We present a word extraction method for the unique wording of each language for lyrics data of different languages in the same song, and its application to decipher cultural nuances in dynamic equivalence. In general, one of the most important tasks is to decipher the differences in cultural nuances between languages for lyrics data in different languages in the same song. To decipher these differences, it is necessary to establish a metric for extracting words with phrases specific to each language, using the example of Let It Go from Frozen, which has been sung in 48 different languages. In this method, words are extracted from lyrics data in each language and a Specific Feature Extraction Score is derived as a measure of the uniqueness of each word in that language. By proposing and introducing an index for extracting words with language-specific phraseology with reference to the F-score of each word in that language, it is possible to extract the language-specific phraseology of each language. By implementing this method, it is possible to derive specific phrases from the lyrics data of each language, and to decipher the differences in cultural nuance between languages by deriving specific phrases from the lyrics data in each language.