Introduction
In the early eleventh century, the central Tibetan region of Ü-Tsang (dbus gtsang) was still in the process of recovering from the collapse of the Tibetan empire which had taken place in the middle of the ninth century. In addition to the official royal patronage of the spread of Buddhism, such as the invitation to Indian pandits to visit the Mnga’ ris region by the royal teacher Lha bla ma Ye shes ‘od (947–1019/24), a band of elite scholars was also quietly taking shape.Footnote 1 This band included a pioneer of the so-called New Translation Tantra movement, namely Smṛtijñānakīrti (hereafter Smṛti),Footnote 2 also known as Smṛti Paṇḍita and Dran pa ye shes grags pa in Tibetan sources, who is accordingly recognized in a number of Tibetan historical records.Footnote 3
The available sources give three main reasons for why he came to the Tibetan region: 1) he had been expelled by a non-Buddhist (mu stegs) from his place of residence,Footnote 4 2) he had been invited by the royal court of Gu ge in Mnga’ ris,Footnote 5 3) he came for the sake of his mother who allegedly had been reborn as a dog in the Tibetan region.Footnote 6 Among these, the first reason is the earliest and perhaps the most credible one. After much hardship in Central Tibet – he was enslaved as a shepherd, bought free, travelled with merchants and learnt Tibetan while doing so – he eventually settled in Danma (’Dan ma) on the border of Sichuan and Qinghai provinces. There, he translated numerous scriptures from Sanskrit into Tibetan, established a teaching lineage of the Abhidharmakośa Footnote 7 and authored the Smra ba’i sgo mtshon cha lta bu (hereafter Smra sgo)Footnote 8 and its commentary.Footnote 9
The Smra sgo had a far-reaching influence on later Tibetan grammatical traditions. It now appears, based on Smṛti’s profound knowledge of Sanskrit and Tibetan grammar, that the Smra sgo inherited some of its contents from the first Tibetan work on the philosophy of language that dates from the late imperial period (c. 800), the Gnas brgyad chen po’i rtsa ba Footnote 10 and its commentary. Further, it was also composed with an eye on the theories of the philosophy of language that is found in the Abhidharma literature – for example, the concept of nāmapadavyañjanakāya, as already noted by Verhagen.Footnote 11 However, the exact texts that inspired Smṛti and the manner in which he adapted those concepts to structure the work all remained subjects for further investigation.
In this article, I will first clarify how Smṛti applied the concept of nāmapadavyañjanakāya to the structure of the Smra sgo and then go on to illustrate the other theories he used.
1. Nāmapadavyañjanakāya in the Smra sgo
The phrase nāmapadavyañjanakāya encapsulates a fundamental theory of the philosophy of language of the Sarvāstivāda school. The corresponding Tibetan expression is ming tshig yi ge’i tshogs,Footnote 12 which is a dvandva compound.Footnote 13 When analysed, the compound yields nāmakāya, padakāya and vyañjanakāya, which most commonly refer to the collection of names, the collection of phrases and the collection of letters, respectively. (For variant definitions of this triad, see below.)
The most common Chinese translation of the phrase is ming ju wen shen (名句文身) which was first used by Xuanzang (玄奘, 600–64). Furthermore, it is noteworthy that, in the Chinese translations of the four Āgamas, this concept only appears in the Za a han jing 雜阿含經 (Saṃyuktāgama),Footnote 14 which is the version belonging to the Sarvāstivāda school.Footnote 15 This fact aids in identifying the school’s affiliation with this concept. It was later adapted by Vasubandhu and Asaṅga and was thus also developed within the Yogācāra school.Footnote 16
It reflects the Sarvāstivādas idea of the structural and epistemological analysis of the words of the Buddha (buddhavacana).Footnote 17 This is how it further explains language in general in terms of nāman, pada and vyañjana.Footnote 18 It is categorized among the conditioned forces disassociated from thought (cittaviprayuktasaṃskāra) as three different entities. This classification grants language an independent status from one’s mind. Furthermore, it is also recognized as the object of unobstructed knowledge of dharma (teachings) (dharmapratisaṃvid)Footnote 19 which, being one of the four types of unobstructed knowledge, makes the triad serve as a fundamental basis for understanding the Buddha’s teachings.Footnote 20 That is to say, the only way to understand the Buddha’s words and to grasp their intended meaning is by means of this triad.
Due to the significance of the triad and the fact that there is no substitute for it, Smṛti adopted this concept from the Abhidharma literature to organize the structure of the Smra sgo. He announces this in the opening remarks of his treatise in the following way:Footnote 21
de la ’di’i gzhung nyid kyang yi ge’i tshogs dang ming gi tshogs dang || tshig gi tshogs rnams ston te | de dag gi yan lag ji skad du bstan pa rnams kyang mdor bsdu na de dag gi rgyu dang || rang gi ngo bo dang || ldang tshad dang | nges pa’i tshig dang | dbye ba rnams kyis ston to ||
Here, the text of this [book] teaches the vyañjanakāya, nāmakāya, padakāya; were we to summarize also the teachings of their stated parts, then these will be taught by their cause, nature, conditions, interpretation (*nirukta), and typology.
Afterwards, Smṛti discusses Sanskrit and Tibetan, from smaller to larger linguistic units in the order of vyañjanakāya (yi ge’i tshogs), nāmakāya (ming gi tshogs) and padakāya (tshig gi tshogs), that is to say, he respectively elaborates on their phonology, nomenclature and syntax in his own way. In terms of their length, those parts of the Smra sgo that deal with the vyañjanakāya and the padakāya occupy the largest portion of the text, while the section on nāmakāya is very brief. Thus, the Smra sgo mainly consists of the phonetic reading methods of Sanskrit centred on the vyañjanakāya and the analyses of different types of compounds and grammatical particles that are discussed in the padakāya. The section on vyañjanakāya deals with the proper recitation of mantras,Footnote 22 and is therefore functionally akin to the Prātiśākhyas that deal with the phonetic aspects of the recitation of the Vedas.
In addition to borrowing the triad as his overall structure of the Smra sgo, Smṛti’s definition of nāmapadavyañjanakāya as it unfolds in his discussion also reflects ideas from the Abhidharma literature. At the end of the first part of the yi ge’i tshogs section, he explains yi ge (*vyañjana) by using an “etymological” interpretation (nges pa’i tshig;*nirukta). He thus writes:Footnote 23
da ni yi ge’i nges pa’i tshig bshad bar ’dod nas|
rnam grangs du ma med pa dang||
ming tshig gsal bar byed pa dang ||
don rnams rjod par byed pa’i phyir ||
yi ge zhes ni brjod pa yin ||
rang bzhin las dang dgos ched kyi ||
nges tshig yin te go rim bzhin ||
zhes bya ba smos te | rnam grangs du ma med pa zhes bya ba ni rang gi ngo bo’i sgo nas nges tshig brjod pa ste | yi ge rnams kyi rang bzhin rnam grangs gzhan du mi ’gyur zhing du ma med pa’i phyir mi ’gyur ba zhes bya’o||
gsal bar byed pa zhes bya ba ni las kyi sgo nas nges pa’i tshig brjod pa ste | mi ’gyur ba nyid kyis ming dang tshig gsal bar byed pas gsal byed ces bya’o ||
rjod par byed pa zhes bya ba ni dgos ched kyi sgo nas nges pa’i tshig brjod pa ste | gsal bar byas pas chos rnams kyi don rjod par byed | dran par byed | shes par byed pas rjod pa [read: rjod par byed pa] zhes bya’o ||
Now, wishing to explain the etymological interpretation (nges pa’i tshig, *nirukta) of the yi ge (*vyañjana), we stated:
Because it is without many synonyms and,
it reveals names and sentences and,
it expresses object-referents,
the “yi ge” is stated.
These involve the etymological interpretation of respectively its
nature, function and purpose.
Without many synonyms it expresses its etymological interpretation from the perspective of its nature (rang bzhin); because the nature of yi ges does not change into another synonym and does not have many [synonyms], the yi ge is called “unchanging” (mi ’gyur ba; *akṣara).
Revealer expresses its etymology from the perspective of function (las); because it reveals names and sentences by virtue of its unchanging nature, the yi ge is called a “revealer” (gsal byed; *vyañjana).
Expresser expresses its etymology from the perspective of its purpose (dgos ched); because the yi ge expresses, conceives and makes known the object-referents of entities, the yi ge is called “expresser/speech” (rjod pa[r byed pa]; *varṇa).
It is difficult to understand Smṛti’s intended meaning solely by means of this terse verse.Footnote 24 Only by further reading the commentary can it be seen that the author aims to analyse the three synonyms (rnam grangs; *paryāya) of the word “yi ge” – “unchanging”, “revealer” and “expresser/speech”. Given that the method of employing etymological interpretations to explain words plays a fundamental role in Sanskrit philosophical analysis and grammar, it is highly unlikely that the composition of Smra sgo was exclusively dependent on Tibetan. Considering Smṛti’s Abhidharma background, this passage is most likely derived from the Abhidharma literature, a point that will be further illustrated later.
Next, the Smra sgo defines ming and tshig as follows:Footnote 25
mdor na chos kyi ngo bo nyid tsam [du] ston nus la khyad par mi ston pa’i yi ge ’dus pa’i gnas skabs la ming zhes bya’o ||
In brief, the circumstance of a combination of yi ges which, while able to indicate the mere intrinsic nature of an entity, [and] does not indicate a [specific] quality, is called a nāman.
gal te ming gnyis ’dus pa’am ||
’on te mang po ’dus kyang rung ||
de dag khyad par rjod byed na ||
tshig ces rab tu brjod pa ste ||
If two nāmans combine,
Or when possibly many combine,
when these express a distinct feature,
they are called pada.
This definition of nāman and pada in the Smra sgo can be divided into two aspects:
First, their composition: the nāman is a combination or aggregation of vyañjanas, and the pada is a combination of nāmans; second, their property: the nāman expresses the nature of entities and the pada expresses the distinction feature among entities.
Depending on the mechanism of composition of the triad, the concluding passage of the Smra sgo discusses how a person can compose a large treatise step by step on the basis of the yi ge (vyañjana):Footnote 26
skyes bu smra bar ’dod pa yis ||
gzhi mtha’i yi ge’i bye brag dag ||
legs par khong du chud pa na ||
de ’dus ming dang de bzhin tshig ||
shlo ka dang ni rab byed sogs ||
ming tshig yi ge mtshungs pa’i phyir ||
tshig sbyor ’bad pa med par ’gyur ||
’di dag de bzhin no || zhes sbyar ro || ’di ltar skyes bu gang zhig don dang ldan pa’i tshig smra bar ’dod pa des dang por ming gi mtha’i yi ge’i bye brag legs par khong du chud par bya’o || de nas yi ge de dag ’dus pas ming ’grub par ’gyur te | de’i mtshan nyid khong du chud par bya’o || de bzhin du ming ’dus pa las tshig yin te | de’i bye brag khong du chud na gzhan tshig ’dus pa shlo ka dang | shlo ka ’dus pa las rab tu byed pa dang | de ’dus pa las bstan bcos chen po la sogs pa rnams kyi tshig gi sbyor ba rnams ’bad pa chung ngus shes par ’gyur ro ||
When a person who desires to speak
thoroughly understands
the particulars of the base and suffix yi ges (vyañjanas),
the ming (nāman) as a combination of the yi ges, and likewise the tshig (pada),
because the verses, chapter, etc.
are like the ming (nāman), tshig (pada) and yi ge (vyañjana),
a composition is effortlessly achieved.
It is so accordingly; we have written: Thus, any person who desires to speak a meaningful tshig (pada) must first thoroughly understand the difference of yi ges (*akṣara/vyañjana) that are the ultimate unit of the ming (nāman). Then, a ming (nāman) is created by a combination of those yi ges, and one will understand its defining characteristic; likewise there is a tshig (pada) from a combination of mings (nāmans). When we understand their features, then furthermore, the composition (tshig sbyor) from the combination of padas, the śloka, the chapter (*prakaraṇa) from the combination of ślokas, the large treatise from the combination of chapters, etc. will be understood with little effort.
Through this passage, Smṛti establishes the sequence of yi ge (vyañjana) combining to form ming (nāman) and ming combining to form tshig (pada). Smṛti clarifies that language learning as well literary creation should follow this sequence of knowing yi ge, ming and tshig.
The term “gzhi mtha’” (base and suffix) in the verse should be understood through the concepts of the nominal stem (prātipatika) and the suffix (anta) of traditional Sanskrit grammar. The idea that ming gi mtha’i yi ge, “the yi ge as the ultimate unit of the ming”, is related to the theory of irreducible particles is discussed in the Abhidharma literature. This notion is detailed in the fourth section of this article.
As noted above by Smṛti, the triad discussed in the Smra sgo possesses three main aspects of their characteristics. First, the term vyañjana has three synonymous expressions, each highlighting one of its qualities, and nāma and pada also serve different functions within language. Second, both nāma and pada are composed of smaller units, with vyañjana being the smallest linguistic unit. Third, grammatical rules must be employed when composing treatises based on this triadic structure.
2. Nāmapadavyañjanakāya in Abhidharma literature
To better understand the meaning of the triad and determine the philosophical foundation on which Smṛti depends and how the notions mentioned above came into being, we need to go back to the relevant Abhidharma literature.
Since the explanation of the triad in Vasubandhu’s (fifth century) Abhidharmakośa represents an intersection of the earlier theoretical developments within the Sarvāstivāda and the later Sautrāntika tradition, this work can function as the starting point of our exploration. The discussion about the triad occurs in the Abhidharmakośa’s second chapter on indriya (faculties):Footnote 27
nāmakāyādayaḥ saṃjñāvākyākṣarasamuktayaḥ |
kāmarūpāptasattvākhyā niḥṣyandāvyākṛtāḥ (tathā) ||
Abhidharmakośa 2.47
Nāmakāya etc. are collections of saṃjñā, vākya and akṣara.
They exist in the kāmadhātu and the rūpadhātu;
they belong to living beings; they are out-flowing; and they are neutral.Footnote 28
The first half of the stanza lists two sets of terms for these three concepts: nāman, pada and vyañjana and its corresponding synonyms saṃjñā, pada and akṣara. The second half summarizes several key characteristics of nāmapadavyañjanakāya such as it belongs to the realms of desire (kāmadhātu) and form (rūpadhātu). This means that they are included within the category of sentient beings (sattva) rather than insentient entities like mountains and rivers, which are out-flowing outcomes produced by homogeneous causesFootnote 29 and are karmically neutral (avyākṛta).Footnote 30 In the Abhidharmakośabhāṣya ad Abhidharmakośa 2.47, Vasubandhu provides the following definitions of the triad:Footnote 31
tatra saṃjñākaraṇaṃ nāma | tadyathā rūpaṃ śabda ity evamādiḥ | vākyaṃ padaṃ yāvatā ’rthaparisamāptis tadyathā anityā vata saṃskārā ity evamādi yena kriyāguṇakālasaṃbandhaviśeṣā gamyante | vyañjanam akṣaraṃ tadyathā a ā ity evamādi |
Nāman refers to making such concepts (samjñākaraṇa) as form (rūpa), sound (śabda), and so on. Pada refers to a phrase (vākya) as far as it completes a meaning, such as “Oh, compounded things are impermanent!” and so on; [or] through this one can understand the related specific qualities of activity, quality and time. Akṣara refers to a letter (vyañjana) like a and ā and so on.Footnote 32
Vasubandhu’s definitions summarize the definitions of nāmapadavyañjanakāya that belong to the Sarvāstivāda school: in brief, nāman means a name, pada refers to a verse line in a stanzaFootnote 33 or a sentence that can distinguish specific qualities and akṣara indicates a letter/phoneme.
Depending on Vasubandhu’s work, SaṅghabhadraFootnote 34 subsequently reinforced the Sarvāstivāda position while also introducing certain amendments by further refining these definitions in the *Abhidharmanyāyānusāra (Apidamo shunzhengli lun 阿毘達磨順正理論) in the following manner:Footnote 35
又名等三相差別者: 謂聲所顯,能顯於義, 已共立為能詮定量, 顯示所解,意樂所生, 能表所知境界自體, 猶如響像, 此相是名; 若能辯析所知境中廣略義門, 此相是句; 於能說者聲已滅位, 猶令繫念,持令不惑,傳寄餘者, 此相是文。
Furthermore, the distinctions among the characteristics (*lakṣana) of nāman, etc. are as follows: nāman refers to [1] what is revealed by sound (*śabda/ghoṣa) and can reveal the object-referent (*artha), [2] [the sound] that has a consensually established (*saṃketa) stipulated delimitation (*kṛtāvadhi) [referring to some object-referents] which is able to make known (*pratyāyayati) [the object-referents], and [3] what manifests understanding and arises from intention, able to convey the nature of the known object, much like an echo or reflection. The one with these characteristics is called nāman. If we can analyse the extensive and concise aspects of the object-referent within the known object, the one with this characteristic is called pada. When the revealer that is the sound has ceased, yet it is still retained in memory, preventing confusion, and conveys the message to others, the one with these characteristics is called vyañjana.Footnote 36
Compared with Vasubandhu’s definitions, Saṅghabhadra’s are more abstract and systematic. In the meantime, they also use similar expressions, but sometimes for different purposes, like the ambiguously worded sentence yi gongli wei nengquan dingliang已共立為能詮定量. It reflects the wording of Abhidharmakośabhāṣya ad Abhidharmakośa 2.47: yo ’rtheṣu kṛtāvadhir vaktṛbhis…, “The speakers created a delimitation of the object-referents of one [ghoṣa/śabda]…”.Footnote 37 Taking this as the proof, Vasubandhu or Sautrāntika do not accept of the real existence of the triad; however, for Saṅghabhadra, the existence of the triad is exactly approved by holding this kind of stipulated delimitation. In Xuanzang’s translation of the passage of the Abhidharmakośabhāṣya, he rendered it by 能說者於諸義中已共立為能詮定量…,Footnote 38 that is, “The speaker referring to some object-referents consensually establish a stipulated delimitation of [a sound] which it is able to make known, …”. Two phrases, namely, gong li共立, “consensually established”, and neng quan能诠, “be able to make known” were added by him. The use of these two additional phrases might be due to differences between the Sanskrit texts used by Xuanzang and by us today, but it is more likely based on the commentary on this sentence in Yaśomitra’s Sphuṭārthā Abhidharmakośavyākhyā.Footnote 39
…yo ’rtheṣu kṛtāvadhiḥ kṛtamaryādaḥ. etena saṃketāpekṣaḥ śabdo ’rthaṃ pratyāyayati. na yaḥ kaścic chabda iti darśayati. kṛtasaṃketaḥ śabdo ’rthaṃ pratyāyayatīti
… “the one (ghoṣa/śabda) owns a delimitation which has been stipulated in respect to some object-referents” means the one has a demarcation which has been created. By this, he shows that it is the śabda that relies on a consensus which makes known an object-referent – not just any śabda whatsoever. Only by the sound which owns a stipulated consensus is able to make known the object-referent.
The definitions of the triad in the Abhidharmasamuccaya and Vasubandhu’s Pañcaskandhaprakaraṇa are largely the same as Saṅghabhadra’s, differing only in their interpretation of vyañjana.Footnote 40 However, concerning the ontology of the triad, Asaṅga, like Vasubandhu, also discounts their existence as real entities. He states in the Abhidharmasamuccaya:Footnote 41
nāmakāyāḥ katame | dharmānāṃ svabhāvādhivacane nāmakāyā iti prajñaptiḥ||
padakāyāḥ katame | dharmāṇāṃ viśeṣādhivacane padakāya iti prajñaptiḥ ||
vyañjanakāyāḥ katame | tadubhayāśrayeṣv akṣareṣu vyañjanakāyā iti prajñaptiḥ| tadubhayābhivyañjanatām upādāya | varṇo’pi saḥ | arthasaṃvarṇanatām upādāya| akṣaraṃ punaḥ paryāyākṣaraṇatām upādāya|| Footnote 42
What are the collections of nāmans? The conventional designation “collections of nāmans” refers to the verbal expression of the nature of dharmas.
What are the collections of padas? The conventional designation “collections of padas” refers to the verbal expression of the distinctions of dharmas.
What are the collections of vyañjanas? The conventional designation “collections of vyañjanas” refers to the letters which are the basis of both, referring to its property of revealing (abhivyañjana) both nāman and pada. It is also varṇa, referring to its property of describing object-referents (arthasaṃvarṇa). Furthermore, [it is called] akṣara, referring to its property of being not flow into any synonym (paryāya-akṣaraṇa).Footnote 43
Here we observe that Asaṅga uses the method of etymologizing (nirukta) to analyse the three synonyms of letters – vyañjana, varṇa and akṣara – by tracing the origin of these three nouns to three verbal adjectives – abhiṿy2ạṇ̃jạṇạ, saṃṿạṛṇạ and ạḳṣạṛạṇa. Asaṅga may have developed this analysis based on the *Mahāvibhāṣā where we read:Footnote 44
文即是字者。巧便顯了, 故名為文(*vyañjana)。此即是字(*akṣara), 無轉盡故。此即顯示 (*varṇa), 能顯頌文, 以字為體。
Vyañjana means akṣara: because it can skilfully illustrate (*abhivyañjana), it is called vyañjana; it is akṣara because it is inexhaustible (*a-kṣara); it is varṇa, it reveals (*saṃvarṇa) verses and has akṣara-s as its essence.
Compared to the explanation within the *Mahāvibhāṣā, Asaṅga notably reinterprets akṣara by abandoning the traditional Indian notion of philosophy of language that akṣara means “imperishable” and adopting a new explanation that akṣara means “being not flow”.
“Being not flow” means that letters, except for taking the function of representing nāmans and padas, are the names of themselves and do not have synonyms. For example, the pronunciation and the name of Sanskrit letter “a” are always the same and it cannot be called as “i”, “u”, “e”, etc. As stated in Sthiramati’s Dacheng apidamo zajilun大乘阿毘达磨杂集论 (*Abhidharmasamuccayavyākhyā):
yathā cakṣuś cakṣur ity etasmāt paryāyād anyeṣv api netrākṣinayanalocanādiṣu paryāyāntareṣu kṣarati tair api tat saṃjñānāt | naivaṃ | a ity etad akṣaraṃ | a ity etaṃ paryāyaṃ muktvā paryāyāntareṇa śakyate jñāpayituṃ | ataḥ paryāyākṣaraṇād akṣaratvaṃ | akṣaraṇaṃ punar agamanam veditavyaṃ || ||Footnote 45
Just as for an eye: from the synonymic term “an eye (cakṣu)”, it flows to other synonymic terms like “a leader (netra)”, “an eye (akṣi)”, “the guiding one (nayana)”, “the illuminating one (locana)”, etc., and also because by them, it (an eye) can be concepted. This is not the case for the letter “a” that apart from the term “a”, by other synonyms it can be made known. Therefore, because of being not flow to other synonyms, it has the essence of “un-flowing”. Moreover, “being not flow” can be understood as “unchanging”.
By explaining akṣara as “being not flow 
”, Asaṅga thus nullifies the notion that the akṣara exists eternally and substantially across the past, present and future.
Meanwhile, Asaṅga completely abandons the description of nāmapadavyañjanakāya from a concrete perspective, and instead defines its component parts by their functions: nāmakāya expresses the intrinsic nature of dharmas; padakāya expresses the differences of dharmas; vyañjanakāya carries three levels – vyañjana, varṇa and akṣara – which reveal both nāman and pada, describing object-referents and have no other synonyms.
3. From Abhidharma to the Smra sgo
Comparing these explanations from Abhidharma literatures with the aforementioned passages of the Smra sgo, it becomes clear that Smṛti primarily drew upon the interpretations found in the Abhidharmasamuccaya or in other materials that contained something similar.
Some might suggest that since the translation of Abhidharmasamuccaya had already been done during the latter part of Tibet’s imperial period, the author of the Smra sgo could also have taken the idea from the Tibetan translation and not from other Sanskrit sources.Footnote 46 Indeed, this analysis of the synonyms for vyañjana were already reflected in the Mahāvyutpatti, a Sanskrit-Tibetan lexicon compiled in the early ninth century. Given that Smṛti translated numerous Sanskrit texts in Tibetan, there can be little doubt that he was familiar with that work. The Mahāvyutpatti lists nāmakāya, padakāya and vyañjanakāya in the section that enumerates those conditioned forces that are disassociated from mentation. It writes:Footnote 47
As shown above, after listing all the items of the conditioned forces dissociated from thought, the Mahāvyutpatti includes the entries vyañjana (gsal byed), akṣara (yi ge) and varṇa (brjod pa), and these terms correspond exactly to the Tibetan translation of the Abhidharmasamuccaya from c. 800:Footnote 48
yi ge’i tshogs gang zhe na | de gnyi ga’i gnas yi ge rnams la yi ge’i tshogs zhes gdags te | de gnyi ga gsal bar byed pa’i phyir ro || rjod pa yang de yin te | don yang dag par brjod pa’i phyir ro || yi ge ni rnam grangs su gtogs pa gzhan du mi ’gyur bar byed pa’i phyir ro ||
Comparing this Tibetan translation with its Sanskrit version shows that the etymologizing analyses of vyañjana and akṣara are not revealed in it, while in the Smra sgo, Smṛti innovatively translated akṣara as mi ’gyur corresponding to its etymological meaning, thus avoiding losing the important meaning of the Sanskrit word. He also illustrated the whole process of analysing the three synonyms. This means that even though the origin of the word list of Mahāvyutpatti tends to be the Abhidharmasamuccaya and the explanation of vyañjana was partly known in Tibetan before the composition of the Smra sgo, due to the inaccuracy and its incompleteness, it is unlikely that the description in the Smra sgo is based on some Tibetan texts.
This conclusion rests on the following: [1] Smṛti also distinguished the three different characteristics of vyañjana reflected in the synonyms vyañjana, akṣara and varṇa, characterizing them in terms of their nature (rang bzhin), function (las) and purpose (dgos ched) of vyañjana, respectively. [2] The order of these terms also differs from the Abhidharma texts. [3] The concluding passage of the Smra sgo – they can reveal, conceive and make known the meanings of all entities – inevitably brings to mind Saṅghabhadra’s description of the function of vyañjana in the *Abhidharmanyāyānusāra, that is, it still retains the memory, preventing confusion, and conveys the message to others.Footnote 49
Therefore, the description of the vyañjana in the Smra sgo is not based on some Tibetan translation texts nor even on a direct citation of Sanskrit Abhidharma texts but rather on a comprehensive reorganization and arrangement of related literature to better serve the textual framework of the Smra sgo itself.
Afterwards, this innovative interpretation of vyañjana (yi ge) came to flourish in Tibetan grammatical works and was frequently cited by subsequent Tibetan scholars such as Sa skya Paṇḍita in his Sgra la ’jug pa, Introduction to Language, and explained by the nineteenth-century scholar Kong sprul Blo gros mtha’ yas (1813–99) in the Shes bya kun khyab, Encyclopedia of Knowledge, where the definition of yi ge is elaborated upon.Footnote 50 This explanation of yi ge appears to have become so commonplace that not one of these scholars mentioned its original source.
With regard to nāman and pada, the definitions offered in the Smra sgo closely correspond to those found in the Abhidharmanyāyānusāra and the Abhidharmasamuccaya. However, a notable point of divergence lies in the Smra sgo’s employment of the seemingly plausible notions of the “combination of vyañjana” and the “combination of nāman”. Neither the Abhidharmakośa, the Abhidharmanyāyānusāra nor the Abhidharmasamuccaya suggest an additive or compositional nature for nāman and pada. This raises significant questions: why did Smṛti adopt such a view, and what may have been his sources for this?
The fundamental reason for this lies in the different understanding of the basic meaning of nāman and pada and also the different linguistic structures of Sanskrit and Tibetan. In Abhidharma philosophy, nāman and pada are parallel concepts on the same level as two individual and distinct entities, and the combination of nāmans does not inherently form a pada. In Sanskrit grammar, furthermore, a nāman is the base form of a noun, and verbs in Sanskrit cannot be referred to as nāman. In addition, it is a highly inflected language and requires morphological changes for each word in a sentence, so there is no standalone nāman in sentences. Thus, many problems would occur were one to suggest the additive nature of the triad as being possible in Sanskrit.
For a nāman to become a component in a sentence, it must first change from a nominal stem (prātipatika) into a pada (a complete inflected form) which usually has a different meaning of the pada in triad.Footnote 51 We could refer to this grammatical pada as pada 2.
In Pāṇini’s Aṣṭādhyāyī, pada 2 is defined as follows:Footnote 52
suptiṅantaṃ padam || 1, 4.14
sup tiṅ iti pratyāhāragrahaṇam subantaṃ tiṅantaṃ ca śabdarūpaṃ padasaṃjñaṃ bhavati.
A pada 2 has sup and tiṅ endings || 1, 4. 14
sup and tiṅ are endings. A form with a sub or tiṅ ending is termed pada 2.
The technical terms sup and tiṅ refer, respectively, to noun case-endings and verb endings, as specified in the Aṣṭādhyāyī. Apart from indeclinable particles, only words with sup and tiṅ endings, having undergone morphological changes, can appear in sentences. Considering this, earlier Abhidharma texts did not explicitly describe the combination of nāmans to form padas.
Then, does the Smra sgo’s theory have part of its basis in Abhidharma literature? Tracing its theory’s origin requires us to turn to the Apitan xinlun jing 阿毘曇心論經 (*Abhidharmahṛdaya) by Youbo shanduo 優波扇多 (Upaśānta, ?third century) which defines nāmapadavyañjanakāya as follows:Footnote 53
……句, 名字集, 隨所欲說義, 分齊究竟名句, 如婆伽羅那[*vaiyakaraṇa] 云言說 [*vākya]。味者字生。名者, 隨義名也, 如牛馬等, 如毘伽羅 [*vyākaraṇa] 論言句。
…a pada is an aggregation of nāmans and vyañjanas. According to the intended object-referent, the completion is determined (?). That is named pada. It is as the same as *vākya which is called by grammarians (*vaiyākaraṇa). The vyañjana arises from the akṣara. The nāman refers to objects such as cow, horse, etc., as in pada 2 according to grammatical conventions (*vyākaraṇa).
This passage presents a theoretical diversion in the definition of the nāmapadavyañjanakāya. The author equates pada with vākya in traditional Indian grammar as sentence and nāman with pada2. With this theoretical background, the possibility of vyañjanas forming nāmans and nāmans forming padas emerges.
Moreover, the concept of analysing language with the structure of the nāmapadavyañjanakāya was mostly used as a framework for Tibetan discussions of the philosophy of language and was never directly applied to the Sanskrit grammatical treatises or, as far as I know, to similar treatises of other Indic languages. At the same time, for a language belonging to the Sino-Tibetan language family like Tibetan, which is without sophisticated or complex inflected systems, the difference between pada1 and pada2 is not that obvious. This relatively independent language structure system provides a fertile ground for the use of triad as a linguistic theory.
Therefore, the Smra sgo could be the first Tibetan grammatical work to use the nāmapadavyañjanakāya as its overall framework for purposes of a grammatical analysis. Smṛti adopted this concept from the Sanskrit Abhidharma literature, while integrating the practical needs of the Tibetan language, by combining various Abhidharma concepts and adapting some Sanskrit grammatical notions into a comprehensive linguistic philosophical theory.
4. The theory of ultimate particles
Beyond the prominent concept of the nāmapadavyañjanakāya, the Smra sgo’s discussions on the composition and formation of language also incorporate Abhidharma theories of atomism and causality.
In the Smra sgo, the theory of atomism of the physical universe is used to explain the similar principle of language by taking yi ge (vyañjana) as the fundamental components of language in parallel with the ultimate particle (rdul phra ba, paramāṇu) of the material world. When discussing the sufficient condition (ldang tshad) of the yi ge, Smṛti says:Footnote 54
shin du phra ba’i rnam pa’i cha ||
de la’ang yi ge gcig ces bya ||
…shin du phra ba’i rnam pa zhes bya ba ni ming gi mtha’i yi ge yin te | dper na gzugs rnams bshigs na rdul phra rab kyi mtha’ thug ste de las lhag par phra ba’i gzugs gzhan med do || de bzhin du ming rnams bshigs na gang du mthar thug par ’gyur zhing de las bye brag tu dbye ba’i bogs dbyung du med pa ni yi ge gcig ces bya ste | gsal byed bco brgyad kyi cha lta bu’o ||
Also, that aspect of the minutest part [of language],
Is called a yi ge.
The so-called minutest part refers to the yi ge as the ultimate unit of nāman. For example, if one analyses form/matter, one will reach the extreme ultimate particle (*paramāṇu), that is, there is no other form/matter that is more minute than that. Similarly, if one analyses a nāman, the extreme, beyond which there is no finer distinction, is the yi ge, such as the eighteen phonetic values (gsal byed).Footnote 55
In this section, Smṛti interprets yi ge (vyañjana) by analogy with rdul phra rab (paramāṇu). This idea was not originally proposed by Smṛti; similar discussions can be found in the Abhidharmakośa, where paramāṇu and akṣara are discussed in parallel. The term gsal byed here does not refer to the vyañjana in the triad or its usual meaning in Tibetan grammar as a consonant, but refers to the different phonetic values of vowels (see the Appendix about this issue).
The Abhidharmakośa states the concept of atomism of the physical world, time and language as follows:Footnote 56
paramāṇvakṣarakṣaṇāḥ | rūpanāmādhvaparyantāḥ || 3.85bc
rūpasyāpacīyamān asya paryantaḥ paramāṇuḥ | kālasya paryantaḥ kṣaṇo nāmnaḥ paryanto ’kṣaraṃ tadyathā gaur iti |
The ultimate particles, akṣaras and instants are the ultimate units of form/matter, nāman and time respectively.
Analysing form, its ultimate unit is the ultimate particle. [Analysing time,] its ultimate unit is the instant. [Analysing nāman,] its ultimate unit is the akṣara, such as gau.Footnote 57
Comparing the two passages, it becomes apparent that the terminology and discussion of the Smra sgo closely mirror those of the Abhidharmakośa. The rare Tibetan term bshigs in the Smra sgo corresponds precisely to Sanskrit apacīyamant, and the phrase ming gi mtha’i yi ge aligns with Sanskrit nāmnaḥ paryanto ’kṣaraṃ. Therefore, it is extremely likely that this section of the Smra sgo is directly derived from the Abhidharmakośa.
Additionally, in the section on nāmakāya, the Smra sgo discusses the cause (rgyu) of nāman as follows:Footnote 58
de la rgyu ni dper na rdul ’dus pa las gong bu grub pa na de nyid dang gzhan du brjod du med pa bzhin yi ge ’dus pa las ming ’grub ste | yi ge nyid kyang ming ma yin | yi ge las gud na yang ming med de | yi ge ’dus pa’i gnas skabs la ming zhes bya’o ||
Here, regarding the cause, for example, when ultimate particles combine to form a lump (gong bu), there is nothing said about whether it is itself or something else. Similarly, yi ges combine to form a nāman; the yi ge as such is not a nāman. Nāman does not exist apart from yi ge. The circumstance of the combination of yi ges is called a nāman.
This passage uses the combination of ultimate particles as an analogy for the aggregation of yi ges. A lump is neither the ultimate particles themselves nor something other than the ultimate particles. Similarly, a nāman is neither a yi ge itself nor something other than a yi ge. In the Abhidharmakośa, the combination of ultimate particles is described as not actually touching but the word touching is used because of their extreme proximity and, apart from ultimate particles, there is no composite form.Footnote 59
At the end of the Smra sgo, Smṛti uses the theory of ultimate particlesFootnote 60 as an example to explain how to learn nāmapadavyañjanakāya:Footnote 61
dper na skyes bu gang zhig gis ||
gzugs kyi bye brag shes ’dod na ||
phra rdul chos la mtshan nyid du ||
de shes che rdul gong bu dang ||
gong chen dag gi mtshan nyid dang ||
las rnams ’bad pa med par shes ||
de shes skabs kyi bya ba gang ||
sbyor ba dag la mi ’jigs bzhin ||
zhes bya ba smos te | dper na skyes bu zhig gzugs kyi bye brag shes par ’dod na dang por rdul phra rab kyi mtshan nyid la mkhas par bya’o || de nas rdul ’dus te rdul chen por ’gyur te de shes par ’gyur ro || rdul chen po ’dus pa las gong bur ’gyur te de shes par ’gyur ro || gong bu las gong chen du ’gyur te de’i mtshan nyid dang las dag kyang shes par ’gyur ro || de ltar las dang mtshan nyid legs par shes pa’i skyes bu de la gang zhig gzugs kyi bye brag ’dri ba na ’jigs pa med pa de bzhin du ||
For example, if a person desires to know the divisions of form/matter, he must understand the characteristics of ultimate particles, then understand the aggregation of large particles and lumps. And the characteristics of large lumps, he knows without effort.
For him, whatever the object of knowledge will be, there is no fear.
For example, if a person wants to know the divisions of form/matter, he must first understand the characteristics of ultimate particles. Then, they must understand how these ultimate particles aggregate to form large particles, how large particles aggregate to form lumps and how lumps aggregate to form large lumps. Similarly, understanding the characteristics and functions of these aggregations enables a person to respond without fear when asked about any division of form…
The Abhidharmakośa describes the size of ultimate particles in terms of the measure (pramāṇa) of form/matter. Every seven ultimate particles form a particle (aṇu), and this aggregatesFootnote 62 successively in multiples of seven up to a digit of a finger (aṅgulīparva).Footnote 63 That system is very similar with what Smṛti tries to explain here; however, apart from ultimate particles and subtle particles, he does not use the terminology found in Abhidharma texts. It appears that Smṛti once again adapted the relevant Abhidharma passage to make it more comprehensible to his readers.
However, there is a fundamental difference between ultimate particles and yi ge, which Smṛti does not mention. As stated in the Abhidharmakośa, ultimate particles are the smallest units of form/matter, whereas vyañjana or akṣara, like moments, can only be applied at the level of time. This issue is extensively discussed in the Abhidharmakośa and its commentaries.Footnote 64 Unlike form, when a person speaks, the sound cannot be aggregated into a single moment, which is a major difference between the material and language. This distinction is a primary argument for the Sautrāntika’s claim that the nāmapadavyañjanakāya is not a real entity, and also a basis for the Sarvāstivāda’s rebuttal. As stated by Saṅghabhadra, the real existence of nāman and pada is based on a kind of integrality, like the shadow of a tree: even though the shadow came into being based on the aggregation of elements of a tree, the shadow itself is a real whole entity.Footnote 65
5. The generation of speech
Another theory of Abhidharma philosophy that Smṛti refers to is the causality of entities illustrated in the Abhidharmakośa. After describing the shapes of letters in the Smra sgo,Footnote 66 Smṛti further explains the cause of verbal speech (ngag) as expressed by the written akṣaras or signs (brda).Footnote 67 Smṛti first elucidates the process of the generation of speech from five aspects and then summarizes it into seven causes (rgyu bdun) and explains their nature one by one.
According to him, the generation of speech proceeds along five phases. The first is the intentional stimulation of the air, which causes the fluctuation of the vocal organs, leading to the production of sound:Footnote 68
da ni brda des bstan pa’i ngag rnams ’byung ba ni rgyu bdun tshogs pa las skye bar ston te | de la bskul ba’i rgyu rnam par rtog pa yin par bstan pa’i phyir |
des bstan brda can ’byung ba na ||
lcags rnams la ni khab len bzhin ||
rnam rtog dbang gis gnas rnams g.yo ||
zhes bya ba smos te | khab len gyi dbang gis lcags g.yo ba bzhin du rnam par rtog pa’i dbang gis bā ya ta zhes bya ba’i rlung lus la gnas pa bskul nas des sgra ’byung ba’i gnas rnams g.yo bar byed do zhes bya bar ’gyur ro ||
Now I will show that the occurrence of the speech that is indicated by these signs (brda, *saṃketa) arises from the collection of seven causes, that is, in order to show that the stimulating cause (bskul ba’i rgyu) is a conceptual activity (rnam par rtog pa, *vikalpa), it is said:
If what has a sign [speech] that is indicated by them occurs,
then just like a magnet that moving iron,
the places of articulation move due to force of conceptual
activity.
Just as iron moves due to the influence of a magnet, the conceptual activity stimulates the wind called *vāyata (vāyava?) within the body, causing the places of articulation that produce sound to move.
Smṛti further explains:Footnote 69
rnam par rtog pas ji ltar rlung skyod ce na |
ji ltar gnas gzhi med bzhin du ||
mkha’ la ’dab bskyod rlung ’byung ltar ||
rnam par rtog pas bskyod pa na ||
bā ya ta zhes bya ba’i rlung ||
lus la rab tu g.yo bar ’gyur ||
zhes bya ba smos te | dper na bya nam mkha’ la ’phur ba’i tshe ’dab ma bskyod pa nyid kyis rlung ’byung ba ltar [1850] ming dang | ming can ’brel ba’i rnam par rtog pa gzhan la go bar byed pa’i blos skyod par byed pa’i tshe lus thams cad la khyab par gnas pa’i rlung g.yo bar byed pa can bā ya ta zhes bya ba’i rlung ldang zhing g.yo bar ’gyur ro || rlung de g.yos pas cir ’gyur zhe na | byed pa rnams kun tu g.yo bar ’gyur te | byed pa rnams bskyod pas ni gnas dang phrad par byed do || de bstan pa’i phyir |
g.yo rlung rab tu g.yos pa yis ||
byed pa kun tu bskyod pas na ||
dog dang yangs pa’i gnas rnams su ||
phrad dang cung zad phrad par byed ||
ces bya ba smos te de la dog pa ni mgul pa dang | sna dang | klad pa’i sbubs so || lhag ma ni yangs pa’o || rkan dang lce sbyar ba dang | so dang lce sbyar ba dang | mchu gnyis sbyar ba ni phrad pa’o || lhag ma ni cung zad phrad pa ste ’di yang’og tu ’chad par ’gyur ro ||
How does the conceptual activity drive the airstream? The answer is:
Just as without a fixed base,
wings moving in the air create wind.
When the conceptual activity moves,
the wind called *vāyata moves within the body.
For example, when a bird flies in the sky, merely by flapping its wings, wind is generated. Similarly, the conceptual activity associated with names and their meanings, driven by the intention to make others understand, causes the wind *vāyata, which pervades the entire body, to move and fluctuate.
If you ask, where does this wind reach? It reaches all the active articulators (byed pa). As these organs are moved, they make contact with the passive articulators that produce sound. To explain this, it is said:
As the wind moves, the active articulators are driven,
in narrow and wide positions, causing contact and
slight contact.
Among them, the narrow positions are the throat, nasal cavity and head cavity; the rest are wide positions. The connection of the soft palate and tongue, the connection of the teeth and tongue, and the connection of the lips are called “contact”; the rest are “slight contact,” which will be further explained below.
When the *vāyata wind causes the active articulators to contact the passive articulators, sound is produced. Accompanying the sound is the niḥśvāsita (exhalation) wind, thus making it a concomitant cause (lhan cig ’byung ba’i rgyu):Footnote 70
de’i tshe lhan cig ’byung ba’i rgyu bstan pa’i phyir | de’i tshe niḥ shvā svā zhes pa’i rlung dang lhan cig ’byung ba ste zhes bya ba smos te | ’di ltar gnas dang byed pa phrad pa’i tshe niḥ shvā svā zhes bya ba’i rlung ni dbugs yin te | ’di dang sgra’i tshogs su ’dres nas ’byung ste | ’di ni lhan cig ’byung ba’i rgyu yin no ||
To explain the concomitant cause, it is said that at that time, the *niḥśvāsvā (niḥśvāsita, exhalation) wind is produced together with sound. When the vocal positions and the active organs make contact, the niḥśvāsita wind is exhaled, mixed with the sound, making it a concomitant cause.
The all-pervasive cause (kun tu ’gro ba’i rgyu) present in all the nāmapadavyañjanakāya is explained by the verse:Footnote 71
kun tu ’gro ba’i rgyu bstan pa’i phyir |
ji ltar gzugs la ’byung ba bzhin ||
nga ro la ni brten bzhin du ||
zhes bya ba smos te | dper na gzugs thams cad ’byung ba chen po bzhis bskyed pa bzhin du yi ge’i tshogs thams cad la dbyangs kyi yi ge bcu drug dang gsal byed bco brgyad kyis khyab ste | de bas na nga ro zhes bya ba dbyangs dang gsal byed la brten nas yi ge dang | ming dang | tshig gi tshogs thams cad ’byung ste ’di ni kun tu ’gro ba’i rgyu yin no ||
To explain the all-pervasive cause, the verse says:
Just as all forms arise from the four great elements,
so do all literary bodies depend on the vowels.
For example, just as all forms arise from the four great elements (’byung ba chen po bzhi), all vyañjanakāya are pervaded by the sixteen vowel characters and their eighteen variations. Therefore, relying on the vowels (nga ro, *svara) and their phonetic values (gsal byed),Footnote 72 the nāmapadavyañjanakāya are produced. This is the all-pervasive cause.
Moreover, before the articulation, the mental image corresponding to the character to be pronounced, analogous to the image in the mind of the speech, is called the homogeneous cause (mnyam pa’i rgyu). This mental impression evokes the concept of sphoṭa by Bhartṛhari (?sixth century) and the universal phonetic image (sgra spyi snang ba) by Sa skya Paṇḍita.Footnote 73 Smṛti writes:Footnote 74
skal ba mnyam pa’i rgyu bstan pa’i phyir |
rnam rtog rang gi snang cha las ||
gzugs brnyan bzhin du cha mnyam skye ||
zhes bya ba smos te | de’i tshe rnam par rtog pa rang gi yul la snang ba’i cha ming dang ming can ’brel par snang bar ’gyur te | ’di ltar ka ba’i dngos po la ka ba zhes ming du btags pa de dang ’brel par brda’i dus na zhen pas phyis don la skyes bu ’jug pas na rang gi snang ba mthun pa’i yi ge ’byung bar ’gyur te | dper na bzhin gyi dkyil ’khor ji ltar gzugs brnyan ’byung ba bzhin no || ’di ni skal ba mnyam pa’i rgyu yin no ||
To explain the homogeneous cause, it is said:
From the conceptual activity’s own image,
An image similar to the reflection arises.
At that time, the conceptual activity produces an image associated with names and their meanings. Just as the designated name ka ba (pillar) corresponds to its named object, the letters corresponding to one’s mental image are produced when the conceptual activity moves. This is like the reflection in a mirror. This is the homogeneous cause.
As described above, the causes of the generation of speech can be categorized into seven kinds:
1. Fundamental cause (gzhi’i rgyu), such as the passive vocal organs (gnas pa), including the palate, throat, teeth, etc.
2. Active cause (’byin par byed pa’i rgyu), such as the active vocal organs (byed pa), including the tongue, lips, etc., and the contact (phrad) between active and passive vocal organs.
3. Stimulating cause (bskul ba’i rgyu), the conceptual activity (rnam rtog, *vikalpa).
4. Driving cause (skyod par byed pa’i rgyu), referring to the wind (vāyata?; rlung) that drives the movement of the active vocal organs.
5. Concomitant cause (lhan cig ’byung ba’i rgyu), the exhalation (niḥśvāsvā/niḥśvāsita, dbugs), which is produced together with sound.
6. Fruition cause (kun tu ’gro ba’i rgyu), referring to the 16 vowels and their 18 phonetic values (gsal byed) (see the Appendix for the meaning of gsal byed here).
7. Homogeneous cause (skal ba mnyam pa’i rgyu), the mental image (snang ba) of the character to be pronounced, which is produced by the conceptual activity.
Abhidharma texts of Sarvāstivāda generally classify causes (rgyu, hetu) into six kinds, known as the six causes (ṣaḍvidho hetuḥ). These first make their appearance in the Abhidharmajñānaprasthāna.Footnote 75 Interestingly, the section on the six causes happens to follow immediately after the discussion of nāmapadavyañjanakāya. According to the Abhdharmakośa, the six causes are:Footnote 76
1. Instrumental cause (kāraṇahetu, byed pa’i rgyu)
2. Concomitant cause (sahabhūhetu, lhan cig ’byung ba’i rgyu)
3. Homogeneous cause (sabhāgahetu, skal ba mnyam pa’i rgyu)
4. Associated cause (saṃprayuktahetu, mthungs par ldan pa’i rgyu)
5. Dominant cause (sarvatragahetu, kun tu ’gro ba’i rgyu)
6. Fruition cause (vipākahetu, rnam par smin pa’i rgyu)
Comparing the seven causes of how speech comes about in Smra sgo with the six causes, it shows that Smṛti does not strictly follow the six causes in his analysis. In his discussion, he uses terms such as instrumental cause, concomitant cause, dominant cause and homogeneous cause which are consistent with four of the six causes. The remaining fundamental cause and stimulating cause can be subsumed under the instrumental cause according to the definitions in the Abhidharmakośabhāṣya.Footnote 77 However, Smṛti’s explanations of concomitant, fruition and homogeneous causes involve literal interpretations that differ somewhat from the definitions in the Abhidharma.
6. Conclusion
As the title of the Smra sgo indicates, Smṛti envisioned that this book would be a gateway into speech that would make people understand the contents, and a weapon that would help them attain wisdom and dispel ignorance.Footnote 78 Within Sarvāstivāda philosophy, the fundamental elements of language are classified as nāmakāya, padakāya and vyañjanakāya. This triad thus functions as a true gateway to language, and Smṛti adopts it as the underlying framework for structuring both the content and organization of the Smra sgo.
At the same time, in order to fulfil the practical goals of the text – not merely as a philosophical treatise, but also as a functional linguistic manual – Smṛti expands upon the traditional concept of nāmapadavyañjanakāya. He does so by integrating philosophical theories from the Sarvāstivāda tradition, such as atomism and causality, with classical Sanskrit grammatical notions like stems and endings, etc.
The Smra sgo thus marks a critical moment in the history of Tibetan linguistic thought: it establishes nāmapadavyañjanakāya as the foundational structure of Tibetan grammatical discourse and initiates the formalization of indigenous Tibetan linguistic theory. One can easily imagine the significant impact such a meticulously crafted and conceptually rich work would have had in its time. As both a philosophical and grammatical innovation, the Smra sgo not only offered new perspectives for the development of Tibetan linguistic literature but also stands as a unique contribution within the broader Abhidharma tradition.
Abbreviations
- Bstan
Bstan ’gyur dpe sdur ma, edited by Krung go’i bod kyi shes rig zhib ’jug lte gnas kyi bka’ bstan dpe sdur khang, 120 vols. Beijing: Krung go’i bod kyi shes rig dpe skrun khang, 1994–2008.
- T
Taishō shinshū daizōkyō 大正新脩大藏經 [Taishō Tripiṭaka], edited by Takakusu Junjirō 高楠順次郎 et. al., 55 vols. Tokyo: Taishō Issaikyō Kankōkai, 1924–1935.
Acknowledgements
This article is a revised and expanded version of my earlier work (Li Reference Li2025). I am sincerely and deeply grateful to Professor Leonard W. J. van der Kuijp and Professor Sa’erji for their generous help and suggestions, all of which have greatly improved this article. I also wish to express my sincere thanks to the three anonymous reviewers of this journal for their careful reading and insightful comments.
Funding statement
This article was supported by the National Funded Postdoctoral Researcher Program (Project No. GZC20230099) and by a major project of the Key Research Base for Humanities and Social Sciences of the Ministry of Education (Institute of Foreign Philosophy, Peking University) (Project No. 22JJD72004).
Competing interests
The author declares none.
Appendix: On the expression of gsal byed bco brgyad in the Smra sgo
In Tibetan grammar, the term gsal ba[bar] byed pa generally refers to the consonant. However, it is different with the Smra sgo, and the Sanskrit original might not be vyañjana but bhinna as indicated in the Kāśikāvṛtti 1.1.9. There it refers to the 18 different phonetic values for each of the three vowels a, i and u. These values are derived from the three different lengths (short, long, prolated), the three pitches (high, middle, low) and the presence or absence of nasalization. When multiplied together, they result in 18 distinct sounds. The Smra sgo thus states:Footnote 79
nga ro dag gi gsal byed ni ||
drang po bkug dang smad pa la ||
’dren pa ring thung gsum yod de ||
de la’ang che ’bring rnam pa gsum ||
de dgu sna ldan bco brgyad do ||
zhes bya ba smos te | dbyangs kyi yi ge de rnams kyi gsal byed ni drang po a dang | bkug pa i dang | smad pa u dang gsum la sgra’i ’dren pa gsum gsum yod de | ’di ltar hra sva zhes bya ba thung ngu ste tha mal pa dang | dī rgha zhes bya ba ring po ste gnyis kyi yun du brjod pa dang | plu ta zhes bya ba ni shin tu ring po ste | gsum gyi yun du brjod pa’o ||
de re re la yang che ’bring gsum gsum ste | u dātta zhes bya ba chen po ste | yi ge gnyis kyi sgra can dang | a nu dātta zhes bya ba chung ngu ste phyed pa dang | sva ri ta zhes bya ba bar ma ste yi ge phyi mo gcig pa yin te | de ltar dgur ‘gyur ro ||
de la yang oṃ zhes bya ba’i sgra sna dang ldan pa zhes bya ba dang | sna dang mi ldan par phye bas ’di ltar bco brgyad du ’gyur ro ||
The gsal byed of vowels:
To draw out the straight (a), tensed (i) and lower (u) [sounds], each has three [forms]: prolated, long and short.
Furthermore, [each has] high, middle [and low pitches].
These nine have nasal [and non-nasal sounds], eighteen in all].
As mentioned above, about the phonetic values (gsal byed) of vowel syllables, in the straight [sound] a, the tensed [sound] i and the lower [sound] u, each of them has three lengths of sound. In this way, these are the short and ordinary sound, known as hrasva; the long sound, known as dīrgha, pronounced with twice the duration; and the prolated sound known as pluta, pronounced with three times the duration.
Each of these types [can further be divided into] high, middle [and low] three [pitches]. The high pitch, known as udātta, has a sound value of two units; the low pitch, known as anudātta, has half that value; and the middle pitch, known as svarita, retains the original sound value of one unit. In this way, there are nine variations [of each vowel syllables].
Furthermore, they can also be differentiated as having a nasal sound, like oṃ, or having no nasal sound. In this way, there are 18 [variations of vowel syllables].
Candragomin’s (?sixth century) Varṇasūtra and the commentary of Aṣtādhyāyī – Kāśikāvṛtti – also applied this kind of division:Footnote 80
atra cāvarṇo hrasvo dīrghaḥ pluta iti tridhā bhinnaḥ || pratyekam udāttānudāttasvaritabhedena sānunāsikaniranunāsikabhedena cāṣṭādaśadhā bhavati || Varṇasūtra 36-37
ekamātriko hrasvaḥ || dvimātriko dīrghaḥ || trimātrikaḥ plutaḥ || Varṇasūtra 41-43
uccair udāttaḥ || nīcair anudāttaḥ || samāhāraḥ svaritaḥ || Varṇasūtra 44-46
Here, the vowel a is categorized into three types: short (hrasva), long (dīrgha) and prolated (pluta). Because each of these is further differentiated into high (udātta), not-high (anudātta) and mixed (svarita) [pitches] and by nasalized (sānunāskika) and non-nasalized (niranunāsika) sounds, there are 18 forms in total.
The short is with one unit of length (mātrika); the long is with two units of length, the prolated is with three units of length.
The high is acute (uccais), the not-high is low (nīcais) and the mixed is assembled (samāhāra).
a a a iti trayo ‘kārā udātta-anudātta-svaritāḥ pratyekaṃ sānunāsikā niranunāsikāś ca hrasva-dīrgha-pluta-bhedād aṣṭādaśadhā bhidyante || Kāśikāvṛtti 1.1.9
“a a a” which represent the high, not-high and mixed [pitches] a syllables are further differentiated into nasalized and non-nasalized sounds and short, long and prolated, so they are divided into 18 types.
One of the earliest Tibetan compositions on linguistics as such, the Gnas brgyad chen po, has also this kind understanding of gsal byed:Footnote 81
gsal ba byed pa ni bco brgyad de | a’i sgra la brten nas thung ngu dang | bar ma dang ring du ’dren pa gsum la che ba dang chung ba dang bar ma gsum gsum du phye bas dgur ’gyur ro || de la sna dang ldan pa dang mi ldan par phye bas bco brgyad … ||
There are clearly 18 phonetic values derived from the vowel a. Basing on the vowel a, there are three lengths: short, medium and long and also low, medium and high pitches. Divided into three times three, there are nine variations. Differentiated them as nasalized and non-nasalized sounds, this totals 18…
