Memorizing a thesaurus

« previous post | next post »

Sounds like fun, doesn't it?

People actually did it in ancient India, and they still do it today.

Here are some passages from the Wikipedia article about the Amarakosha, the most celebrated and most often memorized Indian thesaurus.


The Amarakosha (Devanagari: अमरकोशः, IAST: Amarakośa) is the popular name for Namalinganushasanam (Devanagari: नामलिङ्गानुशासनम्, IAST: Nāmaliṅgānuśāsanam) a thesaurus in Sanskrit written by the ancient Indian scholar Amarasimha. It may be the oldest extant kosha. The author himself mentions 18 prior works, but they have all been lost. There have been more than 40 commentaries on the Amarakosha.


The word "Amarakosha" derives from the Sanskrit words amara ("immortal") and kosha ("treasure, casket, pail, collection, dictionary"). The actual name of the book "Namalinganushasanam" means "instruction concerning nouns and gender".


Amarasimha is said to have been one of the Navaratnas ("nine gems") at the court of Vikramaditya, the legendary king inspired by Chandragupta II, a Gupta king who reigned around AD 400. Some sources indicate that he belonged to the period of Vikramaditya of 7th century

Textual organization

The Amarakosha consists of verses that can be easily memorized. It is divided into three khāṇḍas or chapters. The first, svargādi-khāṇḍa ("heaven and others") has words pertaining to gods and heavens. The second, bhūvargādi-khāṇḍa ("earth and others") deals with words about earth, towns, animals and humans. The third, sāmānyādi-khāṇḍa ("common") has words related to grammar and other miscellaneous words.

Svargadhikhaanda, the first Khaanda of the Amarakosha begins with the verse 'Svaravyam swarganakathridivatrishalaya..' describing various names of Heaven viz. Sva, Avya, swarga, Naka, Tridiva, Tridasalaya etc. The second verse 'Amara, nirjara, deva,’ describes various words that are used for gods and demigods. The fifth and sixth verses give various names of Buddha and Shakyamuni (i.e. Gautama Buddha). The following verses give the different names of Brahma, Vishnu, Vasudeva, Balarama, Kamadeva, Lakshmi, Krishna, Shiva, Indra etc. All these names are treated with great reverence. While Amara Simha is regarded to have been a Buddhist, Amarakosha reflects the period before the rise of sectarianism. Commentaries on Amarakosha have been written by Brahmanical, Jain and well as Buddhist scholars.

The second Kanda, Bhuvargadhikanda, of the Amarakosha is divided into ten Vargas or parts. The ten Vargas are Bhuvarga (Earth), Puravarga (Towns or Cities), Shailavarga (Mountains), Vanoshadivarga (Forests and medicines), Simhadivarga (Lions and other animals), Manushyavarga (Mankind), Bramhavarga (Brahmin), Kshatriyavarga (Kshatriyas), Vysyavarga (Vysyas) and Sudravarga (Sudras).

The Third Kanda, Samanyadhikanda contains Adjectives, Verbs, words related to prayer and business etc. The first verse Kshemankaroristatathi Shivathathi Shivamkara gives the Nanarthas of the word Shubakara or propitious as Kshemankara, Aristathathi, Shivathathi and Shivamkara.


    • Amarakoshodghātana by Kṣīrasvāmin (11th century CE, the earliest commentary)
    • Tīkāsarvasvam by Vandhyaghatīya Sarvānanda (12th century)
    • Rāmāsramī (Vyākhyāsudha) by Bhānuji Dīkshita
    • Padachandrikā by Rāyamukuta
    • Kāshikavivaranapanjikha by Jinendra Bhudhi
    • Pārameśwari by Parameswaran Mōsad in Malayalam
    • A Telugu commentary by Linga Bhatta (12th century)

Most fascinating to me is that "Gunaratha" of Ujjain is said to have translated the Amarakosha into Chinese in the 7th century.  If true, it would be of extraordinary significance, but I have never heard of such a translation, and I am puzzled by how i have been unaware of it for the past half century of my Sinological and Buddhological studies.  I will explore this question further in the remainder of this post.

According to Sanskrit eBooks:

It is of great interest to note that, though the production of a Buddhist, it has been universally accepted as an authority by the Brahmans and the Jainas alike. The fact that it has been commented upon by Buddhists like Subhutichandra, by Jainas like Asadharapandita and Nachiraja, and by Brahmans like Kshirasvamin, Mallinatha and Appayyadikshita testifies to its usefulness to every class of Sanskrit students.  It is a well-known fact that translations of the Amarakosha into Chinese and Thibetan have been recently discovered.

I'm cautious about the veracity of every statement in that paragraph, especially the last sentence, about which I am innately suspicious.  If there is a Tibetan version, I believe that it would not be a recent discovery, and, if there is a Chinese version, in all likelihood I would already have encountered it by now, but I yet to come across a Chinese version of the Amarakosha.

Here's a more professional, scholarly entry on the Amarakosha from the Cambridge University Digital Library:

Amarakoṣa (MS Add.1650)

The Amarakoṣa by Amarasiṃha, probably a Buddhist author, is the most renowned Sanskrit lexicographical work, seemingly composed around the middle of the first millennium CE. "The bulk of the Amarakoṣa is a synonymic dictionary whose articles are grouped subjectwise" (Vogel 1979: 311). The fame of the "Immortal Lexicon" goes far beyond the boundaries of the Indian Subcontinent, as testified by its renderings in Tibetan, Chinese, Mongolian, Sinhalese and Burmese, among others. A further proof of its importance and popularity is the number of commentaries dedicated to it: at least eighty, of which many still remain unpublished. This heavily annotated Nepalese palm-leaf manuscript is probably to be dated to the 14th or the 15th century.

Ramanathan, A. A., Amarakośa of Amarasiṃha, with the unpublished South Indian commentaries, Adyar Library series 101 (Madras: Adyar Library and Research Centre,, 1971) vol. 101.

This extensive article from the Wisdom Library draws on numerous sources and includes many curious facts about the Amarakosha.

Here is the entry on Amarakosha from "A Chronology of Major Events in the History of Lexicography (Oxford Handbooks Online)" which puts it in the overall framework of the development of Indian lexicographical works.

6th cent.?

Amarasiṃha compiles the thematically arranged metrical dictionary known as Amarakoṣa, the most famous of the early lexica of Sanskrit. More than eighty commentaries would be written on it, the earliest around 1000; it would be translated into Burmese, Nevārī, Tibetan, and Mongolian; and it would be mentioned as a forerunner by P. M. Roget in the introduction to his Thesaurus of 1852.

By chance, I came upon this remarkable article that is highly relevant to our current inquiry:

Jens Braarvig, "The Imprint of Buddhist Sanskrit on Chinese and Tibetan: Some Lexical Ontologies and Translation Strategies in the Tang Dynasty", Ch. 16, in: Jens Braarvig and Markham J. Geller: Studies in Multilingualism, Lingua Franca and Lingua Sacra Online version, pp. 427-451.  (pdf)  Max Planck Research Library for the History and Development of Knowledge Studies 10 (2018). 

It provides abundant evidence of a tradition of Sanskritic style synonymicons in Tibetan and to a lesser extent in Chinese, but no evidence of Tibetan and Chinese translations of the Amarakosha per se.

The Amarakosha does exist in Tibetan.  It is titled ‘Chi med mdzod, which literally means "Immortal Treasury," being an abbreviation of Amarasiṃha's name ("Immortal Lion") and the standard Tibetan translation for kośa ("storeroom; treasury; repository").  It was translated into Tibetan by mahapandita Kirticandra and Yar lungs lo tsā ba Grags pa rgyal mtshan.  The text was translated into Tibetan during the second dissemination of the dharma, probably in the late thirteenth century.  It is included in editions of the Tenjur / Tengyur section of the Tibetan canon (e.g., D. 4299), along with Subhūticandra’s commentary (D. 4300), and was widely cited.

The closest we get to an alleged Chinese version of the Amarakosha in Chinese may be had in these remarks of Fabio Rambelli:

The presumed translator of this dictionary was Gunaratha, that is, Paramārtha (Ch. Zhendi 真諦; 499–569 CE), to whom can be attributed a dictionary, the "Translation of Foreign Words (Fan waiguo yu 翻外國語), in seven fascicles (also known as Za shi 雜事 or the Jushe lun yinyuan shi 倶舍論因縁事, T49, 88a). This treatise is listed in the Lidai sanbaoji and the Neidian lu as the last one in the list of Paramārtha’s works […] We know from its name that it was probably not a translation but a composition by Paramārtha." (from Funayama Tōru, (2010). The Work of Paramārtha: An Example of Sino-Indian Cross-cultural Exchange, Journal of the International Association of Buddhist Studies 31, 1/2, 141 – 183, (cit. pp. 149-150)). I am not familiar with this work though.

The fact that the alternative title of  Paramārtha's work begins with jùshě 倶舍, the standard Chinese transcription of Sanskrit kośa, may have led some to confuse it with the Amarakosha.

These remarks of Max Deeg support my contention about kośa and add important information concerning the Tibetan translation of the Amarakosha and its Sinitic analogs:

I would advise to consult Vogel’s small volume in Gonda’s History of Indian Buddhism on Sanskrit dictionaries, including the Namalinganusasana aka Amarakosa. It is still quite valuable and also contains a footnote on the so-called Chinese translation p.313, note 41 to which I would add a reference in the Lidai sanbao ji (T.2034.88a.22ff.). I agree with Vogel that nothing – except for the kosa in the alternative title 俱舍論 – in the catalogue entries indicates that this Fan waiguoyu is a translation of the Amarakosa (the literally translated Chinese name of which would be something like 無死藏, 不死藏 or 甘露藏). The seven juan given in the Chinese catalogues also does not fit the three chapters of the Skt. text. On the Tibetan translation see Sherab Gyaltsen, Amarakoṣa and its Tibetan translation, ʼChi med mdzod: a metrical dictionary of the Sanskrit language by Amarasiṃha with its Tibetan translation made by Mahāpaṇḍita Kīrticandra and Yar-luṅs Lo-tsā-ba Grags-pa-rgyal-mtshan, Gangtok 1984.

These observations by Robert Goldman are also valuable for getting a sense of the general impact of the Amarakosha in India:

The AK has been widely commented upon by scholars of all three religious traditions. Including the ones you mention, I believe.  It also has comms. by  other authors such as the Jain Jinendra Budhi, Rāyamukuta, and Bhānuji Dīkṣita, as well as others in regional languages like Malayalam and Telugu. I have not looked at any of these works so I can’t say too much about them.  But it is not unusual for commentators to use  the works of others from other traditions. In any case the AK, along with other koshas is widely quoted by many scholiasts.

These penetrating insights by Deven Patel provide additional context for understanding why the Amarakosha was such a vital text for Indian scholars and why many of them would have been tempted to memorize it:

I read the Amarakosha with Dr. Ram Karan Sharma (who was a visiting professor of Sanskrit at Penn about six years ago), along with a commentary.  I recorded those sessions because studying this work with a pandit really opens up the lexicon for you.  Professor Sharma would quote numerous sources where a particular word from the Amarakosha occurs since he had thousands of memorized verses at the tip of his tongue.  A student like me could, therefore, not only study (and memorize) the individual verses of the Amarakosha, which is set up like a thesaurus, but also have access to a diversity of references from poems, philosophical works, scientific texts, etc. Sadly, I didn't memorize the Amarakosha.  Most traditionally trained scholars have that work memorized from start to finish.

That the Amarakosha transcends any particular religious orientation is borne out by these perceptive comments by Varun Khanna:

It is true that the Amarakośa is of extreme value to every class of Sanskrit student, and that members of every Sanskritic tradition have commented on it. However, that Amara Simha was a Buddhist is a dubious claim. I'm not sure that we have compelling evidence one way or another. In any case, I would steer clear of any claims about his religious persuasion, and you'll be good.

To sum up, the Amarakośa had broad authority apart from any particular religious tradition. I doubt that it was ever directly translated into Chinese in any recognizable form. Yijing (635-713) discusses Sanskrit grammar studies in his travel account, yet he does not mention this work while mentioning others.  Whether there ever was a direct translation of the Amarakośa into Chinese does not matter so much as that, through the thoughts and works of Chinese pilgrim-monks such as Faxian (337 – c. 422), Xuanzang (fl. 602 – 664), and Yijing, who were learned in Sanskrit, it would have had an impact on Sinitic lexicography and rhetoric.


Selected readings


[Thanks to Donald Lopez, Daniel Boucher, Leonard van der Kuijp, Jonathan Silk, Janet Gyatso, Douglas Duckworth, and Nathan Hill]


  1. Cervantes said,

    October 28, 2020 @ 1:05 pm

    This is as far outside any expertise of mine as magnetohydrodynamics, but if I understand your description of the content correctly I don't see how it could have been translated wholly into Chinese since it largely concerns entities that did not exist in Chinese culture and in any case, it's a thesaurus. The idea of "translating" a thesaurus seems nonsensical in principle. Am I missing something?

  2. Victor Mair said,

    October 28, 2020 @ 1:22 pm


    It was translated into Tibetan and other languages, but your question is thought-provoking nonetheless.

  3. Victor Mair said,

    October 28, 2020 @ 1:32 pm

    From Janet Gyatso:

    Whether it is the first koṡa I don’t know. Don’t forget the great Abhidharmakoṡa, written in the 4th cent.

  4. Victor Mair said,

    October 28, 2020 @ 1:35 pm

    From Patrick Olivelle:

    Amara’s koṡa is cited in all kinds of works, including commentaries on the Dharmaśāśtras. I think the last comment by Varun Khanna is wrong. If you look at the beginning of the description of gods, you have three verses on the Buddha: sarvajñaḥ sugato buddha . . . (1. 13-15) that precede Brahmā etc.

  5. cameron said,

    October 28, 2020 @ 3:36 pm

    My first reaction was similar to that of Cervantes, in the first comment above: what does it even mean to translate a thesaurus?

    To my mind a translation of a lexicon organized as a thesaurus could be an analogous lexicon in the target language, organized along the same principals as the original, but reflecting the vocabulary of the target language. Or it could be effectively a bilingual dictionary, but organized like the original thesaurus – an aid for readers of the target language who wish to study or translate Sanskrit texts.

    Do the extant translations of this work resemble either of those possibilities, or are they something else entirely?

  6. Victor Mair said,

    October 28, 2020 @ 4:35 pm

    From Max Deeg:

    Further to my previous thoughts and spinning off from your blog: I think that the fact that Sanskrit dictionaries were, to a great extent, synonymic makes them indeed not very good targets for a translation into a language like Chinese. There is, however, from the beginning the Indian tradition of nirvacana (“etymology”, semantic analysis) at work when the Buddhist texts are translated from Indic languages (Gandhari, Sanskrit) into Chinese. Paramartha is a good example as he so frequently inserts such analysis into his work. This is much more interesting in terms of linguistic engagement from the Chinese side with Indic languages. I think it was this activity combined with the information brought back by the Tang “pilgrims” that prompted dictionaries like the Yiqiejing yinyi, Anyway, just another direction of thought …

  7. crturang said,

    October 28, 2020 @ 5:34 pm

    >>Amara’s koṡa is cited in all kinds of works, including commentaries on the >>Dharmaśāśtras. I think the last comment by Varun Khanna is wrong. If you look at >>the beginning of the description of gods, you have three verses on the Buddha: >>sarvajñaḥ sugato buddha . . . (1. 13-15) that precede Brahmā etc.

    The first maMgalAcaraNam verse that goes yasya jnAnadayAsindhOH that does not refer to any specific deity suggests a Buddhist origin too.

  8. David Marjanović said,

    October 28, 2020 @ 5:55 pm

    I think the last comment by Varun Khanna is wrong. If you look at the beginning of the description of gods, you have three verses on the Buddha: sarvajñaḥ sugato buddha . . . (1. 13-15) that precede Brahmā etc.

    I agree this is good evidence, and I guess we shouldn't be surprised this wasn't seen as a reason by other traditions to discard the work. In Christianity, a work by a heretic is a Work of the Devil; it is steeped in Evil, and so is everything it touches, so you better leave it alone. In India, wouldn't authors from other traditions be more likely seen as regrettably mistaken about one particular thing but worth listening to on everything else? That would be more like the Christian attitude to pre-Christian philosophers like Aristotle, or even Muslim authors like Avicenna & Averroes whose geographic locations might have been taken to "excuse" not having heard "the Truth" – and there was still quite a lot of resistance to translating their works.

  9. Hiroshi Kumamoto said,

    October 29, 2020 @ 12:31 am

    The Wikipedia article quoted at the top is, as usual, a mixture of useful information and inexactitudes. The first four items in the “Commentaries” section are, according to Claus Vogel, Indian Lexicography (which lists 17 important Sanskrit commentaries with bibliographical details among at least 80 extant):

    Amarakoṣodghāṭana of Bhaṭṭa Kṣīrasvāmin (first half of the 12th c.)
    Ṭīkāsarvasva of Vandyaghaṭīya Sarvānanda (1159/60)
    Vyākhyāsudhā (Rāmāśramī) of Bhānuji Dīkṣita (between 1620 and 1640)
    Padacandrikā of Rāyamukuṭamaṇi (1431/1432)

    Kāśikā-vivaraṇa-pañjikā alias Nyāsa by Jinendrabuddhi (!) (8 or 9 c.) is a well-known commentary of the Kāśikā-vṛtti by Jayāditya and Vāmana (early 7 c.), the oldest surviving running commentary on Pāṇini’s Aṣṭādhyāyī, and has nothing to do with the Amarakoṣa. (Scharfe, Grammatical Literature, 174).

    As Max Deeg points out, Vogel 313, n. 41, has made it clear that the mysterious “Guṇaratha” (or Guṇarata according to Vogel) originates in the remarks of S. Julien, while Bunyiu Nanjio already observed that “there is no trace in these titles of the Amarakoṣa”. Vogel continues that: “Unfortunately his remark has not received all the attention it deserves, the existence of such a translation being still asserted …” So here we have the Wikipedia entry that fascinates Victor.

  10. Harunaga Isaacson said,

    October 29, 2020 @ 2:11 am

    It may just be worth mentioning that there is a second, truly revised and expanded, edition of Vogel's volume on Indian Lexicography, which I see has been referred to by many. Details can be found here, for instance: .

  11. Victor Mair said,

    October 29, 2020 @ 6:33 am

    The Wikipedia article doesn't fascinate me. I posted it as a basis for further discussion and called attention to questionable assertions in it and in other sources. One of the beauties of Language Log is that invariably the true experts, such as Hiroshi and Harunaga, will set things straight.

  12. Terpomo said,

    October 29, 2020 @ 6:58 am

    I notice that the organization of chapters seems quite similar to some of the Chinese dictionaries by category I've seen, such as the 華夷譯語.

  13. Victor Mair said,

    October 29, 2020 @ 7:11 am


    Super good point!

    What you say supports the thrust of the final paragraph of the o.p.

  14. Victor Mair said,

    October 29, 2020 @ 7:13 am

    From Priyamvada Nambrath:

    Thank you for this detailed look at the Amarakosha. I too have been puzzled by the oft-repeated claim that Amarasiṁha was Buddhist – there is no internal evidence for this as far as I can tell apart from one of the initial slokas listing the various names of the Buddha. But taken in perspective, these are slokas #13-15 listing 25 names of the Buddha, and this is followed by 20 names of Brahma in slokas #16-17, 46 names of Viṣṇu in slokas #18-23, and 48 names of Śiva in slokas #32-37, and then follow the the numerous names and various accompanying details of other deities like Balarāma, Kāmadeva, Lakṣmi, Kārtikeya, Pārvati, Indra and so on. Other affinities – the concepts of time would appear to reflect those found in the Purāṇas, and the section on performance has concepts found in the Nāṭyaśāstra. All of this suggests a comprehensive approach that merely included the Buddha in its ambit. Again in the section on intelligence, Amarasiṁha extends this liberating approach to the idea of liberation as well, which is represented by eight synonyms drawn from different schools of thought – mukti, kaivalya, nirvāṇa, śreyas, niḥśreyas, amr̥tam, mokṣa and apavarga. Other such examples abound.

  15. Christopher Nugent said,

    October 29, 2020 @ 8:26 am

    Just to add to what Terpomo noted, this organization is also similar to that of many 類書 and 蒙書 from the Six Dynasties and Tang periods.

  16. Victor Mair said,

    October 29, 2020 @ 10:01 am

    @Christopher Nugent:

    Great comment! I was hoping and waiting for someone to bring up such types of Chinese reference works as these leishu 類書 ("category books") and mengshu 蒙書 ("primers").

  17. Anthony said,

    October 29, 2020 @ 6:49 pm

    I have a wonderful book, "Dictionary of German Synonyms." I had the 2nd edition, have the 3rd, and imagine there are newer ones. One looks up a word in English and get explanations of the differences between the German equivalents. Look up "escape" and you get: entkommen, entspringen, entschlüpfen, entweichen, entwischen, entfliehen, and 11 more. Then there are longer entries covering fewer words at greater length. I would buy similar things for other languages if they exist.

  18. Chas Belov said,

    November 4, 2020 @ 3:25 pm

    @Anthony: Yes! This is exactly the sort of book we need it multiple language-to-language dictionaries. Parker Po-Fei Hwang's Cantonese Dictionary does this for many entries.

  19. Chas Belov said,

    November 4, 2020 @ 3:31 pm


RSS feed for comments on this post