Hasil untuk "Ural-Altaic languages"

Jialu Li, Hao Tan, Mohit Bansal

In Vision-and-Language Navigation (VLN), an agent needs to navigate through the environment based on nat-ural language instructions. Due to limited available data for agent training and finite diversity in navigation environments, it is challenging for the agent to generalize to new, unseen environments. To address this problem, we propose Envedit, a data augmentation method that cre-ates new environments by editing existing environments, which are used to train a more generalizable agent. Our augmented environments can differ from the seen environ-ments in three diverse aspects: style, object appearance, and object classes. Training on these edit-augmented environments prevents the agent from overfitting to existing en-vironments and helps generalize better to new, unseen en-vironments. Empirically, on both the Room-to-Room and the multi-lingual Room-Across-Room datasets, we show that our proposed Envedit method gets significant im-provements in all metrics on both pre-trained and non-pre-trained VLN agents, and achieves the new state-of-the-art on the test leaderboard. We further ensemble the VLN agents augmented on different edited environments and show that these edit methods are complementary.11Code and data are available at https://github.com/jialuli-luka/EnvEdit.

Adele Vaks, Virve-Anneli Vihman, Jan de Jong

This paper investigates the relationship between language exposure and Estonian competence of Estonian-Norwegian bilinguals aged 5–7. The sample includes 29 children acquiring Estonian and Norwegian (16 successive and 8 simultaneous bilinguals acquiring Estonian as a heritage language in Norway, 5 simultaneous bilinguals acquiring Norwegian as a heritage language in Estonia). Their language environments are described using data from the Q-BEx questionnaire, and scores from a Sentence Repetition Task and a Cross-Linguistic Lexical Task are used to describe Estonian ­ competence. A correlation analysis shows significant correlations between exposure quality and amount of current exposure. Conditional inferencing trees and random ­ forest ­ models identify the pro­ portion of current Estonian exposure as the most important predictor of morpho­ syntactic competence. For lexical competence, both proportion and diversity of Estonian exposure play an important role. Kokkuvõte. Adele Vaks, Virve-Anneli Vihman, Jan de Jong: Keelesisend ja -oskus eesti-norra kakskeelsete laste seas. Artiklis kirjeldame 5‒7-aastaste eesti-norra kakskeelsete laste keelesisendit ning keelesisendi suhet eesti keele oskusega. Valimis on 29 last: 16 suktsessiivselt kakskeelset ja kaheksa simultaanselt kakskeelset Norras, ning viis simultaanselt kakskeelset last ­ Eestis. Keelekeskkonna kirjeldamiseks kogusime andmeid laste keelekogemuse kohta Q-BEx ankeediga. Eesti keele oskuse kirjeldamiseks kasutame lausete järelekordmistesti ning sõnavaratesti skoore. Korrelatsioonianalüüsist selgub, et keelesisendi mitmekesisus korreleerub oluliselt ja positiivselt viimase aasta eestikeelse sisendi osakaaluga. Keelesisendi ja -oskuse suhte analüüsimiseks kasutasime tingimuslikke otsustuspuid ning juhumetsa mudeleid. Mudelite põhjal mõjutab laste morfosüntaksi oskust kõige enam käesoleva aasta eestikeelse sisendi osakaal. Sõnavarateadmiste jaoks olid olulised nii eestikeelse sisendi osakaal kui ka sisendi mitmekesisus.

Zhunheng Wang, Xiaoyi Liu, Mengting Hu et al.

The demand for understanding and expressing emotions in the field of natural language processing is growing rapidly. Knowledge graphs, as an important form of knowledge representation, have been widely utilized in various emotion-related tasks. However, existing knowledge graphs mainly focus on the representation and reasoning of general factual knowledge, while there are still significant deficiencies in the understanding and reasoning of emotional knowledge. In this work, we construct a comprehensive and accurate emotional commonsense knowledge graph, ECoK. We integrate cutting-edge theories from multiple disciplines such as psychology, cognitive science, and linguistics, and combine techniques such as large language models and nat-ural language processing. By mining a large amount of text, dialogue, and sentiment analysis data, we construct rich emotional knowledge and establish the knowledge generation model COMET-ECoK. Experimental results show that ECoK contains high-quality emotional reasoning triples, and the performance of our knowledge generation model surpasses GPT-4-Turbo, which can help downstream tasks better understand and reason about emotions. Our data and code is available from https://github.com/ZornWang/ECoK .

Colin Conwell, Emalie McMahon, Kasper Vinken et al.

Recent progress in multimodal AI and “language-aligned” visual representation learning has re-ignited debates about the role of language in shaping the human visual system. In particular, the emergent ability of “language-aligned” vision models (e.g. CLIP) – and even pure language models (e.g. BERT) – to predict image-evoked brain activity has led some to suggest that human visual cortex itself may be “language-aligned” in comparable ways. But what would we make of this claim if the same procedures worked in the modeling of visual activity in a species that has no language? Here, we deploy controlled comparisons of pure-vision, pure-language, and multimodal vision-language models in prediction of human (N=4) and rhesus macaque (N=6, 5:IT, 1:V1) ventral visual activity evoked in response to the same set of 1000 captioned nat-ural images (the “NSD1000”). Preliminary results reveal markedly similar patterns in aggregate model predictivity of early and late ventral visual cortex across both species. Together, these results suggest that language predictivity of the human visual system is not necessarily due to “language-alignment” per se , but rather to the statistical structure of the visual world as reflected in language.

Shahnoza Kaxxarova

Maqolada tasavvuf va unga aloqador tushunchalarning Islom bilan chambarchas bog‘liq ekaniga shubha yo‘q. Tasavvuf falsafasini, ayniqsa, fano va baqo masalasini jiddiy o‘rganish, uning Sharq falsafasi, badiiy adabiyotidagi talqinlari mohiyatini to‘la ochib berish va dunyo ilm-faniga taqdim eta olish juda dolzarbdir. Demak, O‘zbek mumtoz adabiyoti, xususan, Alisher Navoiy ijodidagi fano, baqo talqinlarini maxsus tekshirish ayni nuqtai nazardan ham ahamiyatli.Ushbu maqolada Hazrat Alisher Navoiyning “Hayratu-l-abror” asarida baqo maqomi, solikning baqoga olib boradigan fano martabalari, solikni fano va baqoga olib boruvchi: Iymon, Islom, adolat, karam, adab, qanoat, vafo, ishq, rostlik, ilm kabi bir qancha fazilatlarga alohida-alohida to‘xtalib o‘tildi, ularning mohiyati xususida mutasavvuflarning nazariy qarashlari va ushbu martabalarning badiiy talqini masalasi tahlil etilgan. Shu bilan birga ushbu maqolada fano va baqo talqini masalasi Alisher Navoiyning lirik janrdagi asarlaridan liro-epik asarlaridagi farqlari ochiqlangan.

Shiliang Lyu, Hanyun Qi, Yuwen Min et al.

The articulation of long and short vowels is not only a change in duration, but also includes changes in the shape of the oral cavity to comprehensively affect the timbre. Since the low-resource phonetic data is very limited, especially the pronunciation of long and short vowels will be affected by the surrounding phonemes or syllables, which makes it difficult for the speech synthesis model to adequately learn and capture the differences of vowels, thus affecting the accuracy and naturalness of the synthesis of long and short vowels. The eastern Yugur language in China belongs to the Mongolian group of the Altaic language family, with an endangered language and limited phonetic resource data. If this endangered language is extinct, it can be restored through speech synthesis. In this article, we analyze the acoustics of long and short vowels in the Eastern Yugur language, and find out the differences in the distribution of acoustic parameters, as well as the differences in the parameters of males and females, so as to provide a reference for the synthesis of the Eastern Yugur language.

Aomin1 ts, Dahu Baiyila2, Aijun Li2

Mongolian, a member of the Mongolic branch within the Altaic language family, is an agglutinative language characterized by the absence of tonal distinctions. The issue of word stress in Mongolian has been a hotly debated topic within Mongolian linguistics, with many differing views. Building on existing research on Mongolian word stress, this study employs the PENTA (Parallel Encoding and Target Approximation) intonation model to synthesize word stress patterns, aiming to explore a novel approach to studying Mongolian word stress. The experiments revealed that annotations based on underlying phonetic features are more effective in uncovering the fundamental control factors of surface-level fundamental frequency (FO) manifestations of word stress. Additionally, annotation content varies with different word lengths. Moreover, the PENTA model avoids individual variations lacking communicative function, thereby synthesizing highly consistent word stress patterns. Given the agglutinative nature of Mongolian, categorizing word stress patterns into basic and dynamic patterns more effectively represents the stress patterns of derived words formed by adding affixes to word stems.

Haotian Cui, Chloe Wang, Hassaan Maan et al.

Liina Lindström

Artiklis käsitletakse eesti keele mitmuse 1. isiku keeldkõne moodustamise ja kasutamise varieerumist, keelemuutuse kulgu ning järeldusi keelekorralduse jaoks. Mitmuse 1. isiku keeldkõnet saab moodustada nii imperatiivi keeldkõne vorme kasutades (ärgem tehkem) kui ka indikatiivi lõppe kasutades eitussõnal ja põhisõnal (ärme teeme) või ainult eitussõnal (ärme tee). Eesti keele ühendkorpusest 2019 saadud 4964-lauselise juhuvalimi põhjal selgub, et eitussõna ärme kasutus laieneb, ärgem kahaneb ning lõputa konstruktsiooni (ärme tee) kasutus kasvab. Ärgem- ja ärme-konstruktsiooni valikus on märgata mõningaid kinnistunud väljendeid, mis üht või teist varianti eelistavad. ärgem-konstruktsiooni kasutatakse rohkem retoorilise üleskutsena koos mitteagentiivsete verbidega, ärme-konstruktsiooni kõnelejat kaasavalt ja agentiivsete verbidega. Keelemuutuse suund on indikatiivse lõputa konstruktsiooni kasutuse laienemisele. See on ühtlasi tüpoloogiline muutus, sest sümmeetriline keeldkõnevorm asendub asümmeetrilisega. Seetõttu on keelekorraldus pikalt seda muutust tõrjunud, ent 2022. aasta lõpul siiski aktsepteerinud ka lõputa vormi. Abstract. Liina Lindström: Variation in prohibitive 1Pl forms and typological shift: a problem for language planning? Estonian shows extensive variation in the formation and use of 1Pl prohibitive forms. Older forms use a special imperative (prohibitive) negation verb, which can be conjugated (ärgem); imperative endings are repeated on the main verb (är­gem teh­kem ‘let’s not do it’). New forms use indicative endings on the negation verb (ärme) and main verb; however, the main verb may also occur as a bare stem (ärme teeme or ärme tee ‘let’s not do it’). In this paper, the variation between these three main constructions is analysed on the basis of data taken from the Estonian National Corpus 2019. It appears that use of the imperative construction (ärgem) is decreasing and use of indicative forms (ärme) is increasing; with indicative forms, there is a preference for the main verb to occur as a bare stem. This can also be seen as a typological change: symmetric negation of the imperative is replaced with an asymmetric negation construction. The imperative construction (ärgem), however, is still widely used for rhetorical purposes especially with non-agentive verbs, such as unustama ‘forget’, while the indicative construction (ärme) is typically used speaker-inclusively with agentive verbs. Language planning has resisted the asymmetric negation type so far, but accepted it finally in 2022.

Liina Saarlo

This article discusses the terminology and representations concerning the older Estonian oral song tradition specialists in Estonian folkloristic literature from the late 19th century to the second half of the 20th century. The folk singers’ relationships with their family, community, and folklorists from the perspective of diachronic and synchronic intersubjectivity is also considered. At the heart of the documentation and study of runo songs (regilaulud) are folk singers who actively embody the song culture. They are specialists who, in addition to having knowledge of the lyrics, tunes and performing practices, also have a need to perform in public, learn from other (older) singers, and share their skills with the next generations. At the height of the documentation of folklore at the turn of the 19th and 20th ­centuries, runo songs had receded to the periphery of cultural life and lost their importance in musical self-expression. So, too, the runo singer became the embodiment of the margins of society – someone old and poor. As folklore studies developed, the representation of the folk singer gained agency – the singer was not only a reservoir of knowledge or a repeater of rituals, but also a specialist in communication. In the 1960s, a wedding singer – an improviser who actively used poetic means – took over as the collective image of a folk singer. However, the song tradition is also carried on by people of a different nature, who know a lot of songs and the context in which they are used, but do not wish to perform in public. In this article, I focused on the inactive, modest singers. The main subjects of this article are the folk singers recorded in the 1960s, during the last great wave of runo song documentation. All those studied were „of singers’ blood”: their immediate family were well-known singers. In order to become proficient in performing the runo songs, they had to have grown up in a close relationship with their grandmother. It was the modest singers, in particular, who continued to express themselves by means of the runo songs in the form of autocommunication. The relationship of the modest singers with the collectors of folklore may even have been more positive than that of the active ones; they developed a closer bond with the collectors, were more patient and more accommodating.

Juha Kuokkala, Eino Koponen

The article presents a comparative-historical survey on Saami verbal derivatives with a suffixal *-š- (or *-šš-), based on extensive dictionary data from all Saami languages. The derivatives are divided into several subtypes using structural and functional criteria, and the distribution of each subtype as well as shared derivatives between Saami languages are investigated in order to illuminate the background of each subtype and their interrelations. It is argued that the essive verbs in *-āše̮- (South Saami vealkasjidh ‘appear white’) and denominal verbs in *-uše̮- (North Saami jallošit ‘play the fool’) as well as the diverse *-(e̮)še̮- verbs in the more peripheral languages descend from a single Proto-Saami *-še̮- derivative suffix with a general verbalizer function. The model of this type has probably also contributed to the emergence of the weak-grade *-Všše̮- type, which mainly consists of borrowings adapted from Finnic *-icce- verbs, such as North Saami dárbbašit (~ Finnish tarvitse-) ‘need’. Kokkuvõte. Juha Kuokkala, Eino Koponen: Saami -š-verbituletiste esinemisest ja ajaloost. Artiklis esitatakse võrdlev-ajalooline uurimus saami *-š- või *-šš- sufiksiga verbituletistest. Uurimus põhineb laialdasel sõnaraamatuainestikul kõigist saami keeltest. Tuletised jaotatakse strukturaalsete ja funktionaalsete kriteeriumide järgi alltüüpideks, ning iga alltüübi esinemist ja mitmele keelele ühiseid tuletisi vaadeldakse alltüupide tausta ja nende suhete selgitamiseks. Selle põhjal järeldame, et essiivsete *-āše̮ -verbide (lõunasaami vealkasjidh ‘paistma valgena’) ja denominaalsete *-še̮ -verbide (põhjasaami jallošit ‘veiderdama’) ning perifeersemate keelte erinevate funktsioonidega *-(e̮)še̮ -verbide taust peitub ühes ühises algsaami *-še̮-tuletusliites, mis algselt kandnud üldse verbistaja ja frekventatiivsufiksi funktsioone. Selle tüübi eeskuju on tõenäoliselt mõjutanud ka nõrgaastmelise *-Všše̮-tüübi tekkimist, mis peamiselt koosneb läänemeresoome *-icce-verbidest mugandatud laenudest, nagu põhjasaami dárbbašit (~ soome tarvitse-) ‘vajama’.

Alena Boganeva, Mare Kõiva

The article presents a comparative historical analysis of fairy tales of the Friends in Life and Death (ATU/SUS 470) plot type, their features and distribution among Belarusians, Russians and Estonians. In the Estonian and Belarusian material, stories of this type are in some cases contaminated with the ATU 470A and 471A types. We examine differences in the flow of time in the real world and other worlds, and consider the features of the chronotope of fairy tales and the organization of artistic space within them, as well as ways of transition and communication between the worlds. According to the most common plot of ATU/SUS 470 fairy tales, two friends promise to be guests at each other’s weddings. One of them dies and the living friend invites the dead one to his wedding; however, after receiving a counter-invitation, he finds himself in another world. There he either travels with the deceased friend through his world, where he sees various curiosities and paradoxes, or celebrates at the wedding of the deceased friend. It seems to the living person that he spent several hours with the deceased friend (or with his brother or father in some Estonian versions), but when he returns to his own world, he discovers that decades or centuries have passed on Earth. Most often, after this discovery, the hero dies and turns to dust, but there are some rare Estonian versions where he miraculously restores ­himself and his bride to their original state (and age), after which they get married again.

Ilona Tragel

Artiklis antakse ülevaade Eesti koroonapandeemiaaegsete piirangusiltide kogumisest ja uurimisest. Avaliku ruumi keelelisi sõnumeid on senini uuritud peamiselt keelemaastike suuna raames, fookusega eri keelte kasutusel või siltide keelekasutuse ja visuaalide vahekorral. Koroonapiirangute aegne siltide rohkus võimaldab põhjalikumalt kui kunagi varem uurida siltide keelekasutust, sh grammatilisi nähtusi. Koroonasiltide uurimisel lähtuti eeldusest, et sildil olev tekst, nagu iga teine tekst, on suhtlussündmus, milles osalejatel on kindel roll ja eesmärk. Sildi koostaja on teate vahendaja, kes eeldab, et sildi vastuvõtja käitub vastavalt sildil edastatud nõudele. Seniste uuringute tulemuste põhjal saab väita, et lisaks nõude edastamisele kasutatakse siltidel interpersonaalseid, st saatja ja vastuvõtja vahel seoseid loovaid keelevahendeid, et kõrgemalt tulnud ebamugavusi tekitavaid sõnumeid ja käske pehmendada. Siltidel luuakse ka intertekstuaalseid seoseid, neist sagedasem on seos vahendatava teatega (riiklikku olukorda reguleeriv tekst), kuid leidub ka vihjeid nii lastelaulule kui ka tuntud teatrilavastusele. Linguists meet the pandemic: Crisis communication on COVID-19 signs in Estonia In this article, public messages concerning the pandemic are studied as an example of crisis communication. An overview of the usage of public COVID-19 signs in Estonia during the pandemic is given. The COVID-19 sign is described as a communication act between the participants: the sign’s author (institution) and the addressee (client), who share the discourse of the pandemic including local measures implemented by the Estonian government. The traditional framework for researching public signs has been the Linguistic Landscape. The number of signs that has now emerged, however, allows us to study linguistic means on the signs from many new perspectives. Moreover, the global pandemic discourse provided a unifying feature for public COVID-19 signs all over the world, enabling us to compare the signs across countries. The results of a quantitative analysis of 900 Estonian COVID-19 door signs are presented. Relations between the participants of the communication act are studied from the perspectives of interpersonality (e.g., grammatical person) and intertextuality (means of expressing government regulations in the sign’s text). Furthermore, an overview of the previous studies of COVID-19 signs is given: authority and solidarity in the use of grammatical person, the comparison of signs in China and Estonia, expressing the cause, an experiment of perception of the use of person, speech acts of greetings and wishes. Moreover, various data about COVID-19 signs is presented, such as the most common words per word class: e.g., the most common adjective was hea ‘good’, the most common noun was mask ‘mask’ and the most common verb was sulgema ‘to close’. Quantitative data on the signs’ length is also provided. As might be expected, the overall length of the signs becomes shorter as the pandemic progresses since the authors have no need to explain the already familiar situation on the signs (the median length of a sign was 16 words during the first wave and 10 words during the third wave). Intertextual relations are also discussed in the article, e.g., how government regulations are referred to on the sign. Another interesting manifestation of intertextuality is reference to other genres: a sign could be written in verse or paraphrase a well-known nursery rhyme. Authors’ ideological stance can also be reflected on the sign, e.g., instead of referring to the measures as the cause of closing, one sign announces that the prime minister and minister of social affairs are “bullying”. It can thus be said that public COVID-19 signs are interesting and diverse research material. As messages created spontaneously in a time of crisis, they allow us to analyze how the use of language reflects the authors’ stance, and how grammar and intertextuality influence the perception and effect of the signs.

Orçun Ünal

: The present study focuses on the Proto-Turkic phoneme */d₂/ in intervocalic position, which can be reconstructed only through external data from Mongolic and other Altaic languages. For this phoneme, four examples are presented. These are * kad₂a ‘rock’, * äd₂iä ‘master, lord’, and * k₂ad₂a-‘to save, keep safe, take care of something’. The fourth example is Proto-Turkic * ud₂a-‘to be late’. It yielded the Common Turkic noun * uyag ‘late’ but survived in Mongolic as * uda-.

Mingyuan Shao, Xuna Lin

This article discusses the semantic and syntactic properties of converb markers =Shi, =Zhe, and =Tala in Sinitic varieties in the Qīnghǎi-Gānsù Linguistic Area (qgla), and defines them as converb enclitics rather than conjunctions, modal particles, or other functional words. =Shi is a conditional converb enclitic while =Tala is a terminative converb enclitic; both enclitics derive their respective properties from Altaic languages. The enclitic =Zhe expresses the coordinative and narrative functions, derived from its Altaic source, and the conjunctional function, which is an innovation. The productive use of the converbs coincides with the sharp decline of serial verbs in Sinitic varieties in qgla, and such a phenomenon emerges through intense contact among the Altaic, Sinitic and Tibetan languages in the area.

Jiayi Wu

Cross-border ethnic groups are a group of people living separately in two or more modern countries due to a long history of development. The Oroqen nationality and the Evenki nationality, as homologous ethnic groups living across the border of China and Russia, both belong to the Manchu-Tungusic group of Altaic languages, which are very similar and have great significance for the survival of the two ethnic cultures. With the historical changes and social development, the ethnic languages of the Oroqen nationality and the Evenki nationality are seriously on the danger of disappearing. In this regard, China and Russia have adopted institutionalized and systematic protection measures and formed their own distinctive language protection modes from three aspects: policies and laws, theoretical research and educational practice, which has alleviated the former endangered situation of ethnic languages. Based on the necessity of cross-border ethnic language protection, this paper explores the protection modes of these two ethnic languages and puts forward suggestions for further strengthening the protection in the future, in order to provide help for improving the soft power of national culture and enhancing the friendship between China and Russia.

A. M. Zakharova

The article considers the ways of expressing comparative relations in the Yakut and Altai heroic epics. The related Yakut and Altaic languages are currently being actively studied in comparative terms in order  to  identify  specific  differences.  The  comparative  construction  was  previously  studied  in  the  Yakut language by Yu. I. Vasilyev, theoretical coverage was given by E. I. Ubryatova; in the Altai language the comparative construction was specially studied by L. N. Tybykova. The material of the study was the Yakut heroic epic "Djuluruyar Nyurgun Bootur" ("Nurgun Botur the Swift") and the Altai epic "Maadai Kara".

Paolo Ognibene

In the third part of his Ossetian studies Vsevolod Miller hypothesized that the Alans moved towards the Caucasus following a northern route, that is, without crossing the mountain ridge. In support of his hypothesis he considered the presence in the Ossetian lexicon of many names relating to the terminology of metals of Ural and Altaic origin. In order to consolidate his hypothesis, he also examined various Altaic names of animals and birds present in Ossetian. The diffusion of these names in the Caucasian area allows us to make some considerations on the period and the road through which these words entered Ossetian.

D. Bradley

Six main alternative linkage proposals which involve the Sino-Tibetan family, including Sinitic and other language families of the East Asian area (Miao-Yao, Altaic/Transeurasian, Austroasiatic, Tai-Kadai, Austronesian) are briefly outlined. Using the standard techniques of comparative linguistics, a remote linkage between the Sino-Tibetan languages, including Sinitic, the Yeniseian languages of Siberia, and the Na-Dene languages of northwest North America is demonstrated. This includes cognate core lexicon showing regular sound correspondences, morphological similarities of form and function, as well as similarities in social structure. The other proposals for linkages that connect Sinitic and other languages of the East Asian area appear not to be based on a genetic linguistic relationship but rather due to contact: millennia of loanwords from Sinitic into the languages of those families and some lexicon borrowed into Sinitic. More remains to be done to further document the status of the linkage between Sino-Tibetan and Dene-Yeniseian.

Choi Yoon Seo

Scientist G.I Ramstedt first suggests that the Korean language add to The Altaic language group in 1928, and proposed to add the Korean language to the Altaic language in 1934. But he began to doubt his first suggestion in 1950. To determine the origin and typology of a language, it is necessary to compare it with other languages. Therefore, in order to clarify the doubts of G.I. Ramstedt, I studied the Korean language from a modern point of view, comparing not only one, but also two languages which are Mongolian and Turkish, and tried to identify their similarities and differences. In addition, transitive verbs are inevitably related to vocative, so the distinction between vocative along transitive verbs is reflected in this research. Therefore, the meaning of my research lies in identifying the similarities and differences between the verbs of the three languages and deciphering their meanings.