The four forms of the adverb yahari (yahari, yappari, yappashi and yappa) are used in parallel in native-speaker conversations without being integrated into any one form. Although there have been detailed reports on the number of occurrences of each, there is not enough data on the number of people who use these forms. Therefore, in this paper, we investigate the actual usage of the adverb yahari using the Corpus of Everyday Japanese Conversation (CEJC), focusing on the number of people who used the adverb yahari. As a result, the following three points were observed.

(1) Yappari was used the most, by 80% of the respondents, not only in terms of number of times the word was used but also in terms of the number of people who used it. Yappa is used not only by young people but also by all generations.

(2) The number of respondents who use only one of the four forms of the adverb yahari is greater than 40% for all generations.

(3) In the distinguished use of yappari and yappa, the occurrence of yappari is significantly higher in one-word sentences and yappa is significantly higher in clause heads. The nucleus-free type yappa appears more frequently to mean “as expected”, while the head-high type yappa appears more frequently to mean “in the middle of form/content selection”.

Traditionally, the investigation of the “poetic style” inherent in a collection of Waka was centered around the study of vocabulary distribution and bias. However, with the recent advancements in machine learning, research utilizing word embedding vectors such as word2vec has become feasible. In this study, we demonstrated that by applying principal component analysis to sentence embedding vectors generated by large language models, we can describe the semantic system at the core of Waka's poetic style. Moreover, we observed substantial differences between the “Kokin Wakashū” and “Man'yōshū” and discussed the potential influence of Chinese poetry on these variations.