This research investigates the sociolinguistic patterns characterizing the polarized climate change debate on Facebook, focusing on the communication dynamics occurring within pro-climate action and anti-climate action stakeholders. Our study specifically aims to (1) identify the variations in language codes among these groups, and (2) assess how these linguistic nuances affect the respective audiences. For this goal, we compiled a comprehensive list of relevant English-speaking stakeholders in the climate debate and collected over 2000 of their posts spanning several months. To analyse the textual content they produced, we defined a series of quantitative language code indicators, measuring the readability, concreteness, subjectivity and scientificity of the language used, alongside topic modeling to dissect the discussion themes. Furthermore, we applied regression modeling to assess the impact of language code variations on the audience responses of the two debate groups. The results revealed significant variations in audience reactions across the debate spectrum, with the pro-climate audience responding more to variations in language style, whereas the anti-climate audience exhibited a distinct response to shifts in topic focus.