The aim of the paper is to restate the problem of language dependence as first invented by D. Miller against Tichý’s approach to the problem of verisimilitude. The question is whether the verisimilitude appraisals can be dependent on language in which they are formulated in the sense that this dependence could determine their truth values. If the answer is “Yes”, one of the consequences is that one language, when compared with some other one, may lead to different verisimilitude appraisals in spite of the fact that the situation to be compared is the same. So the answer has to be “No”. The author of the paper tries to explain this by analyzing Miller’s critique of Tichý’s counting method which was a part of Tichý’s theory of verisimilitude.