We present the results of the experiment investigating the mechanism of phonetic ambiguity resolution. A cross-modal priming paradigm was employed. This study takes into account factors potentially influencing the results but irrelevant to the issue itself: a word recognition task (not a lexical decision) was used; the measurements of the same word recognition in control and experimental conditions were compared; the semantic relatedness between the prime and the target word was provided by their common lexical root. The results are discussed from the perspective of activation/suppression approach to the meaning selection process in comparison with the positive/negative choice theory explanation. In contrast to activation-based models, the negative choice is not about the gradual suppression of the activated contextually irrelevant interpretation but aims at keeping it from becoming conscious within the current context. The obtained data is more consistent with the latter explanation and gives rise to some new research questions.