wichaa › Subjects › Grammar & Lexicography
Every other shelf in this archive rests on this one. The catalogue holds 389 witnesses here, all 389 raw-labelled Philology by the source cataloguers, and what they are is the apparatus by which a man who spoke Northern Thai learned to read a sacred language he did not speak.
That is the whole story, and it is worth saying plainly before the counts start. A novice in a Lampang monastery in 1774 could not read Pali. Nobody could — Pali had no native speakers anywhere in the world and had not had any for something like two thousand years. Yet the canon was in Pali, the liturgy was in Pali, and the katha that made a spell work were in Pali. The gap between the boy and the book had to be bridged by technology, and the technology is what these 389 manuscripts are. For every grammar book on these shelves there are roughly six canonical texts and four birth-stories that only open because the grammar book exists.
Thirty-six of the 389 are นิไส / นิสสัย — nissaya. This is the single most consequential invention in the corpus, and it is not a book so much as a format.
Here is the format. You take the Pali sentence. You cut it into its words, in order. After each Pali word you write its vernacular equivalent, introduced by ว่า — "means." Then the next word. Then the next. The Pali is never replaced; it is interleaved, so what the reader has in front of him is the sacred text and a running translation braided into one line, in the same hand, on the same leaf. Read the Pali aloud and you are chanting scripture. Read the glosses and you are reading Northern Thai. Read them alternately, which is what a student did, and you are being taught.
The archive holds one such line, transcribed, and it is worth quoting because it is so ordinary. On a badly damaged mulberry-paper leaf at Wat Ban Ueam in Lampang (ms 5076, p.6) — a manuscript that has fallen into the Other bucket, of all places — almost the entire page has rotted past reading. One clause survives intact:
ปาสาโท ว่า ปราสาท — pāsādo means prasat, a palace.
Four words. Pali noun, the hinge-word "means," vernacular noun. That is the entire mechanism, and it is the reason a village monk could work through a canonical text at all. Multiply that line by the length of a sutta and you have a nissaya; multiply it by a curriculum and you have this genre.
The nissaya witnesses here show how far the format travelled. Most are glosses on grammar itself — นิไสสนธิ at Wat Ban Pang in Lamphun, dated 1727, twelve pages; นิไสการก at Wat Lai Hin Luang, dated 1774; นิไสวุตโตทยะ, a nissaya on the manual of Pali prosody, in three separate copies. Some go up a level: ms 2091 at Wat Sung Men, dated 1836, is นิไสอภิธานสับ, a nissaya on the great Pali thesaurus, 83 pages of glossing a dictionary. Ms 3811 at Wat Duang Di is นิไสสุโพลังการ, a nissaya on the classical manual of poetic ornament — the technique reaching past grammar into literary theory.
And it escaped the monastery's language classes entirely. Over in customary law, ms 3940 at Wat Sung Men, dated 1827, is นิไสคดีโลก — a nissaya of worldly cases. Somebody construed the law code phrase by phrase into Northern Thai so it could be taught the way scripture was taught. Inference — the nissaya is not a genre. It is a verb. Anything hard could be nissaya'd, and the north nissaya'd its law.
Look at what the titles repeat, and the corpus stops looking like a heap and starts looking like a syllabus.
Sixty-three titles begin with สับ / ศัพท์ — sap, from Sanskrit śabda, "word" — and what follows sap is almost always one of the same short list of technical terms:
Tradition holds — those are the divisions of Kaccāyana, the oldest and most widely taught Pali grammar in the Theravada world, inherited by Lanna from Pagan. A student did not read Kaccāyana cover to cover. He worked one division at a time, and each division was copied out as its own fascicle, which is exactly why this genre's titles look the way they do.
The proof is in a single manuscript. Ms 1507, Wat Sung Men, dated 1840, 45 pages, is catalogued as:
ตัวบาฬีสนธิ, ตัวบาฬีนาม, ตัวบาฬีสมาส, ตัวบาฬีตัทธิตตะ, ตัวบาฬีอาขยาต, ตัวบาฬีกิตตะ, ตัวบาฬีอุณณาท, ตัวบาฬีการก
Sandhi, nouns, compounds, secondary derivation, verbs, primary derivation, the irregular suffixes, and the cases. All eight, in one bundle. That is not a manuscript. That is a course, boxed.
It is not alone. Ms 5262 at Wat Ban Luk, Lampang, dated 1756, runs seven of the eight. Ms 4915 at Wat Phra That Lampang Luang, dated 1760, gives five of them each labelled กัปป — kappa, "chapter" — which is Kaccāyana's own word for its divisions. And ms 6246 at Wat Tha Pong in Chiang Mai, dated 1896, does something a teacher will recognise instantly: seven fascicles titled อุททาหอร- uddāharaṇa, "worked examples" — sandhi examples, kita examples, kāraka examples, uṇādi examples, samāsa examples, nāma examples, ākhyāta examples. The exercise book, filed next to the textbook.
Inference — this is the strongest evidence in the whole archive that Lanna's monasteries ran a standardised curriculum. Eight named units, taught in fascicles, with a matching set of exercises, copied in Lampang in 1756 and in Chiang Mai in 1896 under recognisably the same headings. A hundred and forty years apart, the same syllabus.
The second family is the dictionaries, and the north kept the good ones.
Ms 651 at Wat Na Pang, Nan, is อภิธานสับ — the Abhidhānappadīpikā, the standard Pali thesaurus of the Theravada world, arranged by subject rather than alphabet, so that a reader looking for a word for "elephant" finds all of them at once. Wat Sung Men holds it in depth: two ṭīkā (sub-commentary) copies, ms 1449 and ms 1459, both dated 1836, the larger at 91 pages; a further copy as บาลีพระอภิธานปทีปนี (ms 2102); and the nissaya at ms 2091. Four different apparatus-layers on one dictionary, in one library, in one year of copying.
Alongside it sit บาฬีรัตนมาลา Ratanamālā, "the garland of jewels" (ms 1321), นามสังคหะ Nāmasaṅgaha (ms 53), and นามสัพปาลีปะริวาร (ms 52, dated 1590) — the collected vocabularies a student kept at his elbow. This is not lexicography as scholarship. It is lexicography as furniture.
Two hundred and sixty-seven of the 389 carry a date, and they run from 1574 to 1990 — 5 in the sixteenth century, 1 in the seventeenth, 26 in the eighteenth, 203 in the nineteenth, 32 in the twentieth. (An earlier survey of this genre put the nineteenth-century figure at 202; on re-query it is 203.)
The single oldest dated witness is ms 3804 at Wat Duang Di in Chiang Mai, สมาส ผูกสอง, dated 1574, thirteen pages. Translated: Compounds, fascicle two.
Sit with that. The oldest dated grammar in this archive is not a treatise on the nature of language or a preface in praise of the Buddha's speech. It is the second half of the chapter on compound words, thirteen pages long, and somebody kept it for four hundred and fifty years. The next year, 1575, produced ms 5401 at Wat Ban Luk in Lampang — a nissaya on the Saddabhedacintā, 54 pages. Ms 58, dated 1584, holds the Saddabhedacintā itself at Wat Phra Sing Woramahawihan in Chiang Mai. Only five witnesses in this genre are sixteenth-century at all, and here is the whole list: the compounds fascicle (1574), the Saddabhedacintā twice — once nissaya'd (1575) and once bare (1584) — the Yojanā-saddā (1588), and a Pali word-list (1590). Inference — what survived four hundred years here is not grammar in the abstract but the handful of specific manuals the north actually taught from, plus the machinery for construing them.
The largest item in the genre is ms 1547 at Wat Sung Men, โยชนาสัททา Yojanā-saddā, dated 1588, at 155 pages — a full construing commentary on the grammar. Wat Sung Men holds three copies of it, one of them (ms 1505, 1837) at 142 pages. Inference — three heavy copies of the same construing manual is what a working teaching library looks like: you need enough copies that a class can be taught from them simultaneously.
Six witnesses are the pedagogical key, and they are the ones a modern reader can most nearly use:
Inference — the Tai Yuan primers matter disproportionately. Everything else in this genre teaches Pali to people who already read Tham script. These teach the script itself, which makes them the only witnesses here addressed to someone who cannot yet read anything — the first rung of the ladder, and the exact document a modern OCR effort or a modern student needs most.
Two facts about where these things are, both of which cut against a naive reading of the numbers.
Phrae holds 193 of the 389 — 49.6% — and Wat Sung Men alone holds 154, nearly 40% of the entire genre. But Wat Sung Men holds 1,780 manuscripts across all genres; grammar is 8.7% of its shelf, against 5.6% of the collection overall. Inference — Sung Men is not disproportionately grammatical. It is disproportionately large, and it happens to be a teaching monastery, so the teaching corpus came with it. Read these province counts as where the grammars were kept and copied, not where grammar was uniquely studied. After Phrae: Lampang 59, Chiang Mai 51, Nan 30, Phayao 21 — a real spread, and Wat Mueang Mo (32), Wat Lai Hin Luang (25) and Wat Duang Di (15) are substantial second libraries.
Seventy-four witnesses — 19% — are in Tham Lao rather than Tham Lanna, a far higher cross-border share than the canonical genres carry, plus eleven recorded as Pali and Lao, two Shan, and one in the Thai script. Inference — the grammatical curriculum was the most portable thing in the tradition. A chronicle belongs to the muang that wrote it; a jātaka belongs to the sermon; but Kaccāyana's eight divisions are the same in Luang Prabang as in Lampang, and a copy made anywhere works everywhere. This genre's Tham Lao seam is a map of the Lao–Lanna scholarly road, and it is best walked on the Atlas.
376 of 389 are palm leaf — 96.7%, the highest share of any genre in the collection (the canon runs 93.3%, the jātakas 87.9%, and the astrology corpus only 22%). Six are mulberry paper, three are khoi, four unrecorded.
And 146 of 389 — 37.5% — are monolingual Pali, again the highest anywhere; the canon is at 26%, and customary law is at zero.
Inference — those two numbers describe the same thing from two sides. Palm leaf is the archival support, prepared and incised for permanence; paper is what you use for something you expect to consume. A genre at 96.7% leaf is a genre nobody treated as disposable. And a grammar of Pali written in Pali is a book for specialists who are already inside the language — which is exactly the opposite of the law corpus, where a text that is not in the vernacular does not function. Put the two genres side by side and you can see the tradition sorting its books by who has to understand them.
Inference, and a note about the present. The sap manuscripts were built by scanning texts and pulling out the words that needed glossing. This archive's own `vocab_candidates` table currently holds 474 terms harvested the same way from the transcribed printed treatises — flagged, counted, and awaiting a gloss, which is precisely the nāmasaṅgaha procedure with different tools. What comes out the far end is the glossary and the vocabulary. The genre is not only the historical reading apparatus; it is the template for the one being built now.
The live connections graph and any current curiosity-bot findings.