Inferred Language Our inferred language reports look at the Title and Meta Description of a website. If that string of text is longer than 30 characters we will try and infer the language from it.If it is longer than 10 characters and contains unicode character blocks from Hangul, Katakana etc.. we will also characterize them.Inferring a language is different from our technology based language tracking that looks for technical attributes in the code. Instead, this looks at the actual text content of the site as we've found that language codes are sometimes missing and sometimes incorrectly set.Our Inferred language tracking also suffers from inconsistencies, a language might be identified as Serbian when it is actually Russian for example but is a best guess based on the site content and can be used to help identify sites in specific regions.