Lurie 6: kundoku and kunten vis-a-vis Frellesvig, Whitman, and

 Well, as fortune would have it, the work I was expecting to receive this evening didn't arrive, so my rush to get this post up was unnecessary. At any rate, I printed the post out and edited it. There is still some redundancy and I'm not certain about some of the description and characterizations I've made, but this is basically the final form of this post. Thank you.

 Once again, this needs to be printed out and proofed, but I have put too much time into it already (@7,300 words) and need to attend to neglected matters—like work… So I’m throwing this up on the blog as is, and will get around to editing it within a week.

 In the interest of wrapping up this topic for the time being, I’m directing some comments and observations with respect to relevant passages from the recent book by Frellesvig, A History of the Japanese Language, and some information from papers by Whitman and Jorgensen. Reading the work of the aforementioned scholars has spurred me to clarify and further develop ideas that I have articulated in a more nascent form in previous posts.

 If I might be so bold as to affront Harvard, it seems to me that by and large, the referenced writings of Frellesvig and Whitman tend to be aligned with the positions I have staked out as opposed to those espoused by Lurie. 

Furthermore, since I am an America, I would venture to say that John Quincy Adams and Theodore Roosevelt would not approve of the recent trajectory of Harvard. 

In a parallel universe I might have time to go back and integrate all of these entries into a combined book review, reviewing both Lurie and Frellesvig’s books together, or focus a little effort on the relatively unexplored vast stores of kunten materials...

There is a link to Frellesvig's book in the preceding post on this topic. Here are links to the papers that are also discussed below:

John Whitman, etc.

http://conf.ling.cornell.edu/japanese_historical_linguistics/

http://conf.ling.cornell.edu/japanese_historical_linguistics/Scripta3.John_Whitman_1st.pdf

John Jorgensen:

http://www.google.co.jp/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&ved=0CC8QFjAB&url=http%3A%2F%2Fbook.aks.ac.kr%2Flib%2Fdown2.asp%3Fidx%3D2484&ei=I6RqULirMfDImAW1_IGwBg&usg=AFQjCNGE8Bq96vi5bYdoVibrGgTfbC71xQ&sig2=ug1oonPJBH7IPEAXWtts0Q 

p. 258:

… an important aspect of kanbun-kundoku is the notion that it involves verbalizing the original Chinese text in Japanese, and it is popularly thought of as ‘reading’ Chinese text ‘in Japanese’ or ‘with Japanese grammar’. The practice of kanbun-kundoku, understood as ‘the “reading” of Chinese in a local vernacular language’, is not restricted to Japan, but is a common feature of the cultures within the Sinitic cultural sphere, attested and described in the sixth and seventh centuries from places as far-flung as Japan, the Korean peninsula, Vietnam, and Gao Chang…

This is a bit of an aside to a peripheral topic addressed in relation to Lurie’s book, but note that Frellesvig also chooses to use the term “sphere” with respect to the areas and cultures influenced by Chinese writing and culture, adopting the phrase “Sinitic cultural sphere”. Furthermore, Whitman also uses the term sphere, incorporating it into the neologism, “Sinosphere”, which he defines, on p. 2 of “The Ubiquity of the Gloss”, as:

(those places where Chinese was the language of writing, but not of native speaking)

Fellesvig continues on p. 259:

… Thus, kanbun-kundoku involves finding suitable translation equivalents in Japanese for content and function words in the Chinese texts. However, in addition kanbun-kundoku involves a number of processes in order to render (‘read’) the Chinese text in Japanese: transposition (change of word order) and interpolation or specification (of inflectional morphemes or grammatical particles).

In the above-quoted passage, Frellesvig points to what I consider to represent two distinct phases of what I have characterized as “a problematic” in contrast to Lurie’s description of “a compatibility” (refer to earlier post) with respect to the development of the Japanese writing system starting with the adoption of Chinese characters.

Frellesvig continues on p. 259:

9.1.1   Kunten

Kunten (訓点’reading marks, glosses’) is a cover term for a variety of annotations added to Chinese text in order to aid these processes of its rendition in Japanese. The earliest extant kunten, from the late eight century, are punctuation marks, showing phrasing and division of a text, and marks showing how to change the word order when rendering texts in Japanese.

… The development of katakana is closely linked to the practice of glossing, and katakana is traditionally viewed as a type of kunten.

Here, it is necessary to draw attention to the fact that Frellesvig indicates that “katakana is “closely related to the practice of glossing” and “traditionally viewed as a type of kunten”, while describing the earliest materials as being limited to “marks” delineating syntactical structures, and no phonographs (e.g., katakana). Meanwhile, Lurie, in characterizing kundoku “as a method of writing” declares that the use of phonographs represented another, separate method of writing falling outside the scope of his proposed model of kundoku as “a system of writing”.

If we take Frellesvig’s assertion as a statement of fact, that is to say, as a statement that attests to the facticity of what role the Japanese understood themselves to be applying katakana in the eight century, it follows that the Japanese took katakana to be a type of kunten; accordingly, it is incumbent upon scholars to consider the verity of that fact, insofar as it is historically incontrovertible, in the context of its occurrence.

In the following passage, Lurie would seem to imply that all use of phonographs was excluded from his proposed kundoku model as a “method of writing”.

On pp.205-7, Lurie writes:

Kundoku was not the only method of writing I early Japan. In adapting the ‘Chinese’ script to inscribe the Japanese language, another strategy was to use the syllabic values of the Chinese graphs as phonographs for similar Japanese syllables.

However, on p. 184, Lurie states:

…Chinese-style logographic writings (Buddhist and secular classics imported from China) are accompanied by diacritic transposition marks and phonographic annotations (kunten).

And in a footnote on p. 184, associated with the above-quoted passage, Lurie states:

The earliest known kunten were punctuation marks of clause endings and simple marks showing reading order. Over the course of the ninth century, the emergence of systems of abbreviated phonographs and diacritical markings allowed for more extensive annotation of kundoku readings.

Does Lurie contradict himself in the two passages (and footnote) with regard to the use of katakana (abbreviated phonographs) as kunten? Or does the reference to "another strategy" of using phonographs pertain solely to writings of vernacular Japanese uta (songs, poems), in man'yogana, for example? 

Furthermore, he describes the use of kunten as “annotation of kundoku readings”, not as a “method of writing”. What bearing does that have on his characterization of kundoku as “a method of writing”?

Lurie’s characterization becomes even more unclear in consideration of a statement he makes on p. 210:

… By the Heian period, when extensive kunten diacritic annotation revealed its details, the language of kundoku was very different from that of other styles of writing (as recorded in phonographic and mixed styles).

In the above-quoted passage, kunten is again characterized as annotation, but what are we to make of the characterization of kundoku? For example, what is the referent of the phrase “the language of kundoku”? Furthermore, how does kundoku relate to “other styles of writing”? Lurie gives no indication as to the referents of the phrase “other styles of writing”, either. It would seem that the above-quoted statement does nothing but obfuscate the relationship of kundoku reading practices to writing, in its various forms and stages.

These contradictions are perhaps inevitable in light of the attempt to characterize of kundoku as “a method of writing”. I believe that these contradictions undermine that attempted characterization as a simple conceptual fallacy.

Kunten, as discussed below, is not writing but annotation (or mark-ups).

Moreover, in light of Lurie’s characterization of kundoku with respect to "the history of writing” in terms of having “invisibility” as one of four defining features he enumerates, I will address the rendering of kunten with styli and white ink as yet another historical practice that problematizes Lurie’s characterization of kundoku as “a method of writing”.

On p. 263-4, Frellesvig writes:

In China itself, annotating text for interpretation or pronunciation is well established and has a long history. For example, the shoten (‘tone marks’) mentioned earlier (6.1.2.2) are similar to kunten and are part of the same overall phenomenon… In korea marks to annotate text are called kugyol; they are similar to the Japanese kunten, but the earliest have been thought to date from the ninth century. However, especially with the continuing discovery in both Japan and Korea of increasing amounts of kunten material which are annotated not in ink, but by stylus (角筆, kakuhitu) which leaves indentations or scratchings on the paper, but no color, it is gradually becoming clear that techniques for annotation were used both on the Korean peninsula and in Japan at an earlier time than was previously thought, with the oldest such stylus materials in Korea dating from the late seventh century. It now in fact seems overwhelmingly likely that kunten techniques, too, like the Chinese writing and text and kanbun-kundoku were transmitted from the Korean peninsula to Japan. For example, the earliest Japanese materials are far more similar to the Korean materials than are later Japanese materials. 

What does Lurie say about the use of the “stylus” to annotate Chinese texts. Lurie does not mention the topic in the main text of his book, but in an endnote (4.19) on p. 397, he writes:

… An important element of this area of research is the role played by glosses made with a stylus (J. kakuhitsu 角筆) to scratch marks into the surface of the paper. Since the 1960s scholars had been aware that there were numerous Japanese manuscripts with kunten notations made in this manner, but it is only in the past decade that it has become clear that the practice was also widespread in premodern Korea... Most of the materials are later than the period examined in this book, but there are signs that kunten-style glosses were associated with Huayan (J. Kegon 華厳) Buddhism in eighth-century Silla, which was very influential in Nara-period Japan. It is possible that future discoveries will push back the dates of these practices further, or provide more solid evidence of the extent of Korean influence on their development in Japan, but even at this point it seems likely that the idea, at the very least, of using brushes and styli to add glosses to manuscripts was imported to the archipelago from the peninsula in the eight century or earlier (Whitman 2008).

First, it is interesting that in the above-quoted endnote, Lurie adopts a more modest tone in asserting a Korean contribution to the development of the Japanese writing system.

Second, this endnote is instructive with respect to the statement made by Frellesvig--in the preceding passage quoted above--that early Japanese examples of kunten are much more similar to Korean examples than later Japanese examples. That clearly indicates that there was a gradual learning process involved in developing the Japanese writing system over the span of several hundred years. Accordingly, even if there initially had been early input of various sorts from Korea, Lurie’s statement in the above-quoted endnote appealing for recognition that at least "the idea, at the very least, of using styli and brushes to add glosses to manuscripts..." was introduced from Korea is a stark contrast to various exaggerated statements he makes attributing to the Korean contribution an excessive degree of importance, including the statement in the main text (p. 203) to the effect that the writing system was “pre-adapted” by Korean scribes before reaching Japan.

Though I have already quoted the crucial sentence from that passage in an earlier post, the preceding sentence is of particular relevance in the context of the present discussion, so I will quote that followed by the next sentence. On p. 203, Lurie writes:

    This means that we need to revise the traditional assumption that Chinese-language writing was gradually adapted to the Japanese language. Like Buddhism, the technology of writing took off in seventh-century Japan because it was already pre-adapted to both Sinitic and non-Sinitic environments.

Here, it has to be pointed out that while Frellesvig appeals to the existence of similarity between early Japanese examples of kunten and the Korean model of using kunten as evidence that the practice was introduced from Korea, conversely, in pointing out that the Japanese practice of using kunten thereafter diverged from the Korean model, Frellesvig’s statement also supports my assertion that the Japanese applied these techniques in order to solve specific problems as they were encountered along the way to developing higher competence in reading and writing. In other words, “the technology of writing” was not “pre-adapted”, to return to Lurie’s characterization, and it was in fact gradually developed, technique by technique, culminating in the fruition of the Japanese writing system (wabun).

On the other hand, though Lurie does not take up the topic of stylus marks in the main text whatsoever, he does assert the “invisibility” of kundoku as “a method of writing”. That is telling, once again, insofar as the kunten markups performed with stylus or white ink were carried out in a manner such as to preserve the literary Chinese text in its original form; this would seem to indicate that those doing the marking up did not consider kunten to be part of the text per se. Accordingly, if one were intent on attempting to assert some sort of “invisibility” of kundoku as “a method of writing”, it would appear that the issue of using a stylus or white ink to mark Chinese texts for reading according to kundoku would have to be addressed. Furthermore, since those marks are capable of being considered to be invisible, in the above-described sense, their existence would likely give one pause in attempting to assert "invisibility" without taking them into account.

In any case, it might be possible to point out the discrepancy by asking a simple question:

How does Lurie define the relationship between kundoku as "a method of writing" and kunten?

I have pointed out that it is erroneous in terms of the etymology of the Chinese characters forming the character compound "kun[訓]doku[読]" to characterize kundoku as a method of writing, and further evidence, via Frellesvig, found in passages of the Kojiki would seem to support that, as shown below. 

Lurie refers to kunten as annotation a number times, so would not appear to be the case that he intends to characterize kunten itself as writing. On the other hand, I don't recall that Lurie provides an account of the relationship between kundoku as "a method of writing" (a crucial point) to kunten. 

If not for the confusing statements I quoted above, I might be inclined to search the text of his book a little further, but there are other topics awaiting the clatter of the keyboard.

Frellesvig continues:

9.1.2   Kanbun-kundoku and writing in Japanese
A close relation holds between kanbun-kundoku and writing in Japanese. In the course of kanbun-kundoku, fixed habitual renditions of individual kanji arose, resulting in conventional associations of many kanji with specific OJ words, in other words, the establishment of conventional ‘kun-readings’ of kanji.

The above-quoted passage addresses the issue of lexical glossing. Here, Frellesvig points to a close relation between kundoku reading of kanbun literary Chinese texts and the development of Japanese writing with respect to the accumulation of a Japanese glossary of vernacular terms associated with respective Chinese characters, in terms of “conventional kun-readings’ of kanji”. Though it is somewhat unclear as to what exactly he is referring to as “writing in Japanese”, there is no question that the glossing of Chinese characters in vernacular Japanese required some sort of visual marking or annotation (i.e., kunten, as found with kanbun texts that were marked up with kunten to facilitate their verbalization in spoken Japanese), even after some ‘kun-readings’ became conventional, because of the problem that there are multiple glosses for many characters, particularly characters representing commonly used verbs. Even in the modern Japanese writing system (wabun), the reading of a single character used as a verb is indicated by hiragana characters provided as a superscript in smaller print over the verb.

Continuing with the next sentence of the above-quoted paragraph, Frellesvig states:

Once this association of decoding (reading) was established, the next step of reversing the relation to one of encoding (writing) was not a big one.

Bearing in mind that this sentence follows the preceding sentence quoted above, it may be feasible to assume that Frellesvig is focusing on the limited scope of lexical glossing here. Even so, however, my ideas on the processes under examination would seem to diverge somewhat from Frellesvig’s here.

The conceptual jump from reading logographic characters from one language in the spoken vernacular of another language to encoding that speech in an indigenous writing system is not trivial. I would say that perhaps one point to start from, adopting the language of Frellesvig’s above-quoted sentence, would be the question of “encoding”. The Chinese can be said to have encoded their own speech in logographic Chinese characters by composing sentences in literary Chinese syntax exclusively with the logographic characters, but Chinese is a foreign language to the Japanese. For all intents and purposes, it is fair to say that the Japanese could not directly encode their speech with Chinese characters, not in a pracitable manner, at any rate. 

In this respect the use of kunten markups can be seen to represent an intermediate stage, wherein the kunten facilitated the glossing of individual Chinese characters into corresponding Japanese words as well as the parsing of literary Chinese syntax into vernacular Japanese syntax, which in turn afforded the Japanese ample opportunity to learn about the grammatical structure of their own language vis-a-vis the contrast with literary Chinese. Only after the Japanese had attained a somewhat comprehensive understanding of their own grammar could they undertake the task of trying to represent Japanese speech directly, without using the intermediary techniques associated with kunten.

The addition of kunten represents the adding of another layer of code to the classical Chinese text that was subject to kundoku gloss reading. In order to decode the classical Chinese text and parse it into Japanese, a further layer of symbolic representation corresponding to the kun-readings of Chinese characters and the parsing of literary Chinese syntax into spoken Japanese was required. Kunten markups do not directly encode Japanese, but facilitate the decoding of literary Chinese into spoken Japanese by someone literate in Chinese and apprised of the respective kunten protocol.

So it seems to me that “the next step” is not one simple step, but a series of steps, and the gulf between starting point and end point is wide indeed.

Both the process of glossing individual Chinese characters into corresponding Japanese terms and the process of parsing literary Chinese syntax into vernacular Japanese syntax involved the use of kunten (including katakana). Eventually kunten were rendered obsolete by the development of the Japanese (wabun) writing system, which basically transformed the linguistic content represented by the kunten into linguistic content represented primarily by use of the new fully developed phonographic syllabaries (katakana and hiragana). It thereby became possible to represent the full range of semantic-morphological and syntactic-morphological features of Japanese grammar simply with the Japanese writing system proper, including diacritics and punctuation marks. Kunten, on the other hand, continued to be employed in the study of classical Chinese texts to indicate kun-readings and facilitate the parsing of literary Chinese syntax into vernacular Japanese, but the Japanese had succeeded in developing a new writing system for daily use that could fully represent vernacular speech. There were multiple steps to developing the Japanese writing system proper, each of which involved the solving of at least one problem.

Further down on p. 264, Frellesvig makes the following statement, which I have to admit to finding somewhat cryptic.

    The bidirectional reading-writing relationship between kanji and Japanese words and morphemes—and what appears to be an identification of the process of reading and writing—is evident from the use of 訓[kun] in the earliest sources from the Nara period.

Here before moving on to the instructive passages from the Kojiki quoted below by Frellesvig, I’d like to consider briefly his characterization of a “bidirectional reading-writing relationship between kanji and Japanese words and morphemes”. This relates to the notion of ‘decoding’ as reading, too.

If the discussion were limited to the context of writing the sound shapes of Japanese speech using logographic Chinese characters as phonographs, it might be easy to see a relationship akin to a one-to-one correspondence between reading and writing; however, that would not account for kanbun-kundoku. In fact, that is not even what Frellesvig says in the passage. He specifically states that the relationship is “between kanji and Japanese words and morphemes”; therefore, it is clear that we have to contemplate the “bidirectional reading-writing relationship” with respect to kun-readings of logographic Chinese characters used as logographs in Japanese writing.

Although the contrast class of encoding-decoding is useful in illustrating some of the phenomena at hand, it also has limitations. Decoding per say differs from reading in the normal sense insofar as it presupposes an unknown dimension that has to be rendered known before the text is rendered intelligible simply as text; that is to say, before it can be ‘read’ to assess its meaning. Normally, reading is a transparent process that is carried out on the basis of background intelligibility that requires no ‘decoding’ as such in terms of the application of analytical tools in order to carry out reading of the text as text. You the reader of this sentence do not have to decode it because your background knowledge makes the linguistic code embedded in the English language transparent. To produce a complete kunten markup of a classical Chinese text, a person would of necessity have to be able to read the meaning of the Chinese text and comprehend its meaning.

The problem of describing kunten in terms of normal linguistic phenomenon is that it is, in an important sense, outside of normal linguistic phenomenon. Encoding in the sense used by Frellesvig represents the normal representation of spoken language in written language. The ontological status of kunten and kanbun-kundoku in terms of linguistics, however, is in a sort of liminal zone. The Japanese could not encode their speech because they didn’t have a writing system. In order to facilitate the reading of Chinese literary texts in spoken Japanese, the Chinese text first had to be comprehended, and then marked-up for future reference, so as to forego the need to have to analyze the text again at the next reading.

The concept of translation is of use here. Translation represents a kind of re-encoding of a literary text from one written language to another. Since the Japanese didn’t have a writing system, the kunten markups represent an intermediary form of annotation, and it may be useful to think of its function, by way of analogy, as one of enabling the Japanese to interpret literary Chinese texts into spoken Japanese. The term interpretation in the language services field generally relates to rendering speech from one language into the speech of another language, whereas translation, as described above, relates to inter-textual rendering of written language. Considered in these terms, the place of kunten in Japanese is between written and spoken language, where it facilitates the rendering of the written text from one language into the speech of another language.

With regard to reading (verbalizing) a logograph originating in a foreign language in a local vernacular language, the semantic content of the logograph must first be grasped, and its meaning glossed with the corresponding vernacular term. In this regard, it would seem that the concept of ‘glossing’ should be restricted to units of meaning at the semantic level, i.e., words, and phrases. Words have semantic content, that is, meaning, and glossing represents a choosing of a Japanese word or words that approximately corresponds to the meaning of the Chinese character, which is how I generally understand the meaning of gloss.

Rendering a written expression of a complete thought, i.e., a sentence, into another language is not an exercise in glossing, it is translation proper. Accordingly, kunten, which are used to facilitate glossing of individual characters as well as enable the syntax of the literary Chinese text to be transposed into the syntax of spoken Japanese, is annotation, not writing per se. The fact that it was often rendered by stylus or in white ink emphasizes the fact that the annotators themselves did not consider it to be part of the text proper. So it may be proper to refer to it as a practice of glossing, even though it relates primarily to the transposition of syntactical structure, and not to rendering approximations of corresponding meanings per se. But that begs the question of whether there is inherent meaning in syntax, and what happens to the meaning of a sentence when syntactical structures are transposed. That is a question beyond both the scope of this blog and my competence to address it without further study.

My concern is that extending the concept of glossing to cover syntax and grammatically complex units of meaning larger than simple phrases deprives us of a conceptual distinction that I find useful for analyzing these questions and framing the answers. It is important that we are able to discriminate between different stages in the development of the Japanese writing system per se. It could be said to be simply a matter of conceptual categories, but those serve as tools with which to organize and interrelate the pertinent facts and factors involved in a coherent manner.

I make this point simply because my understanding diverges somewhat from how Frellesvig characterizes the phenomena, and because John Whitman prefaces his paper “The Ubiquity of the Gloss” with a disclaimer wherein he acknowledges, in advance, mind you, that he:

…uses the term “glossing” in a somewhat extended sense to refer to a process where a text in one language is prepared (annotated, marked) to be read in another language.

In his preface, Whitman points in the right direction by using the term “marked” next to “annotated” to define kunten; as does Frellesvig, incidentally, in the opening sentence of section 9.11 of his book (p.259, quoted above), in terms of "reading marks, glosses". With respect to the marks used to indicate syntax, as opposed to phonographs (e.g., katakana) used to indicate readings (glosses), kunten can be more aptly characterized as something like a markup language.

The markup language HTML is familiar to many, so I will illustrate the analogy using HTML. Although my programming terminology may be somewhat inaccurate, HTML is basically a set of symbols with which content code is annotated in order to enable a Web browser to parse the code for the purposes of displaying the Web content that has been marked up with HTML. Kunten represents a set of symbols with which Chinese texts were annotated in order to enable a Japanese person that was literate in Chinese to parse the Chinese text into the syntax of vernacular Japanese in order to verbalize the text in spoken Japanese.

This relates to the assertion of a “bidirectional reading-writing relationship between kanji and Japanese” as well as “an identification of the process of reading and writing”. First, it seems to me that once the Japanese attained the ability to render the entire scope of literary Chinese syntax parsable into spoken Japanese with kunten markups, that signals the arrival at stage at which they had, as a result of the processes of decoding the literary Chinese text, etc., achieved enough knowledge about the syntax and morphology of Japanese per se to then reapply that knowledge to developing a system to write (encode) Japanese speech directly in Japanese writing, while making use of the Chinese characters and body of inter-lingual knowledge that had been accumulated over the course of learning the Chinese writing system and how to apply it.

In other words, though the Japanese did not start out studying Chinese texts with the notion of developing a Japanese writing system, once they had inadvertently, as it were, attained a full understanding of the grammar of their own spoken language through the cumbersome exercise of decoding and parsing Chinese texts into spoken Japanese syntax through use of kunten markups, they had basically accumulated enough knowledge to begin writing Japanese language directly in another medium (script). Moreover, doing so basically made sense in terms both of the goals of efficiency and promoting literacy.

Accordingly, rather than a “bi-directional reading-writing relationship” or an “identification of the process of reading and writing”, I see the phenomena as an unfolding of a multistage process that culminates in the realization of the Japanese writing system (for writing “wabun”, to use Ishigami’s term). Once knowledge had been acquired about representing the spoken Japanese language by means of marking up literary texts written in Chinese, it became possible for that knowledge to be retasked and applied to writing spoken Japanese in a system that required nothing extraneous to the text itself. It became possible to develop an indigenous writing system to encode the vernacular language.

In order to more clearly grasp the developmental stages involved, maybe this could be characterized in terms of:

a parallel application of a body of knowledge that has been acquired through decoding literary Chinese texts and annotating/marking-up those texts in a manner such as to facilitate the adoption of the Chinese writing system by enabling individual characters to be glossed in vernacular Japanese equivalents and the syntax of the literary Chinese texts to be parsed into the syntax of spoken Japanese, to the task of solving the related problematic of developing a system for representing (encoding) spoken Japanese directly in an indigenous writing system that adapts the Chinese characters in various ways. 

The development of a writing system could be seen as a natural stage in the progression of the cultural evolution of a people, but I’ll defer to the cultural anthropologists on that, as it would seem that some human societies do not develop a writing system. In any case, the Japanese did show significant initiative in adapting Chinese characters in developing a unique system suited to represent their language in a relatively short period of time.

Whitman approaches a similar position to what I have laid out above in his paper “Ubiquity of the Gloss”. On p. 8, he writes:

The centrality of vernacular reading of cosmopolitan writing calls for a modification of the scheme in (1). In (2) I have shown this, giving what some might consider an exalted status to the role of glossing and the glossed text.

(2)  a. Bilingual speakers of languages A and B become readers of language A.

          ⇩                                  ⇩
b. Graphic Borrowing:                      b′. Glossing:        
Bilingual speakers use                      Bilingual readers adapt
graphs of A to write B.                      texts of A to read in B.

                 ⇩                         ⇩

       c. Writers of B further adapt the graphs and textual

                             practices of A to write B.

(2) has two contemporaneous stages between initial contact with the donor language (2a) and development of vernacular writing (2c). In addition to the conventional stage of graphic borrowing (2b), the stage called “glossing” (2b’), refers to the adaptation of a donor language text to be read in the borrowing (vernacular) language. I use “adaptation” broadly, to include forms of literal annotation (glossing in the narrow sense), pedagogical practices which teach students how to read a cosmopolitan text in the vernacular, or simply, at the hands of a virtuoso reader or in the case of a simpler text, oral reading in the vernacular on the fly. Most often a combination of such techniques must have been used, but I will refer to the entire process of adaptation as glossing. The main argument of this paper is that glossing, in this very broad sense, plays as important a role in the development of a written vernacular as the much better known process of graphic borrowing.

Whitman describes these stages as contemporaneous, whereas I see there being something of a temporal progression involved from the “graphic borrowing” stage and the “glossing” stage”. It seems to me that the above schema again stops short of the ultimate stage representing the fruition of the Japanese writing system, which involves more than simple adaptation, and reaches into applying novel and inventive ideas; in the case of Japanese, this relates to the introduction of two comprehensive sets of phonetic syllabaries, and applying them in novel ways.

Here, with respect to Frellesvig’s characterization of a “bidirectional relationship”, again, the graphic borrowing stage can be seen to represent more of a direct relationship between reading and writing insofar as little addition of graphical representation is required to be added to a single character to enable a particular gloss of the character to be ‘read’ (verbalized) in vernacular Japanese. Although there are morphemes that come into play, the relationship between reading kanji in a kundoku gloss and using that kanji to represent the kundoku gloss at the simple lexical level is still relatively straightforward compared to transposing the syntax of a complete thought expressed in the form of a written sentence.

Whitman describes three stages along the way to “the development of a written vernacular”, to which it would seem at least one more stage could be added.

Before again referring to a passage from Frellesvig quoted above in order to illustrate the point that there are two distinct problematics at the linguistics level, between which there is some overlap due to the inflected nature of Japanese. The two sentences in the passage would seem to indicate that Frellesvig recognizes the dual problematic, too, so this means we should exercise care when considering his characterization of the “bidirectional reading-writing relationship”.

Before quoting Frellesvig, I’ll quote Whitman, on p. 11 of his Ubiquity of the Gloss paper:

The nature of this early material is relevant for considering the origins of glossing in both Japan and Korea. First, the earliest surviving material involves character glosses (Korean chat’o 字토, Japanese kana ten 仮名点). Kasuga (1956: 266) proposes a rough chronology of what he considers the oldest glossed texts known in Japan at the time of his research. The glosses in these texts, all added in white ink, are undated, but Kasuga places the oldest of them in the late 8 century. The four oldest of these texts, as judged by Kasuga, contain only character glosses (kana ten 仮名点), in unabbreviated form (magana 真仮名), inversion glosses, and punctuation marks. These four oldest texts are:

(7) a. The Keiun (768) ms. of the Konponsetsu issai ubu binaya 根本說一切有部毘 奈耶 (Mūlasarvâstivāda vinaya vibhaṅga)

b. The Keiun ms. (768) of the Konponsetsu issai ubu hisshunibinaya 根本說一切有部苾芻尼毘奈耶 (Mūlasarvâstivāda bhikṣuṇī vinaya vibhaṅga)

c. The Keiun ms. (768) of the Jinin bosatsu kyō 持人菩薩經 (Lokadharaparipɾ̣cchā-sūtra)

d. The Keiun ms. (768) of the Ōkutsumara kyō 央掘魔羅經 (Aṅgulimālīya sūtra)

The first two of these texts (7a-b), are vinaya texts, that is, texts laying out the rules and regulations of monastic discipline for monks and nuns respectively. It makes sense that such texts should be glossed for vernacular reading, since they are of practical use. The last text, the Aṅgulimālīya-sūtra, differs from the first three in that it is glossed entirely in Sino-Japanese, and was clearly meant to be read in that form (音読 ondoku/ŭmdok). It therefore contains no inversion glosses, but only Sino-Japanese phonological glosses and linking glosses (gōhu 合符), showing which sequences of characters are to be read together as compound-like units. Kasuga draws two conclusions from these characteristics of the earliest glossed data:

(8) a. In reading Chinese texts in the vernacular (漢文訓讀), morphosyntactic glosses (okoto ten 乎己止點) developed later than readings indicated by phonograms (假名附訓).

b. The “symbolicization” of vernacular glossing (訓點法の符號化) begins with punctuation marks (句點) and inversion glosses (反讀符). (Kasuga 1956: 267)

In the above-quoted passages, the statement made by Kasuga in 1956 regarding glossing would seem to indicate that I should read his book, too. Basically, his statements are in agreement with the notion that development occurred gradually, in stages. 

It is apparent that the difference in the kunten markup between the instructional vinaya texts and the sutra parallels the different roles of the texts. That is to say, the instructional texts are meant to be read and their content comprehended for pedagogical purposes, while the sutra is meant to be chanted, as a component of a ritualized daily regimen, according to the Chinese sound shapes of the characters. This parallel helps to account for the rapid development of the Japanese writing system as a practical tool, and the continuous concurrent study of classical Chinese text in their original form with kunten. The ritual chanting of Buddhist sutras represent the least marked-up texts, because their role is furthest removed from pedagogy or practical matters, relating strictly to the preservation of a religious tradition through practice.

Finally, back to the passage from Frellesvig found on p. 259, which I’ve split in two for emphasis:

… Thus, kanbun-kundoku involves finding suitable translation equivalents in Japanese for content and function words in the Chinese texts.

The first sentence addresses the problem of glossing logographs.

However, in addition kanbun-kundoku involves a number of processes in order to render (‘read’) the Chinese text in Japanese: transposition (change of word order) and interpolation or specification (of inflectional morphemes or grammatical particles).

The second sentence addresses the problem of transposing the syntax of literary Chinese to vernacular Japanese. As I have mentioned in previous a post, there is some overlap between semantics and syntax in Japanese in the form of the morphemes that occur in relation to verbal inflection, and the above-quoted passage by Kasuga (via Whitman) refers to glossing this overlap in terms of "morphosyntactic glosses". At any rate, it would appear to me that the above passage from Frellesvig in fact points to the two tracts of problematic that I have identified as such.

In the following page Frellesvig quotes passages from the Kojiki in which the Chinese character訓[kun] is used in combination with several different Chinese characters to form different compounds. Those compounds all evince a distinct and direct contrast with the compound 訓読[kundoku], or "gloss reading". The usages found in the passages quoted by Frellesvig represent the first instances that I have encountered which I consider to represent direct evidence that contradicts the assertion that kundoku represented a method of writing—both to the Japanese themselves and in the broader linguistics scenario.

In the same paragraph on p. 264, Frellesvig continues:

… In Japan 訓[kun] is used this way in our earliest sources, for example in the ‘reading’ notes inside the main text of the Kojiki, to mean read (out) (a logographically written word)’. The reading notes are instructions written in Chinese, about how to read the main text; they are not later additions, but part of the text.

On p. 265:

However, a quite different use of 訓[kun] is found in the Kojiki preface, which is generally regarded as being written in Chinese, in the paragraph outlining the writing principles employed in the main text of the Kojiki. Here, 訓[kun] is used to mean logographic writing.

He then quotes several passages in which the author or scribe is providing guidance to the reader regarding how to read the characters in which the text is composed; that is to say, whether as logographs or phonographs, basically.

Two of the passages quoted by Frellesvig include the following Chinese character compounds of which the character is a constituent:

訓述 [kunjutsu], which is translated as "logographic writing" in the passage;

and

訓録 [kunroku], which is translated as "recorded logographically" in the passage.

There is one other noteworthy occurrence, still on p. 265:

交用音訓[kouyou onkun], which is translated as "logographic and phonetic writing" in the passage.

This phrase probably translates to something like "logographic and phonographic characters are used in a mixed form", which would likely refer to portions written in the style known as hentai kanbun, and perhaps senmyogaki as well. In this context, this means a mixture of Chinese characters used as logographs and phonographs in the same text, which would still be intended to be verbalized in vernacular Japanese.

The importance of these passages to the present critique is that the term 'kundoku' is not once used to refer to writing, though writing is referred to at least twice with two different Chinese character compounds. This demonstrates that there were various expressions available to the Japanese to refer to logographic writing as such; furthermore, the instances from the Kojiki, which was the first full length tome written by the Japanese for a Japanese audience, testifies to the fact that there were ample resources for differentiating, terminologically, between reading and writing, should the need to do so arise, as in the case of the Kojiki, where a mixture of writing styles are employed.

These passages would appear to be prima facie representations of the fact that the Japanese did not consider the term kundoku[訓読] to represent "a method of writing". Therefore, unless one considers the assertions made by Lurie to represent a meta-historical, that is, theoretical statement of scientific relevance to the field of linguistics, thereby transcending the understanding of the Japanese literati that wrote the Kojiki, for example, these passages from the Kojiki add further substantial weight to the claim I've made to the effect that kundoku is not and never was "a method of writing".

There are various issues regarding the origins of kunten, their introduction into Japan, etc. Without going into much detail, John Jorgensen, in his paper, “Trends in Japanese Research on Korean Buddhism 2000-2005” discusses the history of the use of the stylus to annotate and mark Chinese texts. On pp. 20-1:

… These marks were used to avoid dirtying the precious books with ink marks. First discovered in Japan in 1961, by November 2001 over 2,300 texts using this technique were located throughout Japan. Most were Buddhist texts. Such indentations have been found on Han Dynasty wooden slips and in Dunhuang manuscripts. The first Korean examples were discovered in 2000, and to date fifty-three examples have been located by the author and Korean scholars. They date from the seventh to nineteenth century… Kobayashi argues that the system of reading marks to assist Koreans convert Chinese into Korean word order was invented by Wonhyo’s son, Seol Chong. Chinese did not need these syntax markers, which were indicated by dots and lines around the character, and later in Hangeul and kana, also. This suggest that the system was imported to Japan from Silla, possibly by a Hwaeom monk or Seol Chong’s son who went to Japan in 780… These were likely the origins of Japanese kunten. The earliest Japanese kakuhitsu notations date from 783 in a text at Todaiji in Nara, which may have been brought there by Pyowon of Hwangnyong monastery in Gyeongju.

Here, it is instructive to return to Frellesvig’s passage on p. 263:

… It now in fact seems overwhelmingly likely that kunten techniques, too, like the Chinese writing and text and kanbun-kundoku were transmitted from the Korean peninsula to Japan. For example, the earliest Japanese materials are far more similar to the Korean materials than are later Japanese materials. Both kanbun-kundoku and kunten and their histories must be viewed in a broader pan-East Asian perspective, where, in particular, the spread of Buddhism and Buddhist canonical texts in Chinese translation and commentaries written in Chinese played an important role.

Frellesvig is inclined to concur with Jorgensen that kunten techniques likely first reached Japan via Korea. There is nothing detrimental in considering the history in a manner granting that the Japanese studied the Korean usage of kunten techniques, and then developed them in conjunction with their practical application to actual texts being marked up for parsing into spoken Japanese. We could go back one step earlier and state that the techniques of marking up Chinese literary texts had been introduced to Korea directly from the continent, based on the evidence provided by Jorgensen. 

In closing, although I am inclined to favor limiting the scope of the term "gloss" to the lexical, I'll defer to a passage from Whitman on this point, as per his quote of Wieland's analysis of the traditional use of the term in Western scholarship. There is much to be considered in relation to the subject matter at hand.

Whitam, on pp. 19-20, writes:

The second type of annotation, traditionally labeled glossing, is a type of commentary. Pedagogical glossing as studied by Wieland, however, has properties in common with Korean and Japanese glossing practice. Wieland makes a useful division of glosses by function, which may be applied to glossing traditions West and East:

(14) Wieland’s (1983) classification of glosses by function

(a) Prosodic glosses Mark metrical information: accent or syllable length.

(b) Lexical glosses Give lexical equivalents for a word or phrase.

(c) Grammatical glosses Give morphosyntactic information: case, part of speech, etc.

(d) Syntactical glosses Give syntactic information (mainly word order).

(e) Commentary glosses Summarize content, give etymologies, etc.