{ "paper_id": "2021", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T14:41:37.625167Z" }, "title": "Developing Flashcards for Learning Icelandic", "authors": [ { "first": "Xindan", "middle": [], "last": "Xu", "suffix": "", "affiliation": { "laboratory": "", "institution": "University of Iceland Reykjav\u00edk", "location": { "country": "Iceland" } }, "email": "xindanxu@hi.is" }, { "first": "Anton", "middle": [ "Karl" ], "last": "Ingason", "suffix": "", "affiliation": { "laboratory": "", "institution": "University of Iceland", "location": { "settlement": "Reykjav\u00edk", "country": "Iceland" } }, "email": "antoni@hi.is" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "This paper describes the process of devel oping flashcards for the most frequently used words in Icelandic. The process in volves utilising currently available open source online databases, the Tagged Ice landic Corpus, M\u00cdM, and the Database of Modern Icelandic Inflection, B\u00cdN, to ex tract a list of the most frequently used words, their partofspeech tags, and in flectional forms. This was combined with newly developed language technol ogy tools for Icelandic to generate phonetic and audio transcriptions of the words. The final product is a combination of printable flashcards and digital flashcards which are easily accessible through smart devices.", "pdf_parse": { "paper_id": "2021", "_pdf_hash": "", "abstract": [ { "text": "This paper describes the process of devel oping flashcards for the most frequently used words in Icelandic. The process in volves utilising currently available open source online databases, the Tagged Ice landic Corpus, M\u00cdM, and the Database of Modern Icelandic Inflection, B\u00cdN, to ex tract a list of the most frequently used words, their partofspeech tags, and in flectional forms. This was combined with newly developed language technol ogy tools for Icelandic to generate phonetic and audio transcriptions of the words. The final product is a combination of printable flashcards and digital flashcards which are easily accessible through smart devices.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "Flashcards are a useful tool for learning. They are frequently used for memorising new words when learning a new language. When combined with spaced repetition, they can produce longterm knowledge retention.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "In this project, we created a deck of flashcards that consists of the 4,000 most frequently used words in Icelandic. On the front side of each flash card, a word is shown along with a sample sen tence. On the back of each flashcard, more de tailed information about the word is shown, includ ing the following: its English translation, essential morphosyntactic information (e.g. word class and gender, if applicable), the phonetic transcription, dialectal variation (if applicable), and selected in flectional forms.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "The production of this flashcard dataset was made possible due to the recent developments in language technologies for Icelandic. Twenty years ago, this project would have to be carried out manually because Icelandic language technology resources were almost nonexistent (R\u00f6gnvalds son et al., 2009) . Since 2000, a lot of effort and financial support have been put into develop ing language technologies for Icelandic. This in cluded building online corpora of texts and sound files, e.g. the Tagged Icelandic Corpus M\u00cdM (Helgad\u00f3ttir et al., 2012) , online dictionaries, e.g. The Database of Modern Icelandic Inflection B\u00cdN (Bjarnad\u00f3ttir, 2012) , and basic tools for natural language processing, e.g. IceTagger (Loftsson, 2008) and Lemmald (Ingason et al., 2008) .", "cite_spans": [ { "start": 271, "end": 299, "text": "(R\u00f6gnvalds son et al., 2009)", "ref_id": null }, { "start": 522, "end": 548, "text": "(Helgad\u00f3ttir et al., 2012)", "ref_id": "BIBREF7" }, { "start": 625, "end": 645, "text": "(Bjarnad\u00f3ttir, 2012)", "ref_id": "BIBREF2" }, { "start": 712, "end": 728, "text": "(Loftsson, 2008)", "ref_id": "BIBREF11" }, { "start": 733, "end": 763, "text": "Lemmald (Ingason et al., 2008)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "By utilising these resources, we have compiled a novel dataset that contains a rich variety of in formation for selected words. This information was incorporated into flashcards to create a more detailed and effective learning material. We de veloped two versions of the flashcards: a print able pdfversion and a digital Ankiversion that supports media files and is available on multiple platforms. Both versions of the flashcards will be accessible to the public without charge, and the dataset will be published under an opensource li cense (CC BY 4.0).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "Vocabulary learning is a fundamental aspect of second language acquisition and lasts throughout the learning process. Vocabulary learning involves two scopes: vocabulary size and depth of vocab ulary knowledge (Schmitt, 2008) . Without suf ficient vocabulary size, understanding input and producing satisfactory output in a second language can be frustrating for learners. Furthermore, a lex ical item is learned not only by making a form meaning connection, but also by understanding how it is used in context (Schmitt, 2008) .", "cite_spans": [ { "start": 210, "end": 225, "text": "(Schmitt, 2008)", "ref_id": "BIBREF18" }, { "start": 511, "end": 526, "text": "(Schmitt, 2008)", "ref_id": "BIBREF18" } ], "ref_spans": [], "eq_spans": [], "section": "Flashcards for vocabulary learning", "sec_num": "2" }, { "text": "Flashcards are a learning tool that facilitates the acquisition of vocabulary. Through the use of high frequency words of a second language, flashcards can help acquire sufficient vocabulary size more effectively. Flashcards can also provide lexical items with context, as well as additional informa tion that aids the depth of vocabulary knowledge, for example, word class, pronunciation and inflec tional forms. Furthermore, flashcards can incor porate spaced repetition learning that can produce longterm knowledge retention of the vocabulary. Studies have shown that spaced repetition is one of the most effective learning techniques (Dunlosky et al., 2013\u037e Kang, 2016 . This is a learning tech nique that allows initial study and subsequent re views to be spaced out over time, and that new and more difficult material is reviewed more often than wellknown and easy material.", "cite_spans": [ { "start": 638, "end": 672, "text": "(Dunlosky et al., 2013\u037e Kang, 2016", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Flashcards for vocabulary learning", "sec_num": "2" }, { "text": "Vocabulary and associated morphological infor mation was extracted from two main sources: the Tagged Icelandic Corpus, M\u00cdM (Helgad\u00f3ttir et al., 2012) , and the Database of Mordern Icelandic In flection, B\u00cdN (Bjarnad\u00f3ttir, 2012) .", "cite_spans": [ { "start": 123, "end": 149, "text": "(Helgad\u00f3ttir et al., 2012)", "ref_id": "BIBREF7" }, { "start": 207, "end": 227, "text": "(Bjarnad\u00f3ttir, 2012)", "ref_id": "BIBREF2" } ], "ref_spans": [], "eq_spans": [], "section": "Source of material", "sec_num": "3" }, { "text": "The Tagged Icelandic Corpus (hereafter referred to as M\u00cdM) contains approximately 25 million to kens collected from contemporary Icelandic texts during the period 2006-2010. The texts are se lected from a variety of sources, including pub lished books, newspapers, Icelandic parliament speeches, legal texts, and student essays. These texts are considered to be representative of the Ice landic society's language usage. The texts are mor phosyntactically tagged, lemmatized, and format ted into XMLdocuments defined by TEI (Text En coding Initiative). This makes it possible to ex tract a variety of useful information from the cor pus. In this study, we extracted the frequency of headwords and their partofspeech tags, as well as sample sentences for the selected headwords.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "M\u00cdM corpus", "sec_num": "3.1" }, { "text": "The corpus was tagged and lemmatized auto matically using software IceNLP (Loftsson, 2019) . The accuracy of morphosyntactic tagging was esti mated to be 88.1%-95.1% depending on text type (Loftsson et al., 2010) . The accuracy of lemma tization was estimated to be approximately 90%.", "cite_spans": [ { "start": 74, "end": 90, "text": "(Loftsson, 2019)", "ref_id": "BIBREF12" }, { "start": 189, "end": 212, "text": "(Loftsson et al., 2010)", "ref_id": "BIBREF13" } ], "ref_spans": [], "eq_spans": [], "section": "M\u00cdM corpus", "sec_num": "3.1" }, { "text": "The corpus is available through a special user li cense. 1 An example of entries for the headword \u00e1r (e. year) in the M\u00cdM corpus is shown in Listing 1. The inflectional form of the headword is shown between and : \u00e1rum and \u00e1ra. Type shows the POStag used for the inflectional form, i.e. \"nhf\u00fe\" for \u00e1rum and \"nhfe\" for \u00e1ra. 2 \u00e1rum \u00e1ra Listing 1: Example from the M\u00cdM Corpus The first character in the tag always shows the word class, e.g. \"n\" for \"nafnor\u00f0\" (e. noun), \"s\" for \"sagnor\u00f0\" (e. verb). The number of characters used in the tag depends on the word class. In this case, \"\u00e1rum\" in the first entry was tagged: noun, neutral, plural and dative, whilst \"\u00e1ra\" in the sec ond entry was tagged: noun, neutral, plural and genitive.", "cite_spans": [ { "start": 57, "end": 58, "text": "1", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "M\u00cdM corpus", "sec_num": "3.1" }, { "text": "The Database of Modern Icelandic Inflection (hereafter referred to as B\u00cdN) consists of more than 270,000 headwords with approximately 5.8 mil lion inflectional forms. Language technology data from the database are distributed under a CC BY SA 4.0 license and are available at https://bin. arnastofnun.is/DMII/. The basic version of the database, Sigr\u00fan's format, was used in the de velopment of the flashcards. The data consists of 6 fields: lemma, id, word class, semantic fields, in flectional form, and grammatical tag (see example of \u00e1r in Figure 1 ).", "cite_spans": [], "ref_spans": [ { "start": 544, "end": 552, "text": "Figure 1", "ref_id": "FIGREF1" } ], "eq_spans": [], "section": "B\u00cdN corpus", "sec_num": "3.2" }, { "text": "A Python script was used to parse XML documents and count the frequency of occurrence for each pair of lemma and the first two charac ters of the tag in the M\u00cdM corpus. The resulting dataset was cleaned and expanded upon by com parison with the B\u00cdN corpus.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Data processing", "sec_num": "4" }, { "text": "Unnecessary tokens in the resulting dataset (e.g. symbols and roman numbers) were filtered out by comparing all the entries with the headword entries in the B\u00cdN corpus. Subsequently, since the M\u00cdM corpus was tagged and lemmatized automatically, it was necessary to doublecheck the extracted tags and make corrections where necessary. For exam ple, prepositions and adverbs share the same tag (\"a\") in the M\u00cdM corpus, whilst they have sepa rate tags in the B\u00cdN corpus (\"fs\" for prepositions and \"ao\" for adverbs). Furthermore, a lemma can be two or more separate words from different word classes. For example, lemma sig can be both a neu tral noun meaning \"subsidence\", and a reflexive pronoun referring to oneself. To make sure these instances are tagged correctly, all the headwords and tags extracted from the M\u00cdM corpus were com pared against tags in the B\u00cdN corpus. If the tags did not match, the tags from the B\u00cdN corpus were used. Finally, words were ranked by their frequency of occurrence and the top 4,000 were chosen for the project.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Data processing", "sec_num": "4" }, { "text": "As it would not be beneficial to show all the in flectional forms of a headword at once, selected in flectional forms were chosen based on word class. Selected inflectional forms of the chosen words were retrieved from the B\u00cdN corpus. An example entry for the noun \u00e1r is shown in the Table 1. In this case, the frequency of occurrence of lemma \"\u00e1r\" of class \"hk\" in the M\u00cdM corpus was 96,849 times. This was ranked 29th amongst all the headwords in the M\u00cdM corpus. Its genitive singular form (EFET) is \u00e1rs and nominative plural form (NFFT) is \u00e1r. ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Data processing", "sec_num": "4" }, { "text": "Phonetic transcriptions of the words were gener ated using LSTM encoderdecoder sequenceto sequence models developed by Grammatek ehf. (2021). These models transcribe grapheme to phoneme (g2p) in four pronunciation variants of Icelandic: the standard pronunciation of modern Icelandic, the northern variant (postaspiration), the southern variant (hvpronunciation), and the northeast variant (postaspiration + voiced pronun ciation). 3 The R package ipa (Hayes and Alexan der, 2020) was used to convert the XSAMPA pho netic transcription resulting from the g2p models to ipa transcription. In Icelandic, the pronunciation of a lemma is the same in different word classes. For example, the lemma tala can be used as a feminine noun mean ing \"number, speech\", or as a verb meaning \"talk, speak\". In both instances, pronunciation of the lemma is the same: [t\u02b0a\u02d0la] . Accounting for these duplicates, a total of 3,933 unique lemmas (out of 4,000 in total) was used for phonetic transcription.", "cite_spans": [ { "start": 432, "end": 433, "text": "3", "ref_id": null }, { "start": 851, "end": 859, "text": "[t\u02b0a\u02d0la]", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Phonetic and audio transcription", "sec_num": "4.1" }, { "text": "Audio transcriptions were generated using the Icelandic D\u00f3ra voice included in the Amazon Polly texttospeech service (Amazon Web Ser vices, 2021).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Phonetic and audio transcription", "sec_num": "4.1" }, { "text": "Translation of the Icelandic words was carried out semiautomatically. A list of words was translated automatically using the Google Translate web ser vice. However, the translation accuracy turned out to be poor in some cases. Poor translation accuracy mainly occurs when there is minimal difference in written form between two different words. For ex ample, lemma h\u00e1r can be a noun meaning \"hair\" and an adjective meaning \"high\". In such cases, Google Translate failed to differentiate the word class and their meanings. Furthermore, Google Translate did not recognise the acute accent in some cases. For example, d\u00fdr (e. animal (no.) and expensive (adj.)) and dyr (e. door) are only dis tinguished by the acute accent, but they were both translated into \"animals\" using Google Translate. According to a recent study (Aiken, 2019), Ice landic was among the lowest scoring languages in terms of translation accuracy using Google trans late. Therefore, translations were reviewed man ually using the Concise IcelandicEnglish Dictio nary (H\u00f3lmarsson et al., 2006 ) as a reference.", "cite_spans": [ { "start": 1036, "end": 1060, "text": "(H\u00f3lmarsson et al., 2006", "ref_id": "BIBREF8" } ], "ref_spans": [], "eq_spans": [], "section": "Translation and sample sentence", "sec_num": "4.2" }, { "text": "The process of selecting sample sentences was also carried out semiautomatically. A python script was used to parse the XMLfiles from the M\u00cdM corpus and 10 sentences were selected for each headword. Subsequently, sentences were ar ranged based on their complexity, i.e., length of the sentence and whether there are any uncommon words in the sentence. Finally, the most easily un derstandable sentence was selected manually for each headword to be shown on the flashcards.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Translation and sample sentence", "sec_num": "4.2" }, { "text": "After this step, the data was ready to be used in the production of the flashcards. Table 2 shows a demonstration dataframe with all information ex cluding the sample sentences and selected inflec tional forms.", "cite_spans": [], "ref_spans": [ { "start": 84, "end": 91, "text": "Table 2", "ref_id": null } ], "eq_spans": [], "section": "Translation and sample sentence", "sec_num": "4.2" }, { "text": "Both a printable pdf version and a digital version of the flashcards were made in the project. The pdf version of the flashcards was generated using the R package Knitr (Xie, 2021) and the L A T E Xpackage Flacards (Stuhrmann, 2005) . The main difference between the two versions is that the digital version contains audio files of the selected words so that users can listen to their pronunciation\u037e while the physical flashcards contain the phonetic transcrip tions in regional variants of Icelandic (if applica ble).", "cite_spans": [ { "start": 169, "end": 180, "text": "(Xie, 2021)", "ref_id": "BIBREF20" }, { "start": 215, "end": 232, "text": "(Stuhrmann, 2005)", "ref_id": "BIBREF19" } ], "ref_spans": [], "eq_spans": [], "section": "Printable and digital flashcards", "sec_num": "4.3" }, { "text": "Digital flashcards were made using the Python li brary Genanki (Staley, 2021). The script produces an Ankideck package which can be imported into the Ankiapp. Anki is available on multiple plat forms and supports different media types in the cards. Another advantage of Anki is the inclu sion of spaced repetition, which is considered to be one of the most effective learning techniques (Dunlosky et al., 2013\u037e Kang, 2016 .", "cite_spans": [ { "start": 387, "end": 421, "text": "(Dunlosky et al., 2013\u037e Kang, 2016", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Digital flashcards", "sec_num": null }, { "text": "Basic components of an Anki deck are notes. Each note contains a front (question) and a back (answer) side with information to memorise. The notes in the Genanki library are defined by two components:", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Digital flashcards", "sec_num": null }, { "text": "1. models, which indicate the information to be shown on the card by defining the fields and how the card should look like by defining the templates.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Digital flashcards", "sec_num": null }, { "text": "2. fields, which are the actual information to be shown on the card and should correspond to the fields defined by the model.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Digital flashcards", "sec_num": null }, { "text": "The difference between the fields in the model and the fields in the note is that the fields in the model act like a placeholder for the fields of infor mation to be shown, while the fields in the notes are the actual information. Figure 2 shows an example of the front and back of the Anki flashcard for \u00e1r. The triangle button which is located next to the phonetic transcription is used to replay the audio of the word. At the bot tom of the user interface, the user can choose the interval between repeated viewings. A short in terval should be chosen for flashcards that are dif ficult to memorise so that they are repeated more frequently, whilst a long interval should be cho sen for flashcards that are easy to memorise. This process is done to prioritise the flashcards that are harder to learn and thus to improve the overall ef ficiency of learning. For example, the card would be reviewed immediately by clicking the \"again\" button, after 1 day by clicking the \"Good\" button, and after 4 days by clicking \"Easy\" button. Differ ent interval settings can be selected by the user on their Anki app. ", "cite_spans": [], "ref_spans": [ { "start": 231, "end": 239, "text": "Figure 2", "ref_id": "FIGREF2" } ], "eq_spans": [], "section": "Digital flashcards", "sec_num": null }, { "text": "Despite all the advantages that Anki offers, some studies also showed that physical flashcards may produce learning outcomes similar to those for dig ital flashcards (Sage et al., 2020\u037e Nikoopour and Kazemi, 2014) . Furthermore, studies have shown that digital flashcards on mobile devices have led to distractions (Sage et al., 2020) and low en joyment (Hanson and Brown, 2019) Table 2 : A demonstration dataframe for flashcard production.", "cite_spans": [ { "start": 166, "end": 199, "text": "(Sage et al., 2020\u037e Nikoopour and", "ref_id": null }, { "start": 200, "end": 213, "text": "Kazemi, 2014)", "ref_id": "BIBREF14" }, { "start": 315, "end": 334, "text": "(Sage et al., 2020)", "ref_id": "BIBREF17" }, { "start": 354, "end": 378, "text": "(Hanson and Brown, 2019)", "ref_id": "BIBREF5" } ], "ref_spans": [ { "start": 379, "end": 386, "text": "Table 2", "ref_id": null } ], "eq_spans": [], "section": "Printable flashcards", "sec_num": null }, { "text": "The pdfversion of the flashcards is generated by a mother RNW document and eight child RNW documents. The mother RNW document defines the document class flacards, reads in the dataset (similar to the one shown in Table 2 ), and loops through each row to create the respective flash card. The child RNW documents define differ ent presentations of the cards for different word classes. For example, three inflectional forms were chosen for the word classes noun (lemma, genitive singular and nominative plural), personal pronoun (lemma, genitive singular and nominative plural), and verb (3rd person singular in present tense and past tense, and past participle in neuter singular nominative case). Four child RNW doc uments were created to accommodate different word classes. Subsequently, four corresponding child RNW documents were created to accommo date the regional pronunciation variants. For each row in the dataset, the mother RNW document se lected the child RNW document required to pro duce the flashcard. For example, the child RNW document for the word class noun without pronun ciation variant would be selected for the noun \u00e1r, whilst the child RNW document for adjective with pronunciation variant would be selected for the ad jective mikill (Figure 3) . 4 The front side of the pdfversion (Figure 3) is the same as the Anki version (Figure 2 ). On the back side of the pdfversion, regional variants of pronunciation are shown (Figure 3) as opposed to the audio version of the word in the Ankiversion (Figure 2 ). The noun \u00e1r has the same pronuncia tion across all regions of Iceland. The adjective mikill has regional pronunciation variants in the north and northeast regions of Iceland (Figure 3) .", "cite_spans": [ { "start": 1271, "end": 1272, "text": "4", "ref_id": null } ], "ref_spans": [ { "start": 213, "end": 220, "text": "Table 2", "ref_id": null }, { "start": 1258, "end": 1268, "text": "(Figure 3)", "ref_id": "FIGREF3" }, { "start": 1306, "end": 1316, "text": "(Figure 3)", "ref_id": "FIGREF3" }, { "start": 1349, "end": 1358, "text": "(Figure 2", "ref_id": "FIGREF2" }, { "start": 1443, "end": 1453, "text": "(Figure 3)", "ref_id": "FIGREF3" }, { "start": 1517, "end": 1526, "text": "(Figure 2", "ref_id": "FIGREF2" }, { "start": 1704, "end": 1714, "text": "(Figure 3)", "ref_id": "FIGREF3" } ], "eq_spans": [], "section": "Printable flashcards", "sec_num": null }, { "text": "In this paper, we have described the process of the production of printable and digital flashcards for the most frequently used words in Icelandic (based on the M\u00cdM corpus). The flashcards dataset will be published under an opensource license which means that it will be freely accessible to the pub lic for use and as a template for further flashcard production.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Summary and future implementations", "sec_num": "5" }, { "text": "The flashcards will be useful for anyone who is interested in learning Icelandic, especially at the beginning stage where large quantities of vocab ulary need to be acquired. By learning the high frequency words in the language, learners can un derstand a high percentage of words in common texts such as newspapers and books.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Summary and future implementations", "sec_num": "5" }, { "text": "During the production of the flashcards, all steps were carried out automatically except for transla tion and selecting sample sentences which were both semiautomatic ( Table 3) . The most time con suming parts are, as expected, the manual steps: Table 3 : Summary steps for the production of flashcards in the project.", "cite_spans": [], "ref_spans": [ { "start": 169, "end": 177, "text": "Table 3)", "ref_id": null }, { "start": 247, "end": 254, "text": "Table 3", "ref_id": null } ], "eq_spans": [], "section": "Summary and future implementations", "sec_num": "5" }, { "text": "A complete list of resources used for the devel opment of the flashcards and their respective li censes are shown in Table 4 . In conclusion, we have described the develop ment of a flashcard dataset for leaning Icelandic. The work will serve as a useful template for fur ther development of flashcards as a learning mate rial for Icelandic. For example, a variety of prac tice decks of the Ankiversion can be made so that users can test their learning progress. In Anki, a clozedeletion field or typein text field can be im plemented into the front of a card. The user's an swer will be reviewed automatically and shown in the back (answer) side of the flashcard. This could easily be incorporated into the flashcards so that users can type in the Icelandic words according to the English translation or the phonetic transcrip tion of words with audio display.", "cite_spans": [], "ref_spans": [ { "start": 117, "end": 124, "text": "Table 4", "ref_id": "TABREF5" } ], "eq_spans": [], "section": "Summary and future implementations", "sec_num": "5" }, { "text": "Furthermore, the two flashcard decks will serve as a useful resource for the evaluation of flashcards as a learning material, and to ascertain the rela tive benefits of digital versus physical flashcards for second language learners. We leave that for fu ture work.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Summary and future implementations", "sec_num": "5" }, { "text": "See http://www.malfong.is/files/ userlicense_mim_download_en.pdf.2 See the full list of tagsets used in M\u00cdM corpus: http:// www.malfong.is/files/mim_tagset_files_en.pdf.Proceedings of the 10th Workshop on Natural Language Processing for Computer Assisted Language Learning (NLP4CALL 2021)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "For more information about the regional pronunciation variants of Icelandic, see R\u00f6gnvaldsson (2020).Proceedings of the 10th Workshop on Natural Language Processing for Computer Assisted Language Learning (NLP4CALL 2021)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Proceedings of the 10th Workshop on Natural Language Processing for Computer Assisted Language Learning (NLP4CALL 2021)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "The abbreviations fst, mst and est inFigure 3refer to positive degree, comparative degree and superlative degree respectively.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [ { "text": "We would like to thank Anna Bj\u00f6rk Nikul\u00e1sd\u00f3t tir for the phonetic transcription models, and Atli Jasonarson for writing a script that submitted our word list to Amazon Polly to generate the sound files.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Acknowledgements", "sec_num": null } ], "bib_entries": { "BIBREF0": { "ref_id": "b0", "title": "An updated evaluation of google translate accuracy", "authors": [ { "first": "", "middle": [], "last": "Milam Aiken", "suffix": "" } ], "year": 2019, "venue": "Studies in Linguistics and Liter ature", "volume": "3", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Milam Aiken. 2019. An updated evaluation of google translate accuracy. Studies in Linguistics and Liter ature, 3:p253.", "links": null }, "BIBREF2": { "ref_id": "b2", "title": "The Database of Modern Icelandic Inflection (Beygingarl\u00fdsing \u00edslensks n\u00fat\u00ed mam\u00e1ls)", "authors": [ { "first": "Krist\u00edn", "middle": [], "last": "Bjarnad\u00f3ttir", "suffix": "" } ], "year": 2012, "venue": "LREC 2012 Proceedings: Proceedings of \"Language Technology for Normalisation of Less Resourced Languages", "volume": "", "issue": "", "pages": "13--18", "other_ids": {}, "num": null, "urls": [], "raw_text": "Krist\u00edn Bjarnad\u00f3ttir. 2012. The Database of Modern Icelandic Inflection (Beygingarl\u00fdsing \u00edslensks n\u00fat\u00ed mam\u00e1ls). In LREC 2012 Proceedings: Proceedings of \"Language Technology for Normalisation of Less Resourced Languages\", pages 13-18.", "links": null }, "BIBREF3": { "ref_id": "b3", "title": "Improving students' learning with ef fective learning techniques: Promising directions from cognitive and educational psychology", "authors": [ { "first": "John", "middle": [], "last": "Dunlosky", "suffix": "" }, { "first": "Katherine", "middle": [ "A" ], "last": "Rawson", "suffix": "" }, { "first": "Elizabeth", "middle": [ "J" ], "last": "Marsh", "suffix": "" }, { "first": "Mitchell", "middle": [ "J" ], "last": "Nathan", "suffix": "" }, { "first": "Daniel", "middle": [ "T" ], "last": "Willing", "suffix": "" }, { "first": "", "middle": [], "last": "Ham", "suffix": "" } ], "year": 2013, "venue": "Psy chological Science in the Public Interest", "volume": "14", "issue": "1", "pages": "4--58", "other_ids": { "PMID": [ "26173288" ] }, "num": null, "urls": [], "raw_text": "John Dunlosky, Katherine A. Rawson, Elizabeth J. Marsh, Mitchell J. Nathan, and Daniel T. Willing ham. 2013. Improving students' learning with ef fective learning techniques: Promising directions from cognitive and educational psychology. Psy chological Science in the Public Interest, 14(1):4- 58. PMID: 26173288.", "links": null }, "BIBREF4": { "ref_id": "b4", "title": "Phonetic transcrip tion tool", "authors": [], "year": null, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Grammatek ehf. 2021. g2plstm. Phonetic transcrip tion tool. https://github.com/grammatek/ g2p-lstm.git.", "links": null }, "BIBREF5": { "ref_id": "b5", "title": "Enhanc ing l2 learning through a mobile assisted spaced repetition tool: an effective but bitter pill?", "authors": [ { "first": "Aroline", "middle": [], "last": "Hanson", "suffix": "" }, { "first": "Christina", "middle": [], "last": "Brown", "suffix": "" } ], "year": 2019, "venue": "Com puter Assisted Language Learning", "volume": "33", "issue": "", "pages": "1--23", "other_ids": {}, "num": null, "urls": [], "raw_text": "Aroline Hanson and Christina Brown. 2019. Enhanc ing l2 learning through a mobile assisted spaced repetition tool: an effective but bitter pill? Com puter Assisted Language Learning, 33:1-23.", "links": null }, "BIBREF7": { "ref_id": "b7", "title": "Eir\u00edkur R\u00f6gn valdsson, Krist\u00edn Bjarnad\u00f3ttir, and Hrafn Loftsson", "authors": [ { "first": "Sigr\u00fan", "middle": [], "last": "Helgad\u00f3ttir", "suffix": "" }, { "first": "\u00c1sta", "middle": [], "last": "Svavarsd\u00f3ttir", "suffix": "" } ], "year": 2012, "venue": "Proceedings of the Workshop on Language Tech nology for Normalisation of LessResourced Lan guages", "volume": "", "issue": "", "pages": "67--72", "other_ids": {}, "num": null, "urls": [], "raw_text": "Sigr\u00fan Helgad\u00f3ttir, \u00c1sta Svavarsd\u00f3ttir, Eir\u00edkur R\u00f6gn valdsson, Krist\u00edn Bjarnad\u00f3ttir, and Hrafn Loftsson. 2012. The Tagged Icelandic Corpus (M\u00cdM). In Proceedings of the Workshop on Language Tech nology for Normalisation of LessResourced Lan guages, pages 67-72.", "links": null }, "BIBREF8": { "ref_id": "b8", "title": "\u00cdslenskensk or\u00f0ab\u00f3k / Concise IcelandicEnglish Dictionary", "authors": [ { "first": "Sverrir", "middle": [], "last": "H\u00f3lmarsson", "suffix": "" }, { "first": "Christopher", "middle": [], "last": "Sanders", "suffix": "" }, { "first": "John", "middle": [], "last": "Tucker", "suffix": "" } ], "year": 2006, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Sverrir H\u00f3lmarsson, Christopher Sanders, and John Tucker. 2006. \u00cdslenskensk or\u00f0ab\u00f3k / Concise IcelandicEnglish Dictionary.", "links": null }, "BIBREF9": { "ref_id": "b9", "title": "Hrafn Lofts son, and Eir\u00edkur R\u00f6gnvaldsson", "authors": [ { "first": "Anton", "middle": [], "last": "Karl Ingason", "suffix": "" }, { "first": "Sigr\u00fan", "middle": [], "last": "Helgad\u00f3ttir", "suffix": "" } ], "year": 2008, "venue": "A mixed Proceedings of the 10th Workshop on Natural Language Processing for Computer Assisted Language Learning (NLP4CALL 2021) Processing", "volume": "", "issue": "", "pages": "205--216", "other_ids": {}, "num": null, "urls": [], "raw_text": "Anton Karl Ingason, Sigr\u00fan Helgad\u00f3ttir, Hrafn Lofts son, and Eir\u00edkur R\u00f6gnvaldsson. 2008. A mixed Proceedings of the 10th Workshop on Natural Language Processing for Computer Assisted Language Learning (NLP4CALL 2021) Processing, pages 205-216, Berlin, Heidelberg. Springer Berlin Heidelberg.", "links": null }, "BIBREF10": { "ref_id": "b10", "title": "Spaced repetition promotes efficient and effective learning: Policy implications for instruction", "authors": [ { "first": "H", "middle": [ "K" ], "last": "Sean", "suffix": "" }, { "first": "", "middle": [], "last": "Kang", "suffix": "" } ], "year": 2016, "venue": "Policy Insights from the Behavioral and Brain Sciences", "volume": "3", "issue": "1", "pages": "12--19", "other_ids": {}, "num": null, "urls": [], "raw_text": "Sean H. K. Kang. 2016. Spaced repetition promotes efficient and effective learning: Policy implications for instruction. Policy Insights from the Behavioral and Brain Sciences, 3(1):12-19.", "links": null }, "BIBREF11": { "ref_id": "b11", "title": "Tagging icelandic text: A lin guistic rulebased approach", "authors": [ { "first": "Hrafn", "middle": [], "last": "Loftsson", "suffix": "" } ], "year": 2008, "venue": "Nordic Journal of Lin guistics", "volume": "31", "issue": "1", "pages": "47--72", "other_ids": {}, "num": null, "urls": [], "raw_text": "Hrafn Loftsson. 2008. Tagging icelandic text: A lin guistic rulebased approach. Nordic Journal of Lin guistics, 31(1):47-72.", "links": null }, "BIBREF12": { "ref_id": "b12", "title": "IceNLP natural language pro cessing toolkit. CLARINIS, Stofnun \u00c1rna Magn\u00fas sonar", "authors": [ { "first": "", "middle": [], "last": "Hrafn Loftsson", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Hrafn Loftsson. 2019. IceNLP natural language pro cessing toolkit. CLARINIS, Stofnun \u00c1rna Magn\u00fas sonar.", "links": null }, "BIBREF13": { "ref_id": "b13", "title": "Developing a postagged corpus using existing tools", "authors": [ { "first": "Hrafn", "middle": [], "last": "Loftsson", "suffix": "" }, { "first": "H", "middle": [], "last": "J\u00f6kull", "suffix": "" }, { "first": "", "middle": [], "last": "Yngvason", "suffix": "" } ], "year": 2010, "venue": "LREC 2010 Proceedings: Proceedings of the Workshop on \"Creation and use of basic lexical resources for less resourced languages", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Hrafn Loftsson, J\u00f6kull H. Yngvason, Sigr\u00fan Helgad\u00f3t tir, and Eir\u00edkur R\u00f6gnvaldsson. 2010. Developing a postagged corpus using existing tools. In LREC 2010 Proceedings: Proceedings of the Workshop on \"Creation and use of basic lexical resources for less resourced languages\".", "links": null }, "BIBREF14": { "ref_id": "b14", "title": "Vo cabulary learning through digitized & nondigitized flashcards delivery", "authors": [ { "first": "Jahanbakhsh", "middle": [], "last": "Nikoopour", "suffix": "" }, { "first": "Azin", "middle": [], "last": "Kazemi", "suffix": "" } ], "year": 2014, "venue": "Proceedings of the International Conference on Current Trends in ELT", "volume": "98", "issue": "", "pages": "1366--1373", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jahanbakhsh Nikoopour and Azin Kazemi. 2014. Vo cabulary learning through digitized & nondigitized flashcards delivery. Procedia Social and Behav ioral Sciences, 98:1366-1373. Proceedings of the International Conference on Current Trends in ELT.", "links": null }, "BIBREF15": { "ref_id": "b15", "title": "A Short Overview of the Icelandic Sound System Pronunciation Variants and Phonetic Transcription", "authors": [], "year": null, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Eir\u00edkur R\u00f6gnvaldsson. 2020. A Short Overview of the Icelandic Sound System Pronunciation Variants and Phonetic Transcription.", "links": null }, "BIBREF16": { "ref_id": "b16", "title": "Anna Bj\u00f6rk Nikul\u00e1s d\u00f3ttir, Matthew Whelpton, and Anton Karl Ingason", "authors": [ { "first": "Hrafn", "middle": [], "last": "Eir\u00edkur R\u00f6gnvaldsson", "suffix": "" }, { "first": "Krist\u00edn", "middle": [], "last": "Loftsson", "suffix": "" }, { "first": "Sigr\u00fan", "middle": [], "last": "Bjar Nad\u00f3ttir", "suffix": "" }, { "first": "", "middle": [], "last": "Helgad\u00f3ttir", "suffix": "" } ], "year": 2009, "venue": "Proceedings of the NODALIDA 2009 Workshop Nordic Perspectives on the CLARIN Infrastructure of Language Resources", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Eir\u00edkur R\u00f6gnvaldsson, Hrafn Loftsson, Krist\u00edn Bjar nad\u00f3ttir, Sigr\u00fan Helgad\u00f3ttir, Anna Bj\u00f6rk Nikul\u00e1s d\u00f3ttir, Matthew Whelpton, and Anton Karl Ingason. 2009. Icelandic language resources and technol ogy: status and prospects. In Proceedings of the NODALIDA 2009 Workshop Nordic Perspectives on the CLARIN Infrastructure of Language Resources. Odense, Denmark.", "links": null }, "BIBREF17": { "ref_id": "b17", "title": "Flip it or click it: Equiv alent learning of vocabulary from paper, laptop, and smartphone flashcards", "authors": [ { "first": "Kara", "middle": [], "last": "Sage", "suffix": "" }, { "first": "Michael", "middle": [], "last": "Piazzini", "suffix": "" }, { "first": "John", "middle": [ "Charles" ], "last": "Downey", "suffix": "" }, { "first": "Sydney", "middle": [], "last": "Ewing", "suffix": "" } ], "year": 2020, "venue": "Journal of Educational Technology Systems", "volume": "49", "issue": "2", "pages": "145--169", "other_ids": {}, "num": null, "urls": [], "raw_text": "Kara Sage, Michael Piazzini, IV John Charles Downey, and Sydney Ewing. 2020. Flip it or click it: Equiv alent learning of vocabulary from paper, laptop, and smartphone flashcards. Journal of Educational Technology Systems, 49(2):145-169.", "links": null }, "BIBREF18": { "ref_id": "b18", "title": "Review article: Instructed sec ond language vocabulary learning", "authors": [ { "first": "Norbert", "middle": [], "last": "Schmitt", "suffix": "" } ], "year": 2008, "venue": "Language Teach ing Research", "volume": "12", "issue": "3", "pages": "329--363", "other_ids": {}, "num": null, "urls": [], "raw_text": "Norbert Schmitt. 2008. Review article: Instructed sec ond language vocabulary learning. Language Teach ing Research, 12(3):329-363.", "links": null }, "BIBREF19": { "ref_id": "b19", "title": "The flacards class. CTAN package version 0.1.1b", "authors": [ { "first": "Norbert", "middle": [], "last": "Stuhrmann", "suffix": "" } ], "year": 2005, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Norbert Stuhrmann. 2005. The flacards class. CTAN package version 0.1.1b.", "links": null }, "BIBREF20": { "ref_id": "b20", "title": "Knitr: A GeneralPurpose Package for Dynamic Report Generation in R. R package ver sion 1.31", "authors": [ { "first": "Yihui", "middle": [], "last": "Xie", "suffix": "" } ], "year": 2021, "venue": "Proceedings of the 10th Workshop on Natural Language Processing for Computer Assisted Language Learning", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Yihui Xie. 2021. Knitr: A GeneralPurpose Package for Dynamic Report Generation in R. R package ver sion 1.31. Proceedings of the 10th Workshop on Natural Language Processing for Computer Assisted Language Learning (NLP4CALL 2021)", "links": null } }, "ref_entries": { "FIGREF0": { "type_str": "figure", "num": null, "uris": null, "text": "Xu and Anton Karl Ingason 2021. Developing flashcards for learning Icelandic. Proceedings of the 10th Workshop on Natural Language Processing for Computer Assisted Language Learning (NLP4CALL 2021). Link\u00f6ping Electronic Conference Proceedings 177: 55-61." }, "FIGREF1": { "type_str": "figure", "num": null, "uris": null, "text": "Example of the entry for \u00e1r in the B\u00cdN corpus." }, "FIGREF2": { "type_str": "figure", "num": null, "uris": null, "text": "Example of the front and back of the flashcard for \u00e1r in Anki." }, "FIGREF3": { "type_str": "figure", "num": null, "uris": null, "text": "Example of the front and back side of the pdf flashcard for \u00e1r and mikill." }, "TABREF1": { "num": null, "text": "", "html": null, "content": "", "type_str": "table" }, "TABREF3": { "num": null, "text": "Proceedings of the 10th Workshop on Natural Language Processing for Computer Assisted Language Learning(NLP4CALL 2021)doublechecking the translation accuracy and se lecting sample sentences.", "html": null, "content": "
StepsEfficiency
1 Extract word lists and frequency from M\u00cdMAutomatic
2 Filter out undesirable entries by com paring against lemmas in B\u00cdN corpusAutomatic
3 Extract selected inflectional forms from B\u00cdNAutomatic
4 Phonetic transcriptionAutomatic
5 Audio transcriptionAutomatic
6 TranslationSemiautomatic
7 Sample sentencesSemiautomatic
8 Generate printable flashcardsAutomatic
9 Generate AnkiflashcardsAutomatic
", "type_str": "table" }, "TABREF5": { "num": null, "text": "List of resources used and information about their licences.", "html": null, "content": "", "type_str": "table" } } } }