Search

Speech, speaker, and language systems have traditionally relied on carefully collected speech material for training acoustic models. There is an enormous amount of freely accessible audio content. A major challenge, however, is that such data is not professionally recorded, and therefore may contain a wide diversity of background noise, nonlinear distortions, or other unknown environmental or technology-based contamination or mismatch. There is a crucial need for automatic analysis to screen such unknown datasets before acoustic model development training, or to perform input audio purity screening prior to classification. In this study, we propose a waveform based clipping detection algorithm for naturalistic audio streams and examine the impact of clipping at different severities on speech quality measurements and automatic speaker recognition systems. We use the TIMIT and NIST SRE08 corpora as case studies. The results show, as expected, that clipping introduces a nonlinear distortion into clean speech data, which reduces speech quality and performance for speaker recognition. We also investigate what degree of clipping can be present to sustain effective speech system performance. The proposed detection system, which will be released, could contribute to massive new audio collections for speech and language technology development (e.g. Google Audioset ( Gemmeke et al., 2017 ), CRSS-UTDallas Apollo Fearless-Steps ( Yu et al., 2014 ) (19,000 h naturalistic audio from NASA Apollo missions)).

Yn dit rapport jouwe wy in oersjoch fan besteande taaltechnology (LT) foar it Frysk, mei it each op de taaldata, -middels en -tsjinsten dy’t yn de European Language Grid steane.2 Om’t LT sintraal stiet yn ús deistich libben, lykas yn staveringshifkers, sykmasines, oersetsoftware en firtuele assistintetechnology, is it essinsjeel dat net allinnich nasjonale of mearderheidstalen, mar ek regionale of minderheidstalen genietsje fan digitale lykweardigens en in heech nivo fan digitale fitaliteit – in kearnkomponint fan taalbehâld en -revitalisaasje. It Frysk hat it risiko om dy digitale taallykweardigens net te berikken. Mei it stribjen om de LT-situaasje foar it Frysk te ferbetterjen, presintearje wy in analyze fan de hjoeddeistige data en middels en identifisearje wy de hiaten en útdagings dy’t driuwend it omtinken freegje om derfoar te soargjen dat it Frysk net negatyf beynfloede wurdt troch digitale diglossia (d.w.s. de útsluting fan it Frysk yn digitale fermiddens te’n geunste fan it Nederlânsk). Ut ús analyze fan hjoeddeistige LT-middels foar it Frysk docht bliken dat der wichtige ûntjouwings west hawwe oangeande masine-oersetting, staveringshifkers, ientalige en multymodale korpora – en yn it bysûnder leksikale middels – en dat crowdsourcing-ynset foar inisjativen lykas Mozilla Common Voice relatyf suksesfol binne yn it sammeljen fan brûkbere data foar takomstige stimtechnologyprojekten foar it Frysk. Der is lykwols in tekoart oan, om in pear te neamen, twatalige en meartalige tekstkorpora (parallelle data), taalmodellen, kompûtasjonele grammatikamodellen, tsjinsten foar ynteraksje tusken minsk en kompjûter, sosjale mediadata, as ek transkribearre materialen dy’t brûkt wurde kinne foar de ûntwikkeling fan spraakwerkenners en oare middels. Nettsjinsteande dat de lêste jierren der LT-programma’s foar it Nederlânsk west hawwe, lykas it STEVIN-programma, hawwe de streek- of minderheidstalen yn Nederlân oant no ta gjin omtinken krigen yn dy projekten. Der is ek gjin tydskrift of evenemint dat spesifyk wijd is oan Fryske LT of keunstmjittige yntelliginsje, wylst dy der wol binne foar de Nederlânske taal. Boppedat identifisearje wy hiaten en útdagings dy’t de needsaak toane foar in ferheging fan finansiering en in ferbettering fan it oplieden en oanhâlden fan betûfte LT-ûntwikkelders yn de provinsje Fryslân. Oer it algemien is der, nettsjinsteande de ûntsachlike foarútgong dy’t oant no ta makke is, in dúdlik gebrek oan Fryske LT-middels yn ferliking mei it Nederlânsk en oare (offisjele EU-)talen. Om de digitale fitaliteit en LT-middels fan it Frysk te ferbetterjen riede wy ûnder oaren oan (a) dat in sintrum of netwurk opset wurde moat fan LT-kearnpartijen yn Fryslân om it Frysk yn it digitale tiidrek te beskermjen, (b) dat der no in ferskowing komme moat nei it sammeljen fan data foar it ûntwikkeljen fan en foar oare middels en tsjinsten lykas automatyske ûndertiteling en spraakwerkenningssoftware, lykas skermlêzers foar fisueel beheinde Frysktaligen, en (c) dat belanghawwers fan de Fryske LT inisjativen ûntwikkelje moatte om jongere generaasjes bewust te meitsjen fan de Fryske middels en om de mooglikheden te fergrutsjen om Fryske LT yn it deistich libben te brûken, mei it doel de ynfloed fan digitale diglossia te ferminderjen. Fierders stelle wy út dat ús oanbefellings yntegrearre wurde moatte yn in lanlike digitale taalstrategy op lange termyn, dy’t besiket de ûntwikkeling fan mear (meartalige) LT-middels te fasilitearjen, de kwaliteit fan de hjoeddeistige Fryske LT te ferbetterjen en in bettere gearwurking mooglik te meitsjen tusken de wichtichste LT-belanghawwers yn de provinsje Fryslân, dy’t kollektyf bydrage sil oan it realisearjen fan digitale taallykweardigens foar it Frysk.

We explore the concept of Casual Poetry Creators with the aim of making poetry writing fun and entertaining for the user. We present a simple co-creative interaction design pattern based on constructing poems line by line, suggesting the user a set of line candidates at each step. We also propose objective measures by which a Casual Poetry Creator can evaluate and choose which line candidates to show to the user and sketch out a plan to evaluate the measures and pattern with users.

This textbook approaches second language acquisition from the perspective of generative linguistics. Roumyana Slabakova reviews and discusses paradigms and findings from the last thirty years of research in the field, focussing in particular on how the second or additional language is represented in the mind and how it is used in communication. The adoption and analysis of a specific model of acquisition, the Bottleneck Hypothesis, provides a unifying perspective. The book assumes some non-technical knowledge of linguistics, but important concepts are clearly introduced and defined throughout, making it a valuable resource not only for undergraduate and graduate students of linguistics, but also for researchers in cognitive science and language teachers.

This study aims to identify the role of integrating technology in enhancing the use of language. The impact of utilizing the corpus-based approach (CA) on EFL learners’ academic writing is investigated; highlighting the effects of this integration on L2 learning and writing approaches. The competence development in EFL academic writing skills is asserted by tackling the contextual factors that influence corpus integration. This qualitative study is based on (10) case studies on non-native students in EFL academic writing courses. The corpus approach in this study was implemented during the fourth year of their English undergraduate program in the first semester of the academic year 2019/2020. Five different instruments were used to achieve the study purposes. The findings indicate the positive effect of incorporating corpus and its use in EFL students’ academic writing development and language awareness. The lexico-grammar integration has also been promoted significantly. The corpus-based approach enables participants to solve the problems they encountered while writing independently. Students also gain more confidence and become more autonomous and independent EFL writers. In this study attention is drawn to the participants’ language and learning background experience as factors that can shape the students’ progress and success in corpus implementation. The distinctive role of general corpora in association with the development of the academic writing skills is further highlighted. It is hoped that that the current study will reinforce the tendency of incorporating grammatical language aspects such as collocations and phrasal verbs in the writing pedagogy process.

Many electrical equipment malfunction text messages are collected during power system operation and maintenance procedures. These texts usually contain crucial information for maintenance and condition monitoring. Because these power system malfunction texts are characterized by multidomain vocabularies, complex-syntactic structures, and long sentences, it is challenging to for automated systems to capture their semantic meaning and essential information. To address this issue, we propose a hybrid natural language processing (hybrid-NLP) algorithm to extract entities that represent electrical equipment. This algorithm is composed of a dictionary-based method, a language technology platform (LTP) tool, and the bidirectional encoder representations from a transformers-conditional random field (BERT-CRF) model. Significantly, the softmax output layer of the bidirectional encoder representations from the transformers (BERT) model is replaced by the conditional random field (CRF) algorithm to strengthen the contextual relationships between words and thus solve the local optimization of the preferred word label. The effectiveness of the proposed hybrid-NLP method is verified on a realistic dataset. Moreover, a statistical analysis is conducted to provide a reference for the operation and maintenance of power systems.

Pandemic situations knocks at the door without information. In such situation healthcare is a major concern. Increased count of patients in any adverse condition leads to scarcity of healthcare services. In such situation telemedicine has played a very important role to overcome health problems of patient & to promote the remote healthcare monitoring of patients by physician in online mode. This proved to be very helpful for normal people but physically challenged could not take its advantage because of disabilities. We have implemented a system to reduce the communication barrier. System include many modules most importantly sign recognition & sign interpretation module in combination with ISL database of medical terms. For sign recognition we are using HMM model with Haar cascade classifier. Interpretation system fully uses the language technology including POS tagging, parsing etc. The overall system proved to improve the accuracy & efficiency of online consultation system for Deaf & Dumb people.

There is a contest between the view that the primary function of language is that of an instrument of communication and the view that it is that of a vehicle of thought. But such a dilemma is misconceived. The true opposition is between language as representation and language as activity. The representation picture of language is misleading in so far as it assumes that the representative power can be isolated from other features of language and that it consists in the speaker's being in the correct interior states. That would render understanding completely mysterious, and therefore, we should demand an outward manifestation of understanding expressed in linguistic exchange and other actions.

Službena dvojezičnost pretpostavlja svakodnevno stvaranje usporednih tekstova u dvojezičnim područjima. Slučaj je takav i u Istarskoj županiji, u kojoj se tekstovi obično sastavljaju na hrvatskome, a zatim se prevode na talijanski jezik. Zbog činjenice da je riječ o službenim tekstovima i zbog konteksta uporabe talijanskoga jezika vrlo je važno imati precizno i ujednačeno nazivlje te razvijene jezične tehnologije koje bi omogućile brže i kvalitetnije prevođenje u dvojezičnim institucijama u Istarskoj županiji. Ciljevi su ovoga rada: (1) prikazati ostvarivanje ravnopravne službene uporabe talijanskoga kao manjinskoga jezika u Istarskoj županiji (1a) analizom prevedenih sadržaja na službenim mrežnim stranicama službeno dvojezičnih gradova i općina i (1b) prikazom dosadašnje prakse prevođenja te (2) ukazati na važnost i nužnost razvoja jezičnih tehnologija (2a) prikazom trendova u razvoju jezičnih tehnologija u sličnim dvojezičnim i višejezičnim institucijama, (2b) prikazom pripreme usporednoga korpusa administrativnih tekstova Istarske županije i (2c) analizom postojećega nazivlja provedenom na priređenome korpusu. Rezultati analize dostupnosti dvojezičnih sadržaja na mrežnim stranicama te analize nazivlja provedene na usporednome korpusu pokazali su da je nužno razviti i upotrebljavati prevoditeljske alate i jezične izvore prilagođene talijanskomu kao manjinskomu jeziku kako bi se olakšalo i ubrzalo prevođenje, a time i omogućilo uspješnije ostvarivanje ravnopravne uporabe talijanskoga kao manjinskoga jezika., Official bilingualism assumes the daily creation of parallel texts in bilingual areas. This is also the case in Istria County, where texts are usually written in Croatian and then translated into Italian. Given the official nature of texts and the context of the Italian language usage, it is extremely important to have a precise and uniform terminology, as well as well-developed language technologies, which would allow a faster and more accurate translation processes in bilingual institutions of Croatia. The aims of this paper are (1) to investigate the equal use of both Croatian and Italian as a minority language in the Istrian County through (1a) an analysis of the translated content on the official websites of the officially bilingual cities and municipalities and through (1b) an overview of current translation practices, and (2) to highlight the importance and necessity of the development of language technology through (2a) the presentation of trends in the development of language technology in similar bilingual and multilingual institutions, (2b) the presentation of the parallel corpus compiled from administrative texts of the Istrian County and (2c) the analysis of existing terminology conducted on the compiled corpus. The analysis of the availability of bilingual content on websites and the terminology extracted from the compiled parallel corpus indicates the need to develop and use translation tools and linguistic resources tailored to the Italian language as a minority language, in order to facilitate and accelerate the translation activity, thus allowing a more efficient use of Italian as a minority language.

Human language technology encompasses a wide array of speech and text processing capabilities. The Defense Advanced Research Projects Agency’s pioneering research on automatic transcription, translation, and content analysis were major artificial intelligence success stories that changed science fiction into social fact. During a 40-year period, 10 seminal DARPA programs produced breakthrough capabilities that were further improved and widely deployed in popular consumer products, as well as in many commercial, industrial, and governmental applications. The Defense Advanced Research Projects Agency produced the core enabling technologies by setting crisp, aggressive, and quantitative technical objectives; by providing strong multiyear funding; and by using the Defense Advanced Research Projects Agency’s Common Task Method, which was powerful, efficient, and easy to administer. To achieve these breakthroughs, multidisciplinary academic and industrial research teams working in parallel took advantage of increasingly large and diverse sets of linguistic data and rapidly increasing computational power to develop and use increasingly sophisticated forms of machine learning. This article describes the progression of technical advances underlying key successes and the seminal programs that produced them.

Artikli eesmargiks on selgitada, millises ulatuses mojutab keeleandmetele kohalduv oiguslik režiim keelemudelite arendamist ja kasutamist. Autorid lahtuvad oma kasitluses protsessiskeemist, alustades algandmetest ning lopetades keeletehnoloogiat sisaldavate valmistoodetega (nt koneliidesega kulmik). Keeletehnoloogias kasutatavad algandmed sisaldavad tihti autorioiguslikult kaitstavaid teoseid, autorioigusega kaasnevate oiguste objekte (esitus, salvestus) ja isikuandmeid (isiku haal, isiku kohta kaiv muu info), mida sailitatakse annoteerimata ja annoteeritud andmekogudes. Keelandmete oiguslikke kusimusi on juba varem uuritud. Oiguslikult on labi uurimata aga keelemudelite oiguslikud aspektid. Autorid on seisukohal, et reeglina ei mojuta keelemudelite edasist oiguslikku staatust kasutatud algandmete oiguslik režiim, sest autorioigusega kaitstavad teosed mudelis pigem ei saili. Kull aga voib oiguslikke probleeme tekitada isiku haale kasutamine keelemudelis. Autorid analuusivad voimalikke lahendusvariante nende probleemide uletamiseks. Artiklis vaadeldakse ka uue autorioiguse direktiiviga kehtestatavat andmekaeve regulatsiooni ja selle rakendamist keelemudelite loomise kontekstis. *** Influence of legal regime of language data on language models This article aims to explain the extent to which the legal regime applicable to language data affects the development and use of language models. In their approach, the authors follow a process chart, starting from raw data to finished products containing language technology (eg a refrigerator with a speech interface). The raw data used in language technologies often include copyrighted works, objects of related rights (performances, sound recordings) and personal data (voice, other information about the person) stored in non-annotated and annotated databases. The legal issues of language data have already been studied. However, the legal aspects of language models have not been throughly explored. The authors are of the opinion that, as a rule, the legal status of the language models is not affect by the legal status of the used raw language data, since copyrighted works usually do not remain in the model. However, the use of a person’s voice in a language model can create legal problems. The authors analyze possible solutions to overcome these problems. The article also outlines the regulation of data mining introduced by the new copyright directive and its implementation in the context of development of language models.

Automatic sign language recognition (ASLR) is quite a complex task, not only for the difficulty of dealing with very dynamic video information, but also because almost every sign language (SL) can be considered as an under-resourced language when it comes to language technology. Spanish sign language (LSE) is one of those under-resourced languages. Developing technology for SSL implies a number of technical challenges that must be tackled down in a structured and sequential manner. In this paper, some problems of machine-learning- based ASLR are addressed. A review of publicly available datasets is given and a new one is presented. It is also discussed the current annotations methods and annotation programs. In our review of existing datasets, our main conclusion is that there is a need for more with high-quality data and annotations.

The Twitter discussion with the hashtag #jesuisCharlie was a large-scale social media event commenting on the tragic terrorist attack that took place in Paris in 2015. In this paper, we analyze French tweets compiled with language technology methods from a large dataset. Our qualitative approach determines what types of affectivity are expressed. According to our results, first, core emotions are shared, and they are based on the identification with the internet meme je suis Charlie (I am Charlie). In them, participants show their commitment to democratic values and freedom of speech, as well as grief. They build up a we-agency and togetherness between the networked participants. Second, participants disalign from those who do not share the same values or who are a threat to them. Here, the emotions range from irritation and doubt to anger and disgrace, manifesting awayness. They contain protest against how democratic values are violated.

This introduction presents an overview of the key concepts discussed in the subsequent chapters of this book. The book reflects on the present and future of Translation Studies following the incorporation of terms and perspectives from industrial technology discourses and practices into the discipline. It examines the current status of translation technologies training in translation and interpreting curricula in undergraduate degrees in Spain. The book provides a state-of-the-art review of translation competence, translation revision competence and post-editing competence models, analysing their similarities and differences in the light of recent research on revision and PE training. It proposes a pedagogical approach to future-proofing the translation profession in the 2020s.

Despite great progression within the field of Natural Language Inference, recent studies have revealed that the state-of-the-art neural models suffer from confounding shortcomings. These revelations cast doubt on whether these models possess as deep and robust language understanding as the hype surrounding them suggests. While the field has focused greatly on alleviating these shortcomings through adversarial training, less effort has gone into investigating other, more substantial ways of dealing with the linguistic phenomena that the neural models struggle with. This thesis will therefore propose and explore a novel hybrid approach to NLI, which seeks to combine the strengths of the state-of-the-art neural models with the preciseness of rule-based systems. The method of hybridizing is based on a principle of transitive reasoning, and it is designed to be independent of model choice in order to develop in tandem with an ever-progressing state-of-the-art. Our results indicate that this hybrid approach, dubbed T-NLI, may indeed improve results on datasets that are challenging for typical neural models. The ambition of this project is that these results will incite more optimism regarding further research of hybrid approaches to NLI.

It seems like a simple enough question: Is computer-assisted language learning effective for language learning? Indeed, researchers address this question in a variety of ways, but most language teachers and researchers would point out that the question is considerably more complex than it first appears. Keywords: CALL; computer mediated communication; language teaching; research methods in applied linguistics; evaluation; language learning technology

We apply the Natural Semantic Metalanguage (NSM) approach (Goddard & Wierzbicka 2014) to the lexical-semantic analysis of English evaluational adjectives and compare the results with the picture developed in the Appraisal Framework (Martin & White 2005). The analysis is corpus-assisted, with examples mainly drawn from film and book reviews, and supported by collocational and statistical information from WordBanks Online. We propose NSM explications for 15 evaluational adjectives, arguing that they fall into five groups, each of which corresponds to a distinct semantic template. The groups can be sketched as follows: “First-person thought-plus-affect”, e.g. wonderful; “Experiential”, e.g. entertaining; “Experiential with bodily reaction”, e.g. gripping; “Lasting impact”, e.g. memorable; “Cognitive evaluation”, e.g. complex, excellent. These groupings and semantic templates are compared with the classifications in the Appraisal Framework’s system of Appreciation. In addition, we are particularly interested in sentiment analysis, the automatic identification of evaluation and subjectivity in text. We discuss the relevance of the two frameworks for sentiment analysis and other language technology applications.

Part-of-speech (POS) tagging is an indispensable method of text processing. The main aim is to assign part-of-speech to words after considering their actual contextual syntactic-cum-semantic roles in a piece of text where they occur (Siemund & Claridge 1997). This is a useful strategy in language processing, language technology, machine learning, machine translation, and computational linguistics as it generates a kind of output that enables a system to work with natural language texts with greater accuracy and success. Part-of-speech tagging is also known as ‘grammatical annotation’ and ‘word category disambiguation’ in some area of linguistics where analysis of form and function of words are important avenues for better comprehension and application of texts. Since the primary task of POS tagging involves a process of assigning a tag to each word, manually or automatically, in a piece of natural language text, it has to pay adequate attention to the contexts where words are used. This is a tough challenge for a system as it normally fails to know how word carries specific linguistic information in a text and what kind of larger syntactic frames it requires for its operation. The present paper takes up this issue into consideration and tries to critically explore how some of the well-known POS tagging systems are capable of handling this kind of challenge and if these POS tagging systems are at all successful in assigning appropriate POS tags to words without accessing information from extratextual domains. The novelty of the paper lies in its attempt for looking into some of the POS tagging schemes proposed so far to see if the systems are actually successful in dealing with the complexities involved in tagging words in texts. It also checks if the performance of these systems is better than manual POS tagging and verifies if information and insights gathered from such enterprises are at all useful for enhancing our understanding about identity and function of words used in texts. All these are addressed in this paper with reference to some of the POS taggers available to us. Moreover, the paper tries to see how a POS tagged text is useful in various applications thereby creating a sense of awareness about multifunctionality of tagged texts among language users.

This LIBER workshop (18/11/21) aimed to: (1) tofamiliarise participants withthe work of the European Language Equality Project (ELE) and map out what digital language equality could look like in the library sector; and (2) collect audience input on how digital language equality may affect libraries, their users, and their collections. The session was chaired by Dr Maria Eskevich, Central Office Coordinator of CLARIN. She participates in the ELE project, and has a background in language and speech technologies, information retrieval and evaluation, and digital humanities. She was joined by two speakers: Dr German Rigau and Jean-François Nominé. German Rigau is member of the Ixa research group and deputy director of HiTZ, the Basque Center for Language Technology, based in San Sebastian, Basque Country. He co-leads an established research center that focuses on the development and advancement of language technology. In addition, he is a is core partners of the ELE project. Jean-François Nominé is a translator at the Institute for scientific and technological information (Inist). Inist is a research support unit of the French National Scientific Research Centre. In 2020, he contributed to a working group and report on translation and scholarly communication. This report partly foreshadowed the experimentation with translation services presented in the French Second National Plan for Open Science 2021-2024. This file contains the presentations from the workshop: Slides 1-18: Introduction by Maria Eskevich Slides 19-38: Presentation 1 by German Rigau Slides 39-48: Presentation 2 by Jean-François Nominé Slides 49-50 Summary by Maria Eskevich