This paper investigates bekio, which is constituted by attaching o to beki, the adnominal form of the auxiliary verb beshi, from the viewpoint of its actual condition of use, chronologically, from the Heian period to the Edo period, and considers the relation of bekio to o as a quasi-conjunctive particle attached to a formal noun. Bekio had been frequently used in the Heian period, but it entered into a transition phase in the Kamakura period, and it almost went out of use in the Muromachi period. However, bekio is still somewhat in use in Modern Japanese by lexical cohesion as a form for expressing adversative relations. Whereas bekio went out of use, tokoroo appeared, and tokoroo tends to be more frequently used than bekio in Modern Japanese. It is considered that the appearance of tokoroo correlates with the decline of bekio, and that it occurred with the transformation from ancient Japanese to modern Japanese, as beshi, the conjunctive particle o and quasi-nominal phrases had declined in the Muromachi period.

Proeve eener Japansche Spraakkunst (1857) is a Japanese grammar compiled in Dutch by J.H. Donker Curtius, head of the Dutch factory on Deshima, and edited by Prof. J.J. Hoffmann. It has thus far mostly been treated as a precursor of Hoffmann's own Japansche Spraakleer/A Japanese Grammar (1867-1868), but its value as a Nagasaki dialect source has also been noted. Due to mistranslations of Hoffmann's foreword and a lack of attention to related historical materials, however, many aspects of the background and process of its compilation have remained insufficiently understood. I aim to remedy this by discussing the results of my investigation of the following materials:

1) The manuscript Donker Curtius originally submitted to the Dutch government, as well as one of Hoffmann's personal copies of the published work, replete with his handwritten annotations. Both are kept at the Leiden University Library (shelfmarks LTK 593 and 864 C 4).

2) The archives of the Dutch factory in Japan (Nederlandse Factorij in Japan, 1.04.21) and the Ministry of Colonies (Ministerie van Koloniën, 2.10.02), both kept at the National Archives in the Hague.

Based on these materials, I argue that Donker Curtius was assisted in compiling his grammar mostly by his Japanese servants on Deshima, whereas the Japanese interpreters (tsūji) helped in correcting the manuscript. I also point out that Donker Curtius may have adopted some vocabulary items from the Japanese-Dutch dictionary Rango Yakusen/Nieuw Verzameld Japans en Hollandsch Woordenboek (1810). Additionally, I report the discovery of copies of a number of pages of Donker Curtius's lost manuscript Dutch-Japanese dictionary, a copy of senior interpreter Namura Hachiemon's laudatory declaration regarding Donker Curtius's grammar manuscript, and 49 pages of Katsu Kaishū's autograph comments on this manuscript. Katsu, then a naval student at Nagasaki, wrote these comments in Dutch.

This paper investigates haiwaiqitan (海外奇談), a Japanese-Chinese transliteration, anthologized in wuqijingbu (吾妻鏡補) and assumed to reflect the Nagasaki dialect of the 18th century. This paper argues that the sounds of the Chinese characters, which are applied to transliterating in the document, are Wu. By analyzing them, this paper examines the consonants of the Japanese /h-/ series.

In haiwaiqitan, the consonants in /ha/ and /ho/ are mainly transliterated as the houyin-hekou (喉音合口, a guttural or glottal initial with a rounded medial or vowel) type, while those in /hi/, /hu/ and /he/ are mainly transliterated as the chunyin (唇音, a labial initial) type. In the wide range of Japanese-Wu transliteration documents, this paper confirms the relationship between different types of Chinese characters and the consonants of the Japanese /h-/ series, focusing on the positioning of the houyin-hekou type. The results are as follows:

(1) The chunyin type suggests that /h/ at that time was probably [ɸ].

(2) The houyin-kaikou (喉音開口, a guttural or glottal initial without a rounded medial or vowel) type suggests that /h/ was probably [h(ç)].

(3) The houyin-hekou type suggests that /h/ was probably in the intermediate state of de-labialization [ɸ]>[h]. With the help of modern Wu sounds, this paper presumes that the phonetic value of this intermediate state is [hw].

Accordingly, this paper concludes that haiwaiqitan reflects the de-labialization of the consonants of the Japanese /h-/ series that occurred in the Nagasaki dialect in the 18th century. At that time, the consonants in /hi/, /hu/ and /he/ mainly still maintained [ɸ], while those in /ha/ and /ho/ were in the intermediate state of de-labialization, and the phonetic value is [hw].

This study investigated the usage trends of discourse markers (DM) in Japanese for initiating narratives in casual conversations based on the BTSJ Japanese Natural Conversation Corpus with Transcripts and Recordings (2020), and clarified the following trends:

(1) The number of DMs varies from zero to five and, on average, one DM is used.

(2) In initiating a narrative, nanka is most often used as a linguistic means to convey uncertain events of the past or while searching for a word. When a speaker initiates a narrative, the expression sō is frequently used to show that he or she came up with an event, and demo and de are used to demonstrate an “adversative” or “additive” conjunctive relation with the preceding context, or to change the topic. In initiating a narrative by responding to a question from an interlocutor, interjections like a, ā, un, ano(ō), and iya are often used to express reactions or thoughts.

(3) An analysis of the order in which multiple DMs were used revealed the following pattern: i) “reaction to the preceding context” → ii) “recalling of a narrative” → iii) “showing the connection between the preceding context and the narrative” → iv) “expression of an attitude toward the narrative.” Furthermore, DMs representing a “search” appear anywhere from before i) to after iv).

Speakers use DMs heavily to emphasize their attitude, buy time to think of appropriate words, or repair the content of an utterance. The development of the textual and interpersonal functions of DMs by grammaticalization also leads to their heavy usage.