This chapter explores how sentiment analysis can be applied to unstructured consumer feedback—such as free comments in answer to open-ended questions, reviews, and social media posts—to extract affective meaning relevant to sensory and consumer science. It begins by reviewing lexicon-based approaches, which use predefined emotional vocabularies to interpret sentiment in text. These methods are straightforward and interpretable but often fall short in managing nuance and contextual variation in natural language. The chapter then turns to data-driven approaches, including classical machine learning (unsupervised, supervised and distantly supervised techniques), deep learning, and large language models, which learn patterns of affective expression directly from large datasets. These methods offer greater adaptability, especially when analysing complex, real-world consumer narratives. Hybrid strategies that combine the transparency of rule-based systems with the learning capacity of model-driven techniques are also examined. Throughout, the focus remains on the challenges of interpreting subjective, culturally variable, and often implicit expressions of affective reactions or liking in text. The chapter concludes by discussing ongoing methodological challenges and emerging opportunities, positioning sentiment analysis as a key tool for leveraging the rich, expressive power of consumer text data to better understand sensory experiences and guide product development.