“Tham Annotated Khasi Corpus” By Dr Medari Janai Tham* Natural language processing or NLP enables a computer to analyse and understand human languages in written or spoken form. It is the backbone in the development of machine translation software, voice assistants such as Google Assistant, Siri, chatbots, automatic spelling correction, grammar checking, search engines, and so on. Machine learning is one of the techniques employed by NLP to carry out the above-mentioned tasks. As the name suggests, machine learning implies teaching a machine to learn to execute desired tasks by providing one essential component for learning which is data. This data is formatted according to the task at hand and provide a way for a machine learning model to be generated.This model is a mathematical representation and it is responsible for evaluating new cases which are similar to the data it has seen. Hence, the development of such data is an important step for any NLP task. In NLP literature, this data is called corpus which is a collection of machine-readable text that is sampled to be representative of a particular language. Such corpora exist for languages such as English, German, Chinese, Hindi, Bengali, Punjabi, etc. In English, the most widely used corpora are the British National Corpus (BNC) and it is popular among researchers due to its accessibility. However, not all of these corpora are easily accessible. Khasi Corpus Construction For resource-poor languages like Khasi, an Austro-Asiatic language family spoken mainly in the state of Meghalaya and some parts of Assam and Bangladesh, there is no available corpus till recently. However, now the Khasi annotated corpus titled “Tham Khasi annotated corpus” has been developed and is freely accessible through the European Language Resources Association (ELRA) via the link http://catalog.elra.info/en-us/repository /browse/ELRA-W0321/. The corpus comprises Khasi sentences formatted with parts-of-speech tags using the formulated BIS (Bureau of Indian Standards) POS (Parts-of-Speech) tagset to ensure standardised tagging with other Indian languages. This corpus can serve as a resource for researchers to perform various NLP tasks for the Khasi language. One such outcome is NLP tools for Khasi currently available in https://grammarkhasi.in comprising of POS taggers that automatically tag any given Khasi sentence with its parts of speech and a shallow parser that automatically detects Khasi noun phrases and verb phrases in a sentence. Corpus analysis has revealed the existence of 560 words that are multi-functional which lexicographers or any interested researchers can further investigate. The Tham Khasi annotated corpus is part of the PhD research output of Dr Medari Janai Tham, Associate Professor, Department of Computer Science, St. Anthony’s College, Shillong. Conclusion Developing language technology tools for an under-resourced language such as Khasi has been challenging and simultaneously exhilarating to discover the nitty-gritty of the language in the way studies such as this one exposes. The performance of the HMM tagger conditioned with the features intrinsic in the language has shown that it also provides good performance as reported in the literature relating to HMM POS taggers. This work, being a new initiative, annotating the corpus and developing the tagger, is limited by available resources; however, increasing the size of the annotated corpus for further analysis will be a good step forward. Download Full Post

Meghalaya, in today’s age, has seen a significant increase in social media usage. This has been widely accelerated during the pandemic. Social media is no longer just a tool for socialisation but has embedded itself into various aspects of human life (education, politics, economics). In the case of Shillong, social media activity and engagement have flourished, and individuals are constantly producing and distributing content on multiple platforms. The Rapid Rise of Social Media The rise of social media has led researchers to label the current economy as the “attention economy”. The attention economy is dependent on transforming online attention and online (social) capital into monetary gains. It is commendable to see the youths tap into this emerging industry within the confines of the Shillong Urban Agglomeration. This has led to an emerging online entertainment industry on platforms such as YouTube and Instagram for local content creators. However, not everyone who aspires to become an ‘internet celebrity ‘influencer’ or ‘content creator’ can succeed. Online presence and self-branding are instrumental to capturing attention online. ‘Attention’ as a concept is highly subjective, and individuals differ profoundly in personal interests, values, beliefs, likes and dislikes. Although various positives emerge from social media usage in terms of alternative livelihood opportunities, individuals must remain critical of social media and its relationship with development and employment. Recently, an advertisement in the state for the position of ‘project fellow’ and ‘project associate’ required individuals to ‘shoot a video demonstrating why they were the right candidate; thereafter, individuals were required to post the video on Facebook and Instagram’. The above instance is a PRIME example of how social media has seeped into employment spaces. Similarities can be drawn between the application process and the operations of brand influencers on social media. Both rely on acquiring enough social media capital to secure a paid partnership. Individuals in the state have different relationships with technology, and the digital divide is multifaceted. The digital divide is not only simply determined by ‘access’ in terms of the haves and the have-nots. Multiple inequities exist within access to digital technology, and often structural offline inequities are carried over to the online world. The idea of equating value to social media presence and online capital is capitalistic and follows a neoliberal philosophy which is dictated by market forces. This discredits other skill sets such as critical thinking, community building and bottom-up knowledge production and obsesses over self-presentation, branding and visual aesthetics. The trend of viewing social media content with wide viewership as the ‘social’ norm stems from the collectiveillusion that social media generates. People will have to be critical of the success of these projects and must not solely rely on what is curated on social media. Social media is not always a projection of social or economic reality and the ‘success stories of certain government initiatives are not always replicable on a large scale. As projects begin to recruit individuals with a strong social media presence; citizens will have to be wary of this trend as this could lead to meshing governance with digital entertainment. The Immediate Future The current government has prioritised investing heavily in the visual affordances of social media. This reliance on visual culture can often create an illusion of change and progress. With the state elections coming in 2023, it will be interesting to see if politicians tap into the skill sets of content creators and influencers as a means of strengthening their online presence to grasp the attention of voters through visual optics. Hence, citizens will need to critically evaluate progress through the ‘offline’ rather than depending on ‘online’ mirages. Download Full Post