| { |
| "paper_id": "2021", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T15:31:18.029182Z" |
| }, |
| "title": "SIGMORPHON 2021 Shared Task on Morphological Reinflection: Generalization Across Languages", |
| "authors": [ |
| { |
| "first": "Tiago", |
| "middle": [], |
| "last": "Pimentel", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Maria", |
| "middle": [], |
| "last": "Ryskina", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "mryskina@cs.cmu.edu" |
| }, |
| { |
| "first": "Sabrina", |
| "middle": [], |
| "last": "Mielke", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Shijie", |
| "middle": [], |
| "last": "Wu", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Eleanor", |
| "middle": [], |
| "last": "Chodroff", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of York", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Brian", |
| "middle": [], |
| "last": "Leonard", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Brian Leonard Consulting \u00e1 University of British Columbia AE STKIP Weetebula \u00e8 New York University Abu Dhabi S Australian National University \u0144 Bar-Ilan University R Charles Darwin University", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Garrett", |
| "middle": [], |
| "last": "Nicolai", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Yustinus", |
| "middle": [], |
| "last": "Ghanggo", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Ate", |
| "middle": [], |
| "last": "Ae", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Salam", |
| "middle": [], |
| "last": "Khalifa", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Nizar", |
| "middle": [], |
| "last": "Habash", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Charbel", |
| "middle": [], |
| "last": "El-Khaissi", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Omer", |
| "middle": [], |
| "last": "Goldman", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Michael", |
| "middle": [], |
| "last": "Gasser", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "I", |
| "middle": [], |
| "last": "William", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Matt", |
| "middle": [], |
| "last": "Coler", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of Groningen I Indiana University", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Arturo", |
| "middle": [], |
| "last": "Oncevay", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "ESRC International Centre for Language and Communicative Development (LuCiD) \u00f1 Northeastern Illinois University", |
| "institution": "Russian Academy of Sciences \u0159 University at Buffalo K Karelian Research Centre of the Russian Academy of Sciences", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Jaime", |
| "middle": [], |
| "last": "Rafael", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Montoya", |
| "middle": [], |
| "last": "Samame", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Gema", |
| "middle": [], |
| "last": "Celeste", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Silva", |
| "middle": [], |
| "last": "Villegas", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Adam", |
| "middle": [], |
| "last": "Ek", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Jean-Philippe", |
| "middle": [], |
| "last": "Bernardy", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Andrey", |
| "middle": [], |
| "last": "Shcherbakov", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Aziyana", |
| "middle": [], |
| "last": "Bayyr-Ool", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Karina", |
| "middle": [], |
| "last": "Sheifer", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Sofya", |
| "middle": [], |
| "last": "Ganieva", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Matvey", |
| "middle": [], |
| "last": "Plugaryov", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Elena", |
| "middle": [], |
| "last": "Klyachko", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Ali", |
| "middle": [], |
| "last": "Salehi", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Andrew", |
| "middle": [], |
| "last": "Krizhanovsky", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Natalia", |
| "middle": [], |
| "last": "Krizhanovsky", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Clara", |
| "middle": [], |
| "last": "Vania", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "New York University", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Sardana", |
| "middle": [], |
| "last": "Ivanova", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of Helsinki", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Aelita", |
| "middle": [], |
| "last": "Salchak", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Tuvan State University", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Christopher", |
| "middle": [], |
| "last": "Straughn", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Zoey", |
| "middle": [], |
| "last": "Liu", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Polish Academy of Sciences", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Jonathan", |
| "middle": [], |
| "last": "North", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Washington", |
| "middle": [ |
| "F" |
| ], |
| "last": "Duygu", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Ataman", |
| "middle": [], |
| "last": "Ae", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Witold", |
| "middle": [], |
| "last": "Kiera\u015b", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Marcin", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Woli\u0144ski", |
| "middle": [ |
| "T" |
| ], |
| "last": "Totok", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Bhimrao Ambedkar University", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Niklas", |
| "middle": [], |
| "last": "Stoehr", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Zahroh", |
| "middle": [], |
| "last": "Nuriah", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Bhimrao Ambedkar University", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Shyam", |
| "middle": [], |
| "last": "Ratan", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [ |
| "M" |
| ], |
| "last": "Tyers", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Edoardo", |
| "middle": [ |
| "M" |
| ], |
| "last": "Ponti", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "McGill University Montreal D ETH Z\u00fcrich X University of Colorado Boulder L University of Liverpool", |
| "location": { |
| "settlement": "Mila" |
| } |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Grant", |
| "middle": [], |
| "last": "Aiton", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "S", |
| "middle": [], |
| "last": "Richard", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Hatcher", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Emily", |
| "middle": [], |
| "last": "Prud'hommeaux", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Polish Academy of Sciences", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Ritesh", |
| "middle": [], |
| "last": "Kumar", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "U", |
| "middle": [], |
| "last": "Mans", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Hulden", |
| "middle": [ |
| "X" |
| ], |
| "last": "Botond Barta", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Dorina", |
| "middle": [], |
| "last": "Lakatos", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "A", |
| "middle": [], |
| "last": "G\u00e1bor", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Szolnok", |
| "middle": [ |
| "A" |
| ], |
| "last": "Judit", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "\u00c1cs", |
| "middle": [ |
| "A" |
| ], |
| "last": "Mohit", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Raj", |
| "middle": [ |
| "U" |
| ], |
| "last": "David", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Cotterell", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Ben", |
| "middle": [], |
| "last": "Ambridge", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Ekaterina", |
| "middle": [], |
| "last": "Vylomova", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "vylomovae@unimelb.edu.au" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "This year's iteration of the SIGMORPHON Shared Task on morphological reinflection focuses on typological diversity and crosslingual variation of morphosyntactic features. In terms of the task, we enrich UniMorph with new data for 32 languages from 13 language families, with most of them being under-resourced: Kunwinjku", |
| "pdf_parse": { |
| "paper_id": "2021", |
| "_pdf_hash": "", |
| "abstract": [ |
| { |
| "text": "This year's iteration of the SIGMORPHON Shared Task on morphological reinflection focuses on typological diversity and crosslingual variation of morphosyntactic features. In terms of the task, we enrich UniMorph with new data for 32 languages from 13 language families, with most of them being under-resourced: Kunwinjku", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Abstract", |
| "sec_num": null |
| } |
| ], |
| "body_text": [ |
| { |
| "text": "1 Introduction Chomsky (1995) noted that if a Martian anthropologist were to visit our planet, all of our world's languages would appear as a dialect of a single language, more specifically instances of what he calls a \"universal grammar\". This idea-that all languages have a large inventory of shared sounds, vocabulary, syntactic structures with minor variations-was especially common among cognitive scientists. It was based on highly biased ethnocentric empirical observations, resulting from the fact that a vast majority of cognitive scientists, including linguists, focused only on the familiar European languages. Moreover, as Daniel (2011) notes, many linguistic descriptive traditions of individual languages, even isolated ones such as Russian or German, heavily rely on cross-linguistic assumptions about the structure of human language that are often projected from Latin grammars. Similarly, despite making universalistic claims, generative linguists, for a very long time, have focused on a small number of the world's major languages, typically using English as their departure point. This could be partly attributed to the fact that generative grammar follows a deductive approach where the observed data is conditioned on a general model. However, as linguists explored more languages, descriptions and comparisons of more diverse kinds of languages began to come up, both within the framework of generative syntax as well as that of linguistic typology. Greenberg (1963) presents one of the earliest typologically informed description of \"language universals\" based on an analysis of a relatively larger set of 30 languages, which included a substantial proportion of data from non-European languages. Subsequently, typologists have claimed that it is essential to describe the limits of cross-linguistic variation (Croft, 2002; Comrie, 1989 ) rather than focus only on cross-linguistic similarities. This is especially evident from Evans and Levinson (2009) , where the authors question the notion of \"language universals\", i.e. the existence of a common pattern, or basis, shared across human languages. By looking at cross-linguistic work done by typologists and descriptive linguists, they demonstrate that \"diversity can be found at almost every level of linguistic organization\": languages vary greatly on phonological, morphological, semantic, and syntactic levels. This leads us to p-linguistics (Haspelmath, 2020) , a study of particular languages, including the whole variety of idiosyncratic properties present in them, which makes cross-linguistic comparison challenging.", |
| "cite_spans": [ |
| { |
| "start": 15, |
| "end": 29, |
| "text": "Chomsky (1995)", |
| "ref_id": "BIBREF8" |
| }, |
| { |
| "start": 635, |
| "end": 648, |
| "text": "Daniel (2011)", |
| "ref_id": "BIBREF18" |
| }, |
| { |
| "start": 1473, |
| "end": 1489, |
| "text": "Greenberg (1963)", |
| "ref_id": "BIBREF34" |
| }, |
| { |
| "start": 1834, |
| "end": 1847, |
| "text": "(Croft, 2002;", |
| "ref_id": "BIBREF17" |
| }, |
| { |
| "start": 1848, |
| "end": 1860, |
| "text": "Comrie, 1989", |
| "ref_id": "BIBREF11" |
| }, |
| { |
| "start": 1952, |
| "end": 1977, |
| "text": "Evans and Levinson (2009)", |
| "ref_id": "BIBREF24" |
| }, |
| { |
| "start": 2423, |
| "end": 2441, |
| "text": "(Haspelmath, 2020)", |
| "ref_id": "BIBREF41" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "Haspelmath (2010) suggested a distinction between descriptive categories (specific to languages) and comparative concepts. The idea was then refined and further developed with respect to morphology and realized in the UniMorph schema (Sylak-Glassman et al., 2015b) . Morphosyntactic features (such as \"the dative case\" or \"the past tense\") in the UniMorph occupy an intermediate position between the descriptive categories and comparative concepts. The set of features was initially established on the basis of analysis of typological literature, and refined with the addition of new languages to the UniMorph database (Kirov et al., 2018; McCarthy et al., 2020) . Since 2016, SIG-MORPHON organized shared tasks on morphological reinflection (Cotterell et al., 2016 (Cotterell et al., , 2017 ) that aimed at evaluating contemporary systems. Parallel to that, they also served as a platform for enriching the UniMorph database with new languages. For instance, the 2020 shared task featured 90 typologically diverse languages derived from various linguistic resources.", |
| "cite_spans": [ |
| { |
| "start": 234, |
| "end": 264, |
| "text": "(Sylak-Glassman et al., 2015b)", |
| "ref_id": "BIBREF82" |
| }, |
| { |
| "start": 619, |
| "end": 639, |
| "text": "(Kirov et al., 2018;", |
| "ref_id": "BIBREF52" |
| }, |
| { |
| "start": 640, |
| "end": 662, |
| "text": "McCarthy et al., 2020)", |
| "ref_id": "BIBREF96" |
| }, |
| { |
| "start": 742, |
| "end": 765, |
| "text": "(Cotterell et al., 2016", |
| "ref_id": "BIBREF16" |
| }, |
| { |
| "start": 766, |
| "end": 791, |
| "text": "(Cotterell et al., , 2017", |
| "ref_id": "BIBREF14" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "This year, we are bringing many under-resourced languages (languages of Peru, Russia, India, Australia, Papua New Guinea) and dialects (e.g., for Arabic and Kurdish). The sample is highly diverse: it contains languages with templatic, concatenative (fusional and agglutinative) morphology. In addition, we bring more polysynthetic languages such as Kunwinjku, Chukchi, Ash\u00e1ninka. Unlike previous years, we pay more attention to the conversion of the morphosynactic features of these languages into the UniMorph schema. In addition, for most languages we conduct an extensive error analysis.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "In this shared task, the participants were told to design a model that learns to generate morphological inflections from both a lemma and a set of morphosyntactic features of the target form. Specifically, each language in the task had its own training, development, and test splits. The training and development splits contained triples, with a lemma, a set of morphological features, and the target inflected form, while test splits only provided lemmas and morphological tags: the participants' models needed to predict the missing target form-making this a standard supervised learning task. The target of the task, however, was to analyse how well the current state-of-the-art reinflection models could generalise across a typologically diverse set of languages. These models should, in theory, be general enough to work for natural languages of any typological patterning. 2 As such, we designed the task in three phases: a Development Phase, a Generalization Phase, and an Evaluation Phase. As the phases advanced, more data and more languages were released.", |
| "cite_spans": [ |
| { |
| "start": 879, |
| "end": 880, |
| "text": "2", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Task Description", |
| "sec_num": "2" |
| }, |
| { |
| "text": "In the Development Phase, we provided training and development splits that should be used by participants to develop their systems. Model development, evaluation, and hyper-parameter tuning were, thus, mainly performed on these sets of languages. We will refer to these as the development languages.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Task Description", |
| "sec_num": "2" |
| }, |
| { |
| "text": "In the Generalization Phase, we provided training and development splits for new languages where approximately half were genetically related (belonged to the same family) and half were genetically unrelated (either isolates or belonging to different families) to the development languages. These languages (and their families) were kept as a surprise throughout the first (development) phase and were only announced later on. As the participants were only given a few days with access to these languages before the submission deadline, we expected that the systems couldn't be radically improved to work on them-as such, these languages allowed us to evaluate the generalization capacity of the re-inflection models, and how well they performed on new typologically unrelated languages.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Task Description", |
| "sec_num": "2" |
| }, |
| { |
| "text": "Finally, in the Evaluation Phase, the participants' models were evaluated on held-out test forms from all of the languages of the previous phases. The languages from the Development Phase and the Generalization Phase were evaluated simultaneously. The only difference between the development and generalization languages was that participants had more time to construct their models for the languages released in the Development Phase. It follows that a model could easily favor or overfit to the phenomena that are more frequent in the languages presented in the Development Phase, especially if the parameters were shared across languages. For instance, a model based on the morphological patterning of Indo-European languages may end up with a bias towards suffixing and would struggle to learn prefixing or circumfixation, and the degree of the bias only becomes apparent during experimentation on other languages whose inflectional morphology patterns differ. Further, the model architecture itself could also explicitly or implicitly favor certain word formation types (suffixing, prefixing, etc.) .", |
| "cite_spans": [ |
| { |
| "start": 1075, |
| "end": 1103, |
| "text": "(suffixing, prefixing, etc.)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Task Description", |
| "sec_num": "2" |
| }, |
| { |
| "text": "The Gunwinyguan language family consists of Australian Aboriginal languages spoken in the Arnhem Land region of Australia's Northern Territory.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Gunwinyguan", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "This data set contains one member of this family: a dialect of Bininj Kunwok called Kunwinjku. Kunwinjku is a polysynthetic language with mostly agglutinating verbal morphology. A typical verb there might look like Aban-yawoith-warrgahmarne-ganj-ginje-ng '1/3PL-again-wrong-BENmeat-cook-PP' (\"I cooked the wrong meat for them again\"). As shown, the form has several prefixes and suffixes attached to the stem. As in other Australian languages, long vowels are typically represented by double characters, and trills with \"rr\". 3 According to Evans' (2003) analysis, the verb template contains 12 affix slots which include two incorporated noun classes, and derivational affixes such as the benefactive and comitative. The data included in this set are verbs extracted from the Kunwinjku translation of the Bible using the morphological analyzer from Lane and Bird (2019) and manually verified by human annotators.", |
| "cite_spans": [ |
| { |
| "start": 526, |
| "end": 527, |
| "text": "3", |
| "ref_id": null |
| }, |
| { |
| "start": 541, |
| "end": 554, |
| "text": "Evans' (2003)", |
| "ref_id": "BIBREF23" |
| }, |
| { |
| "start": 849, |
| "end": 869, |
| "text": "Lane and Bird (2019)", |
| "ref_id": "BIBREF56" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Gunwinggic: Kunwinjku", |
| "sec_num": "3.1.1" |
| }, |
| { |
| "text": "The Afro-Asiatic language family is represented by the Semitic subgroup.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Afro-Asiatic", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "Classical Syriac is a dialect of the Aramaic language and is attested as early as the 1st century CE. As with most Semitic languages, it displays non-concatenative morphology involving primarily tri-consonantal roots. Syriac nouns and adjectives are conventionally classified into three 'states'-Emphatic, Absolute, Construct-which loosely correlate with the syntactic features of definiteness, indeterminacy and the genitive. There are over 10 verbal paradigms that combine affixation slots with inflectional templates to reflect tense (past, present, future), person (first, second, third), number (singular, plural), gender (masculine, feminine, common), mood (imperative, infinitive), voice (active, passive), and derivational form (i.e., participles). Paradigmatic rules are determined by a range of linguistic factors, such as root type or phonological properties. The data included in this set was relatively small and consisted of 1,217 attested lexemes in the New Testament, which were extracted from Beth Mardutho: The Syriac Institute's lexical database, SEDRA.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Classical Syriac", |
| "sec_num": "3.2.1" |
| }, |
| { |
| "text": "Modern Standard Arabic (MSA, ara) is the primarily written form of Arabic which is used in all official communication means. In contrast, Arabic dialects are the primarily spoken varieties of Arabic, and the increasingly written varieties on unofficial social media platforms. Dialects have no official status despite being widely used. Both MSA and the dialects coexist in a state of diglossia (Ferguson, 1959) whether in spoken or written form. Arabic dialects vary amongst themselves and are different from MSA in most linguistic aspects (phonology, morphology, syntax, and lexical choice). In this work we provide inflection tables for MSA (ara), Egyptian Arabic (EGY, arz), and Gulf Arabic (GLF, afb). Egyptian Arabic is the variety of Arabic spoken in Egypt. Gulf Arabic refers to the dialects spoken by the indigenous populations of the members of the Gulf Cooperation Council, especially regions on the Arabian Gulf. Similar to other Semitic languages, Arabic is a templatic language. A word consists of a templatic stem (root and pattern) and a number of affixes and clitics. Verb lemmas in Arabic inflect for person, gender, number, voice, mood, and aspect. Nominal lemmas inflect for gender, number, case, and state. Those features are realized through both the templatic patterns and the concatenative affixations. Arabic words also take on a number of clitics: attachable prepositions, conjunctions, determiners, and pronominal objects and possessives. In this work, we do not include clitics as a part of the paradigms, as they heavily increase the size of the paradigms. We made the exception to add the Al determiner particle in order to be consistent with commonly used tokenizations for Arabic treebanks-Penn Arabic Treebank (Maamouri et al., 2004) and Arabic Universal Dependencies (Taji et al., 2017) .", |
| "cite_spans": [ |
| { |
| "start": 395, |
| "end": 411, |
| "text": "(Ferguson, 1959)", |
| "ref_id": null |
| }, |
| { |
| "start": 1743, |
| "end": 1766, |
| "text": "(Maamouri et al., 2004)", |
| "ref_id": "BIBREF59" |
| }, |
| { |
| "start": 1801, |
| "end": 1820, |
| "text": "(Taji et al., 2017)", |
| "ref_id": "BIBREF83" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Arabic", |
| "sec_num": "3.2.2" |
| }, |
| { |
| "text": "For MSA, the paradigms inflect for all the abovementioned features, while for EGY and GLF they inflect for the above-mentioned features except for voice, mood, case, and state. We use the functional (grammatical) gender and number for MSA and GLF, but the form-based gender and number for EGY, since the resources we used did not have EGY functional gender and number (Alkuhlani and Habash, 2011) .", |
| "cite_spans": [ |
| { |
| "start": 368, |
| "end": 396, |
| "text": "(Alkuhlani and Habash, 2011)", |
| "ref_id": "BIBREF3" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Arabic", |
| "sec_num": "3.2.2" |
| }, |
| { |
| "text": "We generated all the inflection tables from the morphological analysis databases using the generation component provided by CamelTools (Obeid et al., 2020) . We extracted all the verb, noun, and adjective lemmas from a number of annotated corpora and selected those that are already in the morphological analysis databases. For MSA, we used the CALIMA-STAR database (Taji et al., 2018) , based on the SAMA database (Maamouri et al., 2010) , and the PATB (Maamouri et al., 2004) as the sources of lemmas. For EGY, we used the CALIMA-EGY database (Habash et al., 2012) and the ARZTB (Maamouri et al., 2012) as the sources of lemmas. For GLF, we used the Gulf verb analyzer (Khalifa et al., 2017) for verbs, and for both nouns and adjectives we extracted all the annotations from the Annotated Gumar Corpus (Khalifa et al., 2018).", |
| "cite_spans": [ |
| { |
| "start": 135, |
| "end": 155, |
| "text": "(Obeid et al., 2020)", |
| "ref_id": "BIBREF72" |
| }, |
| { |
| "start": 366, |
| "end": 385, |
| "text": "(Taji et al., 2018)", |
| "ref_id": "BIBREF84" |
| }, |
| { |
| "start": 415, |
| "end": 438, |
| "text": "(Maamouri et al., 2010)", |
| "ref_id": null |
| }, |
| { |
| "start": 454, |
| "end": 477, |
| "text": "(Maamouri et al., 2004)", |
| "ref_id": "BIBREF59" |
| }, |
| { |
| "start": 545, |
| "end": 566, |
| "text": "(Habash et al., 2012)", |
| "ref_id": "BIBREF38" |
| }, |
| { |
| "start": 581, |
| "end": 604, |
| "text": "(Maamouri et al., 2012)", |
| "ref_id": "BIBREF60" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Arabic", |
| "sec_num": "3.2.2" |
| }, |
| { |
| "text": "As Syriac, Hebrew is a member of the Northwest Semitic branch, and, like Syriac and Arabic, it is written using an abjad where the vowels are sparsely marked in unvocalized text. This fact entails that in unvocalized data the complex ablautextensive non-concatenative Semitic morphology is somewhat watered down as the consonants of the root frequently appear consecutively with the alternating vowel unwritten. In this work we present data in vocalized Hebrew, in order to examine the models' ability to handle Hebrew's full-fledged Semitic morphological system.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Hebrew", |
| "sec_num": "3.2.3" |
| }, |
| { |
| "text": "Hebrew verbs belong to 7 major classes (Binyanim) with many subclasses depending on the phonological features of the root's consonants. Verbs inflect for number, gender, and tense-mood, while the nominal inflection tables include definiteness and possessor.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Hebrew", |
| "sec_num": "3.2.3" |
| }, |
| { |
| "text": "The provided inflection tables are largely identical to those of the past years' shared tasks, scraped from Wiktionary, with the addition of the verbal nouns and all forms being automatically vocalized.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Hebrew", |
| "sec_num": "3.2.3" |
| }, |
| { |
| "text": "Amharic is the most spoken and best-resourced among the roughly 15 languages in the Ethio-Semitic branch of South Semitic. Unlike most other Semitic languages, but like other Ethio-Semitic languages, it is written in the Ge'ez (Ethiopic) script, an abugida in which each character represents either a consonant-vowel sequence or a consonant in the syllable coda position.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Amharic", |
| "sec_num": "3.2.4" |
| }, |
| { |
| "text": "Like other Semitic languages, Amharic displays both affixation and non-concatenative template morphology. Verbs inflect for subject person, gender, and number and tense/aspect/mood. Voice and valence are also marked, both by templates and affixes, but these are treated as separate lemmas in the data. Other verb affixes (or clitics, depending on the analysis) indicate object person, gender, and number; negation; relativization; conjunctions; and, on relativized forms, prepositions and definiteness. None of these are included in the data.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Amharic", |
| "sec_num": "3.2.4" |
| }, |
| { |
| "text": "Nouns and adjectives share most of their morphology and are often not clearly distinguished. Nouns and adjectives inflect for definiteness, number, and possession. Gender is only explicit when the masculine or feminine singular definite suffixes are present; most nouns have no inherent gender. Nouns and adjectives also have prepositional prefixes (or clitics) and accusative suffixes, which are not included in the data.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Amharic", |
| "sec_num": "3.2.4" |
| }, |
| { |
| "text": "The data for the shared task were generated by the HornMorpho generator (Gasser, 2011) , an FST weighted with feature structures. Common orthographic variants of the lemmas and common variant plural forms of nouns are included. In these cases, the variants are distinguished with the LGSPEC1 and LGSPEC2 features. Predictable orthographic variants are not included.", |
| "cite_spans": [ |
| { |
| "start": 72, |
| "end": 86, |
| "text": "(Gasser, 2011)", |
| "ref_id": "BIBREF29" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semitic: Amharic", |
| "sec_num": "3.2.4" |
| }, |
| { |
| "text": "The Aymaran family has two branches: Southern Aymaran (which is the branch described in this contribution, as represented by Mulyaq' Aymara) and Central Aymaran (Jaqaru). 4 Aymaran has no external relatives. The neighboring and overlapping Quechuan family is often erroneously believed to be related.", |
| "cite_spans": [ |
| { |
| "start": 171, |
| "end": 172, |
| "text": "4", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Aymaran", |
| "sec_num": "3.3" |
| }, |
| { |
| "text": "Aymara is spoken mainly in Andean communities in the region encompassing Bolivia and Peru from the north of Lake Titicaca to the south of Lake Poop\u00f3, extending westward to the valleys of the Pacific coast and eastward to the Yunga valleys. It has roughly two million speakers, over half of whom are Bolivian. The rest reside mainly in Peru, with small communities in Chile and Argentina. Aymara is a highly agglutinative, suffix-only language. Nouns are inflected for grammatical number, case, and possessiveness. As Coler (2010) notes, Aymara has 11-12 grammatical cases, depending on the variety (as in some varieties the locative and genitive suffixes have merged and in others they have not). The case suffix is attached to the last element of a noun phrase. Verbs present relatively complex paradigms, with dimensions such as grammatical person (marking both subject and direct object), number, tense (simple, future, recent past, distal past), mood (evidentials, two counterfactual paradigms, and an imperative paradigm). Moreover, Aymara has a variety of suffixes which change the grammatical category of the word. Words can change grammatical category multiple times. 5", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Aymaran: Aymara", |
| "sec_num": "3.3.1" |
| }, |
| { |
| "text": "The Indo-European language family is the parent family of most of the European and Asian languages. In this iteration of the shared task, we enrich the data with languages from Indo-Aryan, Iranian, and Slavic groups. Iranian and Indo-Aryan are recognised as distinct subgroups of Indo-European. Characteristic retentions and innovations make Iranian and Indo-Aryan language families diverged and distinct from each other (Jain and Cardona, 2007) .", |
| "cite_spans": [ |
| { |
| "start": 421, |
| "end": 445, |
| "text": "(Jain and Cardona, 2007)", |
| "ref_id": "BIBREF45" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Indo-European", |
| "sec_num": "3.4" |
| }, |
| { |
| "text": "The Indian subcontinent is the heartland of where the Indo-Aryan languages are spoken. This area is also referred to as South Asia and encompasses India, Pakistan, Bangladesh, Nepal, Bhutan and the islands of Sri Lanka and Maldives (Jain and Cardona, 2007) . Magahi and Braj, which belong to the Indo-Aryan language family, are under our observation.", |
| "cite_spans": [ |
| { |
| "start": 232, |
| "end": 256, |
| "text": "(Jain and Cardona, 2007)", |
| "ref_id": "BIBREF45" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Indo-Aryan, or Indic: Magahi, Braj", |
| "sec_num": "3.4.1" |
| }, |
| { |
| "text": "Magahi comes under the Magadhi group of the middle Indo-Aryan which includes Bhojpuri and Maithili. While the exact classification within this subgroup is still debatable, most accepted analyses put it under one branch of the Eastern group of languages which includes Bangla, Asamiya, and Oriya (Grierson and Konow, 1903) . Magahi speech area is mainly concentrated in the Eastern Indian states of Bihar and Jharkhand, but it also extends to the adjoining regions of Bengal and Odisha (Grierson, 1903) .", |
| "cite_spans": [ |
| { |
| "start": 295, |
| "end": 321, |
| "text": "(Grierson and Konow, 1903)", |
| "ref_id": "BIBREF37" |
| }, |
| { |
| "start": 485, |
| "end": 501, |
| "text": "(Grierson, 1903)", |
| "ref_id": "BIBREF36" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Indo-Aryan, or Indic: Magahi, Braj", |
| "sec_num": "3.4.1" |
| }, |
| { |
| "text": "There is no grammatical gender and number agreement in Magahi, though sex-related gender derivation commonly occurs for animate nouns like /laika/ (boy) and /laiki/ (girl). Number is also marked on nouns, and it affects the form of case markers and postpositions in certain instances (Lahiri, 2021) . Moreover, it has a rich system of verbal morphology to show the tense, aspect, person, and honorific agreement with the subject as well as the addressee.", |
| "cite_spans": [ |
| { |
| "start": 284, |
| "end": 298, |
| "text": "(Lahiri, 2021)", |
| "ref_id": "BIBREF55" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Indo-Aryan, or Indic: Magahi, Braj", |
| "sec_num": "3.4.1" |
| }, |
| { |
| "text": "In the present dataset, the inflectional paradigms for verbs show the honorificity level of both the subjects and the addressees, and also the person of the subject, the tense and aspect markers. The inflectional paradigms for nouns and adjectives are generated on the basis of the inflectional marker used for expressing case, familiarity, plurality, and (sometimes) gender within animate nouns. Pronouns are marked for different cases and honorificity levels. These paradigms are generated on the basis of a manually annotated corpus of Magahi folktales.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Indo-Aryan, or Indic: Magahi, Braj", |
| "sec_num": "3.4.1" |
| }, |
| { |
| "text": "We used a raw dataset from the literary domain. First, we annotated the dataset with the Universal Dependency morphological feature tags at token level using the CoNLL-U editor (Heinecke, 2019) . We then converted the annotated dataset into the UniMorph schema using the script available for converting UD data into the UniMorph tagset (Mc-Carthy et al., 2018) . To finalize the data, we manually validated the dataset against the UniMorph schema (Sylak-Glassman et al., 2015a) .", |
| "cite_spans": [ |
| { |
| "start": 177, |
| "end": 193, |
| "text": "(Heinecke, 2019)", |
| "ref_id": "BIBREF42" |
| }, |
| { |
| "start": 336, |
| "end": 360, |
| "text": "(Mc-Carthy et al., 2018)", |
| "ref_id": "BIBREF52" |
| }, |
| { |
| "start": 447, |
| "end": 477, |
| "text": "(Sylak-Glassman et al., 2015a)", |
| "ref_id": "BIBREF81" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Indo-Aryan, or Indic: Magahi, Braj", |
| "sec_num": "3.4.1" |
| }, |
| { |
| "text": "Brajbhasha, or Braj is one of the Indo-Aryan languages spoken in the Western Indian states of Uttar Pradesh, Madhya Pradesh, and Rajasthan. Grierson (1908) groups Brajbhasha under Western Hindi of the Central Group in the Indo-Aryan family, along with other languages like Hindustani, Bangaru, Kannauji, and Bundeli. Braj is not generally used in education or for any official purposes in any Braj spoken state, but it has a very rich literary tradition. Also in order to preserve, promote, pub-lish and popularise the literary tradition of Braj, the local state government of Rajasthan has set up the Braj Bhasha Akademi (Braj Bhasha Academy) in Jaipur. Along with this, some individuals, local literary and cultural groups, and language enthusiasts at the local level also bring out publications in Braj (Kumar et al., 2018) . In all of the above sources, bhakti poetry 6 constitutes a large proportion of the traditional literature of Braj (Pankaj, 2020) .", |
| "cite_spans": [ |
| { |
| "start": 140, |
| "end": 155, |
| "text": "Grierson (1908)", |
| "ref_id": "BIBREF35" |
| }, |
| { |
| "start": 806, |
| "end": 826, |
| "text": "(Kumar et al., 2018)", |
| "ref_id": "BIBREF54" |
| }, |
| { |
| "start": 943, |
| "end": 957, |
| "text": "(Pankaj, 2020)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Indo-Aryan, or Indic: Magahi, Braj", |
| "sec_num": "3.4.1" |
| }, |
| { |
| "text": "As in the case of other Indo-Aryan languages, Braj is also rich in morphological inflections. The dataset released for the present task contains two sets of inflectional paradigms with morphological features for nouns and verbs. Nominal lemmas in Braj are inflected for gender (masculine and feminine) and number (singular and plural); verb lemmas take gender (masculine and feminine), number (singular and plural), person (first, second and third), politeness/honorificity (formal and informal), tense (present, past and future), and aspect (perfective, progressive, habitual and prospective) markings. Among these, the politeness feature is marked for showing honorificity and formality. More generally, a formal/polite marker is used for strangers and the elite class, while informal/neutral markers are used for family and friends.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Indo-Aryan, or Indic: Magahi, Braj", |
| "sec_num": "3.4.1" |
| }, |
| { |
| "text": "In order to generate the morphological paradigms, we have used the data from the literary domain, annotated at the token level in the CoNLL-U editor (Heinecke, 2019) . The dataset was initially annotated using the Universal Dependencies morphological feature set and then automatically converted to the UniMorph schema using the script provided by McCarthy et al. (2018) . Finally, the converted dataset was manually validated and edited to conform to the constraints and conventions of UniMorph to arrive at the final labels.", |
| "cite_spans": [ |
| { |
| "start": 149, |
| "end": 165, |
| "text": "(Heinecke, 2019)", |
| "ref_id": "BIBREF42" |
| }, |
| { |
| "start": 348, |
| "end": 370, |
| "text": "McCarthy et al. (2018)", |
| "ref_id": "BIBREF13" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Indo-Aryan, or Indic: Magahi, Braj", |
| "sec_num": "3.4.1" |
| }, |
| { |
| "text": "The Iranian branch is represented by Kurdish. Among Western Iranian languages, Kurdish is the term covering the largest group of related dialects. Kurdish comprises three main subgroup dialects, namely Northern Kurdish (including Kurmanji), Central Kurdish (including Sorani), and Southern Kurdish. Sorani Kurdish, spoken in Iran and Iraq, is known for its morphological split ergative system. There are two sets of morphemes traditionally described as agreement markers: clitic markers and verbal affixes, which are verbal agreement markers, or the copula. The distribution of these formatives can be described as ergative alignment, although mandatory agent indexing has led some scholars to refer to the Sorani system as post-or remnantergative (J\u00fcgel, 2009) . Note that Sorani nominals do not feature case marking. The single argument of an intransitive verb is an affix while the transitive verbs have a tense-sensitive alignment. With transitive verbs, agents are indexed by affixes in the present tense and with clitics in the past tense. On the other hand, the object is indexed with a clitic in the present tense and an affix in the past tense. In addition, Sorani also has the so-called experiencersubject verbs, with which both the agent and the object are marked with clitic markers. Like other Iranian languages, Sorani also features a series of light-verb constructions which are composed using the verbs kirdin 'to do' or bun 'to be'. In the light verb constructions, the agent is marked with an affix in the present tense, while a clitic marks the subject in the past tense. Southern Kurdish features all the same verbs types, clitics and affixes, while the alignment pattern can be completely different due to a nominative-accusative alignment system. The usage of agreement markers with affixes is widely predominant in Southern Kurdish and clitics can be used to mark the possessives.", |
| "cite_spans": [ |
| { |
| "start": 748, |
| "end": 761, |
| "text": "(J\u00fcgel, 2009)", |
| "ref_id": "BIBREF46" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Iranian: Kurdish", |
| "sec_num": "3.4.2" |
| }, |
| { |
| "text": "Both dialects of Kurdish allow for clitic and affix stacking marking the agent and the object of a verb. In Sorani, for instance, dit=yan-im 'They saw me' uses a clitic and an affix to mark the agent and the object, and wist=yan=im 'I wanted them' marks both the agent and the object with clitics. Ditransitive verbs can be formed by a transitive verb and an applicative marker. For instance, a ditransitive three-participant verb da-m-\u00een=\u00ee-y\u00ea 'He gave them to me' marks the recipient and the object with affixes, and the agent is marked with a clitic in the presence of an applicative (y\u00ea). A separate set of morphological features is needed to account for such structures, in which the verb dictates the person marker index as subject, agent, object or recipient.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Iranian: Kurdish", |
| "sec_num": "3.4.2" |
| }, |
| { |
| "text": "The Slavic genus comprises a group of fusional languages evolved from Proto-Slavic and spoken in Central and Eastern Europe, the Balkans and the Asian parts of Russia from Siberia to the Far East. Slavic languages are most commonly divided into three major subgroups: East, West, and South. All three are represented in this dataset, with Polish and Czech being the typical West Slavic languages, Russian being the most prominent East Slavic language, and Bulgarian representing the Eastern part of the South Slavic group. Slavic languages are characterized by a rich verbal and nominal inflection system. Typically, verbs mark tense, person, gender, aspect, and mood. Nouns mark gender, number, and case, although in Bulgarian and Macedonian cases are reduced to only nominative and vocative. Masculine nouns additionally mark animacy.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Slavic: Polish", |
| "sec_num": "3.4.3" |
| }, |
| { |
| "text": "Polish data was obtained via a conversion from the largest Polish morphological dictionary (Woli\u0144ski et al., 2020) which is also used as the main data source in the morphological analysis. Table 10 presents a simplified mapping from the original flexemic tagset of Polish (Przepi\u00f3rkowski and Woli\u0144ski, 2003) to the UniMorph schema. The data for the remaining three Slavic languages were obtained from Wiktionary.", |
| "cite_spans": [ |
| { |
| "start": 91, |
| "end": 114, |
| "text": "(Woli\u0144ski et al., 2020)", |
| "ref_id": "BIBREF99" |
| }, |
| { |
| "start": 272, |
| "end": 307, |
| "text": "(Przepi\u00f3rkowski and Woli\u0144ski, 2003)", |
| "ref_id": "BIBREF77" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 189, |
| "end": 197, |
| "text": "Table 10", |
| "ref_id": "TABREF1" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Slavic: Polish", |
| "sec_num": "3.4.3" |
| }, |
| { |
| "text": "The Uralic languages are spoken from the north of Siberia in Russia to Scandinavia and Hungary. They are agglutinating with some subgroups displaying fusional characteristics (e.g., the S\u00e1mi languages). Many of the languages have vowel harmony. Many of the larger case paradigms are made up of spatial cases, sometimes with distinctions for direction and position. Further, most of the languages have possessive suffixes, which can express possession or agreement in non-finite clauses. We use Karelian, Ludic, Livvi, Veps, and V\u00f5ro in the shared task. All the data except V\u00f5ro were exported from the Open corpus of Veps and Karelian languages (VepKar). Veps and Karelian are agglutinative languages with rich suffixal morphology. All inflectional categories in these languages are formed by attaching one or more affixes corresponding to different grammatical categories to the stem.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Uralic: Karelian, Livvi, Ludic, Veps, V\u00f5ro", |
| "sec_num": "3.5" |
| }, |
| { |
| "text": "The presence of one or two stems in the nominal parts of speech and verbs is essential when constructing word forms in the Veps and Karelian languages (Novak, 2019, 57) . In these languages, to build the inflected forms of nouns and verbs, one needs to identify one or two word stems. There are formalized (algorithmic) ways to determine the stem, although not for all words (Novak et al., 2020, 684) .", |
| "cite_spans": [ |
| { |
| "start": 151, |
| "end": 168, |
| "text": "(Novak, 2019, 57)", |
| "ref_id": null |
| }, |
| { |
| "start": 375, |
| "end": 400, |
| "text": "(Novak et al., 2020, 684)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Uralic: Karelian, Livvi, Ludic, Veps, V\u00f5ro", |
| "sec_num": "3.5" |
| }, |
| { |
| "text": "Note that in the Ludic and Livvi dialects of the Karelian language and in the Veps language, reflexive forms of verbs have their own paradigm. Thus, one set of morphological rules is needed for reflexive verbs and another set for non-reflexive verbs.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Uralic: Karelian, Livvi, Ludic, Veps, V\u00f5ro", |
| "sec_num": "3.5" |
| }, |
| { |
| "text": "V\u00f5ro represents the South Estonian dialect group. Similar to other Uralic languages, it has agglutinative, primarily suffixal, morphology. Nouns inflect for grammatical case and number. The current shared task sample contains noun paradigm tables derived from Wiktionary. 7", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Uralic: Karelian, Livvi, Ludic, Veps, V\u00f5ro", |
| "sec_num": "3.5" |
| }, |
| { |
| "text": "The Tungusic genus comprises a group of agglutinative languages spoken from Central and Eastern Siberia to the Far East over the territories of Russia and China. The genus is considered to be a member of the Altaic (or Transeurasian) language family by some researchers, although this is disputed. Tungusic languages are commonly divided into two or three branches (see Oskolskaya et al. (2021) for discussion).", |
| "cite_spans": [ |
| { |
| "start": 370, |
| "end": 394, |
| "text": "Oskolskaya et al. (2021)", |
| "ref_id": "BIBREF73" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Tungusic", |
| "sec_num": "3.6" |
| }, |
| { |
| "text": "The dataset presents two Tungusic languages, namely Evenki and Xibe, belonging to different branches in any approach, with Xibe being quite aberrant from other Tungusic languages. Tungusic languages are characterized by rich verbal and nominal inflection and demonstrate vowel harmony. Typically verbs mark tense, person, aspect, voice and mood. Nouns mark number, case and possession.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Tungusic: Evenki and Xibe", |
| "sec_num": "3.7" |
| }, |
| { |
| "text": "Inflection is achieved through suffixes. Evenki is a typical agglutinative language with almost no fusion whereas Xibe is more fusional.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Tungusic: Evenki and Xibe", |
| "sec_num": "3.7" |
| }, |
| { |
| "text": "The Evenki data was obtained by conversion from a corpus of oral Evenki texts (Kazakevich and Klyachko, 2013) , which uses IPA. The Xibe data was obtained by conversion from a Universal Dependency treebank compiled by Zhou et al. (2020) , which contains textbook and newspaper texts. Xibe texts use the traditional script. 7 The tag conversion schema for Uralic languages is provided here: https://docs. google.com/spreadsheets/d/1RjO_ J22yDB5FH5C24ej7sGGbeFAjcIadJA6ML55tsOI/ edit.", |
| "cite_spans": [ |
| { |
| "start": 78, |
| "end": 109, |
| "text": "(Kazakevich and Klyachko, 2013)", |
| "ref_id": "BIBREF47" |
| }, |
| { |
| "start": 218, |
| "end": 236, |
| "text": "Zhou et al. (2020)", |
| "ref_id": "BIBREF102" |
| }, |
| { |
| "start": 323, |
| "end": 324, |
| "text": "7", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Tungusic: Evenki and Xibe", |
| "sec_num": "3.7" |
| }, |
| { |
| "text": "The Turkic languages of Siberia, spoken mostly within the Russian Federation, range from vulnerable to severely endangered (Eberhard et al., 2021) and represent several branches of Turkic with varying degrees of relatedness (\u0411\u0430\u0441\u043a\u0430\u043a\u043e\u0432, 1969; Tekin, 1990; Sch\u00f6nig, 1999) . They have rich agglutinating morphology, like other Turkic languages, and share many grammatical properties (Washington and Tyers, 2019).", |
| "cite_spans": [ |
| { |
| "start": 123, |
| "end": 146, |
| "text": "(Eberhard et al., 2021)", |
| "ref_id": "BIBREF21" |
| }, |
| { |
| "start": 224, |
| "end": 240, |
| "text": "(\u0411\u0430\u0441\u043a\u0430\u043a\u043e\u0432, 1969;", |
| "ref_id": "BIBREF104" |
| }, |
| { |
| "start": 241, |
| "end": 253, |
| "text": "Tekin, 1990;", |
| "ref_id": "BIBREF85" |
| }, |
| { |
| "start": 254, |
| "end": 268, |
| "text": "Sch\u00f6nig, 1999)", |
| "ref_id": "BIBREF79" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Siberian Turkic: Sakha and Tuvan", |
| "sec_num": "3.8.1" |
| }, |
| { |
| "text": "In this shared task, the Turkic languages of this area are represented by Tuvan (Sayan Turkic) and Sakha (Lena Turkic). For both languages, we make use of the lexicons of the morphological transducers built as part of the Apertium open-source project (Khanna et al., to appear in 2021; Washington et al., to appear in 2021). We use the transducers for Tuvan 8 and Sakha 9 (Ivanova et al., 2019, to appear in 2022) as morphological generators, extracting the paradigms for all the verbs and nouns in the lexicon. We manually design a mapping between the Apertium tagset and the UniMorph schema (Table 8) , based on the system descriptions and additional grammar resources (\u0423\u0431\u0440\u044f\u0442\u043e\u0432\u0430 et al. (1982) for Sakha and \u0418\u0441\u0445\u0430\u043a\u043e\u0432 and \u041f\u0430\u043b\u044c\u043c\u0431\u0430\u0445 (1961) ; Anderson and Harrison (1999) ; Harrison (2000) for Tuvan). Besides the tag mapping, we also include a few conditional rules, such as marking definiteness for nouns in the accusative and genitive cases.", |
| "cite_spans": [ |
| { |
| "start": 251, |
| "end": 266, |
| "text": "(Khanna et al.,", |
| "ref_id": null |
| }, |
| { |
| "start": 671, |
| "end": 694, |
| "text": "(\u0423\u0431\u0440\u044f\u0442\u043e\u0432\u0430 et al. (1982)", |
| "ref_id": null |
| }, |
| { |
| "start": 709, |
| "end": 736, |
| "text": "\u0418\u0441\u0445\u0430\u043a\u043e\u0432 and \u041f\u0430\u043b\u044c\u043c\u0431\u0430\u0445 (1961)", |
| "ref_id": null |
| }, |
| { |
| "start": 752, |
| "end": 767, |
| "text": "Harrison (1999)", |
| "ref_id": "BIBREF5" |
| }, |
| { |
| "start": 770, |
| "end": 785, |
| "text": "Harrison (2000)", |
| "ref_id": "BIBREF39" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 593, |
| "end": 602, |
| "text": "(Table 8)", |
| "ref_id": "TABREF13" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Siberian Turkic: Sakha and Tuvan", |
| "sec_num": "3.8.1" |
| }, |
| { |
| "text": "Since the UniMorph schema in its current version is not well-suited to capture the richness of Turkic morphology, we exclude many forms with morphological attributes that do not have a close equivalent in UniMorph. We also omit forms with affixes that are considered quasi-derivational rather than inflectional, such as the desiderative /-ksA/ in Tuvan (Washington et al., 2016), with the exception of the negative marker. These constraints greatly reduce the sizes of the verbal paradigms: the median number of forms per lemma is 234 and 87 for Tuvan and Sakha respectively, compared to roughly 5,700 forms per lemma produced by either generator. Our tag conversion and paradigm filtering code is publicly released. 10", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Siberian Turkic: Sakha and Tuvan", |
| "sec_num": "3.8.1" |
| }, |
| { |
| "text": "One of the further west Turkic languages, Turkish is part of the Oghuz branch, and, like the other languages of this family, it is highly agglutinative.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Turkic: Turkish", |
| "sec_num": "3.8.2" |
| }, |
| { |
| "text": "In this work, we vastly expanded the existing UniMorph inflection tables. As with the Siberian Turkic languages, it was necessary to omit many forms from the paradigm as the UniMorph schema is not well-suited for Turkic languages. For this reason, we only included the forms that may appear in main clauses. Other than this limitation, we tried to include all possible tense-aspect-mood combinations, resulting in 30 series of forms, each including 3 persons and 2 numbers. The nominal coverage is less comprehensive and includes forms with case and possessive suffixes.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Turkic: Turkish", |
| "sec_num": "3.8.2" |
| }, |
| { |
| "text": "3.9 Austronesian 3.9.1 Malayo-Polynesian: Indonesian Indonesian or Bahasa Indonesia is the official language of Indonesia. It belongs to the Austronesian language family and it is written with the Latin script.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Turkic: Turkish", |
| "sec_num": "3.8.2" |
| }, |
| { |
| "text": "Indonesian does not mark grammatical case, gender, or tense. Words are composed from their roots through affixation, compounding, or reduplication. The four types of Indonesian affixes are prefixes, suffixes, circumfixes (combination of prefixes and suffixes), and infixes (inside the base form). Indonesian uses both full and partial reduplication processes to form words. Full reduplication is often used to express the plural forms of nouns, while partial reduplication is typically used to derive forms that might have a different category than their base forms. Unlike English, the distinction between inflectional and derivational morphological processes in Indonesian is not always clear (Pisceldo et al., 2008) .", |
| "cite_spans": [ |
| { |
| "start": 695, |
| "end": 718, |
| "text": "(Pisceldo et al., 2008)", |
| "ref_id": "BIBREF75" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Turkic: Turkish", |
| "sec_num": "3.8.2" |
| }, |
| { |
| "text": "In this shared task, the Indonesian data is created by bootstrapping the data from an Indonesian Wikipedia dump. Using a list of possible Indonesian affixes, we collect unique word forms from Wikipedia and analyze them using MorphInd (Larasati et al., 2011), a morphological analyzer tool for Indonesian based on an FST. We manually create a mapping between the MorphInd tagset and the UniMorph schema. We then use this mapping and apply some additional rule-based formulas created by Indonesian linguists to build the final dataset (Table 9) . 3.9.2 Malayo-Polynesian: Kodi/Kodhi Kodi or Kodhi [ko\u00e2i] is spoken in Sumba Island, eastern Indonesia (Ghanggo Ate, 2020). Regarding its linguistic classification, Kodi belongs to the Central-Eastern subgroup of Austronesian, related to Sumba-Hawu languages. Based on the linguistic fieldwork observations done by Ghanggo Ate (2020), it may be tentatively concluded that there are only two Kodi dialects: Kodi Bhokolo and Mbangedho-Mbalaghar. Even though some work has been done on Kodi (Ghanggo Ate, to appear in 2021), it remains a largely under-documented language. Further, Kodi is vulnerable or threatened because Indonesian, the prestigious national language, is used in most sociolinguistic domains outside the domestic sphere.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 533, |
| "end": 542, |
| "text": "(Table 9)", |
| "ref_id": "TABREF14" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Turkic: Turkish", |
| "sec_num": "3.8.2" |
| }, |
| { |
| "text": "A prominent linguistic feature of Kodi is its clitic system, which is pervasive in various syntactic categories-verbs, nouns, and adjectives-and marks person (1, 2, 3) and number (SG vs. PL). In addition, Kodi contains four sets of pronominal clitics that agree with their antecedent: NOM(inative) proclitics, ACC(usative) enclitics, DAT(ive) enclitics and GEN(initive) enclitics. Interestingly, these clitic sets are not markers of NOM, ACC, DAT, or GEN grammatical case-as in Malayalam or Latin-but rather identify the head for TERM relations (subject and object). Thus, by default, pronominal clitics are core grammatical arguments reflecting subject and object.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Turkic: Turkish", |
| "sec_num": "3.8.2" |
| }, |
| { |
| "text": "For the analyses of the features of Kodi clitics, the data freshly collected in the fieldwork funded by the Endangered Language Fund is annotated. Then, the collected data is converted to the UniMorph task format, which has the lemmas, the word forms, and the morphosyntactic features of Kodi.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Turkic: Turkish", |
| "sec_num": "3.8.2" |
| }, |
| { |
| "text": "The Seneca language is an indigenous Native American language from the Iroquoian (Hodin\u00f6h\u0161\u00f6ni) language family. Seneca is considered critically endangered and is currently estimated to have fewer than 50 first-language speakers left, most of whom are elders. The language is spoken mainly in three reservations located in Western New York: Allegany, Cattaraugus, and Tonawanda.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Northern Iroquoian: Seneca", |
| "sec_num": "3.10.1" |
| }, |
| { |
| "text": "Seneca possesses highly complex morphological features, with a combination of both agglutinative and fusional properties. The data presented here consists of inflectional paradigms for Seneca verbs, the basic structure of which is composed of a verb base that describes an event or state of action. In virtually all cases, the verb base would be preceded by a pronominal prefix which indicates the agent, the patient, or both for the event or state, and followed by an aspect suffix which usually marks a habitual or a stative state.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Northern Iroquoian: Seneca", |
| "sec_num": "3.10.1" |
| }, |
| { |
| "text": "(1) ha skatkw\u00eb s it he laugh HAB He laughs.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Northern Iroquoian: Seneca", |
| "sec_num": "3.10.1" |
| }, |
| { |
| "text": "In some other scenarios, for instance, when the verb is describing a factual, future or hypothetical event, a modal prefix is attached before the pronominal prefix and the aspect suffix marks a punctual state instead. The structures and orders of the prefixes can be more complicated depending on, e.g., whether the action denoted by the verb is repetitive or negative; these details are realized by adding a prepronominal prefix before the modal prefix.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Northern Iroquoian: Seneca", |
| "sec_num": "3.10.1" |
| }, |
| { |
| "text": "Ash\u00e1ninka is an Arawak language with more than 70,000 speakers in Central and Eastern Peru and in the state of Acre in Eastern Brazil, in a geographical region located between the eastern foothills of the Andes and the western fringe of the Amazon basin (Mihas, 2017; Mayor Aparicio and Bodmer, 2009) . Although it is the most widely spoken Amazonian language in Peru, certain varieties, such as Alto Peren\u00e9, are highly endangered. It is an agglutinating, polysynthetic, verb-initial language. The verb is the most morphologically complex word class, with a rich repertoire of aspectual and modal categories. The language lacks case marking, except for one locative suffix; grammatical relations of subject and object are indexed as affixes on the verb itself. Other notable linguistic features of the language include a distinction between alienably and inalienably possessed nouns, obligatory marking of reality status (realis/irrealis) on the verb, a rich system of applicative suffixes, serial verb constructions, and pragmatically conditioned split intransitivity.", |
| "cite_spans": [ |
| { |
| "start": 254, |
| "end": 267, |
| "text": "(Mihas, 2017;", |
| "ref_id": "BIBREF67" |
| }, |
| { |
| "start": 268, |
| "end": 300, |
| "text": "Mayor Aparicio and Bodmer, 2009)", |
| "ref_id": "BIBREF63" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Southern Arawakan: Ash\u00e1ninka", |
| "sec_num": "3.11.1" |
| }, |
| { |
| "text": "The corpus consists of inflected nouns and verbs from the variety spoken in the Tambo river of Central Peru. The annotated nouns take possessor prefixes, locative case and/or plural marking, while the annotated verbs take subject prefixes, reality status (realis/irrealis), and/or perfective aspect.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Southern Arawakan: Ash\u00e1ninka", |
| "sec_num": "3.11.1" |
| }, |
| { |
| "text": "Yanesha is an Amazonian language from the Pre-Andine subgroup of Arawakan family (Adelaar and Muysken, 2004), spoken in Central Peru by between 3 and 5 thousand people. It has two linguistic variants that correspond to the upriver and downriver areas, both mutually intelligible.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Southern Arawakan: Yanesha", |
| "sec_num": "3.11.2" |
| }, |
| { |
| "text": "Yanesha is an agglutinating, polysynthetic language with a VSO word order. Nouns and verbs are the two major word classes while the adjective word class is questionable due to the absence of non-derived forms. The verb is the most morphologically complex word class and the only obligatory constituent of a clause (Dixon and Aikhenvald, 1999) .", |
| "cite_spans": [ |
| { |
| "start": 314, |
| "end": 342, |
| "text": "(Dixon and Aikhenvald, 1999)", |
| "ref_id": "BIBREF19" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Southern Arawakan: Yanesha", |
| "sec_num": "3.11.2" |
| }, |
| { |
| "text": "Among other typologically remarkable features, we find that the language lacks the distinction in grammatical gender, the subject cross-referencing morphemes and one of the causatives are prefixes; all other verbal affixes are suffixes, and nouns and classifiers may be incorporated in the verb (Wise, 2002) .", |
| "cite_spans": [ |
| { |
| "start": 295, |
| "end": 307, |
| "text": "(Wise, 2002)", |
| "ref_id": "BIBREF97" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Southern Arawakan: Yanesha", |
| "sec_num": "3.11.2" |
| }, |
| { |
| "text": "The corpus consists of inflected nouns and verbs from both dialectal varieties. The annotated nouns take possessor prefixes, plural marking, and locative case, while the annotated verbs take subject prefixes.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Southern Arawakan: Yanesha", |
| "sec_num": "3.11.2" |
| }, |
| { |
| "text": "The Chukotko-Kamchatkan languages, spoken in the far east of the Russian Federation, are represented in this dataset by two endangered languages, Chukchi and Itelmen (Eberhard et al., 2021) .", |
| "cite_spans": [ |
| { |
| "start": 166, |
| "end": 189, |
| "text": "(Eberhard et al., 2021)", |
| "ref_id": "BIBREF21" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Chukotko-Kamchatkan", |
| "sec_num": "3.12" |
| }, |
| { |
| "text": "Chukchi is a polysynthetic language that exhibits polypersonal agreement, ergative-absolutive alignment, and a subject-object-verb basic word order in transitive clauses (Tyers and Mishchenkova, 2020) . We use the data of the Amguema corpus, available through the Chuklang website, 11 comprised of transcriptions of spoken Chukchi in the Amguema variant. The Amguema data had been annotated in the CoNLL-U format by Tyers and Mishchenkova (2020) , and we convert it to the UniMorph format using the conversion system of McCarthy et al. (2018).", |
| "cite_spans": [ |
| { |
| "start": 170, |
| "end": 200, |
| "text": "(Tyers and Mishchenkova, 2020)", |
| "ref_id": "BIBREF87" |
| }, |
| { |
| "start": 416, |
| "end": 445, |
| "text": "Tyers and Mishchenkova (2020)", |
| "ref_id": "BIBREF87" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Chukotko-Kamchatkan: Chukchi", |
| "sec_num": "3.12.1" |
| }, |
| { |
| "text": "Itelmen is a language spoken on the western coast of the Kamchatka Peninsula. The language is con-sidered to be highly endangered since it stopped been transferred from elders to youth \u223c50 years ago (most are Russian-speaking monolinguals). The language is agglutinative and primarily uses suffixes. For instance, the plural form of a noun is expressed by the suffix -Pn. We note that the plural form only exists in four grammatical cases (NOM, DAT, LOC, VOC). 12 The same plural suffix transforms a noun into an adjective. Verbs mark both subjects (with prefixes and suffixes) and objects (with suffixes). For instance, the first person subject is marked by attaching the prefix tand the suffix -\u010den (Volodin, 1976) . 13 The Itelmen data presented in the task was collected through fieldwork and manually annotated according to the UniMorph schema.", |
| "cite_spans": [ |
| { |
| "start": 461, |
| "end": 463, |
| "text": "12", |
| "ref_id": null |
| }, |
| { |
| "start": 701, |
| "end": 716, |
| "text": "(Volodin, 1976)", |
| "ref_id": "BIBREF89" |
| }, |
| { |
| "start": 719, |
| "end": 721, |
| "text": "13", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Chukotko-Kamchatkan: Itelmen", |
| "sec_num": "3.12.2" |
| }, |
| { |
| "text": "3.13 Trans-New Guinea", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Chukotko-Kamchatkan: Itelmen", |
| "sec_num": "3.12.2" |
| }, |
| { |
| "text": "Eibela, or Aimele, is an endangered language spoken by a small (\u223c300 speakers) community in Lake Campbell, Western Province, Papua New Guinea. Eibela morphology is exclusively suffixing. Verbs conjugate for tense, aspect, mood, evidentiality and exhibit complex paradigms with a high degree of irregularity. Generally, verbs cab be grouped into three classes based on their stems. Verbal inflectional classes present various kinds of stem alternations and suppletion. As Aiton (2016b) notes, the present and past forms are produced either through stem changes or by a concatenative suffix. In some cases, the forms can be quite similar (such as na:gla: 'be sick.PST' and na:glE 'be sick.PRS'). The future tense forms are typically inflected using suffixes. The current sample has been derived from interlinear texts from Aiton (2016a) and contains mostly partial paradigms.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Bosavi: Eibela", |
| "sec_num": "3.13.1" |
| }, |
| { |
| "text": "As in the previous editions, each instance in the provided training and development is in a form of a triple (lemma, tag, inflected form). The test set, on the other hand, was released with only lemmas and tags (i.e. without the target inflections). Producing these data sets required a few extra steps, which we discuss in this section.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Data Preparation", |
| "sec_num": "4" |
| }, |
| { |
| "text": "Conversion into the UniMorph schema. After the data collection was finalised for the above languages, we converted them to the UniMorph schema-canonicalising them in the process. 14 This process consisted mainly of typo corrections (e.g. removing an incorrectly placed space in a tag, \"PRIV \" \u2192 \"PRIV\"), removing redundant tags (e.g. duplicated verb annotation, \"V;V.PTCP\" \u2192 \"V.PTCP\"), and fixing tags to conform to the Uni-Morph schema (e.g. \"2;INCL\" \u2192 \"2+INCL\"). These changes were implemented via languagespecific Bash scripts. Given this freshly converted data, we canonicalised its tag annotations, making use of https://github.com/unimorph/ um-canonicalize. This process sorts the inflection tags into their canonical order and verifies that all the used tags are present in the ground truth UniMorph schema, flagging potential data issues in the process.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Data Preparation", |
| "sec_num": "4" |
| }, |
| { |
| "text": "Data splitting. Given the canonicalised data as described above, we removed all instances with duplicated <lemma; tags> pair-as these instances were ambiguous with respect to their target inflected form-and removed all forms other than verbs, nouns, or adjectives. We then capped the dataset sizes to a maximum of 100,000 instances per language, subsampling when necessary. Finally, we create a 70-10-20 train-dev-test split per language, splitting the data across these sets at the instance level (as opposed to, e.g., the lemma one). As such, the information about a lemma's declension or inflection class is spread out across these train, dev and test sets, making this task much simpler than if one had to predict the entire class from the lemma's form alone, as done by, e.g., Williams et al. (2020) and Liu and Hulden (2021) .", |
| "cite_spans": [ |
| { |
| "start": 782, |
| "end": 804, |
| "text": "Williams et al. (2020)", |
| "ref_id": "BIBREF96" |
| }, |
| { |
| "start": 809, |
| "end": 830, |
| "text": "Liu and Hulden (2021)", |
| "ref_id": "BIBREF58" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Data Preparation", |
| "sec_num": "4" |
| }, |
| { |
| "text": "The organizers provide four neural systems as baselines, a product of two models and optional data augmentation. The first model is a transformer (Vaswani et al., 2017, TRM) , and the second model is an adaption of the transformer to character-level transduction tasks (Wu et al., 2021, CHR-TRM) , which holds the state-of-the-art on the 2017 SIG-MORPHON shared task data. Both models follow the hyperparameters of Wu et al. (2021) . The optional data augmentation follows the technique proposed by Anastasopoulos and Neubig (2019) Development afb 19,861 2,184 7,595 2,996 4,208 1,510 -----amh 20,254 670 20,280 1,599 829 195 4,096 668 --668 668 ara 31,002 635 53,365 1,703 58,187 742 -----arz 8,178 1,320 10,533 3,205 6,551 1,771 -----heb 28,635 1,041 3,666 142 ------847 847 syc 596 187 724 329 158 86 --261 77 - Table 2 : Number of samples and unique lemmata (the second number in each column) in each word class in the shared task data, aggregated over all splits. Here: \"V\" -verbs, \"N\" -nouns, \"ADJ\" -adjectives, \"V.CVB\"converbs, \"V.PTCP\" -participles, \"V.MSDR\" -masdars.", |
| "cite_spans": [ |
| { |
| "start": 146, |
| "end": 173, |
| "text": "(Vaswani et al., 2017, TRM)", |
| "ref_id": null |
| }, |
| { |
| "start": 269, |
| "end": 295, |
| "text": "(Wu et al., 2021, CHR-TRM)", |
| "ref_id": null |
| }, |
| { |
| "start": 415, |
| "end": 431, |
| "text": "Wu et al. (2021)", |
| "ref_id": "BIBREF100" |
| }, |
| { |
| "start": 499, |
| "end": 531, |
| "text": "Anastasopoulos and Neubig (2019)", |
| "ref_id": "BIBREF4" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 532, |
| "end": 864, |
| "text": "Development afb 19,861 2,184 7,595 2,996 4,208 1,510 -----amh 20,254 670 20,280 1,599 829 195 4,096 668 --668 668 ara 31,002 635 53,365 1,703 58,187 742 -----arz 8,178 1,320 10,533 3,205 6,551 1,771 -----heb 28,635 1,041 3,666 142 ------847 847 syc 596 187 724 329 158 86 --261 77 -", |
| "ref_id": "TABREF1" |
| }, |
| { |
| "start": 865, |
| "end": 872, |
| "text": "Table 2", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Baseline Systems", |
| "sec_num": "5" |
| }, |
| { |
| "text": ". Rely- L V N ADJ V.CVB V.PTCP V.MSDR", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Baseline Systems", |
| "sec_num": "5" |
| }, |
| { |
| "text": "ing on a simple character-level alignment between the lemma and the form, this technique replaces shared substrings of length > 3 with random characters from the language's alphabet, producing hallucinated lemma-tag-form triples. Data augmentation (+AUG) is applied to languages with fewer than 10K training instances, and 10K examples are generated for each language.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Baseline Systems", |
| "sec_num": "5" |
| }, |
| { |
| "text": "6 Submitted Systems model loss, predicting Levenshtein operations (copy, delete, replace and add) as a multitask objective going from lemma to inflected form, label smoothing based on other characters in the same language (language-wise label smoothing), and scheduled sampling (Bengio et al., 2015) .", |
| "cite_spans": [ |
| { |
| "start": 278, |
| "end": 299, |
| "text": "(Bengio et al., 2015)", |
| "ref_id": "BIBREF7" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Baseline Systems", |
| "sec_num": "5" |
| }, |
| { |
| "text": "BME Team BME's system is an LSTM encoderdecoder model based on the work of Faruqui et al. (2016) , with three-step training where the model is first trained on all languages, then fine-tuned on each language family, and finally fine-tuned on individual languages. A different type of data augmentation technique inspired by Neuvel and Fulop (2002) is also used in the first two steps. Team BME also perform ablation studies and show that the augmentation techniques and the three training steps often help but sometimes have a negative effect.", |
| "cite_spans": [ |
| { |
| "start": 75, |
| "end": 96, |
| "text": "Faruqui et al. (2016)", |
| "ref_id": "BIBREF25" |
| }, |
| { |
| "start": 324, |
| "end": 347, |
| "text": "Neuvel and Fulop (2002)", |
| "ref_id": "BIBREF69" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Baseline Systems", |
| "sec_num": "5" |
| }, |
| { |
| "text": "Following the evaluation procedure established in the previous shared task iterations, we compare all systems in terms of their test set accuracy. In addition, we perform an extensive error analysis for most languages.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Evaluation", |
| "sec_num": "7" |
| }, |
| { |
| "text": "As Analysis for out-of-vocabulary lemmas Table 4 shows the differences in performance between the lemmas present both in training and test sets and the \"unknown\" lemmas. A closer inspection of this table shows that while BME and GUClasp models have an accuracy gap of 5.5% and 3% respectively between previously known and unknown lemmas, the transformer-based architectures show an accuracy gap from 9% to 16%. This larger gap, however, is partly explained by the better performance of the transformer-based models on previously seen lemmas (around 75%, while BME's performance is 71% and GUClasp's is 66%). The performance on the previously unseen lemmas, on the other hand, is mostly driven by data augmentation. The models without data augmentation have an accuracy around 60% on these lemmas, while all other models achieve around 65% on previously unseen lemmas. This is in line with the findings of Liu and Hulden (2021) , who show that the transducer's performance on previously seen words can be greatly improved by simply training the models to perform the trivial task of copying random lemmas during training-a method somewhat related to data augmentation.", |
| "cite_spans": [ |
| { |
| "start": 905, |
| "end": 926, |
| "text": "Liu and Hulden (2021)", |
| "ref_id": "BIBREF58" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 41, |
| "end": 48, |
| "text": "Table 4", |
| "ref_id": "TABREF7" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Results", |
| "sec_num": "8" |
| }, |
| { |
| "text": "Analysis for the most challenging inflections Table 5 shows the accuracy of the submitted systems on the \"most challenging\" test instances, where all four baselines failed to predict the target form correctly. Frequently observed types of such cases include:", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 46, |
| "end": 53, |
| "text": "Table 5", |
| "ref_id": "TABREF9" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Results", |
| "sec_num": "8" |
| }, |
| { |
| "text": "\u2022 Unusual alternations of some letters in particular lexemes which are hard to generalize;", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Results", |
| "sec_num": "8" |
| }, |
| { |
| "text": "\u2022 Ambiguity of the target forms. Certain lemmas allow some variation in forms, while the test set only lists a single exclusive golden form for each (lemma, tags) combination. In most cases, multiple acceptable forms may be hardly distinguishable in spoken language. For instance, they may only differ by an unstressed vowel or be orthographic variants of the same form.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Results", |
| "sec_num": "8" |
| }, |
| { |
| "text": "\u2022 Multiword expressions are challenging when agreement is required. UniMorph does not provide dependency information, however, the information can be inferred from similar samples or other parts of the same lemma paradigm. The system's ability to make generalizations from a sequence down to its subsequences essentially depends on its architecture.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Results", |
| "sec_num": "8" |
| }, |
| { |
| "text": "\u2022 Errors in the test sets. Still, a small percentage of errors come from the data itself.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Results", |
| "sec_num": "8" |
| }, |
| { |
| "text": "As Elsner et al. (2019) note, accuracy-level evaluation might be sufficient to compare model variants but does not provide much insight into the understanding of morphological systems and their learnability. Therefore, we now turn to a more detailed analysis of mispredictions made by the systems. For the purpose of this study, we will rely on the error type taxonomy proposed by Gorman et al. (2019) and .", |
| "cite_spans": [ |
| { |
| "start": 381, |
| "end": 401, |
| "text": "Gorman et al. (2019)", |
| "ref_id": "BIBREF33" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Error Analysis", |
| "sec_num": "9" |
| }, |
| { |
| "text": "For the Evenki language, GUClasp tends to shorten the resulting words, sometimes generating theoretically impossible forms. However, in several cases, the result is practically correct but only in case of a different dialect, such as abull@n instead of abuld@n. The performance is better for nominal wordforms (74.27 accuracy for nouns only vs. 30.55 for verbs only). This is perhaps due to the higher regularity of nominal forms. BME is performing slightly better for the Evenki language, with errors in vowel harmony (such as ahatkanmo instead of ahatkanm@). In contrast with GUClasp, it tends to generate longer forms, adding unnecessary suffixes. The problems with dialectal forms can be found as well. The performance for Xibe is worse for both systems, though BME is better, despite the simpler morphology-perhaps it is due to the complexity of the Xibe script. At least in one instance, one of the systems generated a form with a Latin letter n instead of Xibe letters.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Evenki and Xibe", |
| "sec_num": "9.1" |
| }, |
| { |
| "text": "Both GUClasp and BME generated 350 nominal, verbal and adjectival forms with less than 50% ac-curacy. This includes forms that are hypothetically correct despite being historically unattested (e.g., abydwtkwn 'your (M.PL) loss'). Both systems performed better on nominal and adjectival forms than verbal forms. This may be explained by the higher morphological regularity of nominal forms relative to verbal forms; nominal/adjectival inflections typically follow linear affixation rules (e.g., suffixation) while verbal forms follow the same rules in addition to non-concatenative processes. Further, both systems handle lexemes with two or three letters (e.g., dn 'to judge') poorly compared to longer lexemes (e.g., bt . nwt' 'conception'). Where both systems generate multiple verbal forms for the same lexeme, the consonantal root is inconsistent. Finally, as expected, lexicalised phrases (e.g., kln\u0161 'everyone', derived from the high-frequency contraction kl 'every' and n\u0161 'person') and homomorphs (e.g., ql' 'an expression (n.)' or 'to fry (v.)') are handled poorly. Comparatively, the BME system performed worse than GUClasp, especially in terms of vowel diacritic placement and consonant doubling, which are consistently hypercorrected in both cases (e.g., h . byb' > h . abbbbay; h . yltn' > h . aallto').", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Syriac", |
| "sec_num": "9.2" |
| }, |
| { |
| "text": "Both submitted systems performed well on the Amharic data, BME (98.16% accuracy) somewhat better than GUClasp (93.81% accuracy), though neither outperformed the baseline models. For both systems, target errors represented a significant proportion of the errors, 32.35% for BME, 24.08% for GUClasp. Many of these involved alternative plural forms of nouns. The data included only the most frequent plural forms when there were alternatives, sometimes excluding uncommon but still possible forms. In some cases only an irregular form appeared in the data, and the system \"erroneously\" predicted the regular form with the suffix -(w)o\u010d, which also correct. For example, BME produced hawaryawo\u010d, the regular plural of hawarya 'apostle', instead of the expected irregular plural hawaryat. Another source of target errors was the confusion resulting from multiple representations for the phonemes /h,P,s,s'/ in the Amharic orthography. Again, the data included only the common spelling for a given lemma or inflected form, but alternative spellings are usually also attested. Many of the \"errors\" consisted of predicting correct forms with one of these phonemes spelled differently than in the expected form.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Amharic", |
| "sec_num": "9.3" |
| }, |
| { |
| "text": "The largest category of errors for both systems (unlike the baseline systems) were allomorphy errors, 51.76% for BME, 62.65% for GUClasp. Most of these resulted from the confusion between vowels in verbal templates. Particularly common were vowel errors in jussive-imperative (IMP) forms. Most Amharic verb lemmas belong to one of two inflection classes, each based on roots consisting of three consonants. The vowels in the templates for these categories are identical in the perfective (PRF), imperfective (IPFV), and converb (V.CVB) forms, but differ in the infinitive (V.MSDR) and jussive-imperative, where class A has the vowels .1.@ and class B has the vowels .@.1.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Amharic", |
| "sec_num": "9.3" |
| }, |
| { |
| "text": "Both systems also produced a significant number of silly errors-incorrect forms that could not be explained otherwise. Most of these consisted of consonant deletion, replacing one consonant with another, or repeating a consonant-vowel sequence.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Amharic", |
| "sec_num": "9.3" |
| }, |
| { |
| "text": "Polish is among languages for which both systems and all the baselines achieved the highest accuracy results. BME, with 99.54% accuracy, is doing slightly better than GUClasp (98.52%). However, neither system exceeds the baseline results (99.67-99.70%).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Polish", |
| "sec_num": "9.4" |
| }, |
| { |
| "text": "Most of the errors made by both systems were already noted and classified by Gorman et al. (2019) and follow from typical irregularities in Polish. For example, masculine nouns have two GEN.SG suffixes: -a and -u. The latter is typical for inanimate nouns but the former could be used both with animate and inanimate nouns, which makes it highly unpredictable and causes production of incorrect forms such as negatywa, rabunka instead of negatywu 'negative', rabunku 'robbery'. Both systems are vulnerable to such mistakes. Another example would be the GEN.PL forms of plurale tantum nouns, which could have -\u00f3w or zero suffix, leading to errors such as: tekstyli, wiktua\u0142 instead of tekstyli\u00f3w 'textiles', wiktua\u0142\u00f3w 'victuals'. Some loan words in Polish have fully (mango, marines, monsieur) or partially (millenium, in singular only) syncretic inflectional paradigms. This phenomenon is hard to predict, as the vast majority of Polish nouns inflect regularly. Both systems tend to produce inflected forms of those nouns according to their regular endings, which would be otherwise correct if not for their syncretic paradigms.", |
| "cite_spans": [ |
| { |
| "start": 77, |
| "end": 97, |
| "text": "Gorman et al. (2019)", |
| "ref_id": "BIBREF33" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Polish", |
| "sec_num": "9.4" |
| }, |
| { |
| "text": "One area in which BME returns significantly bet-ter results than GUClasp are imperative forms. Polish imperative forms follow a few different patterns involving some vowel alternations but in general are fairly regular. For the 364 imperative forms in the test data set, BME produced only 12 errors, mostly excusable and concerning existing phonetic alternations which could cause some problems even for native or L2 speakers. GUClasp, however, produced 61 erroneous imperative forms, some of them being examples of overgeneralization of the zero suffix pattern for first person singular imperatives (wyja\u015bn instead of wyja\u015bnij for the verb WY-JA\u015aNI\u0106 'explain').", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Polish", |
| "sec_num": "9.4" |
| }, |
| { |
| "text": "Interestingly, both systems sometimes produce forms that are considered incorrect in standard Polish, but are quite often used colloquially by native speakers. Both BME and GUClasp generated the form podnies\u0105 si\u0119 (instead of podnios\u0105 si\u0119 'they will rise'). Moreover, GUClasp generated the form podesz\u0142e\u015b (instead of podszed\u0142e\u015b 'you came up').", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Polish", |
| "sec_num": "9.4" |
| }, |
| { |
| "text": "Similar to Polish and many other high-resource languages, the accuracy of all systems on Russian is high, with BME being the best-performing model (98.07%). Majority of errors consistently made by all systems (including the baseline ones) are related to the different inflections for animate and inanimate nouns in the accusative case. In particular, UniMorph does not provide the corresponding animacy feature for nouns, an issue that has also been reported previously by Gorman et al. (2019) .", |
| "cite_spans": [ |
| { |
| "start": 473, |
| "end": 493, |
| "text": "Gorman et al. (2019)", |
| "ref_id": "BIBREF33" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Russian", |
| "sec_num": "9.5" |
| }, |
| { |
| "text": "The formation of the short forms of adjectives and participles with -\u0435\u043dand -\u0435\u043d\u043dis another source of misprediction. The systems either generate an incorrect number of \u043d, as in *\u0443\u043c\u0435\u0440\u0435\u043d\u0430 (should be \u0443\u043c\u0435\u0440\u0435\u043d\u043d\u0430 'moderate'), or fail to attach the suffix in cases that require some repetition in the production, as in *\u0436\u0435\u0440\u0442\u0432\u0435\u043d (should be \u0436\u0435\u0440\u0442\u0432\u0435\u043d\u0435\u043d 'sacrificial'), i.e. generation stops after the first \u0435\u043d is produced. In addition to that, the systems often mispredict alternations of \u0435 and \u0451, as in *\u043e\u0448\u0435\u043b\u043e\u043c\u043b\u0451\u043d\u043d\u044b instead of \u043e\u0448\u0435\u043b\u043e\u043c\u043b\u0435\u043d\u044b 'overwhelmed'. The same error also occurs in the formation of past participle forms such as *\u043f\u043e\u043a\u043e\u0440\u043c\u043b\u0451\u043d\u043d\u044b\u0439 (should be \u043f\u043e\u043a\u043e\u0440\u043c\u043b\u0435\u043d\u043d\u044b\u0439 'fed'). Further, we also observe it in noun declension, more specifically, in the prediction of the singular instrumental form: *\u0441\u043b\u0435\u0441\u0430\u0440\u0451\u043c (should be \u0441\u043b\u0435\u0441\u0430\u0440\u0435\u043c 'locksmith'), *\u0433\u043e\u0441\u0442\u044c\u0451\u0439 (should be \u0433\u043e\u0441\u0442\u044c\u0435\u0439, 'female guest'). Additionally, we observe more errors in the prediction of the instrumental case forms, mainly due to allomorphy. In many cases, the systems would have benefited from observing stress patterns or grammatical gender. For instance, consider the feminine \u0430\u043a\u0432\u0430\u0440\u0435\u043b\u044c 'aquarelle' and the masculine \u043f\u0430\u0440\u043e\u043b\u044c 'password'. In order to make a correct prediction, a model should either be explicitly provided with the grammatical gender, or a partial paradigm (e.g., the dative and genitive singular slots) for the corresponding lemma should be observed in the training set. Indeed, the latter is often the case, but the systems still fail to make a correct inference.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Russian", |
| "sec_num": "9.5" |
| }, |
| { |
| "text": "Finally, multiword expressions present themselves as an extra challenge to the models. In most cases, the test lemmas also appeared in the training set, therefore the systems could infer the dependency information from other parts of the same lexeme. Still, Russian multiword expessions appeared to be harder to inflect, probably as they show richer combinatory diversity. For instance, \u044d\u043b\u0435\u043a\u0442\u0440\u043e\u043c\u0430\u0433\u043d\u0438\u0442\u043d\u043e\u0435 \u0432\u0437\u0430\u0438\u043c\u043e\u0434\u0435\u0439\u0441\u0442\u0432\u0438\u0435 'electromagnetic interaction' for the plural instrumental case slot is mispredicted as *\u044d\u043b\u0435\u043a\u0442\u0440\u043e\u043c\u0430\u0433\u043d\u0438\u0442\u043d\u044b\u043c\u0438 \u0432\u0437\u0430\u0438\u043c\u043e\u0434\u0435\u0439\u0441\u0442\u0432\u0438\u044f, i.e the adjective part is correct while the noun form is not. As Table 7 illustrates, the accuracy gap in predicting multiword expressions with lemmas inor out-of-vocabulary is quite large.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 606, |
| "end": 613, |
| "text": "Table 7", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Russian", |
| "sec_num": "9.5" |
| }, |
| { |
| "text": "The Ludic language, in comparison with the Karelian, Livvi and Veps languages, has the smallest number of lemmas (31 verbs, 77 nouns and 1 adjective) and has the lowest accuracy (16-60%). Therefore, the incomplete data is the main cause of errors in the morphological analyzers working with the Ludic dialect ('target errors' in the error taxonomy proposed by Gorman et al.) .", |
| "cite_spans": [ |
| { |
| "start": 360, |
| "end": 374, |
| "text": "Gorman et al.)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Ludic", |
| "sec_num": "9.6" |
| }, |
| { |
| "text": "Testing on the Sorani data yields high accuracy values, although there are errors in some cases. More than 200 lemmas and 16K samples were generated. Both BME and GUClasp generate regular nominal and verbal forms with a high accuracy of 99.46% and 96.6% respectively, although neither system exceeds the baseline results of 99.94% and 99.97%. Kurdish has a complex morphological system with defective paradigms and second-position person markers. Double clitic and affix-clitic construc-tions can mark subjects or objects in a verbal construction and ditranstives are made with applicative markers. Such morphological structures can be the reason for the few issues that still occur.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Kurdish", |
| "sec_num": "9.7" |
| }, |
| { |
| "text": "Both BME and GUClasp predict the majority of the inflected forms correctly, achieving test accuracies of over 99.6% on both Tuvan and Sakha, with BME performing slightly better on both languages. The remaining errors are generally caused by misapplications of morphophonology, either by the prediction system or by the data generator itself.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Tuvan and Sakha", |
| "sec_num": "9.8" |
| }, |
| { |
| "text": "Since the forms treated as ground truth were automatically generated by morphological transducers ( \u00a73.8.1), the mismatch between the prediction and the reference might be due to 'target errors' where the reference itself is wrong (Gorman et al., 2019) . For the BME system, target errors account for 1/8 disagreement cases for Tuvan and 3/13 for Sakha, although for all of them the system's prediction is indeed incorrect as well. For GUClasp, the reference is wrong in 19/62 cases for Tuvan (four of them also have an incorrect lemma, which makes it impossible to judge the correctness of any inflection) and 43/90 for Sakha. Interestingly, GUClasp actually predicts the correct inflected form for 27/43 and 3/15 15 target error cases for Sakha and Tuvan, respectively. The actual failure cases for both BME and GUClasp are largely allomorphy errors, per Gorman et al.'s classification. Common problems include consonant alternation (Sakha *\u043e\u0445\u0441\u0443\u0441\u0443\u04a5 instead of \u043e\u0445\u0441\u0443\u04bb\u0443\u04a5), vowel harmony (Tuvan *\u0438\u0436\u0438\u0430\u0440\u043b\u0435\u0440 instead of \u0438\u0436\u0438\u0433\u0435\u0440\u043b\u0435\u0440) and vowel/null alternation (Tuvan *\u0448\u044b\u043c\u044b\u043d\u0430\u0440 \u0441\u0438\u043b\u0435\u0440 instead of \u0448\u044b\u043c\u043d\u044b\u0440 \u0441\u0438\u043b\u0435\u0440). Unadapted loanwords that entered the languages through Russian (e.g. Sakha \u043f\u0435\u0434\u0430\u0433\u043e\u0433 'pedagogue', \u043f\u0440\u0438\u043d\u0446 'prince', \u043d\u0430\u0441\u043b\u0435\u0434\u0438\u0435 'heritage') are also frequent among the errors for both systems.", |
| "cite_spans": [ |
| { |
| "start": 231, |
| "end": 252, |
| "text": "(Gorman et al., 2019)", |
| "ref_id": "BIBREF33" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Tuvan and Sakha", |
| "sec_num": "9.8" |
| }, |
| { |
| "text": "For Ashaninka, the high baseline scores (over 99.8%) could be attributed to the relatively high regularity of the (morpho)phonological rules in the language. In this language, the BME system achieved comparable performance with 99.5%, whereas GUClasp still achieved a robust accuracy of 93.36%.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Ashaninka and Yanesha", |
| "sec_num": "9.9" |
| }, |
| { |
| "text": "The case of Yanesha is different, as the baseline only peaked at 87.43%, whereas the BME and GUClasp systems underperformed with 82.46% and 55.94%, respectively. The task for Yanesha is harder, as the writing tradition is not transparent enough to predict some rules. For instance, large and short vowels are written in a similar way, always with a single vowel, and the aspirated vowels are optionally marked with a diacritic. These distinctions are essential at the linguistic level, as they allow one to explain the morphophonological processes, such as the syncope of the weak vowels in the flexionated forms (po'kochllet instead of po'kchellet). We also observe allomorphy errors, for instance, predicting phomchoche\u00f1 instead of pomchoche\u00f1 (from mochoche\u00f1ets and V;NO2;FIN;REAL). The singular second person prefix has ph-, peand pas allomorphs, each with different rules to apply. Moreover, there are some spelling issues as well, as the diacritic and the apostrophe are usually optional. For instance, the spellings wapa or w\u00e1pa (to come where someone is located) are both correct. It is important to note that the orthographic standards are going to be revised by the Peruvian government to reduce the ambiguous elements.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Ashaninka and Yanesha", |
| "sec_num": "9.9" |
| }, |
| { |
| "text": "The transformer baseline with data augmentation (TRM+AUG) achieved the highest score, with GUClasp taking the second place (with 72.24%) and the base transformer yielding the lowest score of 66.94%. For Magahi, the results do not vary too much between systems, and the clearest performance boost seems to arise from the use of data augmentation. The low score of the TRM baseline is caused by the scarcity of data and the diversity in the morphophonological structure. Prediction errors on Magahi include incorrect honorificity, mispredicting plural markers, and spelling errors.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Magahi", |
| "sec_num": "9.10" |
| }, |
| { |
| "text": "Honorificity: the systems predict forms lacking the honorific marker /-ak/. For example, /puchhal/ ('asked') is predicted instead of /puchhlak/ ('asked'), or /bital/ ('passed time') instead of /bitlak/ ('passed time').", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Magahi", |
| "sec_num": "9.10" |
| }, |
| { |
| "text": "Plural marker: the systems' predictions omit the plural markers /-van/ and /-yan/, similarly to the case of the honorific markers discussed above. For example, /thag/ ('con') is produced instead of /thagwan/ ('con').", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Magahi", |
| "sec_num": "9.10" |
| }, |
| { |
| "text": "Spelling errors: the predicted words do not oc-cur in the Magahi language. The predictions also do not show any specific error pattern. We thus conclude that the performance of the baseline systems is greatly affected by the morphological structure of Magahi. Also, some languagespecific properties of Magahi are not covered by the scope of the UniMorph tagset. For example, consider the following pair:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Magahi", |
| "sec_num": "9.10" |
| }, |
| { |
| "text": "(/dekh/, /dekhlai/, 'V;3;PRF;INFM;LSGSPEC1', 'see') (/dekh/, /dekhlak/, 'V;3;PRF;INFM;LGSPEC2', 'see')", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Magahi", |
| "sec_num": "9.10" |
| }, |
| { |
| "text": "Here both forms exhibit morphological features that are not defined in the default annotation schema. Morphologically, the first form indicates that the speaker knows the addressee but not intimately (or there is a low level of intimacy), while the second one signals a comparatively higher level of intimacy. Such aspects of the Magahi morphology are challenging for the systems.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Magahi", |
| "sec_num": "9.10" |
| }, |
| { |
| "text": "For the low-resource language Braj, both submitted systems performed worse than the baseline systems. BME achieved 58.52% prediction accuracy, slightly outperforming GUClasp with 56.91%. As for the baseline systems, CHR-TRM scored highest with 59.49% accuracy and TRM scored lowest with 53.38%. Among Indo-European languages, the performance of the BME, GUClasp, and the baseline systems is lowest for Braj. The low accuracy and the larger number of errors are broadly due to misprediction and misrepresentation of the morphological units and the smaller data size. BME, GUClasp, and the baseline systems generated 311 nominal, verbal, and adjectival inflected forms from existing lemmas. In these outputs, the misprediction and misrepresentation errors are morphemic errors, already included/classified by Gorman et al. (2019) . The findings of our analysis of both the gold data and the predictions of all systems highlight several common problems for nominal, verbal, and adjectival inflected forms. Common errors, mispredicted by all models, include morphemes of gender (masculine and feminine: for the noun akhabaaree instead of akhabaar 'newspaper', for the verb arraa instead of arraaii 'shout', and for the adjective mithak instead of mithakeey 'ancient'); morphemes of number (singular and plural: for the noun kahaanee instead of kahaaneen 'story', for the verb utaran instead of utare 'get off', for the adjective achchhe instead of achchhau 'good'); morphemes of honorificity (formal and informal: suni instead of sunikain 'listen', rahee instead of raheen 'be' and karibau instead of kari 'do ', etc.) .", |
| "cite_spans": [ |
| { |
| "start": 807, |
| "end": 827, |
| "text": "Gorman et al. (2019)", |
| "ref_id": "BIBREF33" |
| }, |
| { |
| "start": 1606, |
| "end": 1614, |
| "text": "', etc.)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Braj", |
| "sec_num": "9.11" |
| }, |
| { |
| "text": "A portion of these errors is also caused by the inflection errors in predicting and generating multiword expressions (MWEs) (e.g. aannd instead of aannd-daayak 'comfortable'). Apart from the mentioned error types, the systems also made silly errors (e.g. uthi instead of uthaay 'get up', kathan instead of kathanopakathan 'conversation', karaah instead of karaahatau 'groan', keeee instead of keenee 'do', and grahanave instead of grahan 'accept', etc.) and spelling errors (e.g. dhamaky instead of dhamakii 'threat', laau or laauy instead of liyau 'take', saii instead of saanchii 'correct', and samajhat instead of samajhaayau 'explain', etc.) as classified by Gorman et al. (2019) . Under all of the models, the majority of errors were silly or spelling errors.", |
| "cite_spans": [ |
| { |
| "start": 663, |
| "end": 683, |
| "text": "Gorman et al. (2019)", |
| "ref_id": "BIBREF33" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Braj", |
| "sec_num": "9.11" |
| }, |
| { |
| "text": "All systems achieved high accuracy (99.98%) on this language. The few errors are mainly due to the inconsistency in the initial data annotation. For instance, the form uraqiw is listed as a lemma while it can only be understood as being comprised of two morphemes: uraqi-w(a) 'it (is) land'. The root, or the nominative unmarked form, is uraqi 'land'. The -wa is presumably the declarative suffix. The nucleus of this suffix can be lost owing to the complex morphophonemic rules which operate at the edges of phrases. In addition, the accusative form uraqi is incorrect since the accusative is marked by subtractive disfixation, therefore, uraq is the accusative inflected form.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Aymara", |
| "sec_num": "9.12" |
| }, |
| { |
| "text": "Eibela seems to be one of the most challenging languages, probably due to its small data size and sparsity. Since it has been extracted from interlinear texts, a vast majority of its paradigms are partial, and this certainly makes the task more difficult. A closer look at system outputs reveals that many errors are related to misprediction of vowel length. For instance, to:mulu is inflected in N;ERG as tomulE instead of to:mu:lE:.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Eibela", |
| "sec_num": "9.13" |
| }, |
| { |
| "text": "The data for Kunwinjku is relatively small and contains verbal paradigms only. Test accuracies range from 14.75% (BME) to 63.93% (TRM+AUG).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Kunwinjku", |
| "sec_num": "9.14" |
| }, |
| { |
| "text": "In this language, many errors were due to incorrect spelling and missing parts of transcription. For instance, for the second person plural non-past of the lemma borlbme, TRM predicts *ngurriborlbme instead of ngurriborle. Interestingly, BME mispredicts most forms due to the looping effect described by Shcherbakov et al. (2020) . In particular, it starts producing sequences such as *ngarrrrrrrrrrrrrrmbbbijj (should be karribelbmerrinj) or *ngadjarridarrkddrrdddrrmerri (should be karriyawoyhdjarrkbidyikarrmerrimeninj).", |
| "cite_spans": [ |
| { |
| "start": 304, |
| "end": 329, |
| "text": "Shcherbakov et al. (2020)", |
| "ref_id": "BIBREF80" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Kunwinjku", |
| "sec_num": "9.14" |
| }, |
| { |
| "text": "Reusing transformation patterns", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Discussion", |
| "sec_num": "10" |
| }, |
| { |
| "text": "In most cases, morphological transformations may be properly carried out by matching a lemma against a pattern containing fixed characters and variable (wildcard) character sequences. A morphological inflection may be described in terms of inserting, deleting, and/or replacing fixed characters. When a test sample follows such a regular transformation pattern observed in the training set, it usually becomes significantly easier to track down a correct target form. Table 6 demonstrates the difference in performance w.r.t. whether the required transformation pattern was immediately witnessed in any training sample. To enumerate the possible patterns, we used the technique described by Scherbakov (2020) . It is worth emphasizing that the presence of a matching pattern by itself does not guarantee that achieving high accuracy would be straightforward, because in a vast majority of cases there are multiple alternative patterns. Choosing the correct one is a challenging task.", |
| "cite_spans": [ |
| { |
| "start": 691, |
| "end": 708, |
| "text": "Scherbakov (2020)", |
| "ref_id": "BIBREF78" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 468, |
| "end": 475, |
| "text": "Table 6", |
| "ref_id": "TABREF11" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Discussion", |
| "sec_num": "10" |
| }, |
| { |
| "text": "Inflecting multi-word expressions is one of the most challenging tasks. However, in the the shared task dataset, almost all multi-word lemmas found in the test set are also present in the training set, which made the task easier to solve.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Multi-word forms", |
| "sec_num": null |
| }, |
| { |
| "text": "The systems were quite successful at predicting the multi-word forms if the required transformation was directly observed in a training example. Otherwise, the prediction accuracy significantly degraded. Table 7 shows the multi-word lemma transformation accuracies. From these results we further notice that while all systems' performance degrades on the previously unseen multi-word inflection patterns, this degradation is considerably smaller for the transformer-based baselines (except Table 7 : Accuracy for MWE lemmata in each language on the test data. The numbers in black correspond to fragment substitutions that could be observed in the training set, while the red numbers correspond to more complex transformations.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 204, |
| "end": 211, |
| "text": "Table 7", |
| "ref_id": null |
| }, |
| { |
| "start": 490, |
| "end": 497, |
| "text": "Table 7", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Multi-word forms", |
| "sec_num": null |
| }, |
| { |
| "text": "for Turkish), implying that these models can better generalise to previously unseen patterns.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Multi-word forms", |
| "sec_num": null |
| }, |
| { |
| "text": "The application of wrong (albeit valid) inflectional transformations by the models (allomorphy) is present in most analysed languages. These allomorphy errors can be further divided into two groups:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Allomorphy", |
| "sec_num": null |
| }, |
| { |
| "text": "(1) when an inflectional tag itself allows for multiple inflection patterns which must be distinguished by the declension/inflection class to which the word belongs, and (2) when the model applies an inflectional rule that is simply invalid for that specific tag. These errors are hard to analyse, however. The first is potentially unavoidable without extra information, as declension/inflection classes are not always fully predictable from word forms alone (Williams et al., 2020) . The second type of allomorphic error, on the other hand, is potentially fixable. In our analysis, however, we did not find any concrete patterns to when the models make this second (somewhat arbitrary) type of mistake.", |
| "cite_spans": [ |
| { |
| "start": 459, |
| "end": 482, |
| "text": "(Williams et al., 2020)", |
| "ref_id": "BIBREF96" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Allomorphy", |
| "sec_num": null |
| }, |
| { |
| "text": "Spelling errors are pervasive in most analysed languages, even high-resource ones. These are challenging, as they require a deep understanding of the modelled language in order to be avoided. Spelling errors are especially common in languages with vowel harmony (e.g. Tungusic), as the models have some difficulty in correctly modelling it. Another source of spelling errors are the diacritics. In Por-tuguese, for instance, most of the errors produced by the BME system arise due to missing acute accents, which mark stress; their use is determined by specific (and somewhat idiosyncratic) orthographic rules.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Spelling Errors", |
| "sec_num": null |
| }, |
| { |
| "text": "In the development of this shared task we added new data for 32 languages (13 language families) to UniMorph-most of which are under-resourced. Further, we evaluated the performance of morphological reinflection systems on a typologically diverse set of languages and performed fine-grained analysis of their error patterns in most of these languages. The main challenge for the morphological reinflection systems is still (as expected) handling low-resource scenarios (where there is little training data). We further identified a large gap in these systems' performance between the test lemmas present in the training set and the previously unseen lemmas-the latter are naturally hard test cases, but the work on reinflection models could focus on improving these results going forward, following, for instance, the work of Liu and Hulden (2021) . Further, allomorphy, honorificity and multiword lemmas also pose challenges for the current models. We hope that the analysis presented here, together with the new expansion of the UniMorph resources, will help drive further improvements in morphological reinflection. Following Malouf et al. (2020), we would like to emphasize that linguistic analyses using UniMorph should be performed with some degree of caution, since for many languages it might not provide an exhaustive list of paradigms and variants.", |
| "cite_spans": [ |
| { |
| "start": 826, |
| "end": 847, |
| "text": "Liu and Hulden (2021)", |
| "ref_id": "BIBREF58" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Conclusion", |
| "sec_num": "11" |
| }, |
| { |
| "text": "A Data conversion into UniMorph (Przepi\u00f3rkowski and Woli\u0144ski, 2003) to the UniMorph schema. The mapping contains most of the POS and feature labels and does not allow to reconstruct the full conversion of the original data, as some mappings are conditional. ", |
| "cite_spans": [ |
| { |
| "start": 32, |
| "end": 67, |
| "text": "(Przepi\u00f3rkowski and Woli\u0144ski, 2003)", |
| "ref_id": "BIBREF77" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Conclusion", |
| "sec_num": "11" |
| }, |
| { |
| "text": "The data, systems, and their predictions are available: https://github.com/sigmorphon/2021Task0", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "For example, Tagalog verbs exhibit circumfixation; thus, a model with a strong inductive bias towards suffixing would likely not work well for Tagalog.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "Sometimes Cauqui (also spelled \"Kawki\"), a language spoken by less than ten elders in Cachuy, Canch\u00e1n, Caip\u00e1n, and Chav\u00edn, is considered to be a third Aymaran language but it may be more accurate to consider it a Jaqaru dialect.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "Tags' conversion into UniMorph: https: //github.com/unimorph/aym/blob/main/ Muylaq'AymaraUnimorphConversion.tsv", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "This is dedicated to Indian spiritual and mythological imagination as being associated with Lord Krishna.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "https://github.com/apertium/ apertium-tyv/ 9 https://github.com/apertium/ apertium-sah/ 10 https://github.com/ryskina/ apertium2unimorph", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "https://chuklang.ru/", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "https://postnauka.ru/longreads/156195 13 http://148.202.18.157/sitios/ publicacionesite/pperiod/funcion/pdf/ 11-12/289.pdf", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "The new languages are included into the UniMorph data: https://unimorph.github.io/", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "19 target errors excluding the 4 unjudgeable cases.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| } |
| ], |
| "back_matter": [ |
| { |
| "text": "We would like to thank Dr George Kiraz and Beth Mardutho: the Syriac Institute for their help with Classical Syriac data.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Acknowledgements", |
| "sec_num": null |
| } |
| ], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "The Languages of the Andes. Cambridge Language Surveys", |
| "authors": [ |
| { |
| "first": "F", |
| "middle": [ |
| "H" |
| ], |
| "last": "Willem", |
| "suffix": "" |
| }, |
| { |
| "first": "Pieter", |
| "middle": [ |
| "C" |
| ], |
| "last": "Adelaar", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Muysken", |
| "suffix": "" |
| } |
| ], |
| "year": 2004, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": { |
| "DOI": [ |
| "10.1017/CBO9780511486852" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Willem F. H. Adelaar and Pieter C. Muysken. 2004. The Languages of the Andes. Cambridge Language Surveys. Cambridge University Press.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "The documentation of eibela: An archive of eibela language materials from the bosavi region (western province, papua new guinea)", |
| "authors": [ |
| { |
| "first": "Grant", |
| "middle": [], |
| "last": "Aiton", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Grant Aiton. 2016a. The documentation of eibela: An archive of eibela language materials from the bosavi region (western province, papua new guinea).", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "A grammar of Eibela: a language of the Western Province", |
| "authors": [ |
| { |
| "first": "Grant", |
| "middle": [], |
| "last": "William Aiton", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Grant William Aiton. 2016b. A grammar of Eibela: a language of the Western Province, Papua New Guinea. Ph.D. thesis, James Cook University.", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "A corpus for modeling morpho-syntactic agreement in Arabic: Gender, number and rationality", |
| "authors": [ |
| { |
| "first": "Sarah", |
| "middle": [], |
| "last": "Alkuhlani", |
| "suffix": "" |
| }, |
| { |
| "first": "Nizar", |
| "middle": [], |
| "last": "Habash", |
| "suffix": "" |
| } |
| ], |
| "year": 2011, |
| "venue": "Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies", |
| "volume": "", |
| "issue": "", |
| "pages": "357--362", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Sarah Alkuhlani and Nizar Habash. 2011. A corpus for modeling morpho-syntactic agreement in Arabic: Gender, number and rationality. In Proceedings of the 49th Annual Meeting of the Association for Com- putational Linguistics: Human Language Technolo- gies, pages 357-362, Portland, Oregon, USA. Asso- ciation for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "Pushing the limits of low-resource morphological inflection", |
| "authors": [ |
| { |
| "first": "Antonios", |
| "middle": [], |
| "last": "Anastasopoulos", |
| "suffix": "" |
| }, |
| { |
| "first": "Graham", |
| "middle": [], |
| "last": "Neubig", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)", |
| "volume": "", |
| "issue": "", |
| "pages": "984--996", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/D19-1091" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Antonios Anastasopoulos and Graham Neubig. 2019. Pushing the limits of low-resource morphological in- flection. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natu- ral Language Processing (EMNLP-IJCNLP), pages 984-996, Hong Kong, China. Association for Com- putational Linguistics.", |
| "links": null |
| }, |
| "BIBREF5": { |
| "ref_id": "b5", |
| "title": "Tyvan (Languages of the World/Materials 257)", |
| "authors": [ |
| { |
| "first": "Gregory", |
| "middle": [ |
| "David" |
| ], |
| "last": "", |
| "suffix": "" |
| }, |
| { |
| "first": "Anderson", |
| "middle": [], |
| "last": "", |
| "suffix": "" |
| }, |
| { |
| "first": "K David", |
| "middle": [], |
| "last": "Harrison", |
| "suffix": "" |
| } |
| ], |
| "year": 1999, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Gregory David Anderson and K David Harrison. 1999. Tyvan (Languages of the World/Materials 257). M\u00fcnchen: LINCOM Europa.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "The Seneca Verb: Labeling the Ancient Voice. Seneca Nation Education Department", |
| "authors": [ |
| { |
| "first": "Phyllis", |
| "middle": [ |
| "E" |
| ], |
| "last": "Wms", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Bardeau", |
| "suffix": "" |
| } |
| ], |
| "year": 2007, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Phyllis E. Wms. Bardeau. 2007. The Seneca Verb: La- beling the Ancient Voice. Seneca Nation Education Department, Cattaraugus Territory.", |
| "links": null |
| }, |
| "BIBREF7": { |
| "ref_id": "b7", |
| "title": "Scheduled sampling for sequence prediction with recurrent neural networks", |
| "authors": [ |
| { |
| "first": "Samy", |
| "middle": [], |
| "last": "Bengio", |
| "suffix": "" |
| }, |
| { |
| "first": "Oriol", |
| "middle": [], |
| "last": "Vinyals", |
| "suffix": "" |
| }, |
| { |
| "first": "Navdeep", |
| "middle": [], |
| "last": "Jaitly", |
| "suffix": "" |
| }, |
| { |
| "first": "Noam", |
| "middle": [], |
| "last": "Shazeer", |
| "suffix": "" |
| } |
| ], |
| "year": 2015, |
| "venue": "Advances in Neural Information Processing Systems 28: Annual Conference on Neural Information Processing Systems", |
| "volume": "", |
| "issue": "", |
| "pages": "1171--1179", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Samy Bengio, Oriol Vinyals, Navdeep Jaitly, and Noam Shazeer. 2015. Scheduled sampling for se- quence prediction with recurrent neural networks. In Advances in Neural Information Processing Sys- tems 28: Annual Conference on Neural Informa- tion Processing Systems 2015, December 7-12, 2015, Montreal, Quebec, Canada, pages 1171-1179.", |
| "links": null |
| }, |
| "BIBREF8": { |
| "ref_id": "b8", |
| "title": "Language and nature. Mind", |
| "authors": [ |
| { |
| "first": "Noam", |
| "middle": [], |
| "last": "Chomsky", |
| "suffix": "" |
| } |
| ], |
| "year": 1995, |
| "venue": "", |
| "volume": "104", |
| "issue": "", |
| "pages": "1--61", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Noam Chomsky. 1995. Language and nature. Mind, 104(413):1-61.", |
| "links": null |
| }, |
| "BIBREF9": { |
| "ref_id": "b9", |
| "title": "A grammatical description of Muylaq'Aymara", |
| "authors": [ |
| { |
| "first": "Matt", |
| "middle": [], |
| "last": "Coler", |
| "suffix": "" |
| } |
| ], |
| "year": 2010, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Matt Coler. 2010. A grammatical description of Muy- laq'Aymara. Ph.D. thesis, Vrije Universiteit Amster- dam.", |
| "links": null |
| }, |
| "BIBREF10": { |
| "ref_id": "b10", |
| "title": "A grammar of Muylaq'Aymara: Aymara as spoken in Southern Peru", |
| "authors": [ |
| { |
| "first": "Matt", |
| "middle": [], |
| "last": "Coler", |
| "suffix": "" |
| } |
| ], |
| "year": 2014, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Matt Coler. 2014. A grammar of Muylaq'Aymara: Ay- mara as spoken in Southern Peru. Brill.", |
| "links": null |
| }, |
| "BIBREF11": { |
| "ref_id": "b11", |
| "title": "Language universals and linguistic typology: Syntax and morphology", |
| "authors": [ |
| { |
| "first": "Bernard", |
| "middle": [], |
| "last": "Comrie", |
| "suffix": "" |
| } |
| ], |
| "year": 1989, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Bernard Comrie. 1989. Language universals and lin- guistic typology: Syntax and morphology. Univer- sity of Chicago press.", |
| "links": null |
| }, |
| "BIBREF13": { |
| "ref_id": "b13", |
| "title": "The CoNLL-SIGMORPHON 2018 shared task: Universal morphological reinflection", |
| "authors": [ |
| { |
| "first": "Katharina", |
| "middle": [], |
| "last": "Mccarthy", |
| "suffix": "" |
| }, |
| { |
| "first": "Sabrina", |
| "middle": [ |
| "J" |
| ], |
| "last": "Kann", |
| "suffix": "" |
| }, |
| { |
| "first": "Garrett", |
| "middle": [], |
| "last": "Mielke", |
| "suffix": "" |
| }, |
| { |
| "first": "Miikka", |
| "middle": [], |
| "last": "Nicolai", |
| "suffix": "" |
| }, |
| { |
| "first": "David", |
| "middle": [], |
| "last": "Silfverberg", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Yarowsky", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proceedings of the CoNLL-SIGMORPHON 2018 Shared Task: Universal Morphological Reinflection", |
| "volume": "", |
| "issue": "", |
| "pages": "1--27", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/K18-3001" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "McCarthy, Katharina Kann, Sabrina J. Mielke, Gar- rett Nicolai, Miikka Silfverberg, David Yarowsky, Jason Eisner, and Mans Hulden. 2018. The CoNLL- SIGMORPHON 2018 shared task: Universal mor- phological reinflection. In Proceedings of the CoNLL-SIGMORPHON 2018 Shared Task: Univer- sal Morphological Reinflection, pages 1-27, Brus- sels. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF14": { |
| "ref_id": "b14", |
| "title": "CoNLL-SIGMORPHON 2017 shared task: Universal morphological reinflection in 52 languages", |
| "authors": [ |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Cotterell", |
| "suffix": "" |
| }, |
| { |
| "first": "Christo", |
| "middle": [], |
| "last": "Kirov", |
| "suffix": "" |
| }, |
| { |
| "first": "John", |
| "middle": [], |
| "last": "Sylak-Glassman", |
| "suffix": "" |
| }, |
| { |
| "first": "G\u00e9raldine", |
| "middle": [], |
| "last": "Walther", |
| "suffix": "" |
| }, |
| { |
| "first": "Ekaterina", |
| "middle": [], |
| "last": "Vylomova", |
| "suffix": "" |
| }, |
| { |
| "first": "Patrick", |
| "middle": [], |
| "last": "Xia", |
| "suffix": "" |
| }, |
| { |
| "first": "Manaal", |
| "middle": [], |
| "last": "Faruqui", |
| "suffix": "" |
| }, |
| { |
| "first": "Sandra", |
| "middle": [], |
| "last": "K\u00fcbler", |
| "suffix": "" |
| }, |
| { |
| "first": "David", |
| "middle": [], |
| "last": "Yarowsky", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "Proceedings of the CoNLL SIGMORPHON 2017", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/K17-2001" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ryan Cotterell, Christo Kirov, John Sylak-Glassman, G\u00e9raldine Walther, Ekaterina Vylomova, Patrick Xia, Manaal Faruqui, Sandra K\u00fcbler, David Yarowsky, Jason Eisner, and Mans Hulden. 2017. CoNLL-SIGMORPHON 2017 shared task: Univer- sal morphological reinflection in 52 languages. In Proceedings of the CoNLL SIGMORPHON 2017", |
| "links": null |
| }, |
| "BIBREF15": { |
| "ref_id": "b15", |
| "title": "Shared Task: Universal Morphological Reinflection, pages 1-30, Vancouver. Association for Computational Linguistics", |
| "authors": [], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Shared Task: Universal Morphological Reinflection, pages 1-30, Vancouver. Association for Computa- tional Linguistics.", |
| "links": null |
| }, |
| "BIBREF16": { |
| "ref_id": "b16", |
| "title": "The SIGMORPHON 2016 shared Task-Morphological reinflection", |
| "authors": [ |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Cotterell", |
| "suffix": "" |
| }, |
| { |
| "first": "Christo", |
| "middle": [], |
| "last": "Kirov", |
| "suffix": "" |
| }, |
| { |
| "first": "John", |
| "middle": [], |
| "last": "Sylak-Glassman", |
| "suffix": "" |
| }, |
| { |
| "first": "David", |
| "middle": [], |
| "last": "Yarowsky", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "Proceedings of the 14th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology", |
| "volume": "", |
| "issue": "", |
| "pages": "10--22", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/W16-2002" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ryan Cotterell, Christo Kirov, John Sylak-Glassman, David Yarowsky, Jason Eisner, and Mans Hulden. 2016. The SIGMORPHON 2016 shared Task- Morphological reinflection. In Proceedings of the 14th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphol- ogy, pages 10-22, Berlin, Germany. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF17": { |
| "ref_id": "b17", |
| "title": "Typology and universals", |
| "authors": [ |
| { |
| "first": "William", |
| "middle": [], |
| "last": "Croft", |
| "suffix": "" |
| } |
| ], |
| "year": 2002, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "William Croft. 2002. Typology and universals. Cam- bridge University Press.", |
| "links": null |
| }, |
| "BIBREF18": { |
| "ref_id": "b18", |
| "title": "Linguistic typology and the study of language", |
| "authors": [ |
| { |
| "first": "Michael", |
| "middle": [], |
| "last": "Daniel", |
| "suffix": "" |
| } |
| ], |
| "year": 2011, |
| "venue": "The Oxford handbook of linguistic typology", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Michael Daniel. 2011. Linguistic typology and the study of language. In The Oxford handbook of lin- guistic typology. Oxford University Press.", |
| "links": null |
| }, |
| "BIBREF19": { |
| "ref_id": "b19", |
| "title": "The Amazonian languages (Cambridge Language Surveys)", |
| "authors": [ |
| { |
| "first": "R", |
| "middle": [ |
| "M W" |
| ], |
| "last": "Dixon", |
| "suffix": "" |
| }, |
| { |
| "first": "Alexandra", |
| "middle": [ |
| "Y" |
| ], |
| "last": "Aikhenvald", |
| "suffix": "" |
| } |
| ], |
| "year": 1999, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "R. M. W. Dixon and Alexandra Y. Aikhenvald, editors. 1999. The Amazonian languages (Cambridge Lan- guage Surveys). Cambridge University Press.", |
| "links": null |
| }, |
| "BIBREF20": { |
| "ref_id": "b20", |
| "title": "Diccionario Yanesha' (Amuesha)-Castellano", |
| "authors": [ |
| { |
| "first": "", |
| "middle": [], |
| "last": "Duff-Trip", |
| "suffix": "" |
| } |
| ], |
| "year": 1998, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "M Duff-Trip. 1998. Diccionario Yanesha' (Amuesha)- Castellano. Lima: Instituto Ling\u00fc\u00edstico de Verano.", |
| "links": null |
| }, |
| "BIBREF21": { |
| "ref_id": "b21", |
| "title": "Ethnologue: Languages of the world. Twenty-fourth edition", |
| "authors": [ |
| { |
| "first": "David", |
| "middle": [ |
| "M" |
| ], |
| "last": "Eberhard", |
| "suffix": "" |
| }, |
| { |
| "first": "Gary", |
| "middle": [ |
| "F" |
| ], |
| "last": "Simons", |
| "suffix": "" |
| } |
| ], |
| "year": 2021, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "David M. Eberhard, Gary F. Simons, and Charles D. Fennig (eds.). 2021. Ethnologue: Languages of the world. Twenty-fourth edition. Online version: http://www.ethnologue.com.", |
| "links": null |
| }, |
| "BIBREF22": { |
| "ref_id": "b22", |
| "title": "Modeling morphological learning, typology, and change: What can the neural sequence-to-sequence framework contribute", |
| "authors": [ |
| { |
| "first": "Micha", |
| "middle": [], |
| "last": "Elsner", |
| "suffix": "" |
| }, |
| { |
| "first": "Andrea", |
| "middle": [ |
| "D" |
| ], |
| "last": "Sims", |
| "suffix": "" |
| }, |
| { |
| "first": "Alexander", |
| "middle": [], |
| "last": "Erdmann", |
| "suffix": "" |
| }, |
| { |
| "first": "Antonio", |
| "middle": [], |
| "last": "Hernandez", |
| "suffix": "" |
| }, |
| { |
| "first": "Evan", |
| "middle": [], |
| "last": "Jaffe", |
| "suffix": "" |
| }, |
| { |
| "first": "Lifeng", |
| "middle": [], |
| "last": "Jin", |
| "suffix": "" |
| }, |
| { |
| "first": "Martha", |
| "middle": [ |
| "Booker" |
| ], |
| "last": "Johnson", |
| "suffix": "" |
| }, |
| { |
| "first": "Shuan", |
| "middle": [], |
| "last": "Karim", |
| "suffix": "" |
| }, |
| { |
| "first": "L", |
| "middle": [], |
| "last": "David", |
| "suffix": "" |
| }, |
| { |
| "first": "Luana", |
| "middle": [ |
| "Lamberti" |
| ], |
| "last": "King", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Nunes", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Journal of Language Modelling", |
| "volume": "7", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Micha Elsner, Andrea D Sims, Alexander Erdmann, Antonio Hernandez, Evan Jaffe, Lifeng Jin, Martha Booker Johnson, Shuan Karim, David L King, Lu- ana Lamberti Nunes, et al. 2019. Modeling morpho- logical learning, typology, and change: What can the neural sequence-to-sequence framework contribute? Journal of Language Modelling, 7.", |
| "links": null |
| }, |
| "BIBREF23": { |
| "ref_id": "b23", |
| "title": "Bininj Gun-wok: A Pandialectal Grammar of Mayali", |
| "authors": [ |
| { |
| "first": "Nicholas", |
| "middle": [], |
| "last": "Evans", |
| "suffix": "" |
| } |
| ], |
| "year": 2003, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Nicholas Evans. 2003. Bininj Gun-wok: A Pan- dialectal Grammar of Mayali, Kunwinjku and Kune. Pacific Linguistics. Australian National University.", |
| "links": null |
| }, |
| "BIBREF24": { |
| "ref_id": "b24", |
| "title": "The myth of language universals: Language diversity and its importance for cognitive science", |
| "authors": [ |
| { |
| "first": "Nicholas", |
| "middle": [], |
| "last": "Evans", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Stephen C Levinson", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "Behavioral and brain sciences", |
| "volume": "32", |
| "issue": "5", |
| "pages": "429--448", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Nicholas Evans and Stephen C Levinson. 2009. The myth of language universals: Language diversity and its importance for cognitive science. Behavioral and brain sciences, 32(5):429-448.", |
| "links": null |
| }, |
| "BIBREF25": { |
| "ref_id": "b25", |
| "title": "Morphological inflection generation using character sequence to sequence learning", |
| "authors": [ |
| { |
| "first": "Manaal", |
| "middle": [], |
| "last": "Faruqui", |
| "suffix": "" |
| }, |
| { |
| "first": "Yulia", |
| "middle": [], |
| "last": "Tsvetkov", |
| "suffix": "" |
| }, |
| { |
| "first": "Graham", |
| "middle": [], |
| "last": "Neubig", |
| "suffix": "" |
| }, |
| { |
| "first": "Chris", |
| "middle": [], |
| "last": "Dyer", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies", |
| "volume": "", |
| "issue": "", |
| "pages": "634--643", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/N16-1077" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Manaal Faruqui, Yulia Tsvetkov, Graham Neubig, and Chris Dyer. 2016. Morphological inflection genera- tion using character sequence to sequence learning. In Proceedings of the 2016 Conference of the North American Chapter of the Association for Computa- tional Linguistics: Human Language Technologies, pages 634-643, San Diego, California. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF26": { |
| "ref_id": "b26", |
| "title": "Les dialectes kurdes m\u00e9ridionaux: \u00e9tude linguistique et dialectologique. Acta Iranica : Encyclop\u00e9die permanente des \u00e9tudes iraniennes", |
| "authors": [ |
| { |
| "first": "I", |
| "middle": [ |
| "K" |
| ], |
| "last": "Fattah", |
| "suffix": "" |
| } |
| ], |
| "year": 2000, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "I.K. Fattah. 2000. Les dialectes kurdes m\u00e9ridionaux: \u00e9tude linguistique et dialectologique. Acta Iranica : Encyclop\u00e9die permanente des \u00e9tudes iraniennes. Peeters.", |
| "links": null |
| }, |
| "BIBREF28": { |
| "ref_id": "b28", |
| "title": "Apertium: a free/open-source platform for rule-based machine translation. Machine translation", |
| "authors": [ |
| { |
| "first": "Mireia", |
| "middle": [], |
| "last": "Mikel L Forcada", |
| "suffix": "" |
| }, |
| { |
| "first": "Jacob", |
| "middle": [], |
| "last": "Ginest\u00ed-Rosell", |
| "suffix": "" |
| }, |
| { |
| "first": "Jim", |
| "middle": [ |
| "O" |
| ], |
| "last": "Nordfalk", |
| "suffix": "" |
| }, |
| { |
| "first": "Sergio", |
| "middle": [], |
| "last": "'regan", |
| "suffix": "" |
| }, |
| { |
| "first": "Juan", |
| "middle": [ |
| "Antonio" |
| ], |
| "last": "Ortiz-Rojas", |
| "suffix": "" |
| }, |
| { |
| "first": "Felipe", |
| "middle": [], |
| "last": "P\u00e9rez-Ortiz", |
| "suffix": "" |
| }, |
| { |
| "first": "Gema", |
| "middle": [], |
| "last": "S\u00e1nchez-Mart\u00ednez", |
| "suffix": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [ |
| "M" |
| ], |
| "last": "Ram\u00edrez-S\u00e1nchez", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Tyers", |
| "suffix": "" |
| } |
| ], |
| "year": 2011, |
| "venue": "", |
| "volume": "25", |
| "issue": "", |
| "pages": "127--144", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Mikel L Forcada, Mireia Ginest\u00ed-Rosell, Jacob Nord- falk, Jim O'Regan, Sergio Ortiz-Rojas, Juan An- tonio P\u00e9rez-Ortiz, Felipe S\u00e1nchez-Mart\u00ednez, Gema Ram\u00edrez-S\u00e1nchez, and Francis M Tyers. 2011. Aper- tium: a free/open-source platform for rule-based ma- chine translation. Machine translation, 25(2):127- 144.", |
| "links": null |
| }, |
| "BIBREF29": { |
| "ref_id": "b29", |
| "title": "HornMorpho: a system for morphological processing of Amharic, Oromo, and Tigrinya", |
| "authors": [ |
| { |
| "first": "Michael", |
| "middle": [], |
| "last": "Gasser", |
| "suffix": "" |
| } |
| ], |
| "year": 2011, |
| "venue": "Proceedings of the Conference on Human Language Technology for Development", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Michael Gasser. 2011. HornMorpho: a system for morphological processing of Amharic, Oromo, and Tigrinya. In Proceedings of the Conference on Human Language Technology for Development, Alexandria, Egypt.", |
| "links": null |
| }, |
| "BIBREF30": { |
| "ref_id": "b30", |
| "title": "Kodi (Indonesia) -Language Snapshot. Language Documentation and Description 19", |
| "authors": [ |
| { |
| "first": "Yustinus", |
| "middle": [], |
| "last": "Ghanggo Ate", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "171--180", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Yustinus Ghanggo Ate. 2020. Kodi (Indonesia) -Lan- guage Snapshot. Language Documentation and De- scription 19, pages 171-180.", |
| "links": null |
| }, |
| "BIBREF31": { |
| "ref_id": "b31", |
| "title": "Documentation of Kodi. New Haven: Endangered Language Fund", |
| "authors": [ |
| { |
| "first": "Yustinus", |
| "middle": [], |
| "last": "Ghanggo Ate", |
| "suffix": "" |
| } |
| ], |
| "year": 2021, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Yustinus Ghanggo Ate. 2021. Documentation of Kodi. New Haven: Endangered Language Fund.", |
| "links": null |
| }, |
| "BIBREF32": { |
| "ref_id": "b32", |
| "title": "Reduplication in Kodi: A paradigm function account", |
| "authors": [ |
| { |
| "first": "Yustinus", |
| "middle": [], |
| "last": "Ghanggo Ate", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "Word Structure", |
| "volume": "14", |
| "issue": "3", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Yustinus Ghanggo Ate. to appear in 2021. Reduplica- tion in Kodi: A paradigm function account. Word Structure 14(3).", |
| "links": null |
| }, |
| "BIBREF33": { |
| "ref_id": "b33", |
| "title": "Weird inflects but OK: Making sense of morphological generation errors", |
| "authors": [ |
| { |
| "first": "Kyle", |
| "middle": [], |
| "last": "Gorman", |
| "suffix": "" |
| }, |
| { |
| "first": "Arya", |
| "middle": [ |
| "D" |
| ], |
| "last": "Mccarthy", |
| "suffix": "" |
| }, |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Cotterell", |
| "suffix": "" |
| }, |
| { |
| "first": "Ekaterina", |
| "middle": [], |
| "last": "Vylomova", |
| "suffix": "" |
| }, |
| { |
| "first": "Miikka", |
| "middle": [], |
| "last": "Silfverberg", |
| "suffix": "" |
| }, |
| { |
| "first": "Magdalena", |
| "middle": [], |
| "last": "Markowska", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)", |
| "volume": "", |
| "issue": "", |
| "pages": "140--151", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/K19-1014" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kyle Gorman, Arya D. McCarthy, Ryan Cotterell, Ekaterina Vylomova, Miikka Silfverberg, and Mag- dalena Markowska. 2019. Weird inflects but OK: Making sense of morphological generation errors. In Proceedings of the 23rd Conference on Computa- tional Natural Language Learning (CoNLL), pages 140-151, Hong Kong, China. Association for Com- putational Linguistics.", |
| "links": null |
| }, |
| "BIBREF34": { |
| "ref_id": "b34", |
| "title": "Some universals of grammar with particular reference to the order of meaningful elements", |
| "authors": [ |
| { |
| "first": "Joseph", |
| "middle": [], |
| "last": "Greenberg", |
| "suffix": "" |
| } |
| ], |
| "year": 1963, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "73--113", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Joseph Greenberg. 1963. Some universals of grammar with particular reference to the order of meaningful elements. In Joseph Greenberg, editor, Universals of Language, pages 73-113. MIT Press.", |
| "links": null |
| }, |
| "BIBREF35": { |
| "ref_id": "b35", |
| "title": "Indo-Aryan Family: Central Group: Specimens of the R\u0101jasth\u0101n\u012b and Gujar\u0101t\u012b, volume IX(II) of Linguistic Survey of India. Office of the Superintendent of Government Printing", |
| "authors": [ |
| { |
| "first": "George", |
| "middle": [ |
| "A" |
| ], |
| "last": "Grierson", |
| "suffix": "" |
| } |
| ], |
| "year": 1908, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "George A. Grierson. 1908. Indo-Aryan Family: Cen- tral Group: Specimens of the R\u0101jasth\u0101n\u012b and Gu- jar\u0101t\u012b, volume IX(II) of Linguistic Survey of India. Office of the Superintendent of Government Print- ing, Calcutta.", |
| "links": null |
| }, |
| "BIBREF36": { |
| "ref_id": "b36", |
| "title": "Calcutta: Office of the Superintendent, Government of PRI", |
| "authors": [ |
| { |
| "first": "George", |
| "middle": [], |
| "last": "Abraham Grierson", |
| "suffix": "" |
| } |
| ], |
| "year": 1903, |
| "venue": "Linguistic Survey of India", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "George Abraham Grierson. 1903. Linguistic Survey of India, Vol-III. Calcutta: Office of the Superinten- dent, Government of PRI.", |
| "links": null |
| }, |
| "BIBREF37": { |
| "ref_id": "b37", |
| "title": "Linguistic Survey of India", |
| "authors": [ |
| { |
| "first": "George", |
| "middle": [], |
| "last": "Abraham Grierson", |
| "suffix": "" |
| }, |
| { |
| "first": "Sten", |
| "middle": [], |
| "last": "Konow", |
| "suffix": "" |
| } |
| ], |
| "year": 1903, |
| "venue": "Calcutta Supt., Govt. Printing", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "George Abraham Grierson and Sten Konow. 1903. Lin- guistic Survey of India. Calcutta Supt., Govt. Print- ing.", |
| "links": null |
| }, |
| "BIBREF38": { |
| "ref_id": "b38", |
| "title": "A morphological analyzer for Egyptian Arabic", |
| "authors": [ |
| { |
| "first": "Nizar", |
| "middle": [], |
| "last": "Habash", |
| "suffix": "" |
| }, |
| { |
| "first": "Ramy", |
| "middle": [], |
| "last": "Eskander", |
| "suffix": "" |
| }, |
| { |
| "first": "Abdelati", |
| "middle": [], |
| "last": "Hawwari", |
| "suffix": "" |
| } |
| ], |
| "year": 2012, |
| "venue": "Proceedings of the Twelfth Meeting of the Special Interest Group on Computational Morphology and Phonology", |
| "volume": "", |
| "issue": "", |
| "pages": "1--9", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Nizar Habash, Ramy Eskander, and Abdelati Hawwari. 2012. A morphological analyzer for Egyptian Ara- bic. In Proceedings of the Twelfth Meeting of the Special Interest Group on Computational Morphol- ogy and Phonology, pages 1-9, Montr\u00e9al, Canada. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF39": { |
| "ref_id": "b39", |
| "title": "Topics in the Phonology and Morphology of Tuvan", |
| "authors": [ |
| { |
| "first": "K", |
| "middle": [], |
| "last": "", |
| "suffix": "" |
| }, |
| { |
| "first": "David", |
| "middle": [], |
| "last": "Harrison", |
| "suffix": "" |
| } |
| ], |
| "year": 2000, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "K. David Harrison. 2000. Topics in the Phonology and Morphology of Tuvan. Ph.D. thesis, Yale University.", |
| "links": null |
| }, |
| "BIBREF40": { |
| "ref_id": "b40", |
| "title": "Comparative concepts and descriptive categories in crosslinguistic studies", |
| "authors": [ |
| { |
| "first": "Martin", |
| "middle": [], |
| "last": "Haspelmath", |
| "suffix": "" |
| } |
| ], |
| "year": 2010, |
| "venue": "Language", |
| "volume": "86", |
| "issue": "3", |
| "pages": "663--687", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Martin Haspelmath. 2010. Comparative concepts and descriptive categories in crosslinguistic studies. Language, 86(3):663-687.", |
| "links": null |
| }, |
| "BIBREF41": { |
| "ref_id": "b41", |
| "title": "Human linguisticality and the building blocks of languages", |
| "authors": [ |
| { |
| "first": "Martin", |
| "middle": [], |
| "last": "Haspelmath", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Frontiers in psychology", |
| "volume": "10", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Martin Haspelmath. 2020. Human linguisticality and the building blocks of languages. Frontiers in psy- chology, 10:3056.", |
| "links": null |
| }, |
| "BIBREF42": { |
| "ref_id": "b42", |
| "title": "ConlluEditor: a fully graphical editor for universal dependencies treebank files", |
| "authors": [ |
| { |
| "first": "Johannes", |
| "middle": [], |
| "last": "Heinecke", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the Third Workshop on Universal Dependencies", |
| "volume": "", |
| "issue": "", |
| "pages": "87--93", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/W19-8010" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Johannes Heinecke. 2019. ConlluEditor: a fully graph- ical editor for universal dependencies treebank files. In Proceedings of the Third Workshop on Univer- sal Dependencies (UDW, SyntaxFest 2019), pages 87-93, Paris, France. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF43": { |
| "ref_id": "b43", |
| "title": "Tools for supporting language learning for Sakha", |
| "authors": [ |
| { |
| "first": "Sardana", |
| "middle": [], |
| "last": "Ivanova", |
| "suffix": "" |
| }, |
| { |
| "first": "Anisia", |
| "middle": [], |
| "last": "Katinskaia", |
| "suffix": "" |
| }, |
| { |
| "first": "Roman", |
| "middle": [], |
| "last": "Yangarber", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the 22nd Nordic Conference on Computational Linguistics", |
| "volume": "", |
| "issue": "", |
| "pages": "155--163", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Sardana Ivanova, Anisia Katinskaia, and Roman Yan- garber. 2019. Tools for supporting language learn- ing for Sakha. In Proceedings of the 22nd Nordic Conference on Computational Linguistics, pages 155-163, Turku, Finland. Link\u00f6ping University Electronic Press.", |
| "links": null |
| }, |
| "BIBREF44": { |
| "ref_id": "b44", |
| "title": "A free/open-source morphological analyser and generator for Sakha", |
| "authors": [ |
| { |
| "first": "Sardana", |
| "middle": [], |
| "last": "Ivanova", |
| "suffix": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [ |
| "M" |
| ], |
| "last": "Tyers", |
| "suffix": "" |
| }, |
| { |
| "first": "Jonathan", |
| "middle": [ |
| "N" |
| ], |
| "last": "Washington", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Sardana Ivanova, Francis M. Tyers, and Jonathan N. Washington. to appear in 2022. A free/open-source morphological analyser and generator for Sakha. In preparation.", |
| "links": null |
| }, |
| "BIBREF45": { |
| "ref_id": "b45", |
| "title": "The Indo-Aryan Languages", |
| "authors": [ |
| { |
| "first": "Danesh", |
| "middle": [], |
| "last": "Jain", |
| "suffix": "" |
| }, |
| { |
| "first": "George", |
| "middle": [], |
| "last": "Cardona", |
| "suffix": "" |
| } |
| ], |
| "year": 2007, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Danesh Jain and George Cardona. 2007. The Indo- Aryan Languages. Routledge.", |
| "links": null |
| }, |
| "BIBREF46": { |
| "ref_id": "b46", |
| "title": "Ergative Remnants in Sorani Kurdish?", |
| "authors": [ |
| { |
| "first": "Thomas", |
| "middle": [], |
| "last": "J\u00fcgel", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "Orientalia Suecana", |
| "volume": "58", |
| "issue": "", |
| "pages": "142--158", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Thomas J\u00fcgel. 2009. Ergative Remnants in Sorani Kur- dish? Orientalia Suecana, 58:142-158.", |
| "links": null |
| }, |
| "BIBREF47": { |
| "ref_id": "b47", |
| "title": "Creating a multimedia annotated text corpus: a research task (Sozdaniye multimediynogo annotirovannogo korpusa tekstov kak issledovatelskaya protsedura)", |
| "authors": [ |
| { |
| "first": "Olga", |
| "middle": [], |
| "last": "Kazakevich", |
| "suffix": "" |
| }, |
| { |
| "first": "Elena", |
| "middle": [], |
| "last": "Klyachko", |
| "suffix": "" |
| } |
| ], |
| "year": 2013, |
| "venue": "Proceedings of International Conference Computational linguistics 2013", |
| "volume": "", |
| "issue": "", |
| "pages": "292--300", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Olga Kazakevich and Elena Klyachko. 2013. Creat- ing a multimedia annotated text corpus: a research task (Sozdaniye multimediynogo annotirovannogo korpusa tekstov kak issledovatelskaya protsedura). In Proceedings of International Conference Compu- tational linguistics 2013, pages 292-300.", |
| "links": null |
| }, |
| "BIBREF48": { |
| "ref_id": "b48", |
| "title": "A morphologically annotated corpus of Emirati Arabic", |
| "authors": [ |
| { |
| "first": "Nizar", |
| "middle": [], |
| "last": "Salam Khalifa", |
| "suffix": "" |
| }, |
| { |
| "first": "Fadhl", |
| "middle": [], |
| "last": "Habash", |
| "suffix": "" |
| }, |
| { |
| "first": "Ossama", |
| "middle": [], |
| "last": "Eryani", |
| "suffix": "" |
| }, |
| { |
| "first": "Dana", |
| "middle": [], |
| "last": "Obeid", |
| "suffix": "" |
| }, |
| { |
| "first": "Meera Al", |
| "middle": [], |
| "last": "Abdulrahim", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Kaabi", |
| "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": "Salam Khalifa, Nizar Habash, Fadhl Eryani, Ossama Obeid, Dana Abdulrahim, and Meera Al Kaabi. 2018. A morphologically annotated corpus of Emi- rati Arabic. In Proceedings of the Eleventh Interna- tional Conference on Language Resources and Eval- uation (LREC 2018), Miyazaki, Japan. European Language Resources Association (ELRA).", |
| "links": null |
| }, |
| "BIBREF49": { |
| "ref_id": "b49", |
| "title": "A morphological analyzer for Gulf Arabic verbs", |
| "authors": [ |
| { |
| "first": "Sara", |
| "middle": [], |
| "last": "Salam Khalifa", |
| "suffix": "" |
| }, |
| { |
| "first": "Nizar", |
| "middle": [], |
| "last": "Hassan", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Habash", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "Proceedings of the Third Arabic Natural Language Processing Workshop", |
| "volume": "", |
| "issue": "", |
| "pages": "35--45", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/W17-1305" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Salam Khalifa, Sara Hassan, and Nizar Habash. 2017. A morphological analyzer for Gulf Arabic verbs. In Proceedings of the Third Arabic Natural Language Processing Workshop, pages 35-45, Valencia, Spain. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF50": { |
| "ref_id": "b50", |
| "title": "Irene Tang, and H\u00e8ctor Al\u00f2s i Font. to appear in 2021. Recent advances in Apertium, a free/open-source rule-based machine translation platform for low-resource languages", |
| "authors": [ |
| { |
| "first": "Tanmai", |
| "middle": [], |
| "last": "Khanna", |
| "suffix": "" |
| }, |
| { |
| "first": "Jonathan", |
| "middle": [ |
| "N" |
| ], |
| "last": "Washington", |
| "suffix": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [ |
| "M" |
| ], |
| "last": "Tyers", |
| "suffix": "" |
| }, |
| { |
| "first": "Sevilay", |
| "middle": [], |
| "last": "Bayatl\u0131", |
| "suffix": "" |
| }, |
| { |
| "first": "Daniel", |
| "middle": [ |
| "G" |
| ], |
| "last": "Swanson", |
| "suffix": "" |
| }, |
| { |
| "first": "Tommi", |
| "middle": [ |
| "A" |
| ], |
| "last": "Pirinen", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "Machine Translation", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Tanmai Khanna, Jonathan N. Washington, Fran- cis M. Tyers, Sevilay Bayatl\u0131, Daniel G. Swanson, Tommi A. Pirinen, Irene Tang, and H\u00e8ctor Al\u00f2s i Font. to appear in 2021. Recent advances in Aper- tium, a free/open-source rule-based machine transla- tion platform for low-resource languages. Machine Translation.", |
| "links": null |
| }, |
| "BIBREF51": { |
| "ref_id": "b51", |
| "title": "Diccionario ash\u00e1ninca. Lima: Instituto Ling\u00fc\u00edstico de Verano", |
| "authors": [ |
| { |
| "first": "Lee", |
| "middle": [], |
| "last": "Kindberg", |
| "suffix": "" |
| } |
| ], |
| "year": 1980, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Lee Kindberg. 1980. Diccionario ash\u00e1ninca. Lima: Instituto Ling\u00fc\u00edstico de Verano.", |
| "links": null |
| }, |
| "BIBREF52": { |
| "ref_id": "b52", |
| "title": "UniMorph 2.0: Universal Morphology", |
| "authors": [ |
| { |
| "first": "Christo", |
| "middle": [], |
| "last": "Kirov", |
| "suffix": "" |
| }, |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Cotterell", |
| "suffix": "" |
| }, |
| { |
| "first": "John", |
| "middle": [], |
| "last": "Sylak-Glassman", |
| "suffix": "" |
| }, |
| { |
| "first": "G\u00e9raldine", |
| "middle": [], |
| "last": "Walther", |
| "suffix": "" |
| }, |
| { |
| "first": "Ekaterina", |
| "middle": [], |
| "last": "Vylomova", |
| "suffix": "" |
| }, |
| { |
| "first": "Patrick", |
| "middle": [], |
| "last": "Xia", |
| "suffix": "" |
| }, |
| { |
| "first": "Manaal", |
| "middle": [], |
| "last": "Faruqui", |
| "suffix": "" |
| }, |
| { |
| "first": "Sabrina", |
| "middle": [ |
| "J" |
| ], |
| "last": "Mielke", |
| "suffix": "" |
| }, |
| { |
| "first": "Arya", |
| "middle": [], |
| "last": "Mc-Carthy", |
| "suffix": "" |
| }, |
| { |
| "first": "Sandra", |
| "middle": [], |
| "last": "K\u00fcbler", |
| "suffix": "" |
| }, |
| { |
| "first": "David", |
| "middle": [], |
| "last": "Yarowsky", |
| "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": "Christo Kirov, Ryan Cotterell, John Sylak-Glassman, G\u00e9raldine Walther, Ekaterina Vylomova, Patrick Xia, Manaal Faruqui, Sabrina J. Mielke, Arya Mc- Carthy, Sandra K\u00fcbler, David Yarowsky, Jason Eis- ner, and Mans Hulden. 2018. UniMorph 2.0: Uni- versal Morphology. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. Eu- ropean Language Resources Association (ELRA).", |
| "links": null |
| }, |
| "BIBREF53": { |
| "ref_id": "b53", |
| "title": "Developing LRs for Non-scheduled Indian Languages: A Case of Magahi", |
| "authors": [ |
| { |
| "first": "Ritesh", |
| "middle": [], |
| "last": "Kumar", |
| "suffix": "" |
| }, |
| { |
| "first": "Bornini", |
| "middle": [], |
| "last": "Lahiri", |
| "suffix": "" |
| }, |
| { |
| "first": "Deepak", |
| "middle": [], |
| "last": "Alok", |
| "suffix": "" |
| } |
| ], |
| "year": 2014, |
| "venue": "Human Language Technology Challenges for Computer Science and Linguistics", |
| "volume": "", |
| "issue": "", |
| "pages": "491--501", |
| "other_ids": { |
| "DOI": [ |
| "10.1007/978-3-319-08958-4_40" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ritesh Kumar, Bornini Lahiri, and Deepak Alok. 2014. Developing LRs for Non-scheduled Indian Lan- guages: A Case of Magahi. In Human Language Technology Challenges for Computer Science and Linguistics, Lecture Notes in Computer Science, pages 491-501. Springer International Publishing, Switzerland. Original-date: 2014.", |
| "links": null |
| }, |
| "BIBREF54": { |
| "ref_id": "b54", |
| "title": "Automatic identification of closely-related Indian languages: Resources and experiments", |
| "authors": [ |
| { |
| "first": "Ritesh", |
| "middle": [], |
| "last": "Kumar", |
| "suffix": "" |
| }, |
| { |
| "first": "Bornini", |
| "middle": [], |
| "last": "Lahiri", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Deepak Alok Atul Kr", |
| "suffix": "" |
| }, |
| { |
| "first": "Mayank", |
| "middle": [], |
| "last": "Ojha", |
| "suffix": "" |
| }, |
| { |
| "first": "Abdul", |
| "middle": [], |
| "last": "Jain", |
| "suffix": "" |
| }, |
| { |
| "first": "Yogesh", |
| "middle": [], |
| "last": "Basit", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Dawar", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proceedings of the 4th Workshop on Indian Language Data Resource and Evaluation (WILDRE-4)", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ritesh Kumar, Bornini Lahiri, Deepak Alok Atul Kr. Ojha, Mayank Jain, Abdul Basit, and Yogesh Dawar. 2018. Automatic identification of closely-related In- dian languages: Resources and experiments. In Pro- ceedings of the 4th Workshop on Indian Language Data Resource and Evaluation (WILDRE-4), Paris, France. European Language Resources Association (ELRA).", |
| "links": null |
| }, |
| "BIBREF55": { |
| "ref_id": "b55", |
| "title": "The Case System of Eastern Indo-Aryan Languages: A Typological Overview", |
| "authors": [ |
| { |
| "first": "Bornini", |
| "middle": [], |
| "last": "Lahiri", |
| "suffix": "" |
| } |
| ], |
| "year": 2021, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Bornini Lahiri. 2021. The Case System of Eastern Indo- Aryan Languages: A Typological Overview. Rout- ledge.", |
| "links": null |
| }, |
| "BIBREF56": { |
| "ref_id": "b56", |
| "title": "Towards a robust morphological analyzer for Kunwinjku", |
| "authors": [ |
| { |
| "first": "William", |
| "middle": [], |
| "last": "Lane", |
| "suffix": "" |
| }, |
| { |
| "first": "Steven", |
| "middle": [], |
| "last": "Bird", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the The 17th Annual Workshop of the Australasian Language Technology Association", |
| "volume": "", |
| "issue": "", |
| "pages": "1--9", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "William Lane and Steven Bird. 2019. Towards a robust morphological analyzer for Kunwinjku. In Proceed- ings of the The 17th Annual Workshop of the Aus- tralasian Language Technology Association, pages 1-9, Sydney, Australia. Australasian Language Tech- nology Association.", |
| "links": null |
| }, |
| "BIBREF57": { |
| "ref_id": "b57", |
| "title": "Indonesian morphology tool (Mor-phInd): Towards an Indonesian corpus", |
| "authors": [ |
| { |
| "first": "Vladislav", |
| "middle": [], |
| "last": "Septina Dian Larasati", |
| "suffix": "" |
| }, |
| { |
| "first": "Daniel", |
| "middle": [], |
| "last": "Kubon", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Zeman", |
| "suffix": "" |
| } |
| ], |
| "year": 2011, |
| "venue": "Systems and Frameworks for Computational Morphology -Second International Workshop, SFCM 2011", |
| "volume": "100", |
| "issue": "", |
| "pages": "119--129", |
| "other_ids": { |
| "DOI": [ |
| "10.1007/978-3-642-23138-4_8" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Septina Dian Larasati, Vladislav Kubon, and Daniel Zeman. 2011. Indonesian morphology tool (Mor- phInd): Towards an Indonesian corpus. In Sys- tems and Frameworks for Computational Morphol- ogy -Second International Workshop, SFCM 2011, Zurich, Switzerland, August 26, 2011. Proceedings, volume 100 of Communications in Computer and In- formation Science, pages 119-129. Springer.", |
| "links": null |
| }, |
| "BIBREF58": { |
| "ref_id": "b58", |
| "title": "Can a transformer pass the wug test?", |
| "authors": [ |
| { |
| "first": "Ling", |
| "middle": [], |
| "last": "Liu", |
| "suffix": "" |
| }, |
| { |
| "first": "Mans", |
| "middle": [], |
| "last": "Hulden", |
| "suffix": "" |
| } |
| ], |
| "year": 2021, |
| "venue": "Tuning copying bias in neural morphological inflection models", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": { |
| "arXiv": [ |
| "arXiv:2104.06483" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ling Liu and Mans Hulden. 2021. Can a transformer pass the wug test? Tuning copying bias in neu- ral morphological inflection models. arXiv preprint arXiv:2104.06483.", |
| "links": null |
| }, |
| "BIBREF59": { |
| "ref_id": "b59", |
| "title": "The Penn Arabic Treebank: Building a Large-Scale Annotated Arabic Corpus", |
| "authors": [ |
| { |
| "first": "Mohamed", |
| "middle": [], |
| "last": "Maamouri", |
| "suffix": "" |
| }, |
| { |
| "first": "Ann", |
| "middle": [], |
| "last": "Bies", |
| "suffix": "" |
| }, |
| { |
| "first": "Tim", |
| "middle": [], |
| "last": "Buckwalter", |
| "suffix": "" |
| }, |
| { |
| "first": "Wigdan", |
| "middle": [], |
| "last": "Mekki", |
| "suffix": "" |
| } |
| ], |
| "year": 2004, |
| "venue": "Proceedings of the International Conference on Arabic Language Resources and Tools", |
| "volume": "", |
| "issue": "", |
| "pages": "102--109", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Mohamed Maamouri, Ann Bies, Tim Buckwalter, and Wigdan Mekki. 2004. The Penn Arabic Treebank: Building a Large-Scale Annotated Arabic Corpus. In Proceedings of the International Conference on Arabic Language Resources and Tools, pages 102- 109, Cairo, Egypt.", |
| "links": null |
| }, |
| "BIBREF60": { |
| "ref_id": "b60", |
| "title": "Egyptian Arabic Treebank DF Parts 1-8 V2.0 -LDC catalog numbers LDC2012E93, LDC2012E98, LDC2012E89, LDC2012E99, LDC2012E107, LDC2012E125, LDC2013E12", |
| "authors": [ |
| { |
| "first": "Mohamed", |
| "middle": [], |
| "last": "Maamouri", |
| "suffix": "" |
| }, |
| { |
| "first": "Ann", |
| "middle": [], |
| "last": "Bies", |
| "suffix": "" |
| }, |
| { |
| "first": "Seth", |
| "middle": [], |
| "last": "Kulick", |
| "suffix": "" |
| }, |
| { |
| "first": "Dalila", |
| "middle": [], |
| "last": "Tabessi", |
| "suffix": "" |
| }, |
| { |
| "first": "Sondos", |
| "middle": [], |
| "last": "Krouna", |
| "suffix": "" |
| } |
| ], |
| "year": 2012, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "2013--2034", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Mohamed Maamouri, Ann Bies, Seth Kulick, Dalila Tabessi, and Sondos Krouna. 2012. Egyptian Arabic Treebank DF Parts 1-8 V2.0 -LDC catalog num- bers LDC2012E93, LDC2012E98, LDC2012E89, LDC2012E99, LDC2012E107, LDC2012E125, LDC2013E12, LDC2013E21.", |
| "links": null |
| }, |
| "BIBREF61": { |
| "ref_id": "b61", |
| "title": "Sondos Krouna, Ann Bies, and Seth Kulick. 2010. LDC standard Arabic morphological analyzer (SAMA) version 3.1", |
| "authors": [ |
| { |
| "first": "Mohamed", |
| "middle": [], |
| "last": "Maamouri", |
| "suffix": "" |
| }, |
| { |
| "first": "Dave", |
| "middle": [], |
| "last": "Graff", |
| "suffix": "" |
| }, |
| { |
| "first": "Basma", |
| "middle": [], |
| "last": "Bouziri", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Mohamed Maamouri, Dave Graff, Basma Bouziri, Son- dos Krouna, Ann Bies, and Seth Kulick. 2010. LDC standard Arabic morphological analyzer (SAMA) version 3.1.", |
| "links": null |
| }, |
| "BIBREF62": { |
| "ref_id": "b62", |
| "title": "Lexical databases for computational analyses: A linguistic perspective", |
| "authors": [ |
| { |
| "first": "Robert", |
| "middle": [], |
| "last": "Malouf", |
| "suffix": "" |
| }, |
| { |
| "first": "Farrell", |
| "middle": [], |
| "last": "Ackerman", |
| "suffix": "" |
| }, |
| { |
| "first": "Arturs", |
| "middle": [], |
| "last": "Semenuks", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Proceedings of the Society for Computation in Linguistics 2020", |
| "volume": "", |
| "issue": "", |
| "pages": "446--456", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Robert Malouf, Farrell Ackerman, and Arturs Se- menuks. 2020. Lexical databases for computational analyses: A linguistic perspective. In Proceedings of the Society for Computation in Linguistics 2020, pages 446-456, New York, New York. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF63": { |
| "ref_id": "b63", |
| "title": "Pueblos ind\u00edgenas de la Amazon\u00eda peruana", |
| "authors": [ |
| { |
| "first": "Pedro", |
| "middle": [ |
| "Mayor" |
| ], |
| "last": "Aparicio", |
| "suffix": "" |
| }, |
| { |
| "first": "Richard", |
| "middle": [ |
| "E" |
| ], |
| "last": "Bodmer", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Pedro Mayor Aparicio and Richard E Bodmer. 2009. Pueblos ind\u00edgenas de la Amazon\u00eda peruana. Iquitos: Centro de Estudios Teol\u00f3gicos de la Amazon\u00eda.", |
| "links": null |
| }, |
| "BIBREF64": { |
| "ref_id": "b64", |
| "title": "Cotterell, Mans Hulden, and David Yarowsky. 2020. UniMorph 3.0: Universal Morphology", |
| "authors": [ |
| { |
| "first": "D", |
| "middle": [], |
| "last": "Arya", |
| "suffix": "" |
| }, |
| { |
| "first": "Christo", |
| "middle": [], |
| "last": "Mccarthy", |
| "suffix": "" |
| }, |
| { |
| "first": "Matteo", |
| "middle": [], |
| "last": "Kirov", |
| "suffix": "" |
| }, |
| { |
| "first": "Amrit", |
| "middle": [], |
| "last": "Grella", |
| "suffix": "" |
| }, |
| { |
| "first": "Patrick", |
| "middle": [], |
| "last": "Nidhi", |
| "suffix": "" |
| }, |
| { |
| "first": "Kyle", |
| "middle": [], |
| "last": "Xia", |
| "suffix": "" |
| }, |
| { |
| "first": "Ekaterina", |
| "middle": [], |
| "last": "Gorman", |
| "suffix": "" |
| }, |
| { |
| "first": "Sabrina", |
| "middle": [ |
| "J" |
| ], |
| "last": "Vylomova", |
| "suffix": "" |
| }, |
| { |
| "first": "Garrett", |
| "middle": [], |
| "last": "Mielke", |
| "suffix": "" |
| }, |
| { |
| "first": "Miikka", |
| "middle": [], |
| "last": "Nicolai", |
| "suffix": "" |
| }, |
| { |
| "first": "Timofey", |
| "middle": [], |
| "last": "Silfverberg", |
| "suffix": "" |
| }, |
| { |
| "first": "Nataly", |
| "middle": [], |
| "last": "Arkhangelskiy", |
| "suffix": "" |
| }, |
| { |
| "first": "Andrew", |
| "middle": [], |
| "last": "Krizhanovsky", |
| "suffix": "" |
| }, |
| { |
| "first": "Elena", |
| "middle": [], |
| "last": "Krizhanovsky", |
| "suffix": "" |
| }, |
| { |
| "first": "Alexey", |
| "middle": [], |
| "last": "Klyachko", |
| "suffix": "" |
| }, |
| { |
| "first": "John", |
| "middle": [], |
| "last": "Sorokin", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Mansfield", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "Proceedings of the 12th Language Resources and Evaluation Conference", |
| "volume": "", |
| "issue": "", |
| "pages": "3922--3931", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Arya D. McCarthy, Christo Kirov, Matteo Grella, Amrit Nidhi, Patrick Xia, Kyle Gorman, Ekate- rina Vylomova, Sabrina J. Mielke, Garrett Nico- lai, Miikka Silfverberg, Timofey Arkhangelskiy, Na- taly Krizhanovsky, Andrew Krizhanovsky, Elena Klyachko, Alexey Sorokin, John Mansfield, Valts Ern\u0161treits, Yuval Pinter, Cassandra L. Jacobs, Ryan Cotterell, Mans Hulden, and David Yarowsky. 2020. UniMorph 3.0: Universal Morphology. In Proceed- ings of the 12th Language Resources and Evaluation Conference, pages 3922-3931, Marseille, France. European Language Resources Association.", |
| "links": null |
| }, |
| "BIBREF65": { |
| "ref_id": "b65", |
| "title": "Marrying Universal Dependencies and Universal Morphology", |
| "authors": [ |
| { |
| "first": "D", |
| "middle": [], |
| "last": "Arya", |
| "suffix": "" |
| }, |
| { |
| "first": "Miikka", |
| "middle": [], |
| "last": "Mccarthy", |
| "suffix": "" |
| }, |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Silfverberg", |
| "suffix": "" |
| }, |
| { |
| "first": "Mans", |
| "middle": [], |
| "last": "Cotterell", |
| "suffix": "" |
| }, |
| { |
| "first": "David", |
| "middle": [], |
| "last": "Hulden", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Yarowsky", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proceedings of the Second Workshop on Universal Dependencies (UDW 2018)", |
| "volume": "", |
| "issue": "", |
| "pages": "91--101", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/W18-6011" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Arya D. McCarthy, Miikka Silfverberg, Ryan Cotterell, Mans Hulden, and David Yarowsky. 2018. Marrying Universal Dependencies and Universal Morphology. In Proceedings of the Second Workshop on Univer- sal Dependencies (UDW 2018), pages 91-101, Brus- sels, Belgium. Association for Computational Lin- guistics.", |
| "links": null |
| }, |
| "BIBREF66": { |
| "ref_id": "b66", |
| "title": "The SIGMORPHON 2019 shared task: Morphological analysis in context and cross-lingual transfer for inflection", |
| "authors": [ |
| { |
| "first": "D", |
| "middle": [], |
| "last": "Arya", |
| "suffix": "" |
| }, |
| { |
| "first": "Ekaterina", |
| "middle": [], |
| "last": "Mccarthy", |
| "suffix": "" |
| }, |
| { |
| "first": "Shijie", |
| "middle": [], |
| "last": "Vylomova", |
| "suffix": "" |
| }, |
| { |
| "first": "Chaitanya", |
| "middle": [], |
| "last": "Wu", |
| "suffix": "" |
| }, |
| { |
| "first": "Lawrence", |
| "middle": [], |
| "last": "Malaviya", |
| "suffix": "" |
| }, |
| { |
| "first": "Garrett", |
| "middle": [], |
| "last": "Wolf-Sonkin", |
| "suffix": "" |
| }, |
| { |
| "first": "Christo", |
| "middle": [], |
| "last": "Nicolai", |
| "suffix": "" |
| }, |
| { |
| "first": "Miikka", |
| "middle": [], |
| "last": "Kirov", |
| "suffix": "" |
| }, |
| { |
| "first": "Sabrina", |
| "middle": [ |
| "J" |
| ], |
| "last": "Silfverberg", |
| "suffix": "" |
| }, |
| { |
| "first": "Jeffrey", |
| "middle": [], |
| "last": "Mielke", |
| "suffix": "" |
| }, |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Heinz", |
| "suffix": "" |
| }, |
| { |
| "first": "Mans", |
| "middle": [], |
| "last": "Cotterell", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Hulden", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the 16th Workshop on Computational Research in Phonetics, Phonology, and Morphology", |
| "volume": "", |
| "issue": "", |
| "pages": "229--244", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/W19-4226" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Arya D. McCarthy, Ekaterina Vylomova, Shijie Wu, Chaitanya Malaviya, Lawrence Wolf-Sonkin, Gar- rett Nicolai, Christo Kirov, Miikka Silfverberg, Sab- rina J. Mielke, Jeffrey Heinz, Ryan Cotterell, and Mans Hulden. 2019. The SIGMORPHON 2019 shared task: Morphological analysis in context and cross-lingual transfer for inflection. In Proceedings of the 16th Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 229- 244, Florence, Italy. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF67": { |
| "ref_id": "b67", |
| "title": "The Kampa subgroup of the Arawak language family", |
| "authors": [ |
| { |
| "first": "Elena", |
| "middle": [], |
| "last": "Mihas", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "The Cambridge Handbook of Linguistic Typology, Cambridge Handbooks in Language and Linguistics", |
| "volume": "", |
| "issue": "", |
| "pages": "782--814", |
| "other_ids": { |
| "DOI": [ |
| "10.1017/9781316135716.025" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Elena Mihas. 2017. The Kampa subgroup of the Arawak language family. In Alexandra Y. Aikhen- vald and R. M. W. Dixon, editors, The Cambridge Handbook of Linguistic Typology, Cambridge Hand- books in Language and Linguistics, page 782-814. Cambridge University Press.", |
| "links": null |
| }, |
| "BIBREF68": { |
| "ref_id": "b68", |
| "title": "Modelling verbal morphology in Nen", |
| "authors": [ |
| { |
| "first": "Saliha", |
| "middle": [], |
| "last": "Muradoglu", |
| "suffix": "" |
| }, |
| { |
| "first": "Nicholas", |
| "middle": [], |
| "last": "Evans", |
| "suffix": "" |
| }, |
| { |
| "first": "Ekaterina", |
| "middle": [], |
| "last": "Vylomova", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Proceedings of the The 18th Annual Workshop of the Australasian Language Technology Association", |
| "volume": "", |
| "issue": "", |
| "pages": "43--53", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Saliha Muradoglu, Nicholas Evans, and Ekaterina Vy- lomova. 2020. Modelling verbal morphology in Nen. In Proceedings of the The 18th Annual Work- shop of the Australasian Language Technology As- sociation, pages 43-53, Virtual Workshop. Aus- tralasian Language Technology Association.", |
| "links": null |
| }, |
| "BIBREF69": { |
| "ref_id": "b69", |
| "title": "Unsupervised learning of morphology without morphemes", |
| "authors": [ |
| { |
| "first": "Sylvain", |
| "middle": [], |
| "last": "Neuvel", |
| "suffix": "" |
| }, |
| { |
| "first": "Sean", |
| "middle": [ |
| "A" |
| ], |
| "last": "Fulop", |
| "suffix": "" |
| } |
| ], |
| "year": 2002, |
| "venue": "Proceedings of the ACL-02 Workshop on Morphological and Phonological Learning", |
| "volume": "", |
| "issue": "", |
| "pages": "31--40", |
| "other_ids": { |
| "DOI": [ |
| "10.3115/1118647.1118651" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Sylvain Neuvel and Sean A. Fulop. 2002. Unsuper- vised learning of morphology without morphemes. In Proceedings of the ACL-02 Workshop on Mor- phological and Phonological Learning, pages 31- 40. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF70": { |
| "ref_id": "b70", |
| "title": "Development of rules of generation of nominal word forms for new-written variants of the Karelian language", |
| "authors": [ |
| { |
| "first": "I", |
| "middle": [ |
| "P" |
| ], |
| "last": "Novak", |
| "suffix": "" |
| }, |
| { |
| "first": "N", |
| "middle": [ |
| "B" |
| ], |
| "last": "Krizhanovskaya", |
| "suffix": "" |
| }, |
| { |
| "first": "T", |
| "middle": [ |
| "P" |
| ], |
| "last": "Boiko", |
| "suffix": "" |
| }, |
| { |
| "first": "N", |
| "middle": [ |
| "A" |
| ], |
| "last": "Pellinen", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Vestnik ugrovedenia = Bulletin of Ugric Studies", |
| "volume": "10", |
| "issue": "4", |
| "pages": "679--691", |
| "other_ids": { |
| "DOI": [ |
| "10.30624/2220-4156-2020-10-4-679-691" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "I. P. Novak, N. B. Krizhanovskaya, T. P. Boiko, and N. A. Pellinen. 2020. Development of rules of gen- eration of nominal word forms for new-written vari- ants of the Karelian language. Vestnik ugrovedenia = Bulletin of Ugric Studies, 10(4):679-691.", |
| "links": null |
| }, |
| "BIBREF71": { |
| "ref_id": "b71", |
| "title": "Karelian language and its dialects", |
| "authors": [ |
| { |
| "first": "Irina", |
| "middle": [], |
| "last": "Novak", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Peoples of Karelia: Historical and Ethnographic Essays", |
| "volume": "", |
| "issue": "", |
| "pages": "56--65", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Irina Novak. 2019. Karelian language and its dialects. In I. Vinokurova, editor, Peoples of Karelia: His- torical and Ethnographic Essays, pages 56-65. Pe- riodika.", |
| "links": null |
| }, |
| "BIBREF72": { |
| "ref_id": "b72", |
| "title": "CAMeL tools: An open source python toolkit for Arabic natural language processing", |
| "authors": [ |
| { |
| "first": "Ossama", |
| "middle": [], |
| "last": "Obeid", |
| "suffix": "" |
| }, |
| { |
| "first": "Nasser", |
| "middle": [], |
| "last": "Zalmout", |
| "suffix": "" |
| }, |
| { |
| "first": "Salam", |
| "middle": [], |
| "last": "Khalifa", |
| "suffix": "" |
| }, |
| { |
| "first": "Dima", |
| "middle": [], |
| "last": "Taji", |
| "suffix": "" |
| }, |
| { |
| "first": "Mai", |
| "middle": [], |
| "last": "Oudah", |
| "suffix": "" |
| }, |
| { |
| "first": "Bashar", |
| "middle": [], |
| "last": "Alhafni", |
| "suffix": "" |
| }, |
| { |
| "first": "Go", |
| "middle": [], |
| "last": "Inoue", |
| "suffix": "" |
| }, |
| { |
| "first": "Fadhl", |
| "middle": [], |
| "last": "Eryani", |
| "suffix": "" |
| }, |
| { |
| "first": "Alexander", |
| "middle": [], |
| "last": "Erdmann", |
| "suffix": "" |
| }, |
| { |
| "first": "Nizar", |
| "middle": [], |
| "last": "Habash", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Proceedings of the 12th Language Resources and Evaluation Conference", |
| "volume": "", |
| "issue": "", |
| "pages": "7022--7032", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ossama Obeid, Nasser Zalmout, Salam Khalifa, Dima Taji, Mai Oudah, Bashar Alhafni, Go Inoue, Fadhl Eryani, Alexander Erdmann, and Nizar Habash. 2020. CAMeL tools: An open source python toolkit for Arabic natural language processing. In Proceed- ings of the 12th Language Resources and Evaluation Conference, pages 7022-7032, Marseille, France. European Language Resources Association.", |
| "links": null |
| }, |
| "BIBREF73": { |
| "ref_id": "b73", |
| "title": "A Bayesian approach to the classification of Tungusic languages", |
| "authors": [ |
| { |
| "first": "Sofia", |
| "middle": [], |
| "last": "Oskolskaya", |
| "suffix": "" |
| }, |
| { |
| "first": "Ezequiel", |
| "middle": [], |
| "last": "Koile", |
| "suffix": "" |
| }, |
| { |
| "first": "Martine", |
| "middle": [], |
| "last": "Robbeets", |
| "suffix": "" |
| } |
| ], |
| "year": 2021, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Sofia Oskolskaya, Ezequiel Koile, and Martine Robbeets. 2021. A Bayesian approach to the clas- sification of Tungusic languages. Diachronica.", |
| "links": null |
| }, |
| "BIBREF74": { |
| "ref_id": "b74", |
| "title": "Reconciling Surdas and Keshavdas: A study of commonalities and differences in Brajbhasha literature", |
| "authors": [], |
| "year": null, |
| "venue": "IOSR Journal of Humanities and Social Sciences", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Prateek Pankaj. 2020. Reconciling Surdas and Keshav- das: A study of commonalities and differences in Brajbhasha literature. IOSR Journal of Humanities and Social Sciences, 25.", |
| "links": null |
| }, |
| "BIBREF75": { |
| "ref_id": "b75", |
| "title": "A two-level morphological analyser for the Indonesian language", |
| "authors": [ |
| { |
| "first": "Femphy", |
| "middle": [], |
| "last": "Pisceldo", |
| "suffix": "" |
| }, |
| { |
| "first": "Rahmad", |
| "middle": [], |
| "last": "Mahendra", |
| "suffix": "" |
| }, |
| { |
| "first": "Ruli", |
| "middle": [], |
| "last": "Manurung", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Arka", |
| "suffix": "" |
| } |
| ], |
| "year": 2008, |
| "venue": "Proceedings of the Australasian Language Technology Association Workshop", |
| "volume": "", |
| "issue": "", |
| "pages": "142--150", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Femphy Pisceldo, Rahmad Mahendra, Ruli Manurung, and I Wayan Arka. 2008. A two-level morpholog- ical analyser for the Indonesian language. In Pro- ceedings of the Australasian Language Technology Association Workshop 2008, pages 142-150, Hobart, Australia.", |
| "links": null |
| }, |
| "BIBREF76": { |
| "ref_id": "b76", |
| "title": "Competence-based curriculum learning for neural machine translation", |
| "authors": [ |
| { |
| "first": "Otilia", |
| "middle": [], |
| "last": "Emmanouil Antonios Platanios", |
| "suffix": "" |
| }, |
| { |
| "first": "Graham", |
| "middle": [], |
| "last": "Stretcu", |
| "suffix": "" |
| }, |
| { |
| "first": "Barnab\u00e1s", |
| "middle": [], |
| "last": "Neubig", |
| "suffix": "" |
| }, |
| { |
| "first": "Tom", |
| "middle": [ |
| "M" |
| ], |
| "last": "P\u00f3czos", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Mitchell", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2019", |
| "volume": "1", |
| "issue": "", |
| "pages": "1162--1172", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/n19-1119" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Emmanouil Antonios Platanios, Otilia Stretcu, Graham Neubig, Barnab\u00e1s P\u00f3czos, and Tom M. Mitchell. 2019. Competence-based curriculum learning for neural machine translation. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2019, Minneapolis, MN, USA, June 2-7, 2019, Volume 1 (Long and Short Papers), pages 1162-1172. Associ- ation for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF77": { |
| "ref_id": "b77", |
| "title": "A flexemic tagset for Polish", |
| "authors": [ |
| { |
| "first": "Adam", |
| "middle": [], |
| "last": "Przepi\u00f3rkowski", |
| "suffix": "" |
| }, |
| { |
| "first": "Marcin", |
| "middle": [], |
| "last": "Woli\u0144ski", |
| "suffix": "" |
| } |
| ], |
| "year": 2003, |
| "venue": "Proceedings of the 2003 EACL Workshop on Morphological Processing of Slavic Languages", |
| "volume": "", |
| "issue": "", |
| "pages": "33--40", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Adam Przepi\u00f3rkowski and Marcin Woli\u0144ski. 2003. A flexemic tagset for Polish. In Proceedings of the 2003 EACL Workshop on Morphological Processing of Slavic Languages, pages 33-40, Budapest, Hun- gary. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF78": { |
| "ref_id": "b78", |
| "title": "The UniMelb submission to the SIGMORPHON 2020 shared task 0: Typologically diverse morphological inflection", |
| "authors": [ |
| { |
| "first": "Andreas", |
| "middle": [], |
| "last": "Scherbakov", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Proceedings of the 17th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology", |
| "volume": "", |
| "issue": "", |
| "pages": "177--183", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/2020.sigmorphon-1.20" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Andreas Scherbakov. 2020. The UniMelb submission to the SIGMORPHON 2020 shared task 0: Typo- logically diverse morphological inflection. In Pro- ceedings of the 17th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 177-183, Online. Associa- tion for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF79": { |
| "ref_id": "b79", |
| "title": "The internal division of modern Turkic and its historical implications. Acta Orientalia", |
| "authors": [ |
| { |
| "first": "Claus", |
| "middle": [], |
| "last": "Sch\u00f6nig", |
| "suffix": "" |
| } |
| ], |
| "year": 1999, |
| "venue": "Academiae Scientiarum Hungaricae", |
| "volume": "", |
| "issue": "", |
| "pages": "63--95", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Claus Sch\u00f6nig. 1999. The internal division of modern Turkic and its historical implications. Acta Orien- talia Academiae Scientiarum Hungaricae, pages 63- 95.", |
| "links": null |
| }, |
| "BIBREF80": { |
| "ref_id": "b80", |
| "title": "Exploring looping effects in RNN-based architectures", |
| "authors": [ |
| { |
| "first": "Andrei", |
| "middle": [], |
| "last": "Shcherbakov", |
| "suffix": "" |
| }, |
| { |
| "first": "Saliha", |
| "middle": [], |
| "last": "Muradoglu", |
| "suffix": "" |
| }, |
| { |
| "first": "Ekaterina", |
| "middle": [], |
| "last": "Vylomova", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Proceedings of the The 18th Annual Workshop of the Australasian Language Technology Association", |
| "volume": "", |
| "issue": "", |
| "pages": "115--120", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Andrei Shcherbakov, Saliha Muradoglu, and Ekate- rina Vylomova. 2020. Exploring looping effects in RNN-based architectures. In Proceedings of the The 18th Annual Workshop of the Australasian Lan- guage Technology Association, pages 115-120, Vir- tual Workshop. Australasian Language Technology Association.", |
| "links": null |
| }, |
| "BIBREF81": { |
| "ref_id": "b81", |
| "title": "A universal feature schema for rich morphological annotation and fine-grained cross-lingual part-of-speech tagging", |
| "authors": [ |
| { |
| "first": "John", |
| "middle": [], |
| "last": "Sylak-Glassman", |
| "suffix": "" |
| }, |
| { |
| "first": "Christo", |
| "middle": [], |
| "last": "Kirov", |
| "suffix": "" |
| }, |
| { |
| "first": "Matt", |
| "middle": [], |
| "last": "Post", |
| "suffix": "" |
| }, |
| { |
| "first": "Roger", |
| "middle": [], |
| "last": "Que", |
| "suffix": "" |
| }, |
| { |
| "first": "David", |
| "middle": [], |
| "last": "Yarowsky", |
| "suffix": "" |
| } |
| ], |
| "year": 2015, |
| "venue": "International Workshop on Systems and Frameworks for Computational Morphology", |
| "volume": "", |
| "issue": "", |
| "pages": "72--93", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "John Sylak-Glassman, Christo Kirov, Matt Post, Roger Que, and David Yarowsky. 2015a. A universal feature schema for rich morphological annotation and fine-grained cross-lingual part-of-speech tag- ging. In International Workshop on Systems and Frameworks for Computational Morphology, pages 72-93. Springer.", |
| "links": null |
| }, |
| "BIBREF82": { |
| "ref_id": "b82", |
| "title": "A language-independent feature schema for inflectional morphology", |
| "authors": [ |
| { |
| "first": "John", |
| "middle": [], |
| "last": "Sylak-Glassman", |
| "suffix": "" |
| }, |
| { |
| "first": "Christo", |
| "middle": [], |
| "last": "Kirov", |
| "suffix": "" |
| }, |
| { |
| "first": "David", |
| "middle": [], |
| "last": "Yarowsky", |
| "suffix": "" |
| }, |
| { |
| "first": "Roger", |
| "middle": [], |
| "last": "Que", |
| "suffix": "" |
| } |
| ], |
| "year": 2015, |
| "venue": "Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing", |
| "volume": "2", |
| "issue": "", |
| "pages": "674--680", |
| "other_ids": { |
| "DOI": [ |
| "10.3115/v1/P15-2111" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "John Sylak-Glassman, Christo Kirov, David Yarowsky, and Roger Que. 2015b. A language-independent feature schema for inflectional morphology. In Pro- ceedings of the 53rd Annual Meeting of the Associ- ation for Computational Linguistics and the 7th In- ternational Joint Conference on Natural Language Processing (Volume 2: Short Papers), pages 674- 680, Beijing, China. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF83": { |
| "ref_id": "b83", |
| "title": "Universal Dependencies for Arabic", |
| "authors": [ |
| { |
| "first": "Dima", |
| "middle": [], |
| "last": "Taji", |
| "suffix": "" |
| }, |
| { |
| "first": "Nizar", |
| "middle": [], |
| "last": "Habash", |
| "suffix": "" |
| }, |
| { |
| "first": "Daniel", |
| "middle": [], |
| "last": "Zeman", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "Proceedings of the Third Arabic Natural Language Processing Workshop", |
| "volume": "", |
| "issue": "", |
| "pages": "166--176", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/W17-1320" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Dima Taji, Nizar Habash, and Daniel Zeman. 2017. Universal Dependencies for Arabic. In Proceedings of the Third Arabic Natural Language Processing Workshop, pages 166-176, Valencia, Spain. Associ- ation for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF84": { |
| "ref_id": "b84", |
| "title": "An Arabic morphological analyzer and generator with copious features", |
| "authors": [ |
| { |
| "first": "Dima", |
| "middle": [], |
| "last": "Taji", |
| "suffix": "" |
| }, |
| { |
| "first": "Salam", |
| "middle": [], |
| "last": "Khalifa", |
| "suffix": "" |
| }, |
| { |
| "first": "Ossama", |
| "middle": [], |
| "last": "Obeid", |
| "suffix": "" |
| }, |
| { |
| "first": "Fadhl", |
| "middle": [], |
| "last": "Eryani", |
| "suffix": "" |
| }, |
| { |
| "first": "Nizar", |
| "middle": [], |
| "last": "Habash", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proceedings of the Fifteenth Workshop on Computational Research in Phonetics, Phonology, and Morphology", |
| "volume": "", |
| "issue": "", |
| "pages": "140--150", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/W18-5816" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Dima Taji, Salam Khalifa, Ossama Obeid, Fadhl Eryani, and Nizar Habash. 2018. An Arabic mor- phological analyzer and generator with copious fea- tures. In Proceedings of the Fifteenth Workshop on Computational Research in Phonetics, Phonol- ogy, and Morphology, pages 140-150, Brussels, Bel- gium. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF85": { |
| "ref_id": "b85", |
| "title": "A new classification of the Turkic languages", |
| "authors": [ |
| { |
| "first": "Talat", |
| "middle": [], |
| "last": "Tekin", |
| "suffix": "" |
| } |
| ], |
| "year": 1990, |
| "venue": "T\u00fcrk dilleri ara\u015ft\u0131rmalar\u0131", |
| "volume": "1", |
| "issue": "", |
| "pages": "5--18", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Talat Tekin. 1990. A new classification of the Turkic languages. T\u00fcrk dilleri ara\u015ft\u0131rmalar\u0131, 1:5-18.", |
| "links": null |
| }, |
| "BIBREF86": { |
| "ref_id": "b86", |
| "title": "A finite-state morphological analyser for Tuvan", |
| "authors": [ |
| { |
| "first": "Francis", |
| "middle": [], |
| "last": "Tyers", |
| "suffix": "" |
| }, |
| { |
| "first": "Aziyana", |
| "middle": [], |
| "last": "Bayyr-Ool", |
| "suffix": "" |
| }, |
| { |
| "first": "Aelita", |
| "middle": [], |
| "last": "Salchak", |
| "suffix": "" |
| }, |
| { |
| "first": "Jonathan", |
| "middle": [], |
| "last": "Washington", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)", |
| "volume": "", |
| "issue": "", |
| "pages": "2562--2567", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Francis Tyers, Aziyana Bayyr-ool, Aelita Salchak, and Jonathan Washington. 2016. A finite-state mor- phological analyser for Tuvan. In Proceedings of the Tenth International Conference on Language Re- sources and Evaluation (LREC'16), pages 2562- 2567, Portoro\u017e, Slovenia. European Language Re- sources Association (ELRA).", |
| "links": null |
| }, |
| "BIBREF87": { |
| "ref_id": "b87", |
| "title": "Dependency annotation of noun incorporation in polysynthetic languages", |
| "authors": [ |
| { |
| "first": "Francis", |
| "middle": [], |
| "last": "Tyers", |
| "suffix": "" |
| }, |
| { |
| "first": "Karina", |
| "middle": [], |
| "last": "Mishchenkova", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Proceedings of the Fourth Workshop on Universal Dependencies (UDW 2020)", |
| "volume": "", |
| "issue": "", |
| "pages": "195--204", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Francis Tyers and Karina Mishchenkova. 2020. Depen- dency annotation of noun incorporation in polysyn- thetic languages. In Proceedings of the Fourth Workshop on Universal Dependencies (UDW 2020), pages 195-204, Barcelona, Spain (Online). Associa- tion for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF88": { |
| "ref_id": "b88", |
| "title": "Attention is all you need", |
| "authors": [ |
| { |
| "first": "Ashish", |
| "middle": [], |
| "last": "Vaswani", |
| "suffix": "" |
| }, |
| { |
| "first": "Noam", |
| "middle": [], |
| "last": "Shazeer", |
| "suffix": "" |
| }, |
| { |
| "first": "Niki", |
| "middle": [], |
| "last": "Parmar", |
| "suffix": "" |
| }, |
| { |
| "first": "Jakob", |
| "middle": [], |
| "last": "Uszkoreit", |
| "suffix": "" |
| }, |
| { |
| "first": "Llion", |
| "middle": [], |
| "last": "Jones", |
| "suffix": "" |
| }, |
| { |
| "first": "Aidan", |
| "middle": [ |
| "N" |
| ], |
| "last": "Gomez", |
| "suffix": "" |
| }, |
| { |
| "first": "\u0141ukasz", |
| "middle": [], |
| "last": "Kaiser", |
| "suffix": "" |
| }, |
| { |
| "first": "Illia", |
| "middle": [], |
| "last": "Polosukhin", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "Advances in Neural Information Processing Systems", |
| "volume": "30", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, \u0141ukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Advances in Neural Information Pro- cessing Systems, volume 30.", |
| "links": null |
| }, |
| "BIBREF89": { |
| "ref_id": "b89", |
| "title": "The Itelmen language", |
| "authors": [ |
| { |
| "first": "A", |
| "middle": [ |
| "P" |
| ], |
| "last": "Volodin", |
| "suffix": "" |
| } |
| ], |
| "year": 1976, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "A. P. Volodin. 1976. The Itelmen language.", |
| "links": null |
| }, |
| "BIBREF91": { |
| "ref_id": "b91", |
| "title": "Miikka Silfverberg, and Mans Hulden. 2020. SIGMOR-PHON 2020 shared task 0: Typologically diverse morphological inflection", |
| "authors": [ |
| { |
| "first": "Ekaterina", |
| "middle": [], |
| "last": "Vylomova", |
| "suffix": "" |
| }, |
| { |
| "first": "Jennifer", |
| "middle": [], |
| "last": "White", |
| "suffix": "" |
| }, |
| { |
| "first": "Elizabeth", |
| "middle": [], |
| "last": "Salesky", |
| "suffix": "" |
| }, |
| { |
| "first": "Sabrina", |
| "middle": [ |
| "J" |
| ], |
| "last": "Mielke", |
| "suffix": "" |
| }, |
| { |
| "first": "Shijie", |
| "middle": [], |
| "last": "Wu", |
| "suffix": "" |
| }, |
| { |
| "first": "Maria", |
| "middle": [], |
| "last": "Edoardo", |
| "suffix": "" |
| }, |
| { |
| "first": "Rowan", |
| "middle": [], |
| "last": "Ponti", |
| "suffix": "" |
| }, |
| { |
| "first": "Ran", |
| "middle": [], |
| "last": "Hall Maudslay", |
| "suffix": "" |
| }, |
| { |
| "first": "Josef", |
| "middle": [], |
| "last": "Zmigrod", |
| "suffix": "" |
| }, |
| { |
| "first": "Svetlana", |
| "middle": [], |
| "last": "Valvoda", |
| "suffix": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [], |
| "last": "Toldova", |
| "suffix": "" |
| }, |
| { |
| "first": "Elena", |
| "middle": [], |
| "last": "Tyers", |
| "suffix": "" |
| }, |
| { |
| "first": "Ilya", |
| "middle": [], |
| "last": "Klyachko", |
| "suffix": "" |
| }, |
| { |
| "first": "Natalia", |
| "middle": [], |
| "last": "Yegorov", |
| "suffix": "" |
| }, |
| { |
| "first": "Paula", |
| "middle": [], |
| "last": "Krizhanovsky", |
| "suffix": "" |
| }, |
| { |
| "first": "Irene", |
| "middle": [], |
| "last": "Czarnowska", |
| "suffix": "" |
| }, |
| { |
| "first": "Andrew", |
| "middle": [], |
| "last": "Nikkarinen", |
| "suffix": "" |
| }, |
| { |
| "first": "Tiago", |
| "middle": [], |
| "last": "Krizhanovsky", |
| "suffix": "" |
| }, |
| { |
| "first": "Lucas", |
| "middle": [], |
| "last": "Pimentel", |
| "suffix": "" |
| }, |
| { |
| "first": "Christo", |
| "middle": [], |
| "last": "Torroba Hennigen", |
| "suffix": "" |
| }, |
| { |
| "first": "Garrett", |
| "middle": [], |
| "last": "Kirov", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Nicolai", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "1--39", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/2020.sigmorphon-1.1" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ekaterina Vylomova, Jennifer White, Eliza- beth Salesky, Sabrina J. Mielke, Shijie Wu, Edoardo Maria Ponti, Rowan Hall Maudslay, Ran Zmigrod, Josef Valvoda, Svetlana Toldova, Francis Tyers, Elena Klyachko, Ilya Yegorov, Natalia Krizhanovsky, Paula Czarnowska, Irene Nikkarinen, Andrew Krizhanovsky, Tiago Pimentel, Lucas Torroba Hennigen, Christo Kirov, Garrett Nicolai, Adina Williams, Antonios Anastasopoulos, Hilaria Cruz, Eleanor Chodroff, Ryan Cotterell, Miikka Silfverberg, and Mans Hulden. 2020. SIGMOR- PHON 2020 shared task 0: Typologically diverse morphological inflection. In Proceedings of the 17th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 1-39, Online. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF92": { |
| "ref_id": "b92", |
| "title": "Development of a finite-state model for morphological processing of Tuvan", |
| "authors": [ |
| { |
| "first": "Jonathan", |
| "middle": [], |
| "last": "North Washington", |
| "suffix": "" |
| }, |
| { |
| "first": "Aziyana", |
| "middle": [], |
| "last": "Bayyr-Ool", |
| "suffix": "" |
| }, |
| { |
| "first": "Aelita", |
| "middle": [], |
| "last": "Salchak", |
| "suffix": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [ |
| "M" |
| ], |
| "last": "Tyers", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "Rodnoy Yazyk", |
| "volume": "1", |
| "issue": "", |
| "pages": "156--187", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Jonathan North Washington, Aziyana Bayyr-ool, Aelita Salchak, and Francis M Tyers. 2016. De- velopment of a finite-state model for morphological processing of Tuvan. Rodnoy Yazyk, 1:156-187.", |
| "links": null |
| }, |
| "BIBREF94": { |
| "ref_id": "b94", |
| "title": "Free/open-source technologies for Turkic languages developed in the Apertium project", |
| "authors": [], |
| "year": 2019, |
| "venue": "Proceedings of the International Conference on Turkic Language Processing", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Free/open-source technologies for Turkic languages developed in the Apertium project. In Proceedings of the International Conference on Turkic Language Processing (TURKLANG 2019).", |
| "links": null |
| }, |
| "BIBREF95": { |
| "ref_id": "b95", |
| "title": "Delineating Turkic non-finite verb forms by syntactic function", |
| "authors": [ |
| { |
| "first": "Jonathan", |
| "middle": [], |
| "last": "North Washington", |
| "suffix": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [ |
| "Morton" |
| ], |
| "last": "Tyers", |
| "suffix": "" |
| } |
| ], |
| "year": 2019, |
| "venue": "Proceedings of the Workshop on Turkic and Languages in Contact with Turkic", |
| "volume": "4", |
| "issue": "", |
| "pages": "115--129", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Jonathan North Washington and Francis Morton Tyers. 2019. Delineating Turkic non-finite verb forms by syntactic function. In Proceedings of the Workshop on Turkic and Languages in Contact with Turkic, volume 4, pages 115-129.", |
| "links": null |
| }, |
| "BIBREF96": { |
| "ref_id": "b96", |
| "title": "Predicting declension class from form and meaning", |
| "authors": [ |
| { |
| "first": "Adina", |
| "middle": [], |
| "last": "Williams", |
| "suffix": "" |
| }, |
| { |
| "first": "Tiago", |
| "middle": [], |
| "last": "Pimentel", |
| "suffix": "" |
| }, |
| { |
| "first": "Hagen", |
| "middle": [], |
| "last": "Blix", |
| "suffix": "" |
| }, |
| { |
| "first": "Arya", |
| "middle": [ |
| "D" |
| ], |
| "last": "Mccarthy", |
| "suffix": "" |
| }, |
| { |
| "first": "Eleanor", |
| "middle": [], |
| "last": "Chodroff", |
| "suffix": "" |
| }, |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Cotterell", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics", |
| "volume": "", |
| "issue": "", |
| "pages": "6682--6695", |
| "other_ids": { |
| "DOI": [ |
| "10.18653/v1/2020.acl-main.597" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Adina Williams, Tiago Pimentel, Hagen Blix, Arya D. McCarthy, Eleanor Chodroff, and Ryan Cotterell. 2020. Predicting declension class from form and meaning. In Proceedings of the 58th Annual Meet- ing of the Association for Computational Linguistics, pages 6682-6695, Online. Association for Computa- tional Linguistics.", |
| "links": null |
| }, |
| "BIBREF97": { |
| "ref_id": "b97", |
| "title": "Applicative affixes in Peruvian Amazonian languages. Current Studies on South American Languages", |
| "authors": [ |
| { |
| "first": "Mary", |
| "middle": [ |
| "Ruth" |
| ], |
| "last": "Wise", |
| "suffix": "" |
| } |
| ], |
| "year": 2002, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "329--344", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Mary Ruth Wise. 2002. Applicative affixes in Pe- ruvian Amazonian languages. Current Studies on South American Languages [Indigenous Languages of Latin America, 3], pages 329-344.", |
| "links": null |
| }, |
| "BIBREF98": { |
| "ref_id": "b98", |
| "title": "The online version of grammatical dictionary of Polish", |
| "authors": [ |
| { |
| "first": "Marcin", |
| "middle": [], |
| "last": "Woli\u0144ski", |
| "suffix": "" |
| }, |
| { |
| "first": "Witold", |
| "middle": [], |
| "last": "Kiera\u015b", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)", |
| "volume": "", |
| "issue": "", |
| "pages": "2589--2594", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Marcin Woli\u0144ski and Witold Kiera\u015b. 2016. The on- line version of grammatical dictionary of Polish. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16), pages 2589-2594, Portoro\u017e, Slovenia. European Language Resources Association (ELRA).", |
| "links": null |
| }, |
| "BIBREF99": { |
| "ref_id": "b99", |
| "title": "S\u0142ownik gramatyczny j\u0119zyka polskiego", |
| "authors": [ |
| { |
| "first": "Marcin", |
| "middle": [], |
| "last": "Woli\u0144ski", |
| "suffix": "" |
| }, |
| { |
| "first": "Zygmunt", |
| "middle": [], |
| "last": "Saloni", |
| "suffix": "" |
| }, |
| { |
| "first": "Robert", |
| "middle": [], |
| "last": "Wo\u0142osz", |
| "suffix": "" |
| }, |
| { |
| "first": "W\u0142odzimierz", |
| "middle": [], |
| "last": "Gruszczy\u0144ski", |
| "suffix": "" |
| }, |
| { |
| "first": "Danuta", |
| "middle": [], |
| "last": "Skowro\u0144ska", |
| "suffix": "" |
| }, |
| { |
| "first": "Zbigniew", |
| "middle": [], |
| "last": "Bronk", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Marcin Woli\u0144ski, Zygmunt Saloni, Robert Wo\u0142osz, W\u0142odzimierz Gruszczy\u0144ski, Danuta Skowro\u0144ska, and Zbigniew Bronk. 2020. S\u0142ownik gramatyczny j\u0119zyka polskiego, 4th edition. Warsaw. http:// sgjp.pl.", |
| "links": null |
| }, |
| "BIBREF100": { |
| "ref_id": "b100", |
| "title": "Applying the transformer to character-level transduction", |
| "authors": [ |
| { |
| "first": "Shijie", |
| "middle": [], |
| "last": "Wu", |
| "suffix": "" |
| }, |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Cotterell", |
| "suffix": "" |
| }, |
| { |
| "first": "Mans", |
| "middle": [], |
| "last": "Hulden", |
| "suffix": "" |
| } |
| ], |
| "year": 2021, |
| "venue": "Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume", |
| "volume": "", |
| "issue": "", |
| "pages": "1901--1907", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Shijie Wu, Ryan Cotterell, and Mans Hulden. 2021. Applying the transformer to character-level transduc- tion. In Proceedings of the 16th Conference of the European Chapter of the Association for Computa- tional Linguistics: Main Volume, pages 1901-1907, Online. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF101": { |
| "ref_id": "b101", |
| "title": "Open corpus of Veps and Karelian languages (VepKar): preliminary data collection and dictionaries", |
| "authors": [ |
| { |
| "first": "Nina", |
| "middle": [], |
| "last": "Zaytseva", |
| "suffix": "" |
| }, |
| { |
| "first": "Andrew", |
| "middle": [], |
| "last": "Krizhanovsky", |
| "suffix": "" |
| }, |
| { |
| "first": "Natalia", |
| "middle": [], |
| "last": "Krizhanovsky", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "Corpus Linguistics-2017", |
| "volume": "", |
| "issue": "", |
| "pages": "172--177", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Nina Zaytseva, Andrew Krizhanovsky, Natalia Krizhanovsky, Natalia Pellinen, and Aleksndra Ro- dionova. 2017. Open corpus of Veps and Karelian languages (VepKar): preliminary data collection and dictionaries. In Corpus Linguistics-2017, pages 172-177.", |
| "links": null |
| }, |
| "BIBREF102": { |
| "ref_id": "b102", |
| "title": "Universal Dependency treebank for Xibe", |
| "authors": [ |
| { |
| "first": "He", |
| "middle": [], |
| "last": "Zhou", |
| "suffix": "" |
| }, |
| { |
| "first": "Juyeon", |
| "middle": [], |
| "last": "Chung", |
| "suffix": "" |
| }, |
| { |
| "first": "Sandra", |
| "middle": [], |
| "last": "K\u00fcbler", |
| "suffix": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [], |
| "last": "Tyers", |
| "suffix": "" |
| } |
| ], |
| "year": 2020, |
| "venue": "Proceedings of the Fourth Workshop on Universal Dependencies (UDW 2020)", |
| "volume": "", |
| "issue": "", |
| "pages": "205--215", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "He Zhou, Juyeon Chung, Sandra K\u00fcbler, and Francis Tyers. 2020. Universal Dependency treebank for Xibe. In Proceedings of the Fourth Workshop on Universal Dependencies (UDW 2020), pages 205- 215, Barcelona, Spain (Online). Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF103": { |
| "ref_id": "b103", |
| "title": "Gu\u00eda te\u00f3rica del idioma ash\u00e1ninka. Lima: Universidad Cat\u00f3lica Sedes Sapientiae", |
| "authors": [ |
| { |
| "first": "Zumaeta", |
| "middle": [], |
| "last": "Esa\u00fa", |
| "suffix": "" |
| }, |
| { |
| "first": "Gerardo", |
| "middle": [ |
| "Anton" |
| ], |
| "last": "Rojas", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Zerdin", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Esa\u00fa Zumaeta Rojas and Gerardo Anton Zerdin. 2018. Gu\u00eda te\u00f3rica del idioma ash\u00e1ninka. Lima: Universi- dad Cat\u00f3lica Sedes Sapientiae.", |
| "links": null |
| }, |
| "BIBREF104": { |
| "ref_id": "b104", |
| "title": "\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435 \u0432 \u0438\u0437\u0443\u0447\u0435\u043d\u0438\u0435 \u0442\u044e\u0440\u043a\u0441\u043a\u0438\u0445 \u044f\u0437\u044b\u043a\u043e\u0432 [N. A. Baskakov. An introduction to Turkic language studies", |
| "authors": [ |
| { |
| "first": "\u041d", |
| "middle": [ |
| "\u0410" |
| ], |
| "last": "\u0411\u0430\u0441\u043a\u0430\u043a\u043e\u0432", |
| "suffix": "" |
| } |
| ], |
| "year": 1969, |
| "venue": "\u041c\u043e\u0441\u043a\u0432\u0430: \u0412\u044b\u0441\u0448\u0430\u044f \u0448\u043a\u043e\u043b\u0430", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "\u041d. \u0410. \u0411\u0430\u0441\u043a\u0430\u043a\u043e\u0432. 1969. \u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435 \u0432 \u0438\u0437\u0443\u0447\u0435\u043d\u0438\u0435 \u0442\u044e\u0440\u043a\u0441\u043a\u0438\u0445 \u044f\u0437\u044b\u043a\u043e\u0432 [N. A. Baskakov. An intro- duction to Turkic language studies]. \u041c\u043e\u0441\u043a\u0432\u0430: \u0412\u044b\u0441\u0448\u0430\u044f \u0448\u043a\u043e\u043b\u0430.", |
| "links": null |
| }, |
| "BIBREF105": { |
| "ref_id": "b105", |
| "title": "Iskhakov and A. A. Pal'mbakh. A grammar of Tuvan: Phonetics and morphology", |
| "authors": [ |
| { |
| "first": "\u0424", |
| "middle": [ |
| "\u0413" |
| ], |
| "last": "\u0418\u0441\u0445\u0430\u043a\u043e\u0432", |
| "suffix": "" |
| }, |
| { |
| "first": "\u0410", |
| "middle": [ |
| "\u0410" |
| ], |
| "last": "\u041f\u0430\u043b\u044c\u043c\u0431\u0430\u0445", |
| "suffix": "" |
| } |
| ], |
| "year": 1961, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "\u0424. \u0413. \u0418\u0441\u0445\u0430\u043a\u043e\u0432 and \u0410. \u0410. \u041f\u0430\u043b\u044c\u043c\u0431\u0430\u0445. 1961. \u0413\u0440\u0430\u043c\u043c\u0430\u0442\u0438\u043a\u0430 \u0442\u0443\u0432\u0438\u043d\u0441\u043a\u043e\u0433\u043e \u044f\u0437\u044b\u043a\u0430: \u0424\u043e\u043d\u0435\u0442\u0438\u043a\u0430 \u0438 \u043c\u043e\u0440\u0444\u043e\u043b\u043e\u0433\u0438\u044f [F. G. Iskhakov and A. A. Pal'mbakh. A grammar of Tuvan: Phonetics and morphology].", |
| "links": null |
| }, |
| "BIBREF107": { |
| "ref_id": "b107", |
| "title": "\u0413\u0440\u0430\u043c\u043c\u0430\u0442\u0438\u043a\u0430 \u0441\u043e\u0432\u0440\u0435\u043c\u0435\u043d\u043d\u043e\u0433\u043e \u044f\u043a\u0443\u0442\u0441\u043a\u043e\u0433\u043e \u043b\u0438\u0442\u0435\u0440\u0430\u0442\u0443\u0440\u043d\u043e\u0433\u043e \u044f\u0437\u044b\u043a\u0430: \u0424\u043e\u043d\u0435\u0442\u0438\u043a\u0430 \u0438 \u043c\u043e\u0440\u0444\u043e\u043b\u043e\u0433\u0438\u044f [E. I. Ubryatova et al. Grammar of the modern Yakut literary language: Phonetics and morphology", |
| "authors": [ |
| { |
| "first": "\u0415", |
| "middle": [ |
| "\u0418" |
| ], |
| "last": "\u0423\u0431\u0440\u044f\u0442\u043e\u0432\u0430", |
| "suffix": "" |
| }, |
| { |
| "first": "\u0415", |
| "middle": [ |
| "\u0418" |
| ], |
| "last": "\u041a\u043e\u0440\u043a\u0438\u043d\u0430", |
| "suffix": "" |
| }, |
| { |
| "first": "\u041b", |
| "middle": [ |
| "\u041d" |
| ], |
| "last": "\u0425\u0430\u0440\u0438\u0442\u043e\u043d\u043e\u0432", |
| "suffix": "" |
| }, |
| { |
| "first": "\u041d", |
| "middle": [ |
| "\u0415" |
| ], |
| "last": "\u041f\u0435\u0442\u0440\u043e\u0432", |
| "suffix": "" |
| } |
| ], |
| "year": 1982, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "\u0415. \u0418. \u0423\u0431\u0440\u044f\u0442\u043e\u0432\u0430, \u0415. \u0418. \u041a\u043e\u0440\u043a\u0438\u043d\u0430, \u041b. \u041d. \u0425\u0430\u0440\u0438\u0442\u043e\u043d\u043e\u0432, and \u041d. \u0415. \u041f\u0435\u0442\u0440\u043e\u0432, editors. 1982. \u0413\u0440\u0430\u043c\u043c\u0430\u0442\u0438\u043a\u0430 \u0441\u043e\u0432\u0440\u0435\u043c\u0435\u043d\u043d\u043e\u0433\u043e \u044f\u043a\u0443\u0442\u0441\u043a\u043e\u0433\u043e \u043b\u0438\u0442\u0435\u0440\u0430\u0442\u0443\u0440\u043d\u043e\u0433\u043e \u044f\u0437\u044b\u043a\u0430: \u0424\u043e\u043d\u0435\u0442\u0438\u043a\u0430 \u0438 \u043c\u043e\u0440\u0444\u043e\u043b\u043e\u0433\u0438\u044f [E. I. Ubrya- tova et al. Grammar of the modern Yakut literary language: Phonetics and morphology]. \u041c\u043e\u0441\u043a\u0432\u0430: \u041d\u0430\u0443\u043a\u0430.", |
| "links": null |
| }, |
| "BIBREF108": { |
| "ref_id": "b108", |
| "title": "Accuracy for \"Noun", |
| "authors": [], |
| "year": null, |
| "venue": "", |
| "volume": "18", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Table 18: Accuracy for \"Noun\" in each language on the test data.", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "TABREF1": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table/>", |
| "text": "" |
| }, |
| "TABREF4": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table/>", |
| "text": "Accuracy for each language on the test data." |
| }, |
| "TABREF5": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td>demonstrates, most systems achieve</td></tr><tr><td>over 90% accuracy on most languages, with</td></tr><tr><td>transformer-based baseline models demonstrating</td></tr><tr><td>superior performance on all language families ex-</td></tr><tr><td>cept Uralic. Two Turkic languages, Sakha and Tu-</td></tr><tr><td>van, achieve particularly high accuracy of 99.97%.</td></tr></table>", |
| "text": "" |
| }, |
| "TABREF7": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table/>", |
| "text": "Accuracy comparison for the lemmas known from the training set (black numbers) vs. unknown lem-" |
| }, |
| "TABREF9": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table/>", |
| "text": "Test accuracy for each language for the samples where none of baseline systems succeeds to produce correct prediction." |
| }, |
| "TABREF11": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td>L</td><td>BME</td><td>GUClasp</td><td>TRM</td><td>TRM+ AUG</td><td>CHR-TRM</td><td>CHR-TRM +AUG</td></tr><tr><td colspan=\"2\">aym 100.00</td><td>99.90</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td/><td>-</td><td>-</td><td>-</td><td>-</td><td>-</td><td>-</td></tr><tr><td>bul</td><td>95.00</td><td>93.21</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td/><td>81.81</td><td>63.63</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>ces</td><td>75.00</td><td>55.00</td><td>80.00</td><td>80.00</td><td>80.00</td><td>80.00</td></tr><tr><td/><td>71.42</td><td>57.14</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td colspan=\"2\">ckb 100.00</td><td>100.00</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td/><td>-</td><td>-</td><td>-</td><td>-</td><td>-</td><td>-</td></tr><tr><td>deu</td><td>88.57</td><td>74.28</td><td>97.14</td><td>97.14</td><td>100.00</td><td>100.00</td></tr><tr><td/><td>71.87</td><td>0</td><td>93.75</td><td>93.75</td><td>87.50</td><td>87.50</td></tr><tr><td>kmr</td><td>98.76</td><td>98.71</td><td>95.34</td><td>95.34</td><td>95.51</td><td>95.51</td></tr><tr><td/><td>-</td><td>-</td><td>-</td><td>-</td><td>-</td><td>-</td></tr><tr><td>nld</td><td>50.00</td><td>50.00</td><td>50.00</td><td>50.00</td><td>50.00</td><td>50.00</td></tr><tr><td/><td>-</td><td>-</td><td>-</td><td>-</td><td>-</td><td>-</td></tr><tr><td>pol</td><td>99.85</td><td>99.60</td><td>99.97</td><td>99.97</td><td>99.91</td><td>99.91</td></tr><tr><td/><td>98.28</td><td>92.12</td><td>98.28</td><td>98.28</td><td>98.28</td><td>98.28</td></tr><tr><td>rus</td><td>90.93</td><td>87.91</td><td>96.45</td><td>96.45</td><td>96.05</td><td>96.05</td></tr><tr><td/><td>56.55</td><td>27.04</td><td>72.13</td><td>72.13</td><td>72.13</td><td>72.13</td></tr><tr><td>tur</td><td>99.77</td><td>98.63</td><td>95.67</td><td>95.67</td><td>95.58</td><td>95.58</td></tr><tr><td/><td>90.90</td><td>59.09</td><td>77.27</td><td>77.27</td><td>86.36</td><td>86.36</td></tr></table>", |
| "text": "Accuracy comparison for fragment substitutions that could be observed in the training set (black numbers) vs. more complex transformations (red numbers). Groups having <20 unique lemmas are marked with asterisks." |
| }, |
| "TABREF13": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td/><td>Level-1</td><td/><td>Level-2</td><td/><td>Level-3</td></tr><tr><td colspan=\"6\">MorphInd UniMorph MorphInd UniMorph MorphInd UniMorph</td></tr><tr><td>N</td><td>N</td><td>P</td><td>PL</td><td>F</td><td>FEM</td></tr><tr><td/><td/><td>S</td><td>SG</td><td>M</td><td>MASC</td></tr><tr><td/><td/><td/><td/><td>D</td><td>NEUT</td></tr><tr><td>P</td><td>PROPN</td><td>P</td><td>PL</td><td>1</td><td>1</td></tr><tr><td/><td/><td>S</td><td>SG</td><td>2</td><td>2</td></tr><tr><td/><td/><td/><td/><td>3</td><td>3</td></tr><tr><td>V</td><td>V</td><td>P</td><td>-</td><td>A</td><td>ACT</td></tr><tr><td/><td/><td>S</td><td>-</td><td>P</td><td>PASS</td></tr><tr><td>C</td><td>NUM</td><td>C</td><td>-</td><td>-</td><td>-</td></tr><tr><td/><td/><td>O</td><td>-</td><td>-</td><td>-</td></tr><tr><td/><td/><td>D</td><td>-</td><td>-</td><td>-</td></tr><tr><td>A</td><td>ADJ</td><td>P</td><td>PL</td><td>P</td><td>-</td></tr><tr><td/><td/><td>S</td><td>SG</td><td>S</td><td>-</td></tr></table>", |
| "text": "Apertium tag mapping to the UniMorph schema for Sakha and Tuvan. For the definitions of the Apertium tags, see. This mapping alone is not sufficient to reconstruct the UniMorph annotation, since some conditional rules are applied on top of this conversion (see \u00a73.8.1)" |
| }, |
| "TABREF14": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td colspan=\"2\">PL tag UniMorph tag</td><td>PL tag</td><td>UniMorph tag</td><td>PL tag</td><td>UniMorph tag</td></tr><tr><td>pri</td><td>1</td><td>impt</td><td>IMP</td><td>perf</td><td>PFV</td></tr><tr><td>sec</td><td>2</td><td>imps</td><td>IMPRS</td><td>pl</td><td>PL</td></tr><tr><td>ter</td><td>3</td><td>inst</td><td>INS</td><td>fin</td><td>PRS/FUT</td></tr><tr><td>acc</td><td>ACC</td><td colspan=\"2\">imperf IPFV</td><td colspan=\"2\">praet PRT</td></tr><tr><td>adj</td><td>ADJ</td><td>m2</td><td>MASC;ANIM</td><td>sg</td><td>SG</td></tr><tr><td>adv</td><td>ADV</td><td>m1</td><td>MASC;HUM</td><td>sup</td><td>SPRL</td></tr><tr><td>com</td><td>COMPR</td><td>m3</td><td>MASC;INAN</td><td>pcon</td><td>V.CVB;PRS</td></tr><tr><td>dat</td><td>DAT</td><td>subst</td><td>N</td><td>pant</td><td>V.CVB;PST</td></tr><tr><td>pos</td><td>EQT</td><td>neg</td><td>NEG</td><td>ger</td><td>V.MSDR</td></tr><tr><td>loc</td><td>ESS</td><td>n</td><td>NEUT</td><td>voc</td><td>VOC</td></tr><tr><td>f</td><td>FEM</td><td>inf</td><td>NFIN</td><td>pact</td><td>V.PTCP;ACT</td></tr><tr><td>gen</td><td>GEN</td><td>nom</td><td>NOM</td><td>ppas</td><td>V.PTCP;PASS</td></tr></table>", |
| "text": "A simplified mapping from MorphInd tags to the UniMorph schema for Indonesian data. We follow MorphInd's three-level annotations for the mapping." |
| }, |
| "TABREF15": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table/>", |
| "text": "Simplified mapping from the original flexemic tagset of Polish used in Polish morphological analysers and corpora annotations" |
| }, |
| "TABREF17": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td>Xibe Universal Dependencies feature /</td><td>UniMorph</td><td>Additional rules</td></tr><tr><td>word transliteration</td><td/><td/></tr><tr><td>VerbForm=Conv</td><td>POS=V.CVB</td><td/></tr><tr><td>VerbForm=Fin</td><td>FIN</td><td/></tr><tr><td>VerbForm=Inf</td><td>NFIN</td><td/></tr><tr><td>VerbForm=Part</td><td>POS=V.PTCP</td><td/></tr><tr><td>VerbForm=Vnoun</td><td>POS=V.MSDR</td><td/></tr><tr><td>Voice=Act</td><td>ACT</td><td/></tr><tr><td>Voice=Cau</td><td>CAUS</td><td/></tr><tr><td>Voice=Pass</td><td>PASS</td><td/></tr><tr><td>Voice=Rcp</td><td>RECP</td><td/></tr><tr><td>ateke</td><td>_</td><td/></tr><tr><td>dari</td><td>_</td><td>means 'each, every'</td></tr><tr><td>eiten</td><td>_</td><td>means 'each, every'</td></tr><tr><td>enteke</td><td>_</td><td>means 'like this'</td></tr><tr><td>ere</td><td>PROX</td><td/></tr><tr><td>erebe</td><td>PROX</td><td/></tr><tr><td>ereci</td><td>PROX</td><td/></tr><tr><td>eremu</td><td>PROX</td><td/></tr><tr><td>geren</td><td>_</td><td>means 'all'</td></tr><tr><td>harangga</td><td>_</td><td/></tr><tr><td>tenteke</td><td>_</td><td>means 'like that'</td></tr><tr><td>terali</td><td>_</td><td>means 'like that'</td></tr><tr><td>teralingge</td><td>_</td><td>means 'like that'</td></tr><tr><td>tere</td><td>REMT</td><td/></tr><tr><td>terebe</td><td>REMT</td><td/></tr><tr><td>terei</td><td>REMT</td><td/></tr><tr><td>tesu</td><td>REMT</td><td/></tr><tr><td>tuba</td><td>_</td><td>means 'there'</td></tr><tr><td>tuttu</td><td>_</td><td>means 'like that'</td></tr><tr><td>uba</td><td>_</td><td>means 'here'</td></tr><tr><td>ubaci</td><td>_</td><td>means 'here'</td></tr><tr><td>ubai</td><td>_</td><td>means 'here'</td></tr><tr><td>udu</td><td>_</td><td>means 'some'</td></tr><tr><td>uttu</td><td>_</td><td>means 'like this'</td></tr></table>", |
| "text": "Simplified mapping for the Xibe Universal Dependencies corpus (Pt. 1)" |
| }, |
| "TABREF18": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td colspan=\"4\">B Accuracy trends</td><td/><td/><td/></tr><tr><td>L</td><td>BME</td><td>GUClasp</td><td>TRM</td><td>TRM+ AUG</td><td>CHR-TRM</td><td>CHR-TRM +AUG</td></tr><tr><td>afb</td><td>94.77</td><td>90.26</td><td>95.24</td><td>95.24</td><td>95.84</td><td>95.84</td></tr><tr><td>amh</td><td>89.67</td><td>87.09</td><td>94.83</td><td>94.83</td><td>94.83</td><td>94.83</td></tr><tr><td>ara</td><td>99.87</td><td>98.34</td><td>99.93</td><td>99.93</td><td>99.93</td><td>99.93</td></tr><tr><td>arz</td><td>95.65</td><td>91.39</td><td>97.31</td><td>97.31</td><td>97.07</td><td>97.07</td></tr><tr><td>syc</td><td>10.71</td><td>7.14</td><td>10.71</td><td>17.85</td><td>14.28</td><td>14.28</td></tr><tr><td>ind</td><td>80.15</td><td>69.26</td><td>85.60</td><td>85.60</td><td>84.43</td><td>84.43</td></tr><tr><td>kod</td><td>100.00</td><td>90.90</td><td colspan=\"2\">100.00 100.00</td><td>90.90</td><td>100.00</td></tr><tr><td>ckt</td><td>50.00</td><td>50.00</td><td>50.00</td><td>50.00</td><td>50.00</td><td>50.00</td></tr><tr><td>itl</td><td>50.00</td><td>58.33</td><td>66.66</td><td>58.33</td><td>66.66</td><td>58.33</td></tr><tr><td>bra</td><td>68.75</td><td>65.62</td><td>50.00</td><td>71.87</td><td>65.62</td><td>56.25</td></tr><tr><td>bul</td><td>99.73</td><td>96.85</td><td colspan=\"2\">100.00 100.00</td><td>99.96</td><td>99.96</td></tr><tr><td>ces</td><td>99.49</td><td>97.74</td><td>99.50</td><td>99.50</td><td>99.52</td><td>99.52</td></tr><tr><td>mag</td><td>69.23</td><td>84.61</td><td>76.92</td><td>92.30</td><td>92.30</td><td>84.61</td></tr><tr><td>nld</td><td>97.29</td><td>96.38</td><td>97.85</td><td>97.85</td><td>97.80</td><td>97.80</td></tr><tr><td>pol</td><td>99.91</td><td>99.67</td><td>99.95</td><td>99.95</td><td>100.00</td><td>100.00</td></tr><tr><td>rus</td><td>99.81</td><td>98.88</td><td>99.44</td><td>99.44</td><td>99.44</td><td>99.44</td></tr><tr><td>ail</td><td>12.50</td><td>12.50</td><td>0</td><td>12.50</td><td>12.50</td><td>12.50</td></tr><tr><td>evn</td><td>73.52</td><td>74.50</td><td>78.43</td><td>76.47</td><td>72.54</td><td>77.45</td></tr><tr><td>krl</td><td>100.00</td><td>90.69</td><td>93.02</td><td>93.02</td><td>93.02</td><td>93.02</td></tr><tr><td>olo</td><td>99.80</td><td>98.05</td><td>99.92</td><td>99.92</td><td>99.76</td><td>99.76</td></tr><tr><td>vep</td><td>99.85</td><td>97.86</td><td>99.82</td><td>99.82</td><td>99.88</td><td>99.88</td></tr><tr><td>sjo</td><td>66.66</td><td>66.66</td><td>66.66</td><td>100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>tur</td><td>97.78</td><td>97.41</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr></table>", |
| "text": "Simplified mapping for the Xibe Universal Dependencies corpus (Pt. 2)" |
| }, |
| "TABREF19": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td>L</td><td>BME</td><td>GUClasp</td><td>TRM</td><td>TRM+ AUG</td><td>CHR-TRM</td><td>CHR-TRM +AUG</td></tr><tr><td>syc</td><td>65.21</td><td>6.52</td><td>84.78</td><td>82.60</td><td>86.95</td><td>80.43</td></tr><tr><td>bul</td><td>99.60</td><td>97.90</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>ces</td><td>100.00</td><td>98.40</td><td>99.90</td><td>99.90</td><td>100.00</td><td>100.00</td></tr><tr><td>ckb</td><td>97.91</td><td>95.83</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>deu</td><td>94.89</td><td>91.72</td><td>95.91</td><td>95.91</td><td>96.52</td><td>96.52</td></tr><tr><td colspan=\"2\">kmr 100.00</td><td>100.00</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>nld</td><td>89.17</td><td>78.58</td><td>94.58</td><td>94.58</td><td>96.00</td><td>96.00</td></tr><tr><td>pol</td><td>100.00</td><td>99.92</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>por</td><td>99.83</td><td>98.96</td><td>99.67</td><td>99.67</td><td>99.78</td><td>99.78</td></tr><tr><td>rus</td><td>97.04</td><td>92.06</td><td>96.58</td><td>96.58</td><td>96.67</td><td>96.67</td></tr><tr><td>spa</td><td>99.93</td><td>99.03</td><td>99.35</td><td>99.35</td><td>99.29</td><td>99.29</td></tr><tr><td>evn</td><td>12.76</td><td>7.09</td><td>17.73</td><td>18.43</td><td>14.89</td><td>19.14</td></tr><tr><td>krl</td><td>100.00</td><td>98.18</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>olo</td><td>99.69</td><td>97.83</td><td>99.38</td><td>99.38</td><td>99.69</td><td>99.69</td></tr><tr><td>vep</td><td>99.02</td><td>96.67</td><td>99.21</td><td>99.21</td><td>99.21</td><td>99.21</td></tr><tr><td>sjo</td><td>22.22</td><td>22.22</td><td>27.77</td><td>55.55</td><td>55.55</td><td>50.00</td></tr></table>", |
| "text": "Accuracy for \"Adjective\" on the test data." |
| }, |
| "TABREF20": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td>L</td><td>BME</td><td>GUClasp</td><td>TRM</td><td>TRM+ AUG</td><td>CHR-TRM</td><td>CHR-TRM +AUG</td></tr><tr><td>amh</td><td>98.67</td><td>95.78</td><td>99.75</td><td>99.75</td><td>99.87</td><td>99.87</td></tr><tr><td>itl</td><td>19.04</td><td>13.09</td><td>25.00</td><td>20.23</td><td>21.42</td><td>21.42</td></tr><tr><td>bul</td><td>100.00</td><td>98.57</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>ces</td><td>98.97</td><td>95.47</td><td colspan=\"2\">100.00 100.00</td><td>99.38</td><td>99.38</td></tr><tr><td>pol</td><td>99.22</td><td>99.22</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>rus</td><td>99.21</td><td>97.49</td><td>97.96</td><td>97.96</td><td>99.68</td><td>99.68</td></tr><tr><td>spa</td><td>98.74</td><td>98.23</td><td>99.24</td><td>99.24</td><td>100.00</td><td>100.00</td></tr><tr><td>evn</td><td>23.38</td><td>16.12</td><td>25.00</td><td>32.25</td><td>27.41</td><td>33.06</td></tr><tr><td>sah</td><td>100.00</td><td>100.00</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>tyv</td><td>100.00</td><td>100.00</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>sjo</td><td>54.54</td><td>9.09</td><td>54.54</td><td>54.54</td><td>72.72</td><td>45.45</td></tr></table>", |
| "text": "Accuracy for \"Participle\" on the test data." |
| }, |
| "TABREF21": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td>L</td><td>BME</td><td>GUClasp</td><td>TRM</td><td>TRM+ AUG</td><td>CHR-TRM</td><td>CHR-TRM +AUG</td></tr><tr><td>amh</td><td>88.61</td><td>79.67</td><td>95.12</td><td>95.12</td><td>97.56</td><td>97.56</td></tr><tr><td>heb</td><td>79.53</td><td>73.68</td><td>83.62</td><td>83.62</td><td>83.04</td><td>83.04</td></tr><tr><td colspan=\"2\">aym 100.00</td><td>100.00</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>itl</td><td>33.33</td><td>33.33</td><td>33.33</td><td>50.00</td><td>33.33</td><td>33.33</td></tr><tr><td>bul</td><td>98.85</td><td>98.00</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>kmr</td><td>99.37</td><td>100.00</td><td>98.74</td><td>98.74</td><td>100.00</td><td>100.00</td></tr><tr><td>pol</td><td>99.96</td><td>99.96</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>sjo</td><td>46.15</td><td>0</td><td>46.15</td><td>38.46</td><td>46.15</td><td>30.76</td></tr></table>", |
| "text": "Accuracy for \"Converb\" on the test data." |
| }, |
| "TABREF22": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td>L</td><td>BME</td><td>GUClasp</td><td>TRM</td><td>TRM+ AUG</td><td>CHR-TRM</td><td>CHR-TRM +AUG</td></tr><tr><td>afb</td><td>92.42</td><td>79.83</td><td>95.13</td><td>95.13</td><td>95.43</td><td>95.43</td></tr><tr><td colspan=\"2\">amh 98.36</td><td>91.56</td><td>99.72</td><td>99.72</td><td>99.72</td><td>99.72</td></tr><tr><td>ara</td><td>99.79</td><td>88.63</td><td>99.88</td><td>99.88</td><td>99.87</td><td>99.87</td></tr><tr><td>arz</td><td>93.31</td><td>78.08</td><td>95.16</td><td>95.16</td><td>94.98</td><td>94.98</td></tr><tr><td>heb</td><td>98.41</td><td>92.95</td><td>99.65</td><td>99.65</td><td>99.75</td><td>99.75</td></tr><tr><td>syc</td><td>11.02</td><td>4.41</td><td>25.73</td><td>21.32</td><td>27.94</td><td>25.00</td></tr><tr><td colspan=\"2\">ame 81.30</td><td>57.82</td><td>84.78</td><td>86.52</td><td>86.08</td><td>83.47</td></tr><tr><td>cni</td><td>98.73</td><td>79.51</td><td>99.72</td><td>99.72</td><td>99.63</td><td>99.63</td></tr><tr><td>ind</td><td>83.01</td><td>52.41</td><td>84.47</td><td>84.47</td><td>84.36</td><td>84.36</td></tr><tr><td>kod</td><td>93.65</td><td>92.06</td><td>100.00</td><td>98.41</td><td>100.00</td><td>100.00</td></tr><tr><td colspan=\"2\">aym 99.98</td><td>99.99</td><td colspan=\"2\">100.00 100.00</td><td>100.00</td><td>100.00</td></tr><tr><td>ckt</td><td>25.00</td><td>31.25</td><td>18.75</td><td>37.50</td><td>18.75</td><td>43.75</td></tr><tr><td>itl</td><td>14.96</td><td>12.92</td><td>20.40</td><td>25.17</td><td>21.08</td><td>23.12</td></tr><tr><td>gup</td><td>14.75</td><td>21.31</td><td>59.01</td><td>63.93</td><td>55.73</td><td>60.65</td></tr><tr><td>bra</td><td>31.30</td><td>29.56</td><td>24.34</td><td>27.82</td><td>27.82</td><td>30.43</td></tr><tr><td>bul</td><td>99.51</td><td>98.36</td><td>99.86</td><td>99.86</td><td>99.89</td><td>99.89</td></tr><tr><td>ces</td><td>98.88</td><td>94.97</td><td>99.54</td><td>99.54</td><td>99.40</td><td>99.40</td></tr><tr><td>ckb</td><td>99.72</td><td>96.44</td><td>99.96</td><td>99.96</td><td>99.96</td><td>99.96</td></tr><tr><td>deu</td><td>99.39</td><td>94.55</td><td>99.75</td><td>99.75</td><td>99.73</td><td>99.73</td></tr><tr><td colspan=\"2\">kmr 98.20</td><td>96.97</td><td colspan=\"2\">100.00 100.00</td><td>99.67</td><td>99.67</td></tr><tr><td colspan=\"2\">mag 36.36</td><td>38.63</td><td>42.04</td><td>42.04</td><td>44.31</td><td>39.77</td></tr><tr><td>nld</td><td>99.53</td><td>94.99</td><td>99.86</td><td>99.86</td><td>99.86</td><td>99.86</td></tr><tr><td>pol</td><td>99.57</td><td>98.22</td><td>99.76</td><td>99.76</td><td>99.74</td><td>99.74</td></tr><tr><td>por</td><td>99.84</td><td>99.12</td><td>99.91</td><td>99.91</td><td>99.86</td><td>99.86</td></tr><tr><td>rus</td><td>99.25</td><td>90.31</td><td>97.09</td><td>97.09</td><td>97.38</td><td>97.38</td></tr><tr><td>spa</td><td>99.82</td><td>97.55</td><td>99.90</td><td>99.90</td><td>99.92</td><td>99.92</td></tr><tr><td>see</td><td>78.27</td><td>40.97</td><td>90.65</td><td>89.64</td><td>90.00</td><td>88.63</td></tr><tr><td>ail</td><td>5.69</td><td>6.73</td><td>10.88</td><td>8.80</td><td>9.32</td><td>10.36</td></tr><tr><td>evn</td><td>34.70</td><td>32.03</td><td>44.90</td><td>44.66</td><td>44.90</td><td>46.35</td></tr><tr><td>sah</td><td>99.83</td><td>98.98</td><td>99.61</td><td>99.61</td><td>99.83</td><td>99.83</td></tr><tr><td>tyv</td><td>99.94</td><td>99.50</td><td>99.91</td><td>99.91</td><td>99.95</td><td>99.95</td></tr><tr><td>krl</td><td>99.94</td><td>98.82</td><td>99.94</td><td>99.94</td><td>99.94</td><td>99.94</td></tr><tr><td>lud</td><td>56.25</td><td>56.25</td><td>0</td><td>50.00</td><td>6.25</td><td>50.00</td></tr><tr><td>olo</td><td>99.84</td><td>99.14</td><td>99.71</td><td>99.71</td><td>99.70</td><td>99.70</td></tr><tr><td>vep</td><td>99.71</td><td>97.50</td><td>99.60</td><td>99.60</td><td>99.65</td><td>99.65</td></tr><tr><td>sjo</td><td>18.51</td><td>3.70</td><td>29.62</td><td>33.33</td><td>29.62</td><td>25.92</td></tr><tr><td>tur</td><td>99.96</td><td>99.98</td><td>99.17</td><td>99.17</td><td>99.17</td><td>99.17</td></tr></table>", |
| "text": "Accuracy for \"Masdar\" on the test data." |
| }, |
| "TABREF23": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td>L</td><td>BME</td><td>GUClasp</td><td>TRM</td><td>TRM+ AUG</td><td>CHR-TRM</td><td>CHR-TRM +AUG</td></tr></table>", |
| "text": "Accuracy for \"Verb\" in each language on the test data." |
| } |
| } |
| } |
| } |
