TEI 2019

What is text, really? TEI and beyond

All PapersTEI

An Encoding Strategic Proposal of “Ruby” Texts: Examples from Japanese Texts

Kazuhiro Okada, Satoru Nakamura, Kiyonori Nagasaki

Keywords: Japanese texts, ruby encoding, glosses
Slides: https://docs.google.com/presentation/d/1CNQQSbpdXoE9UzyeRqFw1AkXYyCC6oiz6BwGU2DoRo0/view
Permalink: https://gams.uni-graz.at/o:tei2019.155

We present a novel encoding strategy in ruby texts in Japanese. We propose some elements to encode ruby texts, referring to existing ruby encoding in other schemata: ruby, rb, and rt. Our model differs from existing models in fulfilling historical complex ruby attachments.

In Japanese texts, ruby text is a run of text in phonetic scripts attached to a particular portion of the main body of the line (generally Chinese characters) in order to represent the reading of the portion (W3C, 2001, inter alia). It is named after the body size name of it in the moveable type printing. It is also called furigana as it is allocated to the main texts, which originally developed in East Asian vernacular glossing culture to the Classical Chinese texts.

Ruby today is not, however, a mere annotation to the line: it is more stuck to the particular characters than simple interlinear glosses and, more importantly, it should be read in parallel with the main line, or even as if it were the main line itself. Thus, ruby can be used not only as a simple guide of pronunciation but also as an alternative to the main text. Historically speaking, ruby was also attached to both sides of the main text in order to denote both reading and gloss, especially for borrowing words: even in this case, readers may read any side of ruby in accordance with their preferences. Besides, each side can correspond to different spans of text. These ruby functions are not confined to a simple gloss, which can be encoded within the current TEI framework, and they deserve to be encoded in its own way.

One can find such a two-sided example where a word 打球場, or billiard hall, is attached two rubies (#fig1): "ビリヤード" (phonetic transcription of the word billiard in Japanese katakana) to the whole, and ダキウ (phonetic transcription of the word 打球 in Japanese katakana) to the first two characters.

From f. 28v of Niwa
                     (1878) and its modern rendering. Public Domain.
From f. 28v of Niwa (1878) and its modern rendering. Public Domain.
With this case, we will propose to encode this as follows: 打球 ダキウ 場 ビリヤード
An Encoding Example of the Figure 1.
Here, following existing conventions, ruby element denotes the whole ruby environments; rb element denotes the body texts, where nested ruby elements are allowed inside to encode two-sided rubies; and rt element denotes the ruby texts, where type describes whether the ruby text in question is primary (p) or secondary (s). Primary and secondary rubies are determined according to the text direction: in vertical right-to-left writing, primary rubies are on the right, whereas horizontal left-to-right writing, they will be on the top.

Thus, our encoding strategy encompasses historical varieties. Since ruby has long been one of the major concerns in Japanese texts, its encoding strategy has naturally been proposed from its early days, including W3C’s ruby specifications (W3C, 2001; 2013a; 2013b, Hara & Yasunaga, 2002, WhatWG, 2019). Their considerations are quite broad and informative, regretfully limited to modern usages or either confused on the inclusion of interlinear gloss as ruby solely based on layout similarities. [This is because the existing ruby encodings have focused in realizing the contemporary ruby layouts, especially presented as JIS (2004) or more accessible W3C (2012). See Kawabata (2014) for more references.] Our proposal differentiates interlinear glosses to be encoded by the gloss element, which are attached to a particular place, rather than a particular part of the body (See Estill, 2016 for complexities in encoding of interlinear glosses). This distinguation is of necessary to an encoding to be more proper and simpler.


Hara, S., & Yasunaga, H. (2002). 国文学研究支援のためのSGML/XMSデータシステム: 国文学データ共有のための標準化. 情報知識学会誌. 11 (4): 17–35. doi:10.2964/jsik_KJ00001039453.

JIS (2004). 日本語文書の組版方法. JIS X 4051. Tokyo: Japanese Industrial Standards Committee.

Kawabata, T. (2014). HTMLのルビ標準化の現状と課題. 漢字文献情報処理研究. 15: 4–15.

Estill, L. (2016). Encoding the Edge: Manuscript Marginalia and the TEI. Digital Literary Studies. 1 (1). https://journals.psu.edu/dls/article/view/59715.

Niwa, J. (1878). 竜動新繁昌記 初編. doi:10.11501/767340.

WhatWG (2019). Text-Level Semantics. HTML Living Standard. 18 July 2019. https://html.spec.whatwg.org/multipage/text-level-semantics.html.

W3C (2001). Ruby Annotation. https://www.w3.org/TR/2001/REC-ruby-20010531/.

W3C (2012). Requirements for Japanese Text Layout. https://www.w3.org/TR/jlreq/.

W3C (2013a). HTML Ruby Markup Extensions. 25 February 2013. http://darobin.github.io/html-ruby/snapshot20130225.html.

W3C (2013b). Use Cases & Exploratory Approaches for Ruby Markup. https://www.w3.org/TR/ruby-use-cases/.