The paper presents multi-layer social-semiotic networks as a framework for modeling biases of knowledge communication by means of web-based Communication and Information Technologies (CIT). The network model is exemplified by a range of language-specific releases of Wikipedia in conjunction with a selected set of Wiktionarys. One focus of the paper concerns the underlying social dynamics of CIT. By example of Wiktionary, it examines the existence of a skewed participation dynamics according to which a small set of authors dominates the content of this open, online, collaborative editing system. The paper analyzes the association of users, topics, and lexical knowledge thereby hinting at the Zipfian nature of their tripartite networking. This is complemented by examining a range of fields of knowledge in multiple languages as being manifested by Wikipedia. Using this data, the paper performs a difference analysis showing how languages deviate from each other in representing the same knowledge networks. To this end, it employs methods of Natural Language Processing (NLP), (social) network analysis and computational linguistics.