Search

In this paper, the way religious identity is constructed via languages, with a particular focus on Arabic and Bengali terms originating from Persian, will be explored. It is vital to comprehend how Bengali Muslims create this constellation of languages, recognising that religion also has its linguistic demands, as language assists with making identities (Jaspal and Coyle 2010). The two languages will provide traces of how Muslimness is managed in the East End and consider how historically, the role of these languages have shifted. For example, the Persian-origin terms analysed are crucial in understanding Persian's influence in constructing a particular South Asian Muslim/Islamic expression. I critically examine how Arabic and Bengali intermingle while asserting different socio-religio-positionings. The claims-making qua a religious identity is morphed through various political junctures, particularly while forging a religious identity with other Muslims and how Arabic has become the dominant language in Bengali Muslims' lives. [ABSTRACT FROM AUTHOR]

This study proposes a method to help people with different degrees of hearing impairment to better integrate into society and perform more convenient human-to-human and human-to-robot sign language interaction through computer vision. Traditional sign language recognition methods make it challenging to get good results on scenes with backgrounds close to skin color, background clutter, and partial occlusion. In order to realize faster real-time display, by comparing standard single-target recognition algorithms, we choose the best effect YOLOv8 model, and based on this, we propose a lighter and more accurate SLR-YOLO network model that improves YOLOv8. Firstly, the SPPF module is replaced with RFB module in the backbone network to enhance the feature extraction capability of the network; secondly, in the neck, BiFPN is used to enhance the feature fusion of the network, and the Ghost module is added to make the network lighter; lastly, in order to introduce partial masking during the training process and to improve the data generalization capability, Mixup, Random Erasing and Cutout three data enhancement methods are compared, and finally the Cutout method is selected. The accuracy of the improved SLR-YOLO model on the validation sets of the American Sign Language Letters Dataset and Bengali Sign Language Alphabet Dataset is 90.6% and 98.5%, respectively. Compared with the performance of the original YOLOv8, the accuracy of both is improved by 1.3 percentage points, the amount of parameters is reduced by 11.31%, and FLOPs are reduced by 11.58%. [ABSTRACT FROM AUTHOR]

Zero frequency is a fundamental problem in information retrieval using language models and smoothing is applied to deal with this problem. The cluster-based smoothing method is found to be effective for information retrieval using language models. Since the effectiveness of cluster-based smoothing depends on clustering quality, there is scope for improvement by enhancing the clustering algorithm. In this paper, we present a study on how to improve cluster-based smoothing using a histogram-based incremental clustering algorithm and word embeddings. To our knowledge, this is the first study on the cluster-based smoothing method which is integrated with a language model for developing an effective IR system for the Bengali language which is one of the most spoken Indian languages. The proposed method has been tested on two benchmark Bengali IR datasets. The experimental results show that our proposed model for Bengali document retrieval is effective and it outperforms several baseline IR models. [ABSTRACT FROM AUTHOR]

Named Entity Recognition (NER) plays a significant role in enhancing the performance of all types of domain specific applications in Natural Language Processing (NLP). According to the type of application, the goal of NER is to identify target entities based on the context of other existing entities in a sentence. Numerous architectures have demonstrated good performance for high-resource languages such as English and Chinese NER. However, currently existing NER models for Bengali could not achieve reliable accuracy due to morphological richness of Bengali and limited availability of resources. This work integrates both Data and Model Centric AI concepts to achieve a state-of-the-art performance. A unique dataset was created for this study demonstrating the impact of a good quality dataset on accuracy. We proposed a method for developing a high quality NER dataset for any language. We have used our dataset to evaluate the performance of various Deep Learning models. A hybrid model performed with the exact match F1 score of 87.50%, partial match F1 score of 92.31%, and micro F1 score of 98.32%. Our proposed model reduces the need for feature engineering and utilizes minimal resources. [ABSTRACT FROM AUTHOR]

Adaptive Functioning being a socio-developmental construct varies according to neuro-developmental milestones and equivalent biologically and socio-culturally expected 'standard', thus requiring culturally adapted or culture sensitive tools to measure it. For this reason, aim of the study is to culturally adapt the Vineland Adaptive Behaviour Scale, Second Edition (Vineland-II), Survey Form (Sparrow et al., 2005) into Bengali language considering Bengali context with evaluating the psychometric properties. The research was a cross sectional study design in which current adaptive functioning and intellectual functioning by Binet Kamat Test of Intelligence, (Kamat, 1958) were used. 212 neuro-typically developing children (including 101 boys & 111 girls), with age range from 6 to 18 years, having 90 intellectual quotient score were selected following purposive sampling from five zones of Kolkata. For adaptation purpose, all items were translated in Bengali language and content of 13 items were culturally adapted. The result scale denotes good inter-rater reliability, range lying between 0.72-0.83 (kappa coefficient) and internal consistency reliability, range lying between 0.68 - 0.83 (split half reliability) and good content validity (X-4.3, S.D - 0.79). It is also observed that the social age (X-104.5 months) based on adapted (VABS II) is comparably low than the participants' mental age (X-129 months) and chronological age (X-132 months), (social age < mental age < chronological age) of the same group of neuro-typically developing participants. This finding indicates that adaptation process will only be completed if the standardization procedure is done on the normative population itself along with adaptation of test items; culture specific item is not only required but also development of culture specific age norm is required. [ABSTRACT FROM AUTHOR]

Code-switching, the combination of more than one language within a single utterance is popular in social media sites for informal communication, generating a huge amount of data, though not fully analysed for knowledge extraction due to lack of documentation. Moreover, the code-mixed data is erroneous, and to detect and correct different types of errors, the model requires large erroneous code-mixed language data, not publicly available. This paper first defines generic rules to write Bengali–English code-mixed language in English script, considering the inherent complexity of the Bengali language. Different types of typographical and cognitive errors are induced to obtain the huge erroneous data, based on human behaviour and perception as consulted with the language experts. The errors considered here are applicable to other code-mixed Indic languages and would be beneficial for the researchers. To demonstrate the applicability of the model, we have included the errors in the Hindi–English code-mixed Indic language. An attention-based two-level deep network architecture (uses LSTM as a basic unit) is used for error detection, correction, and translation of code-mixed sentences into a monolingual sentence. Results are reported in terms of accuracy, ROUGE score, and BLEU scores at word level and sentence level for both Bengali–English and Hindi–English code-mixed languages. [ABSTRACT FROM AUTHOR]

Background Chronic urticaria exerts a profound impact on quality of life. Recent guidelines recommend its evaluation in all chronic urticaria patients. Currently, the Chronic Urticaria Quality of Life Questionnaire (CU-Q2oL) is the only validated tool to assess chronic urticaria-specific quality of life. Objective To validate and adapt the CU-Q2oL to the Bengali language for its widespread use. Methods The CU-Q2oL questionnaire was translated into Bengali. Its internal consistency and reliability were tested by asking 42 chronic urticaria patients to complete this version. They completed the validated Bengali Dermatology Life Quality Index and Urticaria Control test questionnaires, and their scores were correlated with CU-Q2oL score to assess the validity of our Bengali version. Results The mean CU-Q2oL score of our patients (mean age 38.41 ± 13.4 years, male: female 29:13) was 48.8 ± 16.5. Domain 4 (sleep problems) was worst affected, followed by domain 1 (pruritus), while domain 2 (swelling) was least affected. We detected an excellent overall internal consistency (Cronbach's alpha = 0.93) of our version and nearly complete agreement (intra-class correlation coefficient = 0.91) between the test-retest scores. We found a significant positive correlation between the overall CU-Q2oL and Dermatology Life Quality Index scores (rs = 0.53, P = 0.0002), thus implying the validity of our version. Additionally, we noted a significant negative correlation between the overall CU-Q2oL and Urticaria Control test scores (rs = -0.48, P = 0.0007), suggestive of a more severe impairment of quality of life with poorer disease control. Limitations Small sample size, observational design and bias in test-retest reliability analysis due to the use of rescue therapy in-between assessment sessions were important limitations of our study. Conclusion The Bengali version of CU-Q2oL questionnaire is a valid and reliable tool suitable for both clinical and research use in Bengali speaking chronic urticaria patients. [ABSTRACT FROM AUTHOR]

Machine translation (MT) is the process of translating text from one language to another using bilingual data sets and grammatical rules. Recent works in the field of MT have popularized sequence-to-sequence models leveraging neural attention and deep learning. The success of neural attention models is yet to be construed into a robust framework for automated English-to-Bangla translation due to a lack of a comprehensive dataset that encompasses the diverse vocabulary of the Bangla language. In this study, we have proposed an English-to-Bangla MT system using an encoder-decoder attention model using the CCMatrix corpus. Our method shows that this model can outperform traditional SMT and RBMT models with a Bilingual Evaluation Understudy (BLEU) score of 15.68 despite being constrained by the limited vocabulary of the corpus. We hypothesize that this model can be used successfully for state-of-the-art machine translation with a more diverse and accurate dataset. This work can be extended further to incorporate several newer datasets using transfer learning techniques. [ABSTRACT FROM AUTHOR]

Although prominent translators, such as Google, Yahoo Babel Fish, Bing, etc., perform better when translating most widely used languages, they tend to commit fundamental mistakes in working with low-resource languages such as Bengali, Romanian, Arabic, etc. Such translators (e.g. Google Translate) use different data-driven translation approaches, such as neural machine translation (NMT), statistical machine translation (SMT), etc., to develop their polyglot translation system. However, performances of these data-driven approaches entirely rely on the attainability of significantly large parallel corpora of the translating language pairs. As a consequence, numerous popular languages, such as Bengali, remain barely explored not only in machine translation but also in other fields of natural language processing. Therefore, the target of this study is to explore effective translation from Bengali to English by accomplishing several Bengali language processing tasks. To be precise, we adopt a basic rule-based machine translator for translating from Bengali to English. Next, we enhance its performance by considering the veracious interpretation of the Bengali names as subjects (and nouns) in a sentence. Besides, we propose a Bengali verb identification and optimization technique by root-word detection (stemming) of the Bengali verbs. Finally, we unfold the efficacy of our proposed techniques through a comparative analysis with popular data-driven translators using a novel customized dataset focusing on Bengali-to-English translation. [ABSTRACT FROM AUTHOR]

In 1890, the Maldivian judge and poet Sheikh Muhammad Jamaluddin connected poetry with linguistic description in two ways. First, when he described features of the Dhivehi language with the aid of Arabic linguistic theory, he used Dhivehi poetry as linguistic evidence for correct usage. Second, he authored Dhivehi-language poetry about Arabic linguistic theory. Cosmopolis scholarship relates a narrative of how the wide circulation of Sanskrit, Arabic, and/or Persian fostered a vast network of writers who authored texts in major vernacular languages like Bengali, Burmese, Javanese, Kannada, Khmer, Malay, Sinhala, Tamil, Telugu, Thai, Tibetan, Turkish, and Urdu. This scholarship suggests that authors living within a particular cosmopolis wrote in divergent vernacular languages yet were, in some sense, connected because they translated and responded creatively to the same widely circulated source texts written in Sanskrit, Arabic, and/or Persian. Yet in cosmopolis scholarship's effort to reveal understudied connections, various degrees of disconnection among writers of vernacular languages within a cosmopolis tend to be missed. One problem of overlooking disconnection among writers of vernacular languages is that readers could mistakenly conflate superculture-subculture interaction with intercultural interaction. In this article, I argue that Dhivehi-language poetry and linguistic description was inside the Arabic cosmopolis but simultaneously outside, because in circa 1890 non-Maldivians in the Arabic cosmopolis of South and Southeast Asia could not even read the Thaana script of the Dhivehi language. [ABSTRACT FROM AUTHOR]

Sign language is the communication process of people with hearing impairments. For hearing-impaired communication in Bangladesh and parts of India, Bangla sign language (BSL) is the standard. While Bangla is one of the most widely spoken languages in the world, there is a scarcity of research in the field of BSL recognition. The few research works done so far focused on detecting BSL alphabets. To the best of our knowledge, no work on detecting BSL words has been conducted till now for the unavailability of BSL word dataset. In this research, a small static-gesture word dataset has been developed, and a deep learning-based method has been introduced that can detect BSL static-gesture words from images. The dataset, “BSLword” contains 30 static-gesture BSL words with 1200 images for training. The training is done using a multi-layered convolutional neural network with the Adam optimizer. OpenCV is used for image processing and TensorFlow is used to build the deep learning models. This system can recognize BSL static-gesture words with 92.50% accuracy on the word dataset. [ABSTRACT FROM AUTHOR]

Objective Bengali, the 6th most spoken language globally with 268 million speakers, demands a culturally appropriate tool for screening any cognitive compromise in this population. Addenbrooke Cognitive Examination-III (ACE-III) is a standardized tool used for screening and/or diagnostic purpose worldwide. The aim of the present study was to adapt and validate ACE-III into Bengali language. Methods The ACE-III UK Version A (2012) was adapted with linguistically and culturally appropriate items and validated on Bengali speakers. The participants consisted of 40 dementia and 22 Mild Cognitive Impairment (MCI) patients and 120 healthy-controls. Reliability and validity were examined. Discriminant function analysis was done. Sensitivity and specificity were evaluated and optimum cut-offs were established for MCI and dementia. Results Both sensitivity and specificity of ACE-III-Bengali of identifying dementia was 1; sensitivity for MCI ranged from 0.83 to 1, specificity from 0.76 to 1. Discriminant function analysis showed a significant difference in all domains of ACE-III-Bengali between healthy individuals and persons with neurocognitive impairment. Separate optimum ACE-III-Bengali cut-off scores were established according to level of education. For low education (

In 2019, the U.S. RISM Office created a catalogue record for a notated manuscript of the opera Satī ki kalaṅkinī (, 'Pure or Tainted?') by Nāgendranath Bandyopādhyāẏ and Modan Mohan Barmaṇa. Written in Calcutta in 1876 and presently housed at Harvard Library, the opera offers passages in Bengali and English script accompanied by music notation in a parallel format. The office chose to pursue Satī ki kalaṅkinī as a test case for exploring opportunities to draw non-European music into a project such as RISM but faced ambivalence about speaking on behalf of its 'Indian' pasts. This study engages particularly with the ethics of transcription and visibility, while also drawing attention to the text's cultural setting. Recently, questions of representation have taken on exceptional significance in U.S. academic libraries, owing to a managerial turn toward diversity, equity, inclusion, and antiracism. We share this case as a contribution to a more robust and realistic understanding of global music cultures, including those from colonial encounters. En 2019, le bureau américain du RISM a créé une notice de catalogue pour un manuscrit musical de l'opéra Satī ki kalaṅkinī (, « Pur ou entaché? ») de Nāgendranath Bandyopādhyāẏ et Modan Mohan Barmaṇa. Écrit à Calcutta en 1876 et actuellement conservé à la bibliothèque de l'université de Harvard, l'opéra comprend des textes en bengali et en anglais accompagnés de notations musicales dans un format parallèle. On a choisi de poursuivre Satī ki kalaṇkinī en tant que test pour explorer les possibilités d'intégrer la musique non-européenne dans le RISM, tout en gardant à l'esprit le défi ambivalent de parler au nom d'un passé « indien ». L'article examine en particulier les questions éthiques liées à la transcription et au mode de présentation et tient également compte de la dimension culturelle de la source. Récemment, les questions de représentation en ce qui concerne la diversité, l'équité, l'inclusion et l'antiracisme ont pris une importance exceptionnelle dans les bibliothèques universitaires américaines. En ce sens, l'étude de cas se veut une contribution à une compréhension plus solide et plus réaliste des cultures musicales mondiales, y compris celles issues de contextes coloniaux. 2019 erstellte die US-amerikanische Arbeitsstelle von RISM einen Katalogeintrag für eine Musikhandschrift der Oper Satī ki kalaṇkinī (, ‚Rein oder befleckt?') von Nāgendranath Bandyopādhyāẏ und Modan Mohan Barmaṇa. Sie entstand 1876 in Kalkutta und wird jetzt in der Bibliothek der Harvard University verwahrt. Das Manuskript präsentiert den Textteil (in Bengalisch und Englisch) mit jeweils zugehöriger musikalischer Notation in synoptischer Darstellung. Man unternahm die Katalogisierung als Testfall, um daran Möglichkeiten zu erproben, wie nicht-europäische Musik in RISM integriert werden kann, sehr wohl eingedenk der ambivalenten Herausforderung, für eine ‚indische' Vergangenheit zu sprechen. Der Beitrag befasst sich insbesondere mit den ethischen Fragen der Transkription und Darstellungsweise und berücksichtigt auch die kulturelle Dimension der Quelle. In letzter Zeit haben aufgrund der Sensibilisierung für Diversität, Gleichstellung, Inklusion und Antirassimus Fragen der adäquaten Repräsentation eine besondere Bedeutung in den wissenschaftlichen Bibliotheken der USA erlangt. Insofern versteht sich die Fallstudie als Beitrag zu einem solideren und realistischeren Verständnis globaler Musikkulturen, auch solchen aus kolonialen Kontexten. [ABSTRACT FROM AUTHOR]

This paper grows from the research conducted on the video industry located in Purulia District, West Bengal, India, namely 'Manbhum videos'. 'Manbhum' signifies both place and idiom, and the videos are imagined and produced in relation as well as in opposition to its big Other(s), which include Bollywood, and reputable Bengali language cinema. Manbhum feature-lenghth videos, comprise discreet episodes, which are intercepted by songs and dances; working within popular narrative strategies and modes of address, some of these videos use unique voices, of both singers and dubbing artists, to tell the story. Such application of music and voice do not only re-present specific conditions of production, but also inform us about the ways in which the actor's 'body' and 'voice' may be used within parallel industrial practices. Therefore, besides addressing the industrial contexts and proliferation of video industries across Global South in the era of the digital, this paper considers subjects of language, community, culture, geography, politics and the problem of Jhumur (songs) transfiguring into a so-called 'vulgar' form, and thereby, examines particular videos to study the characteristics of production, and the question of narration and recounting, in such intermittent videos. [ABSTRACT FROM AUTHOR]

Natural Language Processing (NLP) deals with analysing, understanding and generating languages likes human. One of the challenges of NLP is training computers to understand the way of learning and using a language as human. Every training session consists of several types of sentences with different context and linguistic structures. Meaning of a sentence depends on actual meaning of main words with their correct positions. Same word can be used as a noun or adjective or others based on their position. In NLP, Word Embedding is a powerful method which is trained on large collection of texts and encoded general semantic and syntactic information of words. Choosing a right word embedding generates more efficient result than others. Most of the papers used pretrained word embedding vector in deep learning for NLP processing. But, the major issue of pretrained word embedding vector is that it can‟t use for all types of NLP processing. In this paper, a local word embedding vector formation process have been proposed and shown a comparison between pretrained and local word embedding vectors for Bengali language. The Keras framework is used in Python for local word embedding implementation and analysis section of this paper shows proposed model produced 87.84% accuracy result which is better than fastText pretrained word embedding vectors accuracy 86.75%. Using this proposed method NLP researchers of Bengali language can easily build the specific word embedding vectors for word representation in Natural Language Processing. [ABSTRACT FROM AUTHOR]

The Coronavirus Anxiety Scale (CAS) is among the first few published screening tools for assessing dysfunctional anxiety induced by the current COVID-19 pandemic. The CAS was translated into the Bangla language following the International Test Commission's guidelines for this adaptation study and placed in an online survey (N = 737, with a mean age of 26.55 (SD = 7.166 years) to assess the psychometric properties of the Bangla version of the scale. Results suggested that all items had a good item discrimination index and single-factor structure with good factor loadings. The CAS Bangla version was found to have good internal consistency reliabilities, test-retest reliability, and composite reliability (≥ 0.7). The measurement invariance suggested invariances across age groups and gender. The CAS Bangla version showed a high correlation to the anxiety subscale of the short form of the Depression Anxiety Stress Scale (DASS-21) and a moderate correlation to the depression subscale of the DASS-21 and the COVID-19 Worry Scale. This validation of a Bangla CAS scale would be helpful for mental health practitioners to assess pandemic anxiety among the Bangladeshi people. [ABSTRACT FROM AUTHOR]

In artificial intelligence applications, advanced computational models, such as deep learning, are employed to achieve high accuracy, often requiring the execution of numerous operations. Conversely, lightweight computational models are typically more resource-efficient, making them suitable for various devices, including smartphones, tablets, and wearable technology. This paper presents an ultra-low-computation solution for interpreting sign languages to assist deaf and hard-of-hearing individuals without needing specialized hardware or significant computational resources. The proposed approach initially performs data abstraction on the input data. During this process, the image is systematically scanned from various perspectives, and the collected information is then encoded into a one-dimensional vector. Subsequently, the abstracted information undergoes processing through a Fully Connected Neural Network (FCN), resulting in highly accurate output. We also introduced two abstraction methods, namely Opaque and Glass, inspired by the interaction of light with different types of objects. The proposed abstractions facilitate the comprehension of the hand gesture's outer boundary as well as its row-wise and column-wise density of pixels. Our experiments on three datasets confirm the efficiency of the proposed method, achieving an accuracy of 99.4% in recognizing American Sign Language, 99.96% accuracy in recognizing Indian Sign Language, and 99.95% accuracy in recognizing Bangla Sign Language. Notably, the model size and the number of MAC operations are significantly smaller than state-of-the-art computational models trained on the same datasets. • We introduced an ultra-low-computation method for interpreting sign languages. • We developed a new encoding for the abstraction of hand gesture images. • Our approach significantly reduces both model size and MAC operations. • We achieved high accuracy in understanding American, Indian, and Bangla SLs. [ABSTRACT FROM AUTHOR]

Sign language plays a crucial role as an inclusive mode of communication for individuals experiencing challenges addressing both hearing and speech aspects. Bengali Sign Language serves as a lifeline for those with hearing or speech impairments, enabling effective communication. However, the intricacies of sign language, especially Bengali Sign Language, present unique challenges. Traditional methods of sign language recognition and interpretation involve privacy concerns as they rely on central servers. However, the research on identifying Bengali signs from hand gestures is still largely unexplored, and to the best of current knowledge, no research specifically focuses on safeguarding privacy in this aspect. Therefore, this research introduces a groundbreaking approach utilizing federated learning (FL), ensuring user privacy while classifying and detecting Bengali Signs from hand gestures. After experimenting with various averaging algorithms, employing Visual Geometry Group (VGG19) with Federated averaging (FedAVG) yielded outstanding results. With the utilization of FedAVG, this system achieved a remarkable accuracy rate of 98.36% with the VGG19 model, representing a significant advancement in the realms of sign language interpretation. This research makes substantial contributions to areas related to sign language detection and privacy-aware machine learning techniques, benefiting the deaf community and society as a whole. [ABSTRACT FROM AUTHOR]

The proliferation of fake reviews on various online platforms has created a major concern for both consumers and businesses. Such reviews can deceive customers and cause damage to the reputation of products or services, making it crucial to identify them. Although the detection of fake reviews has been extensively studied in English language, detecting fake reviews in non-English languages such as Bengali is still a relatively unexplored research area. The novelty of this study unfolds on three fronts: (i) a new publicly available dataset called Bengali Fake Review Detection (BFRD) dataset is introduced, (ii) a unique pipeline has been proposed that translates English words to their corresponding Bengali meaning and also back transliterates Romanized Bengali to Bengali, (iii) a weighted ensemble model that combines four pre-trained transformers model is proposed. The developed dataset consists of 7710 non-fake and 1339 fake food-related reviews collected from social media posts. Rigorous experiments have been conducted to compare multiple deep learning and pre-trained transformer language models and our proposed model to identify the best-performing model. According to the experimental results, the proposed ensemble model attained a weighted F1-score of 0.9843 on a dataset of 13,390 reviews, comprising 1339 actual fake reviews, 5,356 augmented fake reviews, and 6695 reviews randomly selected from the 7710 non-fake instances. [ABSTRACT FROM AUTHOR]

• Development of language resource of Bangladeshi bangla spoken language (BBSL). • Development of a large speech corpus named সুবাক্য (SUBAK.KO). • Evaluation of the strength and weakness of SUBAK.KO corpus by comparing it with another similar kind of open-source large corpus. • SUBAK.KO is a more balanced corpus compared to the open-source large corpus by google. • SUBAK.KO contains most of the regional accented speakers' variability of Bangladeshi bangla. This article reports the development of language resource for Bangladeshi Bangla spoken language (BBSL). Bangladeshi Bangla has inadequate large speech corpora for Large Vocabulary Continuous Speech Recognition (LVCSR) system. The accuracy of the automatic speech recognition (ASR) system rests on the quality of the speech corpus. This work discusses the common issues and activities related to the development of a large speech corpus named সুবাক্য (SUBAK.KO). This corpus is designed to support ASR research in Bangladeshi Bangla. It has been labeled sentence-wise. We have trained this corpus with one of the well-known current End-to-End ASR algorithms, Recurrent Neural Networks (RNNs) with Connectionist Temporal Classification (CTC). To know the strengths and weaknesses, the CER (Character Error Rate) and the WER (Word Error Rate) of the trained RNN CTC model have been observed. Another open-source large Bangla ASR corpus has been trained using the same ASR algorithm. Both trained models have been compared to assess the quality of these corpora. It has been found that SUBAK.KO is a more balanced corpus and considered more regional accented speech variability for a LVCSR system compared to that open-source large Bangla ASR corpus. [ABSTRACT FROM AUTHOR]

Many Japanese literary texts have been translated recently into Bangla. However, nobody has yet identified the first translation of Japanese literary text in the periodicals until today. The purpose of this paper is first to map the early phase of Japanese literature in Bangla periodicals; second, it attempts to distinguish the first piece of Japanese literature in Bangla; and third, it strives to ascertain the trends in interpreting Japanese literature in the second half of the 19th century. At least four pieces of literature are unearthed for the first time in this article. Despite not embracing the original Japanese or English name in the title in Bangla, Gonpachikomurasakihiyokufun (The Loves of Gompachi and Komurasaki) and Hone Kawa (Bones and Ribs) became the earliest Japanese literary specimens in Bangla. Colonial influence and preference for works of religious significance pertinent to Bengal society -- are pinned down as the trends in interpreting Japanese literature in periodicals in the 19th century. [ABSTRACT FROM AUTHOR]

The present paper aims to determine whether the Bengali book Londoner Naksa ebong France Bhraman (Wondrous Capers at London and Travelling in France) was written by the geologist Pramathanath Bose (P.N. Bose). To find it out, two well-established style markers often used in authorship attribution studies; namely, function words and punctuation marks, are used here. The result shows that possibly this book was penned by the geologist P.N. Bose. As a corollary, it may also be added that this approach may be used in future authorship attribution studies involving Bengali writings. [ABSTRACT FROM AUTHOR]

The paper reports on the changing phase of sociolinguistics in India. The early phase of sociolinguistically oriented research is located within the philological tradition of India which ruled the early phase of the modern linguistics activities. The next phase was led by Gumperz and Pandit, who were deeply influenced by Emeneau, particularly with reference to outcomes of stable bi/multilingualism. Post Pandit, sociolinguistics in India took a major applied turn which lasted for nearly three decades. Beginning 2000 saw a variationist wave in sociolinguistics in India which has got stronger over the years. The paper analyses the factors that were responsible for the various turns that Sociolinguistics took over time. [ABSTRACT FROM AUTHOR]

The collection of essays in this volume examines forms of business documentation in the late Persianate world and the Indian Ocean, between the seventeenth and twentieth centuries. Looking upon business in its broadest sense, the themes range from property disputes within families to inter-polity and inter-imperial deals, all of which is captured within the notion of the bazaar. Presenting documents and documentary forms written in Persian, but also the associated languages of Arabic, Bengali, Gujarati, Hindi, Marathi and Rajasthani, the articles collectively enrich the idea of the Persianate, delineating its specific dispensations within regional contexts, and also its boundaries and limitations. This is also a contribution to the study of Persographia, in this case Persianate rather than just Persian writing. The articles study specific language combinations, lexical elements and usages that came to be deployed in different areas and the legal cultures they provide evidence for. [ABSTRACT FROM AUTHOR]

• Introduced and formulated the problem of Bangla sign language (BdSL) alphabet recognition in Zero-shot Learning. • Created a new BdSL semantic attribute representation. • Proposed a seen/unseen split of BdSL signs for the Zero-shot learning approach. • Developed a transfer learning based approach to recognize both one-handed and two handed BdSL. • Compiled and evaluated our model on a large-scale dataset and made this publicly available on GitHub. Bangla, being the fifth most spoken language in the world has its own distinct sign language with two methods (one-handed and two-handed) of representation. However, a standard automatic recognition system of Bangla sign language (BdSL) is still to be achieved. Though widely studied and explored by researchers in the past years, certain unaddressed issues like identifying unseen signs and both types of BdSL or lack of evaluation of the models in versatile environmental conditions demarcate the real-world implementation of the automatic recognition of BdSL. To find a probable solution to the shortcomings in the existing works, this paper proposes two approaches based on conventional transfer learning and contemporary Zero-shot learning (ZSL) for automatic BdSL alphabet recognition of both seen and unseen data. The performance of the proposed system is evaluated for both types of Bangla sign representations as well as on a large dataset with 35,149 images from over 350 subjects, varying in terms of backgrounds, camera angle, light contrast, skin tone, hand size, and orientation. For the ZSL approach, a new semantic descriptor dedicated to BdSL is created and a split of the dataset into seen and unseen classes is proposed. Our model achieved 68.21%, 91.57%, and 54.34% of harmonic mean accuracy, seen accuracy, and zero-shot accuracy with six unseen classes respectively. For the transfer learning-based approach, we found pre-trained DenseNet201 architecture to be the best performing feature extractor and Linear Discriminant Analysis as the best classifier with an overall accuracy of 93.68% on the large dataset after conducting quantitative experimentation on 18 CNN architectures and 21 classifiers. The satisfactory result from our models supports its very probative potential to serve extensively for the hearing and speaking impaired community. [ABSTRACT FROM AUTHOR]

Popular translators such as Google, Bing, etc., perform quite well when translating among the popular languages such as English, French, etc.; however, they make elementary mistakes when translating the low-resource languages such as Bengali, Arabic, etc. Google uses Neural Machine Translation (NMT) approach to build its multilingual translation system. Prior to NMT, Google used Statistical Machine Translation (SMT) approach. However, these approaches solely depend on the availability of a large parallel corpus of the translating language pairs. As a result, a good number of widely spoken languages such as Bengali, remain little explored in the research arena of artificial intelligence. Hence, the goal of this study is to explore improvized translation from Bengali to English. To do so, we study both the rule-based translator and the corpus-based machine translators (NMT and SMT) in isolation, and in combination with different approaches of blending between them. More specifically, first, we adopt popular corpus-based machine translators (NMT and SMT) and a rule-based machine translator for Bengali to English translation. Next, we integrate the rule-based translator with each of the corpus-based machine translators separately using different approaches. Besides, we perform rigorous experimentation over different datasets to report the best performance score for Bengali to English translation till today by revealing a comparison among the different approaches in terms of translation performance. Finally, we discuss how our different blending approaches can be re-used for other low-resource languages. [ABSTRACT FROM AUTHOR]

Background: The Community‐Oriented Program for Control of Rheumatic Diseases (COPCORD) is intended to reduce the information deficit about the epidemiology of rheumatic diseases, particularly in rural communities in high‐income countries. Multiple studies have been conducted using the WHO‐International League of Associations for Rheumatology (ILAR)‐COPCORD core questionnaire in Bangladesh using the translated version in the Bangla language. Cross‐cultural adaptation and validation of the questionnaire are important to achieve better outcomes. Aim: To develop a culturally adapted, valid, and reliable Bangla version of the WHO‐ILAR‐COPCORD Core English Questionnaire to use as a rheumatic screening instrument among Bangla‐speaking people. Methods: The original English WHO‐ILAR‐COPCORD questionnaire was translated into Bangla and adapted in the local socio‐cultural context maintaining idiomatic, semantic, experiential, and conceptual equivalence between the English and Bangla versions. Pretesting was carried out among 30 patients and healthy attendants, following standard international recommendations. Content validity of the adapted Bangla version was assessed by the item‐ and scale‐level content validity indices (I‐CVI and S‐CVI). The adapted Bangla version of the WHO‐ILAR‐COPCORD questionnaire was used to assess 120 patients with rheumatic problems and healthy individuals. Test‐retest reliability was assessed using intraclass correlation coefficients. Results: The Bangla version of the WHO‐ILAR‐COPCORD questionnaire showed excellent content validity (I‐CVI = 1, S‐CVI = 1). The test‐retest reliability was also acceptable (intraclass correlation coefficient > 0.7). Conclusion: The adapted Bangla version of the WHO‐ILAR‐COPCORD questionnaire demonstrated acceptable psychometric properties, in terms of content validity and test‐retest reliability, for evaluating rheumatic problems in Bangladeshi patients. [ABSTRACT FROM AUTHOR]

Borders are not merely assertions of sovereign territorial demarcations, but indicative of cultural boundaries. This article discusses how the India-Bangladesh Land Boundary Agreement (2015) led to reorganisation of territorial boundaries, whereby the inhabitants had the choice of citizenship between India and Bangladesh that reaffirmed their identity. The subjective differentiation between 'us' and 'them' was negotiated by various cultural elements such as Bangla language; an everyday language symbolizing membership in the constituting communities, kinship and citizenship in the nation. Language was an important attribute that identified them as 'Bengali'/'Bangladeshi' while navigating the trans-territorial national identity. While arguing that the new citizens' construction of 'acts of citizenship' was based on a linguistic identity and everyday languages made claims to citizenship possible, the paper also explores: how does language help negotiate the cultural boundaries of 'us' or 'them' in post Land Border Agreement? How has language (re)shaped identity towards attainment of citizenship? [ABSTRACT FROM AUTHOR]

Purpose: Indigenous people have their faith on traditional living cultures. But in the context of dynamic changed situation and on the perspectives of globalization in the world the concepts of transformation, absorption, integration etc have been coming in the forefront. The merging of Small into the Big is a phenomenon of the society. In this study the social and cultural life along with the demographic situation of the Manipuri peoples (Bishnupriya) have been described. The main purpose of the study is to find out the nature of the absorbing process into the use of major language by the small Bishnupriya Manipuri community people. Moreover, an attempt has been made to identify the role of public libraries in the development and continuation of traditional language spoke by Bishnupriya Manipuri, an indigenous tribal community in Bangladesh. Scope and Coverage: Identification of the merger of traditional language of Bishnupriya Manipuri into a major language in Bangladesh and to safeguard the traditional indigenous language and culture the role of libraries has been assessed. Bishnupriya Manipuri peoples had concentrated themselves in Sylhet division and in this division the preferred place for living of Manipuri peoples is Adampur union of Kamalganj sub-division. Methods used: Basically the descriptive method was followed in this study. The geographical area was surveyed along with observation techniques to identify the social and cultural aspects of the Manipuri peoples with special reference to their use of language as a communication medium in the society. To identify the nature of language used by Bishnupriya Manipuri peoples, the rural villages and the local market areas were surveyed during the stay at Kamalganj in Bangladesh in 2019. The public libraries were identified and described to find out their role to develop the own indigenous languages in Bangladesh. Moreover, a number of documents were searched on the issue and after reviewing those published literature an attempt of discussion was made in this study which has been concentrated and focussed. Findings: As it is found from the survey that Bishnupriya Manipuri people in Bangladesh are habituated with the use of Bengali language, an attempt was made to find out the nature of origin of the language. Hence, from the discussion in the study it is revealed that a number of researches have been made in this regard. It is found that the Bishnupriya Manipuri Language may be either from the offspring of Bengali Language or from the Assamese language or Oriya language or may be regarded as own indigenous tribal language. About the role and functions of Libraries in the indigenous language development, it is seen that there is a need to establish public libraries in the rural areas. As there is no public library in the rural areas where Bishnupriya Manipuri people reside, discussion on the benefits of the public library in the language development process can not be made. Originality / Value: The article is entirely original in nature. No other research works have been done in this regard so far. [ABSTRACT FROM AUTHOR]

This article problematizes the relationship between neoliberal economic policies and the resurgence of a certain kind of Islamic religiosity in contemporary Bangladesh. It reflects the findings of an empirical study based on a multi-method approach to explore the macro-cultural impacts of the relationship between neoliberalism and Islam in the country and the embedded, nuanced, and complex set of economic and political relationships that appear to fuel social inequality, engender inequitable distribution and growth, and facilitate an ongoing marginalization of minority groups, in which "othering" among the equals emerge as a cemented outcome. Empirical findings suggest that premature or unplanned applications of neoliberal economic policies based on client-patron relations are the main precursors behind these outcomes and they continue to adversely affect Bangla language, Bengali cultures, the country's advanced education sector, and the overall ability for Bangladeshis to think critically and form and produce social relations. Evidence further indicates Bengali cultures appear to become increasingly embedded in a number of rituals in the name of Islamic principles and philosophy. The article refers to this phenomenon as the "great" transformation of contemporary Bangladesh. [ABSTRACT FROM AUTHOR]

Signature verification has been one of the major researched areas in the field of computer vision. Many financial and legal organizations use signature verification as an access control and authentication. Signature images are not rich in texture; however, they have much vital geometrical information. Through this work, we have proposed a signature verification methodology that is simple yet effective. The technique presented in this paper harnesses the geometrical features of a signature image like center, isolated points, connected components, etc. and, with the power of artificial neural network classifier, classifies the signature image based on their geometrical features. Publicly available dataset MCYT, BHSig260 (contains the image of two regional languages Bengali and Hindi) has been used in this paper to test the effectiveness of the proposed method. We have received a lower equal error rate on MCYT 100 dataset and higher accuracy on the BHSig260 dataset. [ABSTRACT FROM AUTHOR]

This study was conducted to understand the fertility behavior between the two population groups speaking different languages: Bengali-speaking Hindus and Hindi-speaking Hindus. The study further attempts to study the perception and practice regarding contraception use and their decision-making ability. The present study was conducted in Kolkata, West Bengal. A total of 64 women (Bengali-speaking Hindus: 34 and Hindi-speaking Hindus: 30) ranging between 15 years and 44 years were chosen by convenience sampling method. Semistructured interview schedule was employed to record the response of the participants. Sociodemographic profile of the participants and of their husbands and fertility history were collected using structured interview. Open-ended questions were asked to the participants to understand their perception about family planning. Case studies were also taken from each participant to know about their decision-making process of family planning. Descriptive statistic was carried out to analyze quantitative data, and for qualitative data, thematic analysis was carried out manually. Mean age of the participants was 33.1 years and that of their husbands was 40.2 years. Age at marriage was low in both the groups. Source of knowledge on contraceptives was mostly husband in both the populations. The husband (100%) played a dominant role in decision-making in both the groups. The role of in-laws played a dominant role in decision-making among the Hindi-speaking population. Withdrawal method was a common method of contraceptives in both the groups. [ABSTRACT FROM AUTHOR]

Context: Hyperuricemia and gout has been found to be associated with reduced health-related quality of life (HRQoL); however, there is no available Bengali questionnaire assessing the same. Aims: We aimed to develop the Bengali version of a questionnaire and examine its cross-cultural adaptability considering linguistic equivalence. Settings and Design: A multicentric, mixed methods, cross-sectional study was conducted through consecutive sampling at the outpatients of three homeopathic hospitals in West Bengal. Subjects and Methods: The Bengali version of the questionnaire was produced by standardized forward–backward translations. Psychometric analysis was run to examine its factor structure, validity, and reliability. Statistical Analysis Used: Reliability was examined using internal consistency (n = 210). Construct validity was examined by exploratory factor analysis (n = 105) using principal component analysis (PCA; varimax rotation). Subsequently, confirmatory factor analysis (CFA; n = 105) was performed to verify the model fit. Results: The internal consistency (Cronbach's α =0.880; 95% confidence interval 0.855–0.902), test–retest reliability and concurrent validity of the questionnaire– all were within acceptable limits. The (Kaiser–Meyer–Olkin = 0.832) and Bartlett's test of sphericity (Chi-square: 1644.344 at df = 210, P < 0.001) both suggested adequacy of the sample. In factor analysis using varimax, all the items loaded above the prespecified value of 0.4 and identified 6 components, explaining 77% of the variation. One item revealed a negative variance; hence the whole component of 2 items was removed from further evaluation. The goodness-of-fit of the 5-components model in CFA was also acceptable (Comparative fit index = 0.702, tucker Lewis index = 0.641, Root Mean Square Error of Approximation = 0.156, and Standardized Root Mean Square Residual = 0.123). Conclusions: The developed Bengali version of the questionnaire consisting of 19 items and framed within 5 components, appeared to be a valid and reliable instrument measuring HRQoL in patients suffering from hyperuricemia. [ABSTRACT FROM AUTHOR]

SUBESCO is an audio-only emotional speech corpus for Bangla language. The total duration of the corpus is in excess of 7 hours containing 7000 utterances, and it is the largest emotional speech corpus available for this language. Twenty native speakers participated in the gender-balanced set, each recording of 10 sentences simulating seven targeted emotions. Fifty university students participated in the evaluation of this corpus. Each audio clip of this corpus, except those of Disgust emotion, was validated four times by male and female raters. Raw hit rates and unbiased rates were calculated producing scores above chance level of responses. Overall recognition rate was reported to be above 70% for human perception tests. Kappa statistics and intra-class correlation coefficient scores indicated high-level of inter-rater reliability and consistency of this corpus evaluation. SUBESCO is an Open Access database, licensed under Creative Common Attribution 4.0 International, and can be downloaded free of charge from the web link: https://doi.org/10.5281/zenodo.4526477. [ABSTRACT FROM AUTHOR]

With the accelerated evolution of the internet in the form of web-sites, social networks, microblogs, and online portals, a large number of reviews, opinions, recommendations, ratings, and feedback are generated by writers or users. This user-generated sentiment content can be about books, people, hotels, products, research, events, etc. These sentiments become very beneficial for businesses, governments, and individuals. While this content is meant to be useful, a bulk of this writer-generated content requires using text mining techniques and sentiment analysis. However, there are several challenges facing the sentiment analysis and evaluation process. These challenges become obstacles in analyzing the accurate meaning of sentiments and detecting suitable sentiment polarity specifically in the Bangla language. Sentiment analysis is the practice of applying natural language processing and text analysis techniques to identify and extract subjective information from text. This paper presents how the attention mechanism could be incorporated effectively and efficiently in analyzing the Bangla sentiment or opinion. [ABSTRACT FROM AUTHOR]

What do language policy in Pakistan and the drive to Islamise the state have in common? In the wake of independence, Pakistan emerged as a state striving to create a nation and it looked both to language and religion in search of constructing its Islamic national identity. This paper looks at the darker side of the nation-building process in the country, with a specific focus on the role of language in the struggle to purify Pakistan of its un-Islamic elements and at the shifting nature on the discourse of Islamic nationhood in the country. In particular, it spotlights how politics and law function not only to determine what constitutes an Islamic language and blasphemous speech but, in doing so, also construct the Islamic nation and its 'Other'. [ABSTRACT FROM AUTHOR]

The main concern of this study is to conduct an evaluation of Bengali newspaper websites from West Bengal, Tripura and Assam in India based on the eight Alexa indexes from Alexa Internet. A total of 30 Bengali newspaper websites are selected from web directory of world newspapers, magazines, news sites, and newspaper publishers (w3newslive.com) for evaluation in the present study. Each newspaper websites were searched in Alexa databank and relevant data using Alexa indices including traffic rank (India and Global), pages viewed, links, bounce percentage, time on site, search percentage, and percentage of Indian/foreign users were collected. The collection of data were tabulated and analysed to disclose findings in accordance with the desired objectives of this study. Based on data obtained from Alexa Internet tools, the results of this study show that Ei Samay Sangbadpatra and Anandabazar Patrika have the highest traffic rank. Rashtriyo Khobor Bangla Dainik has the highest number of average pages viewed per day and estimated daily time spent on site by the visitors. Ei Samay Sangbadpatra has the maximum number of links. Aajkal has the highest reach amongst the global internet users; whereas Puber Kalom has the lowest bounce percentage. The highest percentage of visits that came from search engines is for Gati Dainik. The highest number of foreign users is for Aajkal Patrika. Most of the foreign users to Indian newspapers come from the Bangladesh. [ABSTRACT FROM AUTHOR]

Scene texts are the type of texts that we see in our surroundings when we are outside. They have custom fonts, sizes, shapes and colors that are often different from the fonts or styles typically found in documents. Scene text recognition has grown in popularity due to its significance in the development of smarter automated systems. Despite the fact that numerous types of research on scene text detection and recognition have been conducted in different languages, similar work on the Bengali language is limited by the lack of large-scale usable datasets. In this paper, we propose a multipurpose system for Bengali Scene text that allows for easy dataset collection via crowdsourcing as well as annotation, classification and detection tasks to be performed on the same platform. It comes with an Android-based application that can be used to capture or load images from the device, select Bengali texts over the images and annotate them. The images and labels are stored in the cloud. A Python-based script is used to perform real-time data processing, analysis, and detection tasks. The proposed system has also been tested on a crowdsourced Bengali scene text dataset collected by this system. The classification model achieved an accuracy score of 97%, and the detection model achieved a Mean Average Precision score of 92 on the on these crowd contributed dataset. [ABSTRACT FROM AUTHOR]

It reports the development of an enhanced library OPAC prototype through integration of language analysis tool and book reader in the retrieval interface. Language analysis or text analytics is considered as one of the components of language documentation and when integrated with library OPAC can extend supports to analyse corpus of the retrieved document in terms of word/phrase frequency, term circus, term links, term context etc through visual representation in a single-window along with the other datasets generally expected in a typical library OPAC. The open source software based integration mechanism is tested with English and Bengali as mainstream languages and a Unicode-compliant Indian official tribal language Santali (Ol Chiki script) as minority language. [ABSTRACT FROM AUTHOR]

In this paper we introduce a real Pashtu handwritten numerals dataset (PHND) having 50,000 scanned images and make publicly available for research and scientific use. Although more than fifty million people in the world use this language for written and oral communication, no significant efforts are devoted to the Pashtu Optical Character Recognition (POCR). We present a new approach for Pahstu handwritten numerals recognition (PHNR) based on deep neural networks. We train Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) on high-frequency numerals for feature extraction and classification. We evaluated the performance of the proposed algorithm on the newly introduced Pashtu handwritten numerals database PHND and Bangla language number database CMATERDB 3.1.1. We obtained best recognition rate of 98.00% and 98.64% on PHND and CMATERDB 3.1.1. respectively. [ABSTRACT FROM AUTHOR]

Background: The pattern of post-stroke aphasia in speakers of Bengali language has not previously been reported in the literature. Furthermore, the inter-relationship between age, gender, and level of education and aphasia typology has remained unsettled thus far. Aims: To investigate the incidence and type of aphasia in first-ever acute stroke patients who were speakers of Bengali language. Methods & Procedures: Bengali version of Western Aphasia Battery was used for language assessment in our study participants. Lesion localization was done by using magnetic resonance imaging(3T) for ischemic stroke (if not contraindicated) and computed tomography for hemorrhagic stroke. Among 515 screened cases of first-ever acute stroke, 208 presented with aphasia. Language assessment was done between 7 and 14 days in all study participants. Outcomes & Results: The incidence of post-stroke aphasia in our sample was found at 40.39%. Types of aphasia were – Broca's (38.5%) followed by global (27.9%); Wernicke's (12.5%); transcortical motor (9.6%); anomic (4.3%); transcortical sensory (3.8%); isolation (1.9%); and conduction (1.4%). Mean number of years of formal education was significantly higher in fluent aphasia group in comparison with non-fluent group (10.51 years versus 7.01 years, p = 0.003). In logistic regression analysis, location of lesion (posterior perisylvian) (p = 0.007, OR = 5.406, 95% CI, 1.602–18.240) and education (p = 0.044, OR = 1.097, 95% CI, 1.003–1.199) were two independent predictors favoring fluent aphasia. Conclusions: Aphasia among post-stroke Bengali patients is quite frequent. The commonest type of aphasia in our sample was Broca's aphasia. Bengali-speaking people with higher education were more likely to present fluent aphasia. [ABSTRACT FROM AUTHOR]

Individuals with hearing and speaking impairment communicate using sign language. The movement of hand, body and expressions of face are the means by which the people, who are unable to hear and speak, can communicate. Bangla sign alphabets are formed with one or two hand movements. There are some features which differentiates the signs. To detect and recognize the signs, analyzing its shape and comparing its features is necessary. This paper aims to propose a model and build a computer systemthat can recognize Bangla Sign Lanugage alphabets and translate them to corresponding Bangla letters by means of deep convolutional neural network (CNN). CNN has been introduced in this model in form of a pre-trained model called "MobileNet" which produced an average accuracy of 95.71% in recognizing 36 Bangla Sign Language alphabets. [ABSTRACT FROM AUTHOR]

• We introduced a flexible framework capable of automatically recognizing phonetic units present in a speech utterance of any language spoken in any mode. • Multilingual speech mode classification (MSMC) model is proposed for classyfing modes of input speech. • A heirarchial framework is proposed to build the MSMC model using vocal tract and excitation source features. • DNN based Multilingual phone recognition systems (MPRS) are developed using vocal tract, tandem and excitation source features. • Jointly considered the conversation, and read modes from four Indian languages: Bengali, Odia, Telugu and Kannada for evaluation of MSMC and MPRS performances. The aim of this paper is to develop a flexible framework capable of automatically recognizing phonetic units present in a speech utterance of any language spoken in any mode. In this study, we considered two modes of speech: conversation and read modes in four Indian languages, namely, Telugu, Kannada, Odia, and Bengali. The proposed approach consists of two stages: (i) Automatic speech mode classification (SMC) and (ii) Automatic phoneme recognition using mode-specific multilingual phone recognition system (MMPRS). The vocal tract and excitation source features are considered for classifying speech modes using feed forward neural networks (FFNNs). The vocal tract, excitation source, and tandem features are used in training deep neural network (DNN)-based multilingual phone recognition systems (MPRSs). The performance of the proposed approach is compared with baseline mode-dependent and mode-independent MPRSs. Experimental results show that the proposed approach which combines both SMC and MMPRS into a single system outperforms the baseline phone recognition systems. [ABSTRACT FROM AUTHOR]

Background: School failure or poor academic performance is often found in neurodevelopmental disorders (NDD); however, there is a dearth of a comprehensive assessment tool to evaluate various underlying deficits, including perceptual-motor, cognitive, language, and scholastic skills of those who have NDD. The study aimed to develop a test to fill-up this gap. Materials and Methods: The study followed three phases: the construction of an assessment battery in both Bengali and English languages, separately, incorporating tasks on perceptual-motor, cognitive, language, and scholastic skills; doing a pilot study, and finally, standardization. Standardization was done on 91 normal children (NC) aged 4.5 to 9.5 years, from four districts of West Bengal. The test was applied to 57 children with poor school performance across various NDD, including specific learning disorder, autism spectrum disorder, attention deficit/hyperactivity disorder, and communication disorder. Binet Kamat Test (BKT) of intelligence, National Institute of Mental Health and Neuro-Sciences (NIMHANS) Index for specific learning disability (SLD), Childhood Autism Rating Scale (CARS), Conner's Abbreviated Rating Scale- Parent Report, Linguistic Profile Test, and Test of Pragmatic Language were used as screening tools to identify children with various NDD. The psychometric properties of the tool were assessed. Results: The factor analysis suggested four-factor solution named scholastic–cognitive–motor, attention, auditory–verbal, and perceptual skill. The internal consistency of the test was found to be higher (Cronbach's α >0.70 for most tests), indicating high reliability. Discriminant validity revealed significant score differences between NC and children with NDD (P <.01), suggesting that the new tool can differentiate children with NDD from healthy NC. Conclusion: The results favor the new tool as a psychometrically strong tool to assess the scholastic backwardness of children with NDD. It may be further used to create specific profiles for different categories of NDD. [ABSTRACT FROM AUTHOR]

This paper analyses some key historical developments (from early twentieth century to present) in Bangladesh from a socio-historical and language ideological perspective to explore the social injustice faced by the non-dominant and linguistically diverse population. These developments sparked language debates involving questions of identity, nationhood, and political hegemony. The Bengal partition in 1905, for example, flared up the Bengali monolingualism and the territorial principle so strongly that the British had to give in, putting the non-dominant Bengali-speaking Muslims at a socio-economic disadvantage. The 1947 partition brought to the forefront the ideologies of an Islamic language, a Hindu language, and linguistic diversity as a threat to national unity, making both the Bengali-speaking Muslims and the hill people face social and political injustice. The birth of Bangladesh in 1971 was a welcome change but its monolingual Bengali ideology put the speakers of minority languages at a long-term disadvantage. The 1997 Peace Accord, UNESCOS's 1999 declaration of 21 February as the International Mother Language Day, the 2010 educational policy, and the creation of an International Mother Language Institute offered some flickers of hope, but the underlying ideology remains Bengali monolingual, still creating barriers for a socially just future for minority language speakers. [ABSTRACT FROM AUTHOR]

In this article, I foreground the granular movements that determine the ubiquitous nature of India's new media economy created by the advent of streaming media platforms and the emergence of regional online content creators in India. I argue that the increasing preference of Indian audiences to consume online content in their own language has led to a demand for 'regional' content, whereby streaming platforms and online creators are increasingly investing in 'non-Hindi' and 'non-English' language content to cater to the linguistically and culturally diverse Indian population. Through a primary focus on online content creation practices in Bengali and Marathi languages, the article explores their 'local', 'regional', 'transnational' and 'global' appeal, and subsequent blurring of boundaries between 'regionalization' and 'localization'. The remainder of the article focuses on the emerging diaspora of regional online content creators who are adopting distinct content strategies to develop relationships with online communities based on commonalities of language and culture. [ABSTRACT FROM AUTHOR]

Understanding the context with generation of textual description from an input image is an active and challenging research topic in computer vision and natural language processing. However, in the case of Bengali language, the problem is still unexplored. In this paper, we address a standard approach for Bengali image caption generation though subsampling the machine translated dataset. Later, we use several pre-processing techniques with the state-of-the-art CNN-LSTM architecture-based models. The experiment is conducted on standard Flickr-8K dataset, along with several modifications applied to adapt with the Bengali language. The training caption subsampled dataset is computed for both Bengali and English languages for further experiments with 16 distinct models developed in the entire training process. The trained models for both languages are analyzed with respect to several caption evaluation metrics. Further, we establish a baseline performance in Bengali image captioning defining the limitation of current word embedding approaches compared to internal local embedding. [ABSTRACT FROM AUTHOR]

Information Extraction allows machines to decipher natural language through using two tasks: Named Entity Recognition and Relation Extraction. In order to build such a system for Bangla Language, in this work a Named Entity Recognition (NER) System is proposed, which requires a minimum information to deliver a decent performance having less dependency on handcrafted features. The proposed model is based on Deep Learning, which is accomplished through the use of a Densely Connected Network (DCN) in collaboration with a Bidirectional-LSTM (BiLSTM) and word embedding, i.e., DCN-BiLSTM. Such a system, specific to the Bangla language, has never been done before. Furthermore, a unique dataset was made since no Named Entity Recognition dataset exists for Bangla language till date. In the dataset, over 71 thousand Bangla sentences have been collected, annotated, and classified into four different groups using IOB tagging scheme. Those groups are person, location, organization, and object. Due to Bangla's morphological structure, character level feature extraction is also applied so that we can access more features to determine relational structure between different words. This is initially done with the use of a Convolutional Neural Network but is later outperformed by our second approach which is through the use of a Densely Connected Network (DCN). As for the training portion, it has been done for two variations of word embedding which are word2vec and glove, the outcome being the largest vocabulary size known to both models. A detailed discussion in regard to the methodology of the NER system is explained in a comprehensive manner followed by an examination of the various evaluation scores achieved. The proposed model in this work resulted in having a F1 score of 63.37, which is evaluated at Named Entity Level. [ABSTRACT FROM AUTHOR]

Question classification (QC) is a prime constituent of an automated question answering system. The work presented here demonstrates that a combination of multiple models achieves better classification performance than those obtained with existing individual models for the QC task in Bengali. We have exploited state-of-the-art multiple model combination techniques, i.e., ensemble, stacking and voting, to increase QC accuracy. Lexical, syntactic and semantic features of Bengali questions are used for four well-known classifiers, namely Naïve Bayes, kernel Naïve Bayes, Rule Induction and Decision Tree, which serve as our base learners. Single-layer question-class taxonomy with 8 coarse-grained classes is extended to two-layer taxonomy by adding 69 fine-grained classes. We carried out the experiments both on single-layer and two-layer taxonomies. Experimental results confirmed that classifier combination approaches outperform single-classifier classification approaches by 4.02% for coarse-grained question classes. Overall, the stacking approach produces the best results for fine-grained classification and achieves 87.79% of accuracy. The approach presented here could be used in other Indo-Aryan or Indic languages to develop a question answering system. [ABSTRACT FROM AUTHOR]