We analyze the effect on the yield curve of Banco de la Republica's communication through two specific outlets, the minutes of the monetary policy meetings and the inflation reports during the period 2011-Q2 to 2018-Q4. We extract numeric information from the inflation reports' fan charts, and narrative information -using Latent Dirichlet Allocation, a computational linguistics tool- from the text of both outlets. We use an event-study approach to analyze the impact on four specific maturities: one-year spot, three-year forward, five-year forward and five-year ahead five-year forward rates. We find no evidence that numeric information has any effect on market yields. Regarding narrative variables we find that (i) for the inflation report, there is a significant effect on just two yields (one-year spot and five-year forward), and (ii) for the minute, there is a significant effect on all yields. We believe that these results may be explained by the publication lag of the inflation report during the period of analysis.