Skaitmeninių technologijų keliami iššūkiai ir siūlomi būdai juos įveikti

BENDRINĖ KALBA 90 (2017) www.bendrinekalba.lt ISSN 2351-7204 RAMUNĖ VASKELAITĖ DIGITAL TECHNOLOGIES CHALLENGES AND PROPOSED WAYS TO OVERCOME THEM 24th Jon Jablonskis Conference The autumn series of conferences organized by the Lithuanian Language Institute and the Department of Lithuanian Language of Vilnius University since 1993 reveals that perhaps the most frequently chosen topics were rather specific problems of norming and usage or the relationship between norm and usage (e.g.: Language norms and usage, 1994; New language phenomena and their evaluation, 1998; Change of norms and usage, 1999; Equations of language norms and usage, 2000; Language usage and management, 2001). However, there was probably not a single conference that did 1 not raise more fundamental questions in one way or another – the concept of common language, its boundaries, purpose, relationship with society, change in changing circumstances, survival. And since the turn of the century, as the various circumstances affecting language have changed more and more fundamentally, such questions are purposefully raised as the main object of consideration (cf. : Common Language Limits, 2002; Common Language and Society, 2005; Common Language and Other Language Variants, 2006; Language Variability and Linguistic Attitudes, 2009; The Existence and Renewal of the Lithuanian Language in Modern Society, 2013). The evolution of conferences reveals another tendency: as the gaze turns to the fundamentals, and thus this approach deepens, it also encompasses an increasingly broader perspective. Already four conferences have been held as international, which means that they have discussed not only the state or concept of the Lithuanian 2 common language, but also the concepts of common language developed in other countries or, if such concepts are not applied, the directions of language planning, policy, management. This year’s 24th conference opened up a particularly broad perspective – a European one, covering the multilingual European Union and the problem of linguistic diversity. According to the focus area, the conference DIGITAL LANGUAGE RESOURCES, THEIR DEVELOPMENT DIRECTIONS AND USE OPPORTUNITIES (September 29, 2017), held at the Lithuanian Language Institute, could also be linked to an even wider dimension – the global one. And although the adjective globalus, -i is proposed to be replaced by Lithuanian words such as pasaulinis, -ė; universal, -ė; huge, -ė 1 The theses of some of J. Jablonski’s conference papers can be found on the website of the Lithuanian Language Institute at http://lki.lt/tarptautin%C4%97-jono-jablonskio-konferencija/#1501063755178-1d439107-84e9. Conference reports are published in Common Language (formerly Language Culture). 2 International conferences were organized in 2005, 2009, 2010 and 2016. Ramune Vaskelaite. Challenges posed by digital technologies and proposed ways to overcome them | 2 BENDRINĖ KALBA 90 (2017) www.bendrinekalba.lt ISSN 2351-7204 or huge, -a, this year’s discussions can most accurately be called international, encompassing 3 several semantic nuances. First, the digital space is not only global, but also universal, huge computers, but also smart watches, phones, voice-controlled vacuum cleaners, robots, artificial ar net beribė, juolab kad pranešėjų kaip skaitmeninių technologijų sritis minėti ne tik intelligence in general. Secondly, the word global, -i immediately shows the connections with globalization, globalization, and such obvious connections have not only the chosen areas of language use to be examined, but also the fundamental question raised at the conference. The question is summarized by Laimutis Telksnis, Ph.D. in mathematics: “How do we, a few million Lithuanians, act so that we do not disappear in the ocean of more than 7 billion writing and speaking people and the coming smart machines?” However, both the specific and objective answer given by the academician in the plenary session and the joint projects presented one after another by philologists, mathematicians, programmers and engineers in the other four sessions of the conference showed that the general strategic direction was well known. It is the introduction of language and its structure into the digital space and active care for its functioning there – both in written and oral form. The strategic direction was not discussed at the conference, but rather considered as the only possibility for the survival of a small country’s language in the digital age, the only possible choice. Algirdas Saudargas, MEP, compared it to the choice made by book carriers in a Cyrillic-writing environment to choose the Lithuanian script with Latin letters, and Dr. Audrius Valotka, project manager of the Faculty of Philology of Vilnius University, mentioned a sequential chain of choices: nations that chose to write something on clay or parchment survived, then those nations that printed texts with Gutenberg’s invented printing press survived on the path of history; now it's time for the inevitable gathering to move into the digital space. Difficulties arising from going in a clearly perceived direction were mentioned somewhat more often than the strategic direction, but they were not the basis of the conference’s deliberations. Many have identified as a problem the fact that the technology world is dominated by companies from the United States of America (hereinafter referred to as the USA), such as Microsoft, Amazon or Google, while the market of small countries, especially their language, is dominated by them. Therefore, it has been concluded several times that the use of the language in the digital space and its survival must be taken care of by oneself. However, many of those already involved have mentioned the lack of resources, including human resources. Understandably, for a small country to compete in the field of language technology with market sharks or huge so-called US private equity companies, there are not many opportunities, but from the European Union (hereinafter – the EU) that discussed the situation 3 For example, such substitutions are listed in the Consultation Bank provided by the State Lithuanian Language Commission: http://www.vlkk.lt/konsultacijos/5531-globalus. Ramune Vaskelaite. The presentation by the Spanish researcher Rafael Rivera showed that BENDRINĖ KALBA 90 (2017) www.bendrinekalba.lt ISSN 2351-7204 the dominance of these companies in the information technology world is also a problem at EU level. On the other hand, the EU itself is also moving towards the English language, which should be a particularly complicating factor for the implementation and existence of the Lithuanian language in the digital space. However, the conference opened up a completely different perspective. Algirdas Saudargas, who kalba dirbtiniame prote pradėjęs Europos Parlamente dirbantis Nepriklausomybės Akto signataras initiated the study Language Equality in the Digital Age in the European Parliament several years ago, was inspired by a study by META-NET, a network of excellence comprising 60 research centres in 34 countries, which showed that the Lithuanian language scores the lowest in terms of language technology status, i.e. it is considered “to have little or no technological support”, which means that the language is at risk of extinction. The study revealed that the prerequisite for true integration in the EU is not monolingualism, but rather linguistic diversity. It was later supplemented by the Mother Language project, which proposes cross-cutting strategic directions for achieving a technological breakthrough in the EU that would allow every citizen of the Union to access and communicate in the single digital space regardless of language. One of the researchers, Rafael Rivera, director of the consulting company iClaves, came to present the research and the study prepared on its basis. He pointed out the contradiction of linguistic diversity in the EU: on the one hand, it is considered one of the greatest assets of the Union with 24 official languages and more than 60 national and regional minority languages, as an expression of European cultural diversity, and on the other hand, the resulting language barriers are one of the main obstacles to the creation of a truly integrated Union. They particularly affect the less educated and the older people, as well as the speakers of smaller and regional minority languages. This creates a linguistic separation. The economic and social consequences of language barriers are considerable: the full potential of the single market, the free movement of goods and labour, the participation of citizens in political processes, and e-commerce are not exploited, and the citizens who are entitled to them are not able to take full advantage of them. And especially urgent this problem becomes in the digital age. New technologies, such as self-paying neural networks, are emerging to overcome language barriers, but are being hampered by the fragmentation of the European language technology market, uncoordinated research, insufficient funding and a lack of priority on the EU policy agenda. All these causes and problems explain Ramune Vaskelaite. Digital challenges and proposed solutions | 4 European mother tongue BENDRINĖ KALBA 90 (2017) www.bendrinekalba.lt ISSN 2351-7204 project envisioning the emergence of a comprehensive set of tools covering institutional, academic, industrial, market and public sector areas. 4 Professor of the Institute of Mathematics and Informatics of Vilnius University, pioneer of the development of the Internet in Lithuania, researcher who is famous not only in Lithuania for his inventions. In his presentation Written and spoken Lithuanian in conventional and electronic media, Laimutis Telksnys, answering the question raised by himself about the possibility of surviving in a 5 billion-billion ocean, formulated quite specific works as the necessary conditions for survival: that written and spoken language rules and technical and software equipment, transcribers, which popieriniuose dokumentuose ir elektroninėje terpėje išliktų taisyklinga lietuvių kalba, reikia present spoken language sounds in Lithuanian spoken language signs, suitable for automatic synthesis leistų nelietuvių rašytinės kalbos ženklus pavaizduoti lietuviškais rašytinės kalbos ženklais, o of spoken Lithuanian words sounds. This would ensure that both computers and phones, other talking smart objects and robots would communicate in the correct Lithuanian written and spoken language, and in Lithuanian electronic data repositories non-Lithuanian personal names would be presented in Lithuanian script character codes and original characters; their pronunciation by voice would be presented by audio recordings made by pronouncing them with phonemic sounds of the Lithuanian language and audio recordings in the original language. The professor himself and other specialists of the Institute of Mathematics and Informatics have already taken care of the correct use of spoken Lithuanian language in the digital space – this became clear from the presentations read at the first meeting Speech Recognition and Synthesis. In cooperation with philologists – the Institute of Lithuanian Language, the Faculty of Philology of Vilnius University, the Vilnius University of Education and the University of Šiauliai – the group of specialists of the Institute of Mathematics and Informatics of Vilnius University and the Faculty of Mathematics and Informatics, led by L. Telksnis, Ph.D., implemented the project Lithuanian Speaking Managed Services (LIEPA) in 2013–2015. The result is seven services available to all, provided in correct spoken Lithuanian: the Pažintuvas helping to explain the structure of the cell in schools, the Ieškatuvas introducing UNESCO heritage objects, the voice commands allowing to control the computer and search 4 Studija Language equality in the digital age. Towards a Human Language Project and its English summary are publicly available. They are available on the European Parliament's website (http://www.europarl.europa.eu/thinktank/en/document.html?reference=EPRS_STU(2017)598621), but reproduction or translation for non-commercial purposes requires the consent of the European Parliament. 5 Pagal pranešimą parengtą straipsnį žr. Bendrinės kalbos šio tomo „Apžvalgų“ skyriuje. Ramune Vaskelaite. Challenges posed by digital technologies and suggested ways to overcome them | 5 BENDRINĖ KALBA 90 (2017) www.bendrinekalba.lt ISSN 2351-7204 The Controller and the Browser, the Helper and the Tartvo who speak proper Lithuanian language and a computer synthesizer for the blind who reads text on the computer screen. Infrastructure services were created together: electronic text reader, web page opening program with voice, Lithuanian speech recognition engine, speaker, voice recording laboratory. It would seem that a lot has been done, 6 but L. Telksnys, reading together with the Projects Department of the same institute Padėtis. Perspektyvos, these works were named only as opening the door for the Lithuanian spoken language to vyresniuoju specialistu Gediminu Navicku parengtą pranešimą Lietuvių šneka valdomos paslaugos. enter the electronic space. As a result of the project, the speaker was inclined to mention not so much the created services, as the emphasized importance of the “fusion” of the knowledge of IT specialists and philologists, as well as the example significance – the possibilities of using Lithuanian-speaking managed services are shown to manufacturers and service providers. The scientist, who sees the global directions in the field of technology, considered the inclusion of robots in educational curricula, but before that – also the training of teachers, as a rather urgent task that is currently arising, which no longer falls to the creators and developers of language technologies. Project manager of Vilnius University, Dr. Audrius Valotka, also described the LIEPA project as a gateway. In his presentation Far from Gutenberg’s printing press: Lithuanian speech in the latest technologies (co-author G. Navickas) he stated that “currently Lithuanian speech in information technologies is in a similar situation as 30 years ago, when the issues of the use of Lithuanian diacritics in computing were being solved”. This condition encourages to take further work – project LIEPA 2. The result of this project, which has already begun, should be infrastructure solutions: a 1,000-hour Lithuanian speech speaker, a 3,000-word Lithuanian speech recognizer, a more advanced Lithuanian speech synthesizer. New services will also be created: a humanoid robot that develops children’s decision-making skills, a voice control of personal contacts on a mobile phone, a taxi caller, an internet news reader, an interlingual communicator covering Lithuanian and Chinese languages. However, the most important result of the new project is the ability to communicate with objects (mobile phones, tablets, smart watches, robots) using natural speech, giving commands in human voice and understanding their answers. Another representative of the Institute of Mathematics and Informatics – associate professor Pijus Kasparaitis in the presentation Present and prospects of Lithuanian computer voice synthesis (co-authors: assistant professor Gintaras Skersys, PhD, Faculty of Mathematics and Informatics, Vilnius University, and lecturer 6 They are described in more detail on the website of the Integrated Lithuanian Language and Writing Resources Raštija.lt: https://liepa.rastija.lt. RAMUNĖ VASKELAITĖ. Skaitmeninių technologijų keliami iššūkiai ir siūlomi būdai juos įveikti | 6 BENDRINĖ KALBA 90 (2017) www.bendrinekalba.lt ISSN 2351-7204 Margarita Beniušė), using video material and 7 websites, illustrated the project The possibilities of using the synthesizer created during LIEPA: the synthesizer was used during the 2014 theatre festival Sirenos in the performance Remote Vilnius (the audience, who also played the role of actors, walked through the streets of Vilnius listening to synthetic voices through headphones); articles of some online newspapers (lzinios.lt, ukininkopatarejas.lt, vilnius.lt, delfi.lt) can be not only read, but also listened to; Vilkaviškis bus station synthetic voice reads messages to passengers; the text-to-speech service RoboBraille allows you to automatically convert any text document into an audio file; NAO invited to visit the Lithuanian stand at the EU countries presentation event in Brussels, recited the Year of Kristijonas Donelaitis at the conference “Will robots speak Lithuanian?” held in the Seimas? Speaking about the future possibilities of using voice synthesis, which have already taken certain steps, P. Kasparaitis mentioned the voice-over of articles on Linux-based websites, virtual assistants, the mobile vending machine MIO mobile kiosk that can offer snacks or drinks in a park. The papers presented in the other three sessions, mainly related to the efforts to preserve written language, gave the impression that from the transfer of dictionaries, textbooks, databases, archives to the digital space, there is an increasing trend towards such digitization of language that would allow language resources to be used not only by language users and researchers, but also by language technology applications, artificial intelligence. Such a strategic direction, which is becoming more relevant with the increasing progress of artificial 8 intelligence, was emphasized both in articles in the press and at this conference by language strategist A. Saudargas. In the theses submitted to the conference it is stated as follows: “the development of artificial intelligence converges into a hybrid form, in which neural networks correspond to unconscious brain mechanisms, and conscious mind activity is modelled by traditional, so-called symbolic artificial intelligence. The same applies to language technologies, if we understand them as models of language mechanisms in the human brain (living language). Therefore, each language community must take care that language technologies corresponding to symbolic artificial intelligence (so-called “rule-based” programs) fully and accurately reflect the entire structure of the native language, and that a rich environment of the native language (and native culture) is created for the self-learning of neural networks.” 7 Pagal pranešimą parengtą straipsnį žr. Bendrinės kalbos šio tomo „Apžvalgų“ skyriuje. 8 For example, in the article Fulgura plango, published in the daily Bernardinai.lt kalba: ką Google nespręs, published in (http://www.bernardinai.lt/straipsnis/2012-12-05-algirdas-saudargas-fulgura-plango/91787), straipsnyje Gimtoji Lietuvos žinios (http://lzinios.lt/lzinios/Mokslas-ir-svietimas/gimtoji-kalbako-google-neisspres/246316). Ramune Vaskelaite. 7 In the second session of the conference, DIGITAL LITHUANIAN LANGUAGE BENDRINĖ KALBA 90 (2017) www.bendrinekalba.lt ISSN 2351-7204 RESOURCES, the presentation of a new project carried out by the Lithuanian Language Institute, it was also stated that language resources, as new needs, challenges and expectations arise, must change: new ideologies and concepts of digital language resources must be found, meaning corridors, conceptual engineering, variety of information search possibilities, etc. In the presentation “E-language – an innovation in the use of digital language resources” presented by Deimantė Budriūnaitė, Head of the General Affairs Infrastructure Department, the new project of the Institute, which she is implementing together with Director Jolanta Zabarskaitė and Netcode Project Manager Skirmantas Šermukšnis, is in fact a continuation of the previous language digitization project. At that time, 11 monolingual and bilingual dictionaries and 5 archives were digitized, thus creating a Lithuanian language resource information system, and the new project 9 aims not only to integrate new resources, but also to create technologies, solutions, tools for semantic search of resources, visualize its results, create services based on technology for users Search in the word network, E. marketing, E. concepts and E. advice. The project aims to adapt the resources for marketing purposes, providing public and private electronic services, as well as ensuring greater availability of word-making tools and language advice on the Internet, creating opportunities to link the Lithuanian word network with more than 200 other language word networks. The care for the use of Lithuanian language resources already installed on the Internet was also shown by two databases presented by representatives of the Lithuanian Language Institute, one of which is integrated into the Lithuanian Language Resources Information System, and the other is being prepared to be integrated. The initiator and creator of the database of Lithuanian new words, Dr. Rita Miliūnaitė, senior researcher at the General Language Research Center of the Lithuanian Language Institute, described the database as an online reference for recording and lexicographically 10 describing Lithuanian new words from the beginning of the 21st century, but her presentation revealed that at the current stage of its creation, the most important thing is the usefulness of this database for users. In order to achieve this, the search is adapted by title word, origin of the new word, original form, spelling variants, scope of use, some more special features, by the name of the most famous person who used the new word or its author, and having in mind individual target groups of users, an advanced search system is created. Interest in both the database and new words is high, so the only problem mentioned so far is that the data sent by users cannot be processed. On the other hand, the abundance of parcels denies two visuomenėje gajus mitus – kad lietuvių kalba miršta ir kad žodžius kuria kalbininkai. Vis dėlto 9 The Lithuanian Language Resource Information System is available at http://lkiis.lki.lt/home. 10 The Lithuanian language new words database is accessible at http://naujazodziai.lki.lt. Ramune Vaskelaite. Challenges posed by digital technologies and suggested ways to overcome them | 8 BENDRINĖ KALBA 90 (2017) www.bendrinekalba.lt ISSN 2351-7204 Dr. Daiva Murmulaitytė, also working at the General Language Research Centre, in her presentation Perspectives of Research on the Formation of New Words (The Case of the Lithuanian Language New Words Database) suggested that this database should also be made more useful for linguistic researchers. It should be adapted accordingly. It was proposed to mark new words with indexes – to record the language part of the derived base word, changes in the base of the verb, analogies, the role of translation in making a new word, atypical expressions of the verb, etc. According to the speaker, the attempt to do this revealed such phenomena of word formation that are little or not studied at all. Dr. Laimutis Bilkis, senior researcher at the Baltic Languages and Names Research Centre of the 11 Lithuanian Language Institute, also mentioned both practical objectives, i.e. teaching correct usage, and language research (including language preservation and dissemination), when listing the objectives of another database – the Lithuanian geo-informatic database of place names. The fact that this database is oriented to both language users and researchers is shown by the possibilities of searching and presenting information: information about place names can be searched by name, gender, number, pronunciation, conjugation, origin according to the belonging of the base word to the language and origin according to the belonging of the base word to the lexical group, and in the area of place name analysis the pronunciation of all syllables of place names, origin and conjugation interpretations are provided. About 25,000 names have been uploaded to the database. Some interfaces have already been taken care of: by integrating the database into the Lithuanian Language Resources Information System, links have been created to another placename from which the searcher originated, to the name in the Lithuanian Names Database from which the placename originated, to the historical records of that placename in the Historical Placenames Database. There are also envisaged links with general or dialectal lexicon, especially semantic, functional and local, but several problems are mentioned – not only the abundance of place names, but also the fragmentation of funding. The work of textbook linguists also showed an obvious shift from the collection of language data to the search for their use. In the third session of the conference, the representatives of Vytautas Magnus University, which mainly works in Lithuania in this field, presented two group projects, Kompiuterinės lingvistikos centro mokslininkai kartu su Baltijos pažangių technologijų instituto which revealed the desire to annotate texts and thus use them for language research and other language resources. The first group of textbook linguists – Erika Rimkutė, Agnė Bielinskienė, Loïc Boizou and Andrius Utkas – presented the morphologically annotated textbook MATAS and the syntactically annotated textbook ALKSNIS. Textbook MATAS prepared on the basis of 1 million 12 words textbook, This database is also available at http://lkiis.lki.lt/lietuvos-vietovardziu-geoinformacine-duomenu-baze. 11 12 Pagal pranešimą parengtą straipsnį žr. Bendrinės kalbos šio tomo „Kalbos technologijų“ skyriuje. Ramune Vaskelaite. Digital technology challenges and proposed ways to overcome them | 9 The BENDRINĖ KALBA 90 (2017) www.bendrinekalba.lt ISSN 2351-7204 morphological annotator prepared by the Computer Linguistics Center was used to annotate the text after applying statistical models. ALKSNIO annotation is based on the principles of automatic morphological and syntactic annotation, syntactic dependencies model was applied. The second group of researchers – Erika Rimkutė, Agnė Bielinskienė, Loïc Boizou, Ieva Bumbulienė, Jolanta Kovalevskaitė, Tomas Krilavičius, Justina Mandravickaitė and Laura Vilkaitė – revealed that, after annoting the Delfi.lt texts, they are moving to the development of a methodology for the study of the current written Lithuanian language’s constant word combinations and the preparation of a Lithuanian language collocations dictionary. The database of constant word combinations based on the textbook is intended to provide grammatical, lexical information, frequency of use, text rubric, concordance examples, etc. Currently, the database contains about 2,700 lemas and more than 35,000 functional forms. The practical possibilities of using textbooks were also revealed in the presentation of Gintarė Judžentytė and Dr. Vilma Zubaitienė, lecturers at the Department of Lithuanian Language of Vilnius University, entitled Textbook-based research of academic phrases: formal structure and semantics. After reviewing the recent attempts to investigate academic language on the basis of textbooks, the authors presented a project, which, based on the textbook of student papers created for this purpose at Vilnius University, aims to investigate the structure and semantics of academic language phrases. Using the textbook material and the results of the research, a summary of academic language phrases will be compiled, useful for students of various degrees and specialties. The fourth session of the conference AUTOMATIZED LANGUAGE CONTRACT AND TEXTS ANALYSIS revealed even more complex possibilities of using new technologies for language analysis. Virginijus Dadurkevičius, engineer at the Institute of Applied Sciences of Vilnius University, in his presentation “Grammar of the Lithuanian language in the digital open source world” reviewed the 13 work of Vytautas Zinkevičius, programmer at the Institute of Lithuanian Language, on analizės srities darbus, beveik du dešimtmečius buvusius nepamainoma ir vienintele modernaus morphological research of the Lithuanian language and computer applications. He presented and kompiuterinės morfologijos srityje, kurią leido pasiekti Hunspell platformos panaudojimas. described a certain shift in the morphology of the language in the platform. The morphology of the language is described in two text files, the basis of both of which is the Grammar of the Contemporary Lithuanian Language and textbooks of Vilnius University, Vytautas the Great University and the Seimas of the Republic of Lithuania. According to the rapporteur, in total, about 5 000 groups of addressable rules were recorded, more than 170 000 lems were selected. The selection of lemas was limited to the correct modern Lithuanian language. According to the speaker, this morphological analyzer correctly 13 Pagal pranešimą parengtą straipsnį žr. Bendrinės kalbos šio tomo „Kalbos technologijų“ skyriuje.