Warning: This paper is based on hate speech detection and may contain examples of abusive/ offensive phrases.

Cyberbullying, online harassment, etc., via offensive comments are pervasive across different social media platforms like ™Twitter, ™Facebook, ™YouTube, etc. Hateful comments must be detected and eradicated to prevent harassment and violence on social media. In the Natural Language Processing (NLP) domain, the most prevalent task is comment classification, which is challenging, and language models based on transformers are at the forefront of this advancement. This paper intends to analyze the performance of language models based on transformers like BERT, ALBERT, RoBERTa, and DistilBERT on the Indian hate speech datasets over binary classification. Here, we utilize the existing datasets, i.e., HASOC (Hindi and Marathi) and HS-Bangla. So, we evaluate several multilingual language models like MuRIL-BERT, XLM-RoBERTa, etc., few monolingual language models like RoBERTa-Hindi, Maha-BERT (Marathi), Bangla-BERT (Bangla), Assamese-BERT (Assamese), etc., and perform cross-lingual experiment also. For further analyses, we perform multilingual, monolingual, and cross-lingual experiments on our Hate Speech Assamese (HS-Assamese) (Indo-Aryan language family) and Hate Speech Bodo (HS-Bodo) (Sino-Tibetan language family) dataset (HS dataset version 2) also and achieved a promising result. The motivation of the cross-lingual experiment is to encourage researchers to learn about the power of the transformer. Note that no pre-trained language models are currently available for Bodo or any other Sino-Tibetan languages.

The idea for this special issue of Journal of Classics Teaching arose from the conference ‘Monsters in the classroom: Latin and Greek at primary school’ which Steve Hunt (Cambridge), Lidewij Van Gils (Amsterdam), and myself (Ghent) co-organised in January of 2022.1 This conference gathered teaching expertise from eight countries and attracted more than 120 participants from 20 countries to discuss both the successes and challenges related to current international practice in the teaching of Classical languages at primary school. It became a constructive and fruitful event, where participants from different countries shared good practice in order to learn from each other and formulate steps forward.

This article provides a textured history of the multivalent term “hindu” over 2,500 years, with the goal of productively unsettling what we think we know. “Hindu” is a ubiquitous word in modern times, used by scholars and practitioners in dozens of languages to denote members of a religious tradition. But the religious meaning of “hindu” and its common use are quite new. Here I trace the layered history of “hindu,” part of an array of shifting identities in early and medieval India. In so doing, I draw upon an archive of primary sources—in Old Persian, New Persian, Sanskrit, Prakrit, Hindi, Marathi, Bengali, and more—that offers the kind of multilingual story needed to understand a term that has long cut across languages in South Asia. Also, I do not treat premodernity as a prelude but rather recognize it as the heart of this tale. So much of South Asian history—including over two thousand years of using the term “hindu”—has been misconstrued by those who focus only on British colonialism and later. We need a deeper consideration of South Asian pasts if we are to think more fruitfully about the terms and concepts that order our knowledge. Here, I offer one such contribution that marshals historical material on the multiform and fluid word “hindu” that can help us think more critically and precisely about this discursive category.

In multilingual people, semantic knowledge is predominantly shared across languages. Providing semantic-focused treatment to people with aphasia has been posited to strengthen connectivity within association cortices that subserve semantic knowledge. In multilingual people, such treatment should result in within- and cross-language generalisation to all languages, although not equally. We investigated treatment effects in two multilingual participants with aphasia who received verb-based semantic treatment in two pre-stroke highly proficient languages. We compared within- and cross-language generalisation patterns across languages, finding within- and cross-language generalisation after treatment in the less-impaired, pre-morbidly more-proficient first-acquired language (L1). This observation supports the theory that connectivity is greater between the lexicon of a pre-morbidly more-proficient L1 and the shared semantic system than the lexicon of a pre-morbidly less-proficient later-acquired language. Our findings of within- and cross-language generalisation patterns could also be explained by both the Competing Mechanisms Theory and the theory of lingering suppression.

Word-level language identification is an essential prerequisite for extracting useful information from code-mixed social media content. Previous studies in word-level language identification show two important observations. First, the local context is an important indicator of the language of a word when a word is valid in multiple languages. Second, considering the word in isolation from its context leads to more effective language classification when a word is borrowed or embedded into sentences of other languages. In this paper, we propose a framework for language identification that makes use of a dynamic switching mechanism for effective language classification of both words that are borrowed or embedded from other languages as well as words that are valid in multiple languages. For a given input, the proposed switching mechanism makes a dynamic decision to bias its prediction either towards the prediction obtained by the contextual information or that obtained by the word in isolation. In contrast to existing studies that rely upon large amounts of annotated data for robust performance in a multilingual environment, the proposed approach uses minimal annotated resources and no external resources, making it easily extendible to newer languages. Evaluation over a corpus of transliterated Facebook comments shows that the proposed approach outperforms its baseline counterparts: classification based on the contextual information, classification based on the word in isolation, as well as an ensemble of the two classifiers.

Since 2015, more than 58,000 Syrian refugees have settled in Canada and, at the time of the 2016 national census, more than a fifth had settled in the province of Quebec. The rising numbers of refugees and the risks associated with families’ forced displacement have underscored the need to better understand and support the language of refugee children. The article reports on the oral language of three Syrian children ages five and six years, drawing on data from parent interviews, teacher reports, measures of the children’s language, and observations of their language use in a dual-language stimulation group, StimuLER. By triangulating this data, we were able to develop a rich and realistic portrait of each child’s language abilities. For these three boys, we observed that the home language was vulnerable to delays and weaknesses, and that learning the language of school was a drawn-out process. We also documented that parents and teachers had difficulties communicating with one another, and thus had difficulty meeting the educational needs of these children. We conclude that to foster resiliency in these children who are refugees, schools must find a way to build bridges with the parents to support the children’s language learning in both the language of school and at home.

The purpose of this research was to investigate factors that influence professionals’ guidance of parents of children with hearing loss regarding spoken language multilingualism and spoken language choice. Sixteen professionals who provide services to children and young people with hearing loss completed an online survey, rating the importance of a range of potential influences on the guidance they provide to parents. These participants were invited to comment on the importance of these influences. Participants included teachers of the deaf, speech-language pathologists, special education teachers, psychologists, auditory-verbal therapists, Auslan interpreters, and curriculum coordinators. All participants had experience working with multilingual families and reported that they would sometimes or always recommend multilingualism for children with hearing loss, with fewer reporting that they would sometimes recommend monolingualism. Professionals placed greater importance on factors relating to family and community considerations (e.g., family language models, communication within the family, community engagement), and less importance on organisational policy and children's characteristics. This research provides an initial insight into the factors that professionals consider when guiding parents around spoken language and spoken language multilingualism decision-making for their children with hearing loss.

The first part of this paper reviews research evidence for typological similarity and/or L2 status as determinants of cross-linguistic influence (CLI) in multilingual lexical production. The second part presents a model of vocabulary acquisition as a framework to explain CLI at the levels of form, (syntactic) frame, and meaning, as well as some of the developmental changes that have been reported for CLI patterns in relation to L3 learners’ proficiency. It is suggested that these patterns can be related to default processes and stages involved in the acquisition of individual word forms and their integration into networks of existing lexical triads, as described in the Parasitic Model of vocabulary acquisition. The third part of this paper points to research into the complexity and non-linearity of multilingual lexical development and the need to learn more about it.