Three experiments were conducted to test the psychological relevance of objectively quantified word collocations. The first experiment showed that perceived frequency of word combinations roughly followed the objective count. Another recurrent quality of words, constructional tendency, was supplemented as independent variable in the two following experiments. This variable reflects a word's tendency to appear in word combinations and it was found to interact with frequency when subjects rated frequency and comprehensibility. The experiments showed that word collocations, defined at the levels of combinations and constructional tendency of individual words, can be supposed to have psychological counterparts. The linguistic recurrence seems to have cognitive representations. (Author)