Search

Reports on a small-scale ethnographic study carried out with lower secondary students of Urdu and Bengali at a mainstream school in East London. Focused on the use of the Internet and multi-lingual word processing within a Web publishing project and examined the role played by information and communications technologies at different stages. (Author/VWL)

A study examined the effects of differing degrees of bilingualism (English monolingual, English-French balanced bilingual, Bengali-English partial bilingual) on nonverbal problem-solving abilities of 71 grade-three children. The groups were compared on control of attention and analysis of knowledge. Results indicate the balanced bilinguals performed better on non-linguistic tasks requiring control of attention, but no other significant differences. (Author/MSE)

Safder Alladina's analysis of the position of South Asian languages in Britain leaves much to be desired, concerning itself largely with Urdu, Hindi to a lesser extent, and hardly at all with Bengali, Gujarati, and Panjabi. The problems and solutions of devising policies appropriate to these languages need to be discussed in greater detail. (CB)

Discusses the syntactic properties of participial adjuncts and the relevant general syntactic principles in Eastern Bengali, specifically those responsible for case-assignment, word order, and binding properties. (40 references) (Author/VWL)

Argues against the use of an alternative model of Language Awareness (LAw), which rejects Transformational Grammar-based applications of language study in favor of a semiotics-based approach, and presents examples of teaching English to native Bengali speakers to support the use of the more traditional LAw. (CB)

Cues from multiple modalities have been successfully applied in several fields of natural language processing including machine translation (MT). However, the application of multimodal cues in low-resource MT (LRMT) is still an open research problem. The main challenge of LRMT is the lack of abundant parallel data which makes it difficult to build MT systems for a reasonable output. Using multimodal cues can provide additional context and information that can help to mitigate this challenge. To address this challenge, we present a multimodal machine translation (MMT) dataset of low-resource languages. The dataset consists of images, audio and corresponding parallel text for a low-resource language pair that is Manipuri–English. The text dataset is collected from the news articles of local daily newspapers and subsequently translated into the target language by translators of the native speakers. The audio version by native speakers for the Manipuri text is recorded for the experiments. The study also investigates whether the correlated audio-visual cues enhance the performance of the machine translation system. Several experiments are conducted for a systematic evaluation of the effectiveness utilizing multiple modalities. With the help of automatic metrics and human evaluation, a detailed analysis of the MT systems trained with text-only and multimodal inputs is carried out. Experimental results attest that the MT systems in low-resource settings could be significantly improved up to +2.7 BLEU score by incorporating correlated modalities. The human evaluation reveals that the type of correlated auxiliary modality affects the adequacy and fluency performance in the MMT systems. Our results emphasize the potential of using cues from auxiliary modalities to enhance machine translation systems, particularly in situations with limited resources. [ABSTRACT FROM AUTHOR]

Bengali is the official language of Bangladesh and is widely used in Bangladesh and West Bengal in India. Due to the growing accessibility of the internet and smart devices, the use of digital text material and documents in Bengali is growing with time. An automated Bengali Sentence Validation System is proposed in this study to effectively determine the correctness of sentences in such extensively available Bengali content. As far as we know, no substantial work has been done in the field of Bengali Sentence Validation utilizing deep learning approaches. Due to the lack of linguistic resources, sophisticated Natural Language Processing tools, and benchmark datasets, developing an automated Sentence Validation System for a limited-resource language like Bengali is challenging. Additionally, Bengali Sentences come in two morphological varieties (Sadhu-bhasha and Cholito-bhasha), making the validation process more challenging. The proposed automated Bengali Sentence Validation system contains the CNN-BiLSTM hybrid classifier model. As of now, there is no standard dataset for Bengali sentence validation. Due to the lack of a standard dataset, we collected Bengali sentences from different sources in Bangladesh and developed a Bengali Sentence Validation (BSV) Dataset with around 5000 labelled sentences arranged into two categories such as correct and incorrect. Experimental results demonstrate that the proposed system outperformed other classifier models and existing approaches for Bengali Sentence Validation and is able to categorize a wide range of Bengali sentences based on their correctness. The system’s F1 score for the Bengali Sentence Validation is 98%. [ABSTRACT FROM AUTHOR]

With the increasing practice of using regional languages in social media platforms, hate speech detection in regional languages has received the attention of researchers. In India, hundreds of languages are spoken in various forms, which are dependent on their geography, culture, etc. Recently the number of active internet users has been rapidly increasing in India, and therefore social media has penetrated the common Indian population. Though the need for proper detection and timely removal of abusive or offensive texts has increased, well-organized and labeled data for Indian languages are scarce. Almost all the regional languages in India are low-resource languages. Hence, the objective of this study is to develop an approach that will learn from relatively small volumes of Indian language data and provide state-of-the-art results. A fusion of features extracted from a fined-tuned multilingual BERT (Bidirectional Encoder Representations from Transformers) and a fine-tuned Indic BERT has been proposed in this study. Since the BERT models that we have used for this work are pre-trained using a large volume of texts in multiple Indian languages, transfer learning solves the problem of low training data volume, and this makes the proposed model more generic. Three datasets for three different Indian languages namely, Bengali, Marathi, and Hindi have been considered in this study to evaluate the proposed approach. The proposed model achieved a weighted F1 score of 0.923, 0.815, and 0.924 for the Bengali, Hindi, and Marathi datasets respectively. In the Bengali and Marathi datasets, the obtained results are better than the existing best results. [ABSTRACT FROM AUTHOR]

An English-Bengali machine translation (MT) application can convert an English text into a corresponding Bengali translation. To build a better model for this task, we can optimize English-Bengali MT. MT for languages with rich resources, like English-German, started decades ago. However, MT for languages lacking many parallel corpora remains challenging. In our study, we employed back-translation to improve the translation accuracy. With back-translation, we can have a pseudo-parallel corpus, and the generated (pseudo) corpus can be added to the original dataset to obtain an augmented dataset. However, the new data can be regarded as noisy data because they are generated by models that may not be trained very well or not evaluated well, like human translators. Since the original output of a translation model is a probability distribution of candidate words, to make the model more robust, different decoding methods are used, such as beam search, top-k random sampling and random sampling with temperature T, and others. Notably, top-k random sampling and random sampling with temperature T are more commonly used and more optimal decoding methods than the beam search. To this end, our study compares LSTM (Long-Short Term Memory, as a baseline) and Transformer. Our results show that Transformer (BLEU: 27.80 in validation, 1.33 in test) outperforms LSTM ( 3.62 in validation, 0.00 in test) by a large margin in the English-Bengali translation task. (Evaluating LSTM and Transformer without any augmented data is our baseline study.) We also incorporate two decoding methods, top-k random sampling and random sampling with temperature T, for back-translation that help improve the translation accuracy of the model. The results show that data generated by back-translation without top-k or temperature sampling ("no strategy") help improve the accuracy (BLEU 38.22 , + 10.42 on validation, 2.07 , + 0.74 on test). Specifically, back-translation with top-k sampling is less effective ( k = 10 , BLEU 29.43 , + 1.83 on validation, 1.36 , + 0.03 on test), while sampling with a proper value of T, T = 0.5 makes the model achieve a higher score ( T = 0.5 , BLEU 35.02 , + 7.22 on validation, 2.35 , + 1.02 on test). This implies that in English-Bengali MT, we can augment the training set through back-translation using random sampling with a proper temperature T. [ABSTRACT FROM AUTHOR]

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]

In the sentiment analysis (SA) task, we can obtain a positive or negative-typed comment or feedback from an online user or a customer about any object, such as a movie, drama, food, and others. This user’s sentiment may positively impact various decision-making processes. In this regard, a lot of studies have been done on identifying sentiments from a text in high-resource languages like English. However, a small number of studies are detected in the under-resource Bengali language because of the unavailability of the benchmark corpus, limitations of text processing application software, and so on. Furthermore, there is still enough space to enhance the classification performance of the SA task. In this research, we experiment on a recognized Bengali dataset of 11,807 comments to find positive or negative sentiments. We employ five state-of-the-art transformer-based pretrained models, such as multilingual Bidirectional Encoder Representations from Transformers (mBERT), BanglaBERT, Bangla-Bert-Base, DistilmBERT, and XLM-RoBERTa-base (XLM-R-base), with tuning of the hyperparameters. After that, we propose a combined model named Transformer-ensemble that presents outstanding detection performance with an accuracy of 95.97% and an F1-score of 95.96% compared to the existing recent methods in the Bengali SA task.

The cause(s) of missing inflectional morphology in obligatory contexts by adult speakers of second language (L2) English is subject to ongoing discussion. Whatever the specific theory, however, the apparent asymmetrical production of the morpheme '-s' in the marking of number on plural nouns versus that on third person singular agreement has to be accounted for. This study adopts the theoretical approach put forward by the Prosodic Transfer Hypothesis, whereby the prosodic representation of inflectional morphology in the first language (L1) can, to some extent, account for differences in the suppliance of inflectional morphology in L2 English within and across L1s. It is in this context that the production of third person singular agreement, and, for contrastive purposes, number on plural nouns, by L1 Bengali speakers of L2 English, is considered in relation to available prosodic representation in the L1, as well as against phonological processes attested in L1 acquisition. More specifically, with respect to covert contrasts. An inspection of spectrograms from instances of the apparent omission of inflection by L1 Bengali speakers of L2 English at Beginner to Intermediate proficiency levels does not, however, indicate that learners are covertly supplying agreement on the third person singular (or plural number on nouns). This finding does not necessarily rule out the occurrence of covert contrasts in L2 production of inflectional morphology; alternative techniques may detect a systematic difference between bare verbs and non-audible (to the listener) inflection. [ABSTRACT FROM AUTHOR]