<p>Unlike earlier times, the modern era is seeing the common man’s voice and opinions being accessible to the world through digital social media platforms such as Facebook, Twitter (now X), Instagram, etc., thereby making it very relevant and impactful to be considered by all nations in affairs like socio-economic planning, opinion making and formulation, policy making, law and order, instant feedback on relevant issues of public importance, etc. Hence, sentiment analysis emerged as a very sought-after research area to cater to this need, but it has its own limitations. This paper provides a scoping review on sentiment analysis in low-resource languages from the prism of two of its key challenges, i.e., Fake news or rumour detection and Sarcasm or Irony detection. The review elaborates on the state-of-the-art machine learning (ML) and deep learning (DL) techniques used for fake news/rumour and sarcasm/irony detection in existing literature for low-resource languages along with the performance parameters, techniques used, and their limitations. It is identified that there is significant research work which is done in fake news detection using ML and DL techniques on low-resource languages but existing literature has very little to offer when it comes to sarcasm detection for such languages. The study reveals several research gaps and highlights the future direction for fake news, rumour, and sarcasm detection in low-resource languages.</p>