
| { |
| "text": "We have discussed a methodology to extract with better accuracy word order patterns from CoNLL-U files obtained from the automatic parsing of raw texts; the methodology, which exploits different UD annotation layers and language-specific list of lemmata, is exemplified on the heterogenous lexical categories of determiners and numerals, whose word order patterns are analyzed in a parallel corpus of 10 Indo-European languages. We have shown that the methodology is able to correct some of the errors introduced by the automatic parsing and inconsistent use of UPOS tags and UD Relations, thus improving the quality of the analysis, as shown with the category of numerals. Furthermore, the methodology sheds light on areas of variability, which were previously hidden by the UD lumping of articles, demonstratives and quantifiers into a unitary category. Given the very high frequency of articles, whose variability is close to zero, this unitary category displays very low values of entropy across languages; once this unitary category is split into its three components, some languages show moderate-to-high levels of entropy with respect to demonstratives (Greek) and quantifiers (Irish and Portuguese).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Conclusion", |
| "sec_num": "5" |
| }, |
| { |
| "text": "Version 1.3.0 https://stanfordnlp.github. io/stanza/ 2 Version 2.8 https://stanfordnlp.github.io/ stanza/available_models.html 3 https://pyconll.readthedocs.io/en/ stable/ 4 https://doi.org/10.5281/zenodo. 6580701", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "https://universaldependencies.org/ guidelines.html", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| } |
| ], |
| "back_matter": [], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "Simonetta Montemagni, Petya Osenova, Kiril Simov, and Giulia Venturi. 2020. Quantitative linguistic investigations across universal dependencies treebanks", |
| "authors": [ |
| { |
| "first": "Chiara", |
| "middle": [], |
| "last": "Alzetta", |
| "suffix": "" |
| }, |
| { |
| "first": "Felice", |
| "middle": [], |
| "last": "Dell'orletta", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "CLiCit", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Chiara Alzetta, Felice Dell'Orletta, Simonetta Mon- temagni, Petya Osenova, Kiril Simov, and Giulia Venturi. 2020. Quantitative linguistic investigations across universal dependencies treebanks. In CLiC- it.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "Universal Dependencies and quantitative typological trends. a case study on word order", |
| "authors": [ |
| { |
| "first": "Chiara", |
| "middle": [], |
| "last": "Alzetta", |
| "suffix": "" |
| }, |
| { |
| "first": "Felice", |
| "middle": [], |
| "last": "Dell'orletta", |
| "suffix": "" |
| }, |
| { |
| "first": "Simonetta", |
| "middle": [], |
| "last": "Montemagni", |
| "suffix": "" |
| }, |
| { |
| "first": "Giulia", |
| "middle": [], |
| "last": "Venturi", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Chiara Alzetta, Felice Dell'Orletta, Simonetta Monte- magni, and Giulia Venturi. 2018. Universal Depen- dencies and quantitative typological trends. a case study on word order. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. Eu- ropean Language Resources Association (ELRA).", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "Universal Dependencies", |
| "authors": [ |
| { |
| "first": "Marie-Catherine", |
| "middle": [], |
| "last": "De Marneffe", |
| "suffix": "" |
| }, |
| { |
| "first": "Christopher", |
| "middle": [ |
| "D" |
| ], |
| "last": "Manning", |
| "suffix": "" |
| }, |
| { |
| "first": "Joakim", |
| "middle": [], |
| "last": "Nivre", |
| "suffix": "" |
| }, |
| { |
| "first": "Daniel", |
| "middle": [], |
| "last": "Zeman", |
| "suffix": "" |
| } |
| ], |
| "year": 2021, |
| "venue": "Computational Linguistics", |
| "volume": "47", |
| "issue": "2", |
| "pages": "255--308", |
| "other_ids": { |
| "DOI": [ |
| "10.1162/coli_a_00402" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Marie-Catherine de Marneffe, Christopher D. Man- ning, Joakim Nivre, and Daniel Zeman. 2021. Uni- versal Dependencies. Computational Linguistics, 47(2):255-308.", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "Linguistically-driven Selection of Correct Arcs for Dependency Parsing", |
| "authors": [ |
| { |
| "first": "Felice", |
| "middle": [], |
| "last": "Dell'orletta", |
| "suffix": "" |
| }, |
| { |
| "first": "Giulia", |
| "middle": [], |
| "last": "Venturi", |
| "suffix": "" |
| }, |
| { |
| "first": "Simonetta", |
| "middle": [], |
| "last": "Montemagni", |
| "suffix": "" |
| } |
| ], |
| "year": 2013, |
| "venue": "Computaci\u00f3n y Sistemas", |
| "volume": "17", |
| "issue": "", |
| "pages": "125--136", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Felice Dell'Orletta, Giulia Venturi, and Simonetta Montemagni. 2013. Linguistically-driven Selection of Correct Arcs for Dependency Parsing. Com- putaci\u00f3n y Sistemas, 17:125 -136.", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "Dependency locality as an explanatory principle for word order", |
| "authors": [ |
| { |
| "first": "Richard", |
| "middle": [], |
| "last": "Futrell", |
| "suffix": "" |
| }, |
| { |
| "first": "Roger", |
| "middle": [ |
| "P" |
| ], |
| "last": "Levy", |
| "suffix": "" |
| }, |
| { |
| "first": "Edward", |
| "middle": [], |
| "last": "Gibson", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Language", |
| "volume": "96", |
| "issue": "2", |
| "pages": "371--412", |
| "other_ids": { |
| "DOI": [ |
| "10.1353/lan.2020.0024" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Richard Futrell, Roger P. Levy, and Edward Gibson. 2020. Dependency locality as an explanatory prin- ciple for word order. Language, 96(2):371-412.", |
| "links": null |
| }, |
| "BIBREF5": { |
| "ref_id": "b5", |
| "title": "Quantifying word order freedom in dependency corpora", |
| "authors": [ |
| { |
| "first": "Richard", |
| "middle": [], |
| "last": "Futrell", |
| "suffix": "" |
| }, |
| { |
| "first": "Kyle", |
| "middle": [], |
| "last": "Mahowald", |
| "suffix": "" |
| }, |
| { |
| "first": "Edward", |
| "middle": [], |
| "last": "Gibson", |
| "suffix": "" |
| } |
| ], |
| "year": 2015, |
| "venue": "Proceedings of the Third International Conference on Dependency Linguistics", |
| "volume": "", |
| "issue": "", |
| "pages": "91--100", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Richard Futrell, Kyle Mahowald, and Edward Gibson. 2015. Quantifying word order freedom in depen- dency corpora. In Proceedings of the Third In- ternational Conference on Dependency Linguistics (Depling 2015), pages 91-100, Uppsala, Sweden. Uppsala University, Uppsala, Sweden.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "Rediscovering greenberg's word order universals in UD", |
| "authors": [ |
| { |
| "first": "Kim", |
| "middle": [], |
| "last": "Gerdes", |
| "suffix": "" |
| }, |
| { |
| "first": "Xinying", |
| "middle": [], |
| "last": "Sylvain Kahane", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Chen", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the Third Workshop on Universal Dependencies", |
| "volume": "", |
| "issue": "", |
| "pages": "124--131", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/W19-8015" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kim Gerdes, Sylvain Kahane, and Xinying Chen. 2019. Rediscovering greenberg's word order universals in UD. In Proceedings of the Third Workshop on Universal Dependencies (UDW, SyntaxFest 2019), pages 124-131, Paris, France. Association for Com- putational Linguistics.", |
| "links": null |
| }, |
| "BIBREF7": { |
| "ref_id": "b7", |
| "title": "Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies. Association for Computational Linguistics", |
| "authors": [ |
| { |
| "first": "Jan", |
| "middle": [], |
| "last": "Haji\u010d", |
| "suffix": "" |
| }, |
| { |
| "first": "Dan", |
| "middle": [], |
| "last": "Zeman", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/K17-3" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Jan Haji\u010d and Dan Zeman, editors. 2017. Proceedings of the CoNLL 2017 Shared Task: Multilingual Pars- ing from Raw Text to Universal Dependencies. As- sociation for Computational Linguistics, Vancouver, Canada.", |
| "links": null |
| }, |
| "BIBREF8": { |
| "ref_id": "b8", |
| "title": "Comparative concepts and descriptive categories in cross-linguistic studies", |
| "authors": [ |
| { |
| "first": "Martin", |
| "middle": [], |
| "last": "Haspelmath", |
| "suffix": "" |
| } |
| ], |
| "year": 2010, |
| "venue": "Language", |
| "volume": "", |
| "issue": "4", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Martin Haspelmath. 2010. Comparative concepts and descriptive categories in cross-linguistic studies. Language, 86(4).", |
| "links": null |
| }, |
| "BIBREF9": { |
| "ref_id": "b9", |
| "title": "Basic characteristics of modern greek word order", |
| "authors": [ |
| { |
| "first": "Chryssoula", |
| "middle": [], |
| "last": "Lascaratou", |
| "suffix": "" |
| } |
| ], |
| "year": 1998, |
| "venue": "Constituent Order in the Language of Europe", |
| "volume": "", |
| "issue": "", |
| "pages": "151--171", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Chryssoula Lascaratou. 1998. Basic characteristics of modern greek word order. In Anna Siewierska, ed- itor, Constituent Order in the Language of Europe, pages 151-171. Mouton de Gruyter, Berlin.", |
| "links": null |
| }, |
| "BIBREF10": { |
| "ref_id": "b10", |
| "title": "Token-based typology and word order entropy: A study based on universal dependencies", |
| "authors": [ |
| { |
| "first": "Natalia", |
| "middle": [], |
| "last": "Levshina", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Linguistic Typology", |
| "volume": "23", |
| "issue": "3", |
| "pages": "533--572", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Natalia Levshina. 2019. Token-based typology and word order entropy: A study based on universal de- pendencies. Linguistic Typology, 23(3):533-572.", |
| "links": null |
| }, |
| "BIBREF12": { |
| "ref_id": "b12", |
| "title": "Quantitative word order typology with UD", |
| "authors": [ |
| { |
| "first": "Matias-Guzm\u00e1n", |
| "middle": [], |
| "last": "Naranjo", |
| "suffix": "" |
| }, |
| { |
| "first": "Laura", |
| "middle": [], |
| "last": "Becker", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proceedings of the 17th International Workshop on Treebanks and Linguistic Theories (TLT 2018)", |
| "volume": "155", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Matias-Guzm\u00e1n Naranjo and Laura Becker. 2018. Quantitative word order typology with UD. In Pro- ceedings of the 17th International Workshop on Tree- banks and Linguistic Theories (TLT 2018), pages 91-104, Oslo, Norway. Link\u00f6ping Electronic Con- ference Proceedings 155:10.", |
| "links": null |
| }, |
| "BIBREF13": { |
| "ref_id": "b13", |
| "title": "Faculty of Mathematics and Physics", |
| "authors": [ |
| { |
| "first": "Joakim", |
| "middle": [], |
| "last": "Nivre", |
| "suffix": "" |
| }, |
| { |
| "first": "\u017deljko", |
| "middle": [], |
| "last": "Agi\u0107", |
| "suffix": "" |
| }, |
| { |
| "first": "Maria", |
| "middle": [ |
| "Jesus" |
| ], |
| "last": "Aranzabe", |
| "suffix": "" |
| }, |
| { |
| "first": "Masayuki", |
| "middle": [], |
| "last": "Asahara", |
| "suffix": "" |
| }, |
| { |
| "first": "Aitziber", |
| "middle": [], |
| "last": "Atutxa", |
| "suffix": "" |
| }, |
| { |
| "first": "Miguel", |
| "middle": [], |
| "last": "Ballesteros", |
| "suffix": "" |
| }, |
| { |
| "first": "John", |
| "middle": [], |
| "last": "Bauer", |
| "suffix": "" |
| }, |
| { |
| "first": "Kepa", |
| "middle": [], |
| "last": "Bengoetxea", |
| "suffix": "" |
| }, |
| { |
| "first": "Riyaz", |
| "middle": [ |
| "Ahmad" |
| ], |
| "last": "Bhat", |
| "suffix": "" |
| }, |
| { |
| "first": "Cristina", |
| "middle": [], |
| "last": "Bosco", |
| "suffix": "" |
| }, |
| { |
| "first": "Sam", |
| "middle": [], |
| "last": "Bowman", |
| "suffix": "" |
| }, |
| { |
| "first": "Giuseppe", |
| "middle": [ |
| "G A" |
| ], |
| "last": "Celano", |
| "suffix": "" |
| }, |
| { |
| "first": "Miriam", |
| "middle": [], |
| "last": "Connor", |
| "suffix": "" |
| }, |
| { |
| "first": "Marie-Catherine", |
| "middle": [], |
| "last": "De Marneffe", |
| "suffix": "" |
| }, |
| { |
| "first": "Arantza", |
| "middle": [], |
| "last": "Diaz De Ilarraza", |
| "suffix": "" |
| }, |
| { |
| "first": "Kaja", |
| "middle": [], |
| "last": "Dobrovoljc", |
| "suffix": "" |
| }, |
| { |
| "first": "Timothy", |
| "middle": [], |
| "last": "Dozat", |
| "suffix": "" |
| }, |
| { |
| "first": "Toma\u017e", |
| "middle": [], |
| "last": "Erjavec", |
| "suffix": "" |
| }, |
| { |
| "first": "Rich\u00e1rd", |
| "middle": [], |
| "last": "Farkas", |
| "suffix": "" |
| }, |
| { |
| "first": "Jennifer", |
| "middle": [], |
| "last": "Foster", |
| "suffix": "" |
| }, |
| { |
| "first": "Daniel", |
| "middle": [], |
| "last": "Galbraith", |
| "suffix": "" |
| }, |
| { |
| "first": "Filip", |
| "middle": [], |
| "last": "Ginter", |
| "suffix": "" |
| }, |
| { |
| "first": "Iakes", |
| "middle": [], |
| "last": "Goenaga", |
| "suffix": "" |
| }, |
| { |
| "first": "Koldo", |
| "middle": [], |
| "last": "Gojenola", |
| "suffix": "" |
| }, |
| { |
| "first": "Yoav", |
| "middle": [], |
| "last": "Goldberg", |
| "suffix": "" |
| }, |
| { |
| "first": "Berta", |
| "middle": [], |
| "last": "Gonzales", |
| "suffix": "" |
| }, |
| { |
| "first": "Bruno", |
| "middle": [], |
| "last": "Guillaume", |
| "suffix": "" |
| }, |
| { |
| "first": "Jan", |
| "middle": [], |
| "last": "Haji\u010d", |
| "suffix": "" |
| }, |
| { |
| "first": "Dag", |
| "middle": [], |
| "last": "Haug", |
| "suffix": "" |
| }, |
| { |
| "first": "Radu", |
| "middle": [], |
| "last": "Ion", |
| "suffix": "" |
| }, |
| { |
| "first": "Elena", |
| "middle": [], |
| "last": "Irimia", |
| "suffix": "" |
| }, |
| { |
| "first": "Anders", |
| "middle": [], |
| "last": "Johannsen", |
| "suffix": "" |
| }, |
| { |
| "first": "Hiroshi", |
| "middle": [], |
| "last": "Kanayama", |
| "suffix": "" |
| }, |
| { |
| "first": "Jenna", |
| "middle": [], |
| "last": "Kanerva", |
| "suffix": "" |
| }, |
| { |
| "first": "Simon", |
| "middle": [], |
| "last": "Krek", |
| "suffix": "" |
| }, |
| { |
| "first": "Veronika", |
| "middle": [], |
| "last": "Laippala", |
| "suffix": "" |
| }, |
| { |
| "first": "Alessandro", |
| "middle": [], |
| "last": "Lenci", |
| "suffix": "" |
| }, |
| { |
| "first": "Nikola", |
| "middle": [], |
| "last": "Ljube\u0161i\u0107", |
| "suffix": "" |
| }, |
| { |
| "first": "Teresa", |
| "middle": [], |
| "last": "Lynn", |
| "suffix": "" |
| }, |
| { |
| "first": "Christopher", |
| "middle": [], |
| "last": "Manning", |
| "suffix": "" |
| }, |
| { |
| "first": "C\u0203t\u0203lina", |
| "middle": [], |
| "last": "M\u0203r\u0203nduc", |
| "suffix": "" |
| }, |
| { |
| "first": "David", |
| "middle": [], |
| "last": "Mare\u010dek", |
| "suffix": "" |
| }, |
| { |
| "first": "Jan", |
| "middle": [], |
| "last": "H\u00e9ctor Mart\u00ednez Alonso", |
| "suffix": "" |
| }, |
| { |
| "first": "Yuji", |
| "middle": [], |
| "last": "Ma\u0161ek", |
| "suffix": "" |
| }, |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Matsumoto", |
| "suffix": "" |
| }, |
| { |
| "first": "Anna", |
| "middle": [], |
| "last": "Mcdonald", |
| "suffix": "" |
| }, |
| { |
| "first": "Verginica", |
| "middle": [], |
| "last": "Missil\u00e4", |
| "suffix": "" |
| }, |
| { |
| "first": "Yusuke", |
| "middle": [], |
| "last": "Mititelu", |
| "suffix": "" |
| }, |
| { |
| "first": "Simonetta", |
| "middle": [], |
| "last": "Miyao", |
| "suffix": "" |
| }, |
| { |
| "first": "Shunsuke", |
| "middle": [], |
| "last": "Montemagni", |
| "suffix": "" |
| }, |
| { |
| "first": "Hanna", |
| "middle": [], |
| "last": "Mori", |
| "suffix": "" |
| }, |
| { |
| "first": "Petya", |
| "middle": [], |
| "last": "Nurmi", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Osenova", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Joakim Nivre, \u017deljko Agi\u0107, Maria Jesus Aranzabe, Masayuki Asahara, Aitziber Atutxa, Miguel Balles- teros, John Bauer, Kepa Bengoetxea, Riyaz Ah- mad Bhat, Cristina Bosco, Sam Bowman, Giuseppe G. A. Celano, Miriam Connor, Marie-Catherine de Marneffe, Arantza Diaz de Ilarraza, Kaja Do- brovoljc, Timothy Dozat, Toma\u017e Erjavec, Rich\u00e1rd Farkas, Jennifer Foster, Daniel Galbraith, Filip Gin- ter, Iakes Goenaga, Koldo Gojenola, Yoav Gold- berg, Berta Gonzales, Bruno Guillaume, Jan Ha- ji\u010d, Dag Haug, Radu Ion, Elena Irimia, Anders Johannsen, Hiroshi Kanayama, Jenna Kanerva, Si- mon Krek, Veronika Laippala, Alessandro Lenci, Nikola Ljube\u0161i\u0107, Teresa Lynn, Christopher Man- ning, C\u0203t\u0203lina M\u0203r\u0203nduc, David Mare\u010dek, H\u00e9c- tor Mart\u00ednez Alonso, Jan Ma\u0161ek, Yuji Matsumoto, Ryan McDonald, Anna Missil\u00e4, Verginica Mititelu, Yusuke Miyao, Simonetta Montemagni, Shunsuke Mori, Hanna Nurmi, Petya Osenova, Lilja \u00d8vre- lid, Elena Pascual, Marco Passarotti, Cenel-Augusto Perez, Slav Petrov, Jussi Piitulainen, Barbara Plank, Martin Popel, Prokopis Prokopidis, Sampo Pyysalo, Loganathan Ramasamy, Rudolf Rosa, Shadi Saleh, Sebastian Schuster, Wolfgang Seeker, Mojgan Ser- aji, Natalia Silveira, Maria Simi, Radu Simionescu, Katalin Simk\u00f3, Kiril Simov, Aaron Smith, Jan \u0160t\u011bp\u00e1nek, Alane Suhr, Zsolt Sz\u00e1nt\u00f3, Takaaki Tanaka, Reut Tsarfaty, Sumire Uematsu, Larraitz Uria, Vik- tor Varga, Veronika Vincze, Zden\u011bk \u017dabokrtsk\u00fd, Daniel Zeman, and Hanzhi Zhu. 2015. Universal de- pendencies 1.2. LINDAT/CLARIAH-CZ digital li- brary at the Institute of Formal and Applied Linguis- tics (\u00daFAL), Faculty of Mathematics and Physics, Charles University.", |
| "links": null |
| }, |
| "BIBREF14": { |
| "ref_id": "b14", |
| "title": "Stanza: A Python natural language processing toolkit for many human languages", |
| "authors": [ |
| { |
| "first": "Peng", |
| "middle": [], |
| "last": "Qi", |
| "suffix": "" |
| }, |
| { |
| "first": "Yuhao", |
| "middle": [], |
| "last": "Zhang", |
| "suffix": "" |
| }, |
| { |
| "first": "Yuhui", |
| "middle": [], |
| "last": "Zhang", |
| "suffix": "" |
| }, |
| { |
| "first": "Jason", |
| "middle": [], |
| "last": "Bolton", |
| "suffix": "" |
| }, |
| { |
| "first": "Christopher", |
| "middle": [ |
| "D" |
| ], |
| "last": "Manning", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Peng Qi, Yuhao Zhang, Yuhui Zhang, Jason Bolton, and Christopher D. Manning. 2020. Stanza: A Python natural language processing toolkit for many human languages. In Proceedings of the 58th An- nual Meeting of the Association for Computational Linguistics: System Demonstrations.", |
| "links": null |
| }, |
| "BIBREF15": { |
| "ref_id": "b15", |
| "title": "Modern Irish: A Comprehensive Grammar. Comprehensive grammars", |
| "authors": [ |
| { |
| "first": "Nancy", |
| "middle": [], |
| "last": "Stenson", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Nancy Stenson. 2020. Modern Irish: A Comprehen- sive Grammar. Comprehensive grammars. London & New York: Routledge.", |
| "links": null |
| }, |
| "BIBREF16": { |
| "ref_id": "b16", |
| "title": "Evaluating contextualized embeddings on 54 languages in POS tagging, lemmatization and dependency parsing", |
| "authors": [ |
| { |
| "first": "Milan", |
| "middle": [], |
| "last": "Straka", |
| "suffix": "" |
| }, |
| { |
| "first": "Jana", |
| "middle": [], |
| "last": "Strakov\u00e1", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Milan Straka, Jana Strakov\u00e1, and Jan Hajic. 2019. Eval- uating contextualized embeddings on 54 languages in POS tagging, lemmatization and dependency pars- ing. CoRR, abs/1908.07448.", |
| "links": null |
| }, |
| "BIBREF17": { |
| "ref_id": "b17", |
| "title": "A new methodology for an old problem: a corpusbased typology of adnominal word order in European languages", |
| "authors": [ |
| { |
| "first": "Luigi", |
| "middle": [], |
| "last": "Talamo", |
| "suffix": "" |
| }, |
| { |
| "first": "Annemarie", |
| "middle": [], |
| "last": "Verkerk", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "Italian Journal of Linguistics", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Luigi Talamo and Annemarie Verkerk. to appear. A new methodology for an old problem: a corpus- based typology of adnominal word order in Euro- pean languages. Italian Journal of Linguistics.", |
| "links": null |
| }, |
| "BIBREF18": { |
| "ref_id": "b18", |
| "title": "Dependency length minimization vs. word order constraints: An empirical study on 55 treebanks", |
| "authors": [ |
| { |
| "first": "Xiang", |
| "middle": [], |
| "last": "Yu", |
| "suffix": "" |
| }, |
| { |
| "first": "Agnieszka", |
| "middle": [], |
| "last": "Falenska", |
| "suffix": "" |
| }, |
| { |
| "first": "Jonas", |
| "middle": [], |
| "last": "Kuhn", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the First Workshop on Quantitative Syntax (Quasy, SyntaxFest 2019)", |
| "volume": "", |
| "issue": "", |
| "pages": "89--97", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/W19-7911" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Xiang Yu, Agnieszka Falenska, and Jonas Kuhn. 2019. Dependency length minimization vs. word order constraints: An empirical study on 55 treebanks. In Proceedings of the First Workshop on Quantita- tive Syntax (Quasy, SyntaxFest 2019), pages 89-97, Paris, France. Association for Computational Lin- guistics.", |
| "links": null |
| }, |
| "BIBREF19": { |
| "ref_id": "b19", |
| "title": "Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies", |
| "authors": [ |
| { |
| "first": "Daniel", |
| "middle": [], |
| "last": "Zeman", |
| "suffix": "" |
| }, |
| { |
| "first": "Jan", |
| "middle": [], |
| "last": "Haji\u010d", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Daniel Zeman and Jan Haji\u010d, editors. 2018. Proceed- ings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies. Association for Computational Linguistics, Brussels, Belgium.", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "FIGREF0": { |
| "text": "Left: entropy values for the categories of determiners & quantifiers and numerals, as captured by the rel combination and the rel+upos combination. Right: entropy values for the categories of articles, demonstratives, quantifiers and numerals, as captured by rel+pos+lemma combination.", |
| "uris": null, |
| "num": null, |
| "type_str": "figure" |
| }, |
| "TABREF1": { |
| "content": "<table/>", |
| "num": null, |
| "text": "Values used to capture the categories of nominal heads, articles, demonstratives, quantifiers and numerals. Non-specific values are given in italics.", |
| "html": null, |
| "type_str": "table" |
| } |
| } |
| } |
| } |
