Abstract
Commencing in the sixteenth century, texts in romanized Japanese by the Jesuits present us with the earliest largely consistent romanization scheme for the language. The value of such transcriptions for the study of Japanese phonetics and phonology through time is generally acknowledged. At the same time, these texts also invite us to reflect on the still vexing issue of spacing between orthographic words. What were the basic rules established by the Jesuits, how can we explain exceptions to these rules, and what are the sometimes conflicting motivations behind their approach?
The remarkable change in the spacing of enclitics in the middle of the Jesuits’ Feiqe no monogatari (1592) has long captured scholarly attention. In-depth studies of the spacing behavior seen in different works began with Inoue (1968) at the latest. Studies in more recent years (e.g., Chiba 2009) have achieved considerable progress in that they not only provide detailed statistics but also consider related phenomena such as hyphenation and take a larger text corpus into account. Research to date, however, lacks transparency in that the underlying (digital) text data usually remains unpublished, making claims difficult to verify independently. Based on a quantitative analysis of the freely accessible digital text of the Contemptus mundi (1596), this paper offers an overview of its manifold approaches to providing readers with segmentation cues. The result is contrasted with the earliest extant grammatical descriptions of Japanese in the Arte da lingoa de Iapam (1604–1608) by João Rodriguez.
https://ror.org/04tsk2644 
