There is also kana NI iu, kana ni ōsu, and kana ni kaku.

史記抄 (1477): いかに仮名にかくとも、なにかかくは云われうぞ
杜詩抄 19 (mid-Muromachi): 太易腰折詩ヲ作テ訪甫於船中求添削ゲナホドニ、如此云タ、ナヲイテクレヨト、カナニ云た
日葡辞書 (1603): [Cananiyù] 「仮名に言ウ」皆が理解するように、わかりやすく話す
一休はなし (early? Edo): 今まで引とどめ、あまつさへこよひは一夜とまれと、かなに仰られける
蘆屋道満大内鑑 (1734): この子爰帰ぐと、かたいやうに聞ゆれど、仮名ていへばつい嫁入

I wonder if there was any association here between onnade and kana. So this kana de iu could also have a subtext of "so simple even a woman could understand it."

Kana = onnade as late as Edo? Dunno man, I wouldn’t bet on it…

Yeah, I don't get the impression that it's a negative thing, either. More like "saying it in kanji" is unnecessarily fancy.

"Kana = onnade as late as Edo?"

Echoing leoboiko, I'd like to point out that this association's been questioned for the Heian Period as well.

So, you know.

Yes, example: http://positions.dukejournals.org/content/8/2/465.full.pdf

By the way, a quote from this paper that’s related to the original theme of the post:

> Today’s definition of Heian kana as a phonetic system of writing has been deeply inflected by modern linguistic ideology and its reification of spoken national language. In the Heian period, however, there was yet to be a concept of Japanese language that hierarchically unifies the diverse practices of speaking and writing; neither do we find in that context an epistemological basis for valorizing phonetic writing as a medium of native voice. Heian poetics strongly suggests that the relationship between sound and script were not grasped instrumentally, with writing reduced to the signifier of speech. In the Heian period, for instance, the most popular type of Japanese poetry, short poems (tanka) made up of thirty-one syllables, was commonly referred to as thirty-one *graphs* (misohito-moji). In other words, the poem’s syllabic the poem’s syllabic pattern was measured in terms of the unit of writing rather than sound.


> Describing the origin of tanka, the Kana Preface of Kokinwakashū remarks, “In the age of gods, the poetry did not have set (number of) graphs. In the human era, the composition of thirty-one graphs began with the poems of Susanowo no Mikoto.” The readers of the preface would have known that,strictly speaking, there were no graphs to write down the first song uttered in their land (the history of a writing system entering Japan from China is mentioned elsewhere in the anthology). The passage suggests that from the point of view of Heian poetics, however, whether or not poems were actually written down, graphs served as the conceptual apparatus through which the formal convention of yamato uta was recognized. Writing, as it were, anchored poetic discourse to its proper form.

And then there’s Chad Hansen’s infamous essay http://www.jstor.org/stable/2059652 .

I could easily be missing something there, but wouldn't the embrace of graph-count as a representation of poem length suggest the opposite, that sound and script were in fact perceived as a unity, with the hierarchy perhaps going the other way?

I think you’re saying the same thing as her, only from another angle. Yes, characters and sounds as isomorphic (which is kind of funny, when one thinks about it, given that kundoku, early ongana and kungana all allowed a character to represent more than one syllable… anyway).

But her point is that the characters, in the popular conceptual model of the time, seem to have had a role of “anchor” or nexus for language (cf. Hansen). Recall that poetry was intended to be read aloud—the characters are not unlike sheet music, and as such they’re closer to the “poem itself” than the aural renderings. I think? (Consider the part in the Tosa where the author hears a bad poem, and says that “even if written down, I doubt it could be read”—but she just heard it! Compare: “The melody that he composed was so awful that, even if written down, I doubt it could be played”—setting it down means you fully grok it; alas, the poor poet was so bad that even this legitimizing proccess wouldn’t save him.)
At any rate we know from the literature of the time (Genji, Makura, Tosa) that knowing only how to compose poems, without knowing how to write them down, didn’t count. So it’s hard to think of a native hierarchy that classifies sounds as more important, in this specific culture/context (Heian) at least. Though I don’t think a “hierarchy” prizing characters would be a good model either – instead, I’d prefer something like LaMarre’s proposal that the poems were a deliberate alignment, a meaningful juxtaposition of different levels.

--

I wanted to bring this thread an example I saw when browsing Penguin’s Short Stories in Japanese, but I couldn’t find the exact quote. It was in Koike Masayo’s Genjitsu House げんじつ荘。 The narrator is talking about pregnancy with a friend, who is quite proud because she just had a natural delivery, and not a teiousekkai テイオウセッカイ—written in the story like this, in kana. The narrator keeps talking for a while, puzzled by the unknown word in the back of her mind, until she “assembles the sounds into kanji” as 帝王切開 “Imperial incision” and (only _then_) wonders why it has such an impressive name. In this case, being “in kana” has the opposite meaning of “kana de iu”!

(帝王切開 is the Japanese for “Cæsarean section”; like so much medical stuff it came from German, Kaiserschnitt – « The Germanic and Slavic peoples seem to have called all Roman emperors "caesar" », so that Cæsar “Roman family name” generalized to Kaiser “emperor”, so Kaiserschnitt became the somewhat opaque 帝王切開。)

I’m sure most of you must have witnessed something like this. It’s certainly not the case that they “think in characters” (the illiterate are perfectly able to think!), but literate Japanese do occasionally use “knowing the kanji” as a shorthand or mental model for “understanding the meaning and nuance of these morphemes”—even in spoken conversation! I wonder if there’s a detailed study of this phenomenon.

I *like* that sheet music metaphor. Especially since our own spoken usage ("play a note", etc.) is a good parallel to the "31 characters" bit.

That Tosa example is really intriguing (had totally forgotten about it--thanks!). Looking it up, it seems that the problem with the poem was not necessarily its quality (though the laughter it aroused can't be a good sign), but its length, at 37 syllables. I wonder how far we can read into this? Mere practical experience that writing can fail as a medium in specialized situations--such as poetry recital--where reperformance of the speech act requires more information than writing-as-notation can convey? Or can we see it as an example of a deeper awareness of the gap between speech and writing? (Especially, as you said, given the fact of multiple readings per character.)


----
On the use of "understanding kanji"="grasp the meaning", I know I've heard people saying "変換できなかった" on the analogy of computer input when they have an experience like the person in your anecdote with the Caesarian section.

"Today’s definition of Heian kana as a phonetic system of writing has been deeply inflected by modern linguistic ideology and its reification of spoken national language". "In the Heian period, for instance, the most popular type of Japanese poetry, short poems (tanka) made up of thirty-one syllables, was commonly referred to as thirty-one *graphs* (misohito-moji)."

So we are talking of reinterpretations of reinterpretations?