Food security is a major concern in many countries in West Africa, particularly Burkina Faso. Early warning systems for food security and famines rely primarily on numerical data for analysis, while textual data, which is more complex to process, is seldom used. In this paper, we propose a textual analysis approach using text mining techniques on French language corpus (press papers and YouTube video transcripts) to enrich the lexicon related to food security. This study involves the extraction of food security domain terminology to enhance surveillance systems. The process is conducted in three steps: initially, relevant documents are selected based on an expert lexicon; then, terms are extracted from these pertinent documents; finally, the initial lexicon is enriched using the newly extracted terms. This methodology ensures a comprehensive and up-to-date lexicon for improved food security monitoring and analysis.