Abstract
The paper describes comparability of normalised lexical frequencies based on comparable corpora. Statistical significance measures are used for the evaluation and both possibilities and limitations of any frequency-based approach are explored.