Abstract
This work consists of a corpus-based study on the frequency of the use of numerals in Lithuanian and Latvian. The collected data display the following features: a) the high frequency of the lowest numerals; b) a generally decreasing trend from the first to the last element of each series (units, tens, etc.); c) peaks of frequency often corresponding to important structural positions in the system. In line with the data available for other languages, these features seem to confirm that the universal cognitive abilities and limitations of the speakers play a major role in structuring the frequency of the use of numerals. At the same time, cultural factors are also involved in the process of creating a cognitive hierarchy among number concepts. I suggest distinguishing between innate (cognitive) and non-innate (linguistic and cultural) saliency. The latter is responsible for the identification of reference numbers, thus giving greater credence to the partially relativistic hypothesis that the linguistic system influences the mental mapping of the speakers in the cognitive domain of numbers.