The Recitation of the Heart Sutra in Japan

I've written before about how Japanese Buddhist recitation of Classical Chinese imposes a metrical equivalent on bisyllabic Chinese readings of words, e.g. 色 shiki is pronounced twice as fast as 自 ji, clearly in an attempt to retain some of the monosyllabic rhythm that the Middle Chinese originally had (色 is MC ʂik and 自 is dzi^H).

The other day I was reminded of this again, because someone on Twitter posted a photo of the Heart Sutra which was written not in Classical Chinese, as one might expect but instead it was transcribed phonetically in hiragana along with recitational notes (thank you Sven Osterkamp, I would have certainly missed that if you hadn't pointed it out!). What's cool about the text is that it uses the sign 引 to point out that short syllables should be stretched out to be just as long as typically bimoraic syllables. 

Something I hadn't quite realized until I saw it written out like this, is that this recitational practice actually destroys  some contrasts as well, since a word like  空 kuu (MCkʰuŋ) and 究 ku (MC kjuw^H) end up becoming homophonous. Historically these were probably kept distinct, since the Sino-Japanese pronunciation of old -ŋ final words, which have now become long vowels, appear to have retained some trace of the final nasal, transcribed by Frellesvig as Ũ and Ĩ. Thus these two in recitation were probably pronounced kuũ and kuu, thus distinct.

What else was interesting about this transcription is that it is written in historical spelling, which reflect Middle Japanese pronunciation rather than modern pronunciation, giving us some insight into what early recitation may have sounded like. Because I'm interested in artful linguistic use, I couldn't resist to listen to some recitations of the Heart Sutra of which there are plenty of Japanese recitations on YouTube. As I was doing it though, I ran into all kinds of surprising things. Let me just post some observations.

The title of the Sutra, in our hiragana transcription does not seem to have been considered part of the recitation, since it lacks the 'lengthening' instructions. But in recitation it appears to generally be read long like that.

Another puzzling thing is that the transcription of 般若波羅蜜多 is hannyaharamitta, that is, with the expected reading of 蜜 mitsu assimilating to the next ta. That is what we would expect, and in non-recited Japanese, indeed seems to be how it is pronounced, but it's very clear if you listen to the recitations that the syllable is simply pronounced mī. It strikes me that the recitation is actually remembering an ancient irregular reading of this character, as it is a better match for the Prakrit that it is supposed to represent, namely: paññapāramitā 'perfection'of wisdom'< prajñapāramitā. In fact, if we turn back the sound law of *p> h, the Japanese pronunciation is quite an excellent match for the Prakrit pronunciation: pannyāpārāmītā, but the match is better without the reflex of the final *-t of 蜜.

Notice also the automatic voicing of shin-gyou. 経 MJ kyaũ is typically voiceless, but due to liaison with preceding nasals ends up voiced.

This is the first thing that really made me raise my eyebrows. The hiragana transcription of 菩薩 has bōsatsu, and that is exactly what we would expect, and indeed bosatsu is the modern Japanese pronunciation for 'Bodhisattva', but time and time again as I looked for recitations, I keep on hearing bōsā, lacking the final syllable. What's going on? I did find one recitation that has the pronunciation as I expected, but clearly that is not the norm.

My guess of what is going on, as Frellesvig (2010: 316ff.) points out, Japanese, in Sino-Japanese vocabulary, actually posessed a word-final -t. How widespread that was is not so clear, but Christian transcriptions like fotnet suru 'get fever' (Modern hatsunetsu suru) and connit 'today' (Modern konnichi) leave little room for doubt. Could it be that the traditional pronunciation here was bōsat, and that the t was simply lost rather than reinstated as a full syllable in the reading tradition?

Just want to point out here that gyoujin has voiced the initial consonant of 深 due to the lost final nasal of gyou. Historically: gyaũ-sin > gyaũ-zin.

The fact that issai simply retains its geminate makes it clear that the recitation style of hannyaharamitta really is strange.

The combination of 身意 read as shinnī, is interesting. In modern Japanese the reading of 意 is i, not ni. This is clearly a case of historical liaison of a syllable final n to a word that historically started with w. There are several examples of this in Japanese, for example 天王tennou 'heavenly king' is irregular from the modern readings of the character ten + ou. However, ou comes from MJ waũ ultimately from MC ɦuaŋ. Japanese has a fairly regular rule nw > nn. Thus here too we have a historical memory in the recitation from a time that this verse was approably recited mū-gen-nī-bī-zet-sin-wi!

I haven't found any note anywhere that explicitly notes that this character was pronounced with initial w however, and historically in Middle Chinese it appears to have been ʔi. Such a w-less reading also seems to explain the irregular reading of 新発意 'new monk/convert (to Buddhism)' as shinbochi< sin-bot-i, although it is also read sinbotchi which perhaps can be derived from sin-bot-wi. The Sino-Korean pronunciation is ui which really does seem to suggest the existence of a w- initial reading.

Nothing too special with these lines, but I liked the alternation of the assimilated final -k stem in tokkō besides toku in the previous line. But that's in line in what you would simply expect if it were natively pronounced.

While 薩 was pronounce sā in the two character version of the word for bodhisattva, here with the full four character transcription, 薩 simply behaves as expected, and we end up with a fairly good rendering of the Prakrit bodhisatta.

Here I thought it was nice to see the archaic go-on pronunciation of 有 as u surface rather than the kan-on yū. In fact, all readings of this sutra are go-on (this is typical for Buddhist chants), but for this character, in Sino-Japanese vocabulary the kan-on pronunciation is much more common, I know words like 有名 yuumei 'famous' (which I suppose would be umyou in go-on :-)) etc. 

It's also cool to see that 有 read as yū has the general meaning 'existence', but the moment you read it as u it means bhava, i.e. the Buddhist technical term for 'existence'.

Evidence that the bōsā at the beginning really is weird, because here we have sanzeshobutsu 'All buddhas of past, present and future' not bū or something like that. (Again notice also the neat voicing of san+se > sanze).

Once again a nice bit of Sanskrit/Prakrit phonetic spelling: anuttarā-samyak-saṃbodhi "The supreme perfect enlightenment"

I was surprised by the reading watsu for 曰. I looked in some dictionaries and could only find Go-on woti > ochi and Kan-on wetu> etsu. I'm not sure what the story is here. The middle chinese rhyme class is 月, which gives go-on gwati (and gwatu) and Kan-on gwetu. But I see now that another go-on that I was unaware of is goti. So I suppose oti next to watu makes sense even though no dictionary I could check records it.

Again a transcription of Sanskrit: gate gate pāragate pārasaṃgate bodhi svāhā

Some interesting things happen here. In this recitation we've seen 菩提 "bodhi" already, and every time it was pronounced bōdai, but in this Mantra it becomes bōji. Why? Wiktionary tells me it is a 'corruption' of bodai, but boji < bodi is obviously a better match for bodhi than bodai is. Is this, like the mī reading of 蜜 mitsu a case where the reading tradition retains a more accurate pronunciation of what it should be, where the information of the characters could not have given that?

There is good reason to think that that is the case, because there is yet anore place where something really cool happens, namely the word that comes right after 薩婆訶 which is read sowakā, that is with two SHORT syllables rather than the stretched out sound for CV syllables we have seen so far. In other words, the pronunciation seems to retain an actual memory that 薩婆 represents only a single Sanskrit syllable, and is therefore made metrically equivalent to the syllable that follows: 訶.

It is also worth noting that not only is that 'rythm' irregular, also the reading of 薩 as so is irregular. This is the same character we find in the second syllable of our vexing bosatsu~bōsā 'Bodhisattva' word. so is in no way a typical or reasonable reading of the character, whose go-on reading is sati and kan-on reading is satu from Middle Chinese sat.

We therefore see that, despite the Classical Chinese composition of this Sutra, the Japanese Buddhist reading tradition seems to have some memory of the proper Sanskrit pronunciation that isn't filtered through a spelling pronunciation of the Chinese characters! Instead it is a genuine memory of an oral tradition!

That's it for now. I know nothing about Classical Chinese (I literally have no clue what the heart sutra says), I probably made some silly mistakes here and there, and I have no idea how much of this is actually known already... but it seems really worthwhile to study the interaction of reading tradition and writing of Buddhist recitation, perhaps even in a more comparative scope, comparing it to thinks we see in Quranic or Hebrew recitation etc.