Friday, March 29, 2024


March 30 is World Idli Day, a holiday started nine years ago by M. Eniyavan in Chennai, who runs a catering business specializing in idli. There are newspaper columns and videos touting the innovative varieties they sell.

The same batter can be used to make (the ordinary varieties of) idli, dosa, and uttapam.. This is made from urad dal,(उड़द दाल), that is, dehulled beans (cotyledons) of black gram (Vigna mungo), soaked and ground, then mixed with soaked and ground polished rice.

Vigna mungo used to belong to the Phaseolus genus. One might wonder why urad dal / mash dal is mungo, but mung dal (मूँग mūṅg दाल) is radiata. It starts with Linnaeus. He gave the name P. radiatus to a plant described by Dillenius, which was mung. But the type specimen he grew wasn't the same, but P. sublobatus. Then later he gave the name P. mungo to another plant grown at Upsala, believing that it was a mung bean, when it was urad. Roxburgh tried to make P. mungo mung; but that just caused confusion. Hobson-Jobson gives moong as P. mungo and oord as P. radiatus. Prain pointed all this out, calling it “unfortunate.” A good one-page summary is here.

The batter is fermented overnight or up to a day: acid and gas for leavening are produced by lactic acid bacteria like Leuconostoc mesenteroides, naturally present in black gram. Idlis are steamed in special pans (like a dumpling), dosas are griddled and folded and stuffed (like a crepe), and uttapam are filled and cooked and flipped (like a pancake).

A Dravidian Etymological Dictionary only lists cognates, all referring to the same dishes, for iṭṭali 'idli' and tōcai 'dosa'. Wiktionary for Tamil தோசை tōcai gives an etymology from தோய் tōy 'soak; curdle', that is, 'ferment', citing a 1967 article in செந்தமிழ்ச்செல்வி Senthamilchelvi (that does not seem to be anywhere on the site to which it links or among the scanned issues). Devaneya Pavanar's A Comprehensive Etymological Dictionary of the Tamil Language gives a similar derivation: தோய் tōy → தோயை tōyai → தோசை tōcai. Wikipedia for தோசை tōcai adds a couple that sure look like folk etymologies: தேய் tēy 'rub' + செய் cey 'do', on account of how dosas are cooked; and ஸ்ஸை ssai, the hissing noise dosas make when cooking, prefixed by தோ < Hindi दो 'two' because you hear it twice. This latter is even cited by Pavanar as the perfect example of a “Playful Etymology”, that is, a joke. Pavanar for இட்டளி iṭṭaḷi again lists cognates: ம. இட்டலி Ma. iṭṭali (ഇട്ടലി); க. இட்டலி Ka. iḍḍali (ಇಡ್ಡಲಿ); தெ. இட்டென Te. iḍḍena (ఇడ్డెన). And some relationship with இட்டம் iṭṭam which I am not sure I get. Kamil Zvelebil's Comparative Dravidian Phonology proposes that, for these idli words, the -ṭṭ- in the Literary Tamil indicates a loanword from Kannada through a Colloquial Tamil -ḍḍ-; the same lack of orthographic -ḍḍ- is seen above in Pavanar's Tamil spellings of the cognates. For ஊத்தப்பம் ūttappam, Pavanar gives ஊற்று ūṟṟu → ஊத்து ūttu + அப்பம்‌ appam, that is, ūṟṟu 'pour' and appam, a sweet rice-flour cake. On the other hand, OED says, “ūttappam, lit. 'raised pancake' < ūttu- 'to blow, swell' + appam.” Similarly, it seems that of the Combined Tamil Dictionaries with entries for ஊத்தப்பம் ūttappam, Cre-A's Tamil-Tamil and Madras Lexicon, propose 'pour' and 'blow', respectively. Both alternatives involve a transitivizing gemination, 'flow' into 'pour' and 'blow (wind)' into 'blow (horn)', from roots ūṟu and ūtu-, covered by Schiffman's Reference Grammar Of Spoken Tamil 3.7.6.

More than fifteen years ago, when this blog was just getting going, LanguageHat had a post about the then newly released The Tenth Rasa: An Anthology of Indian Nonsense, quoting a poem by Sampurna Chattarji.

Idli lost its fiddli
Dosa lost its crown
Wada lost its wiolin
And let the whole band down.

This poem was originally written in English, but Chattarji also translates from Bengali: many of the translations in the anthology are her work, including some by Sukumar Ray. These and more are in Wordygurdyboom! The Nonsense World Of Sukumar Ray. Chattarji's series of sixteen Indian food poems was titled, “The Food Finagle: A Culinary Caper.” I believe they are all in the children's book The Fried Frog and Other Funny Freaky Foodie Feisty Poems, but without some footnotes that were printed elsewhere. Another one included in the anthology is relevant to this post.

Idiyappam keeps yapping
Puttu plays golf
Utthapam's my girlfriend
Mutthu's real name is Rolf.
Menus in South Indian restaurants detail their selections in roman script, and the lurking presence of English words within these essentially South Indian words was really the starting point of this verse.

I think the note refers to menus all in Roman type, versus having the names that way within overall explanations in, say, Bengali or Hindi or Marathi. Although naturally menus mostly consist of just the names of dishes and prices.

There are a number of South Indian restaurants around here (San Francisco, Berkeley, Oakland) and even more a little bit South (Sunnyvale, Cupertino), where we used to need to be more often before the pandemic. Even before moving from Boston, we would occasionally make our own dosas from store-bought batter. And, like everyone else, during the pandemic we did more of that out here, since they tend to get soggy as take-out, even if home-made is never as crispy as in a restaurant. The Indian market walking distance from home has recently expanded, and there are several more along possible bus routes from work. All stock fresh, locally produced (Fremont, I think) batters. But even more lately, in order to reduce salt intake (some of these mixes are like .2g / idli, and that's before any chutneys), we have been making it from scratch. Since we haven't sprung for a proper wet grinder, we use a Vitamix. And since it is never warm for a whole day in San Francisco (it is never really cold, either), we ferment in an Instant Pot on the yogurt setting. Naturally it took a few tries. The only gotcha seems to be that if the blender motor gets too hot it can kill the bacteria in the urad dal. So we use ice instead of just water while grinding it. The rice doesn't seem to care. We usually use idli rice. The cashier at the market will sometimes confirm that this is what we want, versus “regular” basmati rice, which is thoughtful. They may also want to protect their limited stocks, since last fall India banned the export of Non-Basmati White (NBW) Rice, including idli and sona masoori, which are used in South Indian cooking. Basmati might not even be a problem, since plenty of it is grown in California and Texas. We do add fenugreek (मेथी mēthi < Drav.), since we like its taste. Experiments by one of the prolific posters on The Fresh Loaf, a website for bread nerds, have determined that it doesn't add anything to the biochemistry. We don't mix by hand, though, which is another traditional, supposedly necessary, step. I always reasoned that this was not to introduce more bugs, but to start the warming process. And that's unnecessary when using a yogurt maker. When the batter is ready, we first make idli. Leftover steamed idli can be fried, although we sometimes cheat and do that to some right away. After a few days, we make dosas. And, then, finally, at the end of the week, with the last of the batter, an uttapam or two.

Read More

The annual Idli Day newspaper pieces and blog posts typically include a summary history based on the work of K. T. Achaya, an oilseed chemist and culinary historian. Specifically, his Indian Food: a Historical Companion and a rearranged and somewhat shortened version of mostly the same material, A Historical Dictionary of Indian Food. (In what follows, some glosses are takens from the “Glossary of Non-English Words” in the former. When the source may not be obvious, such glosses here may have his initials, KTA. Similarly the transliteration of single words, but not source quotations, may be chosen to match, even though it is not entirely self-consistent. Otherwise, book titles will follow any romanization printed on the title page or the ALA-LC Tables. Everything else should be ISO 15919, although there are without doubt accidental deviations due to my careless copy / pasting.) Some of the major points available about idli are:

  • The first mention is in the 920 Jain work in Kannada, Vaddaradhane, in the variant form iddaligē, as one of eighteen items served to a brahmachāri who visits the home of a lady.
  • The earliest recipe is from Chavundaraya's 1025 Lokopakara.
  • It is described in the 1130 Sanskrit work Manasollasa.
  • These early versions differ from modern idli in three important ways:
    • They do not include rice grits along with the urad dhal.
    • They do not call for a long fermentation.
    • They do not specify cooking by streaming to fluffiness.
  • Somewhat later references compare idli to the moon, which might suggest whiteness from rice.
  • The fermentation technique used for idli might have originated in Indonesia:
    • There is a long tradition of fermented grains there.
    • It would be brought by the chefs who accompanied the Hindu kings of Indonesia on their visits to South India in search of brides.
    • In fact, there is an Indonesia dish like idli called kedli.
  • The Chinese traveller Xuan Zang noted that India lacked vessels for steaming.

The last two are naturally fuel for a nationalistic take in the piece, should that be desired, as it is more and more these days.

The Vaddaradhane Wikipedia entry is just a stub. For example, it does not reference any English translation, like Veneration to the Elders. Nor any secondary works like this study (originally a PhD thesis). But still an idli-lover has inserted a sentence and footnote citing the mention of iḍḍalige there. The reference to the eighteen dishes, of which iḍḍalige is one, is in the chapter on Bhadrabāhu, but not in his story itself, but rather in an inserted sub-story with a flashback to an earlier Śruta Kevalī, Nandimitra. Nandimitra's father is killed by robbers before he is born and his mother dies shortly afterwards. The relatives he then lives with also die and the townspeople decide this is a bad omen and the orphan is sent wandering. As a teenager, he is taken in by a fisherman who, seeing his Herculean strength, uses him to gather firewood that they sell. The fisherman instructs his wife not to feed the boy too much, but she does so once at festival time. This leads to Nandimitra asking for more than rags as clothing, which alerts the fisherman, who accuses and beats his wife before throwing her out. Nandimitra then begins independently selling firewood and supporting his adoptive mother. It is at this time that Nandimitra encounters the Bhaṭṭāraka Śivagupta returning from the forest after a long fast and decides to follow him. The king has proclaimed that no one should stop Śivagupta from coming to the palace to break his fast and the king and queen come out and greet him. They mistake Nandimitra for Śivagupta's disciple and feed both of them the wide variety of rich foods in question. This motivates Nandimitra to ask Śivagupta to also become a monk. The king, seeking further merit, tries to entice Nandimitra to break his fast at the royal table again, but Nandimitra reasons that by fasting longer he will get something even better. He repeats this with the queen the next day and so on for a week. And then Nandimitra is accepted by Śivagupta and renounces food altogether.

Before getting to the list of eighteen items, it is worth noting a few amplifications of the half-sentence context summary given by Achaya and copied into most citations since. The hosts are the king and queen in the palace, not just a lady at home. The guests are a bhaṭṭāraka and Nandimitra, who takes a vow of brahmacāri as a result of meeting him and the feast. Moreover, fasting and feasting, or the promise thereof, drive his spiritual development toward a śrutakevalī.

In any case, only twelve of the eighteen dishes are enumerated:

ರಾಜಾನ್ನದ ಕೂೞುಂ ಪೆಸಱತೊವೆಯುಂ ಬೆಣ್ಣೆಗಾಸಿದಾಮೋದ ಸುಗಂಧ ಪರಿಮಳಂ ನಾರ್ಪ ತುಪ್ಪಮುಂ ಪಲವುಂ ತೆಱದ ಬಾಡುಗಳುಂ ತುಯ್ಯಲುಂ ಪೂರಿಗೆಯಿಡ್ಡಲಿಗೆ ಸೋದಿಗೆ ಲಾವಣಿಗೆ ಘೃತಪೂರಂ ಲಡ್ಡುಗೆ ಮಂಡಗೆ ಮೊದಲಾಗೊಡೆಯ ಪದಿನೆಂಟುಂ ತೆಱದ ಭಕ್ಷ್ಯರೂಪಂಗಳುಮಂ
rājānnada kūḻuṁ pesaṟatoveyuṁ beṇṇegāsidāmōda sugaṁdha parimaḷaṁ nārpa tuppamuṁ palavuṁ teṟada bāḍugaḷuṁ tuyyaluṁ pūrigeyiḍḍalige sōdige lāvaṇige ghr̥tapūraṁ laḍḍuge maṁḍage modalāgoḍeya padineṁṭuṁ teṟada bhakṣyarūpaṁgaḷumaṁ

(The text here differs only in ಪ್ರ pra for ಪೂ pu.)

Achaya says that this is the first mention of the following foods: iddaligē, pūrigē, sōdhigē, lāvangē, ghratapūran, mandigē. Likewise Kittell's Kannada dictionary in the Digital Dictionaries of South Asia cites this passage for a number of words.

  • rājānna kūl 'royal variety of rice'.
  • pesaṟa tove 'green gram boiled dish' (Tel. పెసలు pesalu).
  • beṇṇegāsida āmōda sugaṁdha parimaḷaṁ nārpa tuppa 'fragrant ghee obtained by melting perfumed butter'.
  • palavu teṟada bāḍugaḷ 'various kinds of side dishes'. (? This, with the preceding item, is quoted, s.v. ಬಾಡು bāḍu. 2. 'flesh', but it cannot be meat. A modern Kannada translation has ಹಲವು ರೀತಿಯ ಕಾಯಿಪಲ್ಲೆಗಳು halavu rītiya kāyipallegaḷu 'various kinds of nuts'.)
  • tuyyal, a sweet milk dish.
  • pūrigē 'sweet stuffed wheat circlet (KTA)' poori.
  • iddaligē idli.
  • sōdhigē 'sweet vermicelli dish (KTA)'.
  • lāvangē 'perhaps a wheat dish (KTA)'. Others take ಲಾವಣಿಗೈ lāvaṇigē to be from ಲವಣ lavaṇa 'salt' and so 'pickles'.
  • ghratapūran 'ghee-filled wheat flour confection (KTA)' (Skt. घृतपूर ghṛtapūra).
  • laddugē 'fried globules of ground pulses, or sesame seeds, or rava moulded into balls with sugar or jaggery syrup (KTA)' (Hin. लड्डू laḍḍū) laddu.
  • mandigē 'sweet stuffed wheat circlets (KTA)'.
  • 'totaling eighteen kinds of edible'.

There is no indication of what the iḍḍalige is made from there. The Lōkōpakāra (ಲೋಕೋಪಕಾರ. Note that that Wikipedia page has an incorrect link to the wrong ಚಾವುಂಡರಾಯನ Cāvuṇḍarāyana, the ruler and not the poet. The Kannada literature page is correct.) passage is more specific.

ಅರೆದುರ್ದಿನ ಬೇಳೆಯ ನೊಂ ।
ದಿರೆ ಮೊಸರಧಿ ನೀರೊಳಿಂಗು ಜೀರಗೆ ಕೊತ್ತುಂ ॥
ಬರಿ ಮೆಣಸಲ್ಲಮಿವಿನಿತಂ ।
ಬೆರಸಿಡ್ಡಲಿಗೆಯನಡಲ್ಕೆ ಕಂಪಂ ಕುಡುಗುಂ ॥೧೧॥
aredurdina bēḷeya noṁ- |
dire mosaradhi nīroḷiṁgu jīrage kottuṁ- ‖
bari meṇasallamivinitaṁ |
berasiḍḍaligeyanaḍalke kaṁpaṁ kuḍuguṁ ‖ 11 ‖
Grind the washed split blackgram dhal and add clear water obtained from the surface of the curd to it. Add asafetida, cumin seeds, coriander, and black pepper to it. Idlis prepared from this ground paste will be highly delicious. (tr. Ayangara; see also Peppertrail)

Manasollasa (मानसोल्लास Mānasōllāsa) has recipes for both dhōsakā and iddarikā.

विदलं चणकस्यैवं पूर्वसम्भारसंस्कृतम् ॥९२॥
ताप्यां तैले (ल) विलिप्त्यायां धोसकान्विपचेद्बुधः ।
माषस्य राजमाषस्य वट्टाणस्य च धोसकान् ॥९३॥
अनेनैव प्रकारेण विपचेत्पाकतत्त्ववित् ।
vidalaṁ caṇakasyaivaṁ pūrvasambhārasaṁskr̥tam ‖ 92 ‖
tāpyāṁ tailē (la) viliptyāyāṁ dhōsakānvipacēdbudhaḥ |
māṣasya rājamāṣasya vaṭṭāṇasya ca dhōsakān ‖ 93 ‖
anēnaiva prakārēṇa vipacētpākatattvavit |
A batter of ground chickpeas (चणकः chanaka) is prepared first.
The dhosaka are smeared on a hot pan along with oil and cooked thoroughly.
Dhosaka can also be made from urad dal (माषः māsha 'Vigna mungo (KTA)'), red beans (राजमाष rājmāsha 'Phaseolus vulgaris, earlier perhaps Vigna mungo (KTA)'), or dried peas (वट्टाण vatāna cf. Guj. વટાણા vaṭāṇā).
True cooking should cook in this same way.
आम्लीभूतं माषपिष्टं वटिकासु विनिक्षिपेत् ।
वस्त्रगर्भाभिरन्याभिः पिधाय परिपाचयेत् ॥९९॥
आव्तार्यात्र मरिचं चूर्णितं विकिरेदनु ।
घृताक्ता हिङ्गुसर्पिभ्यां जीरकेण च धूपयेत् ॥१००॥
सुशीता धवला (ः) श्लक्ष्णा एता इडरिका वराः ।
āmlībhūtaṁ māṣapiṣṭaṁ vaṭikāsu vinikṣipēt |
vastragarbhābhiranyābhiḥ pidhāya paripācayēt ‖ 99 ‖
āvtāryātra maricaṁ cūrṇitaṁ vikirēdanu |
ghr̥tāktā hiṅgusarpibhyāṁ jīrakēṇa ca dhūpayēt ‖ 100 ‖
suśītā dhavalā (ḥ) ślakṣṇā ētā iḍarikā varāḥ |
Form the fermented ground urad dal into cakes (वटिका vatikā).
Cover with a cloth and cook.
Sprinkle with crushed black pepper (मरिचः maricha).
Make fragrant with ghee, asafetida (हिङ्गु hingu), and cumin (जीरकः jīraka).
The idli (इड्डरिका iddarikā) will be cooling, white, and soft.

(See also Peppertrail for translations.) If I understand āmlībhūtaṁ 'become acidic', could that imply fermentation? Also could the cover aid steaming? There is definitely no sign of rice yet.

The Saundara Vilāsa (ಸೌಂದರ ವಿಲಾಸ), a poetic work of about 1600, includes among the wares of a ಮಿಠಾಯಿಯಂಗಡಿ miṭhāyiyaṅgaḍi 'sweets shop', ಹಿಮಕರನಂತೆ ರಾಜಿಸುವ ಇಡ್ಡಲಿಗೆ himakaranante rājisuva iḍḍalige 'idli shining like the moon' (in Wikisource; the Internet Archive has several scans, but they all have OCR problems). Achays's other such comparison, from 1485, is less straightforward. It is from ಸನತ್ಕುಮಾರ ಚರಿತೆ Sanatkumara Charithe by Bommarasa, a Jain Vijayanagara poet. Note that the note number for this is accidentally printed 2k instead of 2h, which would make it be the 1584 ಚನ್ನಬಸವ ಪುರಾಣ Cannabasava Purāṇa. This typo leads to this misattribution in a conference paper. As to the verse in question,

ಬಟ್ಟವೆಱೆಯೊ ಮಂಜಿನೊಬ್ಬುಳಿ
ಬಟ್ಟಿತಾದುದೊ ಅಮೃತರಸವಳ
ವಟ್ಟು ವೃತ್ತದ ಪಿಂಡವಾದುದೊ ಚಂದ್ರಿಕೆಯೆ ಬಂದು
ಘಟ್ಟಿಗೊಂಡಿತೊ ಎನಲು ನೋಳ್ಪರ
ದಿಟ್ಟಿಗೊಲವನು ಮನಕೆ ಹರುಷವ
ಪುಟ್ಟೆಪುದ್ದಿನ ಕಡುಬ ಸವಿದರು ನೃಪರು ಮನನಲಿಯೆ
baṭṭaveṟeyo mañjinobbuḷi
baṭṭitādudo amr̥tarasavaḷa
vaṭṭu vr̥ttada piṇḍavādudo candrikeye bandu
ghaṭṭigoṇḍito enalu nōḷpara
diṭṭigolavanu manake haruṣava
puṭṭepuddina kaḍuba savidaru nr̥paru mananaliye
The Kings are relishing the kadubu made of black gram: it looked like a full moon; like a mass of mist set together; as if heavenly nectar had solidified into circles; or as if a drop of moonlight had hardened. The kadubu was attractive to the eye and pleasing to the mind. (tr. Achaya)

Kadubu is explained in the glossary as 'steamed slab of fermented rice-pulse'. Kotte kadubu or kadubu idli are ones steamed / served in jackfruit leaves. There are other kinds with fillings like coconut or lentils, but I suppose they don't come out moon-shaped.

Achaya glosses kedli as '(Indonesia): steamed rice-pulse patty'. An endnote credits at least the idea that Indonesian cooks brought fermentation suitable for idli to India, and perhaps that further name of a specific source dish (I cannot tell), to C. R. Krishna Murti personal communication, March 1985. I am pretty sure that this CKR, another chemist, is intended. The first question is whether it is even necessary to find a source for natural fermentation. As pointed out by Feasts and Fasts: A History of Food in India, nearly all cultures have some form of fermentation. Other dishs require yeast or a starter. The advanced fermentation techniques of Indonesia involve growing the mold first and then innoculating the legumes. But the microflora for idli are already present on the urad dal. Even for experimental variations using other grains, such as soyidli, there are enough. Idli batter can be used as a starter for a new batch, or buttermilk or water from yogurt, or even yeast added, but that is not the traditional recipe. If South India needed to learn how to ferment idli, then wouldn't Gujarat have needed it for dhoklā (ઢોકળાં)? The fermentation process is not very different, just with chickpeas. Next, as far as I can make out, no one who has looked has been able to identify kedli with an Indonesian dish, ideally a cereal cake leavened by fermented legumes. The best explanation seems to be that it is a misunderstanding of kedelai, the Indonesian name for 'soybean', from Tamil கடலை kaṭalai 'chickpea' (but now usually 'peanut', following a common pattern where a New World food starts out with a qualified name for an existing one and then takes over that name unqualified as well). Soybeans are fermented into tempeh. The Soyinfo Center books related to tempeh have a good summary of a comprehensive but hard to find book on its history. By default, tempeh is made from soybeans, but there are other varieties, and, when it is necessary to disambiguate, the full name is tempe kedelai 'soy tempeh', versus tempe gembus 'okara tempeh' or any of the others listed in Appendix C in the Professional Edition of The Book of Tempeh and summarized here. Wilkinson's Malay Dictionary does have Phaseolus vigna, the former classification of black gram, for کدلي kĕdĕlai. I do not know whether he is mistaken or it could mean that instead of or in addition to 'soybean' then or there, though either way it does not seem to really help for Indonesian idli.

The Great Tang Records on the Western Regions (大唐西域記 Dà Táng Xīyù Jì) has Xuanzang's (玄奘) report about steaming.

Suī fǔ huò sī yòng. Ér chuī zèng mò zhī.
While they use kettles and woks, they do not know cooking with a steamer.

Of course, steaming is still possible without special vessels, using cloth or banana leaves.

Many of Achaya's references are in [Old] Kannada. This is not altogether inappropriate: the case for Karnataka as the source of these dishes is pretty good. And even for their restaurant forms to have originated with restauranteurs from Udupi, spreading first to other Indian urban centers, and then to the world. Most of Kannada passages can evidently be found in the appendix to a 1969 edition of Sūpaśāstra. Another source is a pair of short papers in Indian Linguistics, one by P. K. Gode in 1955 and another by H. G. Narahari in 1957.

There are also similar references from not too much later in other languages. For instance, the Marathi dictionary Śabdakōśa, included in the DDSA, s.v. इडरी, इडली iḍarī, iḍalī, has a quotation from R̥ddhipuravarṇana, where idli is again compared to the moon. (snippet; scanned without OCR)

पूर्ण चंद्राचा अनुकारी : चोखाळपणें भजिजे इडरीं :
pūrṇa candrācā anukārī : cōkhāḷapaṇēm bhajijē iḍarīm :
resembling the full moon: purely revered idli

Another Sanskrit source, besides Mānasōllāsa above, is Sri Ramanuja Campu, a life of Ramanuja in mixed prose and verse. It contains a poetic description of both idli and dosa (3.29).

अभ्यागम्य पदे पदे सविनयं संप्रार्थितो गेहिभिः
दोशामण्डलमिन्दुबिम्बधवलं सद्योघृतेनाप्लुतं
भक्तं स्वर्णसवर्णसूपसहितं सामोदमास्वादयन् ॥ २९ ॥
abhyāgamya padē padē savinayaṁ saṁprārthitō gēhibhiḥ
śuṇṭhījīrakarāmaṭhādisurabhīrgaṇḍākr̥tīriḍḍalīḥ |
dōśāmaṇḍalamindubimbadhavalaṁ sadyōghr̥tēnāplutaṁ
bhaktaṁ svarṇasavarṇasūpasahitaṁ sāmōdamāsvādayan ‖ 29 ‖
He approaches step-by-step and modestly greets the householders
The round-shaped idli is fragrant with ginger, cumin, and asafetida
The circle of the dosa, resembling the white disc of the moon, is instantly bathed in ghee
Served together with gold-colored soup, it is delicious eating

शुण्ठी śuṇṭhī 'dried', specifically, 'dried ginger' has been subtituted for शोणी śōṇī 'red' here, following the note at the end. This seems more suited to the context and follows everyone that quotes this passage.

Here it is dosa that is compared to the moon, which would seem to imply some other form than we get in restaurants. Of course, with food history, there are cases where something similar used to have a different name, or a name used to be applied to a something somewhat different.

Keeping that in mind, for earlier dosa, Achaya writes, in the same Snacks of the South box in the Regional Cuisines chapter, that tōsai (dōsai) is “noted” in Tamil Sangam literature. The endnote for this sentence references the Food section of P. T. Srinivasa Iyengar's Pre-Aryan Tamil Culture. This does not seem to mention tōsai by name, though. It does have a list of cakes, including appam (அப்பம்) and melladai (மெல்லடை), that is, மெல் mel 'soft' + அடை aṭai 'wafer'. Achaya glosses adai as 'fried pulse snack'. Elsewhere, Achaya says that in Chōlā times, pan-fried snacks were dōsai and adai, both based on rice. The endnote for this points here, which says that āppam rice cake soaked in milk was a luxury. Achaya goes on to explain adai as a mix of rice and no less than four pulses ground together. Back in the Southern Snack box, the next paragraph says that circular āppam was mentioned in Perumpānūru, along with idi-āppam, and that these remain unchanged today. (Idiyappam was included above in one of Chattarji's nonsense poems.) No specific reference is given. But in the previous chapter, in a section on professional cooking, he refers to, “… kaazhiyar and kuuviyar, vendors of snacks like āppam, idi-āppam, adai, …” with an endnote giving three actual Sangam citations.

  1. Silappaddikāram, Chapter 5. Line 24 of a street scene: காழியர் கூவியர் கண்ணொடை யாட்டியர் kāḻiyar kūviyar kaṇṇoṭai yāṭṭiyar 'washermen, makers of muffins, wine-sellers' (tr. Dikshitar). The English just has 'washermen' for the first. Even though Achaya's apposition quoted above might be ambiguous, the glossary is not, giving both “kaazhiyar (T): snack vendor,” and “kuuviyar (T): vendor of snacks.” The note for the source text explains, காழியர்‌ - பிட்டுவிற்பார்‌ ; வண்ணாருமாம்‌. கூவியர்‌ - அப்பம்‌ சுடுவார்‌. kāḻiyar - piṭṭuviṟpār ; vaṇṇārumām. kūviyar - appam cuṭuvār. 'kaazhiyar - pittu sellers; Vannar. kuuviyar - appam steamers.' It is not clear to me whether that means that the word has two senses or that pittu-sellers belonged to that particular caste. Dictionaries with just the collective காழியர் kāḻiyar agree on வண்ணார் vaṇṇār 'washermen'; one adds அப்பவாணிகர் appavāṇikar 'cake sellers'. Those that instead have the singular காழியன் kāḻiyaṉ give both senses; the Madras Lexicon has two headwords, with slightly different proposed derivations, washer from क्षाल kṣāl 'washing' via காழ்- kāḻ-, sweets vendor from क्षार kṣāra 'something caustic or salty or sugary'. (Those two Sanskrit roots being generally supposed to be related somehow, I believe.) Pavanar also has two lemmas, but with the identical-looking etymologies from காழ் kāḻ. None of which resolves the question, I don't think. காழியர் kāḻiyar and கூவியர்‌ kūviyar recur together in the next chapter (6.137-8; translation), in a scene where vendors are burning their lamps; for the first, it has, காழியர் மோதகத் தூழுறு விளக்கமும் kāḻiyar mōtakat tūḻuṟu viḷakkamum 'lamps of kaazhiyars scooping-out modaka'; again there is a note, மோதகம்‌ - ஈண்டுப்‌ பிட்டு mōtakam - īṇṭup piṭṭu 'modaka in place of pittu'. (If appams are muffins, what are pittu? Buns?)
  2. Perumpānūru, line 377. காரகற் கூவியர் பாகொடு பிடித்த / இழைசூழ் வட்டம் பால்கலந் தவைபோல் kārakaṟ kūviyar pākoṭu piṭitta / iḻaicūḻ vaṭṭam pālkalan tavaipōl 'These look like rice-cakes placed in milk, prepared in pans with thread-like paste with jelly mixed by those who trade in cakes'. (The English translation does not have parallel line numbering; this starts at 438. Also, this text has been OCRed with tesseract -l san, so for Devanagari, and, as a result, both the Tamil and the Roman are garbage.) This passage is quoted in dictionary entries for அகல் akal, the pot in which the cake is being cooked, and வட்டம் vaṭṭam 'circle', here referring to the cake formed by winding threads (இழை iḻai) of batter into that shape. The image here is of blossoms falling into the water and looking like idiyappam being cooked.
  3. Mathuraikkānchi, lines 624/7. நல்வரி இறாஅல் புரையு மெல்லடை / அயிருருப் புற்ற ஆடமை விசயங் / கவவொடு பிடித்த வகையமை மோதகந் / தீஞ்சேற்றுக் கூவியர் தூங்குவனர் உறங்க nalvari iṟāal puraiyu mellaṭai / ayirurup puṟṟa āṭamai vicayaṅ / kavavoṭu piṭitta vakaiyamai mōtakan / tīñcēṟṟuk kūviyar tūṅkuvaṉar uṟaṅka 'Breadmakers that sold jelly wafers soft that look like striped honeycombs and cakes with coconut sugar sweet and pulse stuffed in and flour with jelly mixed, are now asleep.' (English starting at 688.)

As we see, these early references are circumstantial: they describe sellers of round cakes, some of which might be like dosa. More recent and specialized than Achaya's works is சங்ககாலத் தமிழர் உணவு Caṅkakālat tamil̲ar uṇavu, 'Sangam Tamil Food', which was “free” for Kindle with “points,” although it lacks search. Among these dishes, it seems to just list (p. 198) அப்பம்‌ (இனிப்பு - பெரும்பாண்‌. 377-378) āppam (iṉippu 'dessert' perumpāṇ‌), those same lines as 2 above.

For the word dosa itself, there is சேந்தன் திவாகரம் Cēntaṉ Tivākaram, a Tamil lexicographical work of the 10th century. It is most like what we might call a thesaurus, a kind of work known in Sanskrit as कोश kośa, using the same treasure metaphor. Classes of things are described by words for specific members of the class. See the front matter to the Madras Lexicon, around page xxvi, or Gregory's Colporuḷ: A History of Tamil Dictionaries, which compares it to Ælfric's Glossary. In section 6, பல்‌ பொருள்‌ பெயர்த்‌ தொ pal poruḷ peyart to 'collection of multiple-meaning names', entry 92, அப்ப வர்க்கத்தின் பெயர் appa varkkattin peyar 'cake: sorts of edible names', பேத வகைப் பெயர் pēta vakaip peyar 'sorts of division names', are given: பூரிகம், நொலையல், கஞ்சம், தோசை pūrikam, nolaiyal, kañcam, tōcai.

  • பூரிகம் pūrikam All these dictionaries agree that பூரிகம் - அப்பவருக்கம் pūrikam - appavarukkam 'sort of cake'. The Madras Lexicon specifically cites திவா. Tivā for this. Its < pūrikā and 'Pastry or cake full in appearance but hollow inside' suggest that this is a kind of pūri. Winslow's points to பூரிகை pūrikai 'A kind of unleavened cake'. Achaya's Glossary has an entry for pūrika 'crisp wheat snack', but the page to which it points, discussing the Mānasōllāsa, and so Sanskrit, specifically says, “Pūrika were small fried cakes of gram flour: not the pūri of the present, but the pāpadi.” See Peppertrail again for a translation of that Purika recipe. It seems perhaps two dishes have been confused here.
  • நொலை, நொலையல் nolai, nolaiyal 'Pastry, confectionary' or 'A kind of unleavened cake'. Not the same as நோலை nōlai, a sesame seed confection (note vowel length).
  • கஞ்சம் kañcam 'A kind of pastry'. The referenced Telugu cognate కజ్జము kajjamu is likewise 'A sort of sweet cakes'. But one of the forms in that latter headword, కజ్జికాయలు kajjikāyalu, has lots of online recipes and is included in Wikipedia for Gujia. Apparently not related to கஞ்சி kañci congee. Or, for that matter, to கஞ்சா kañcā < गांजा ganja.
  • தோசை tōcai dosa.

தமிழர் உணவு / Tamil̲ar uṇavu 'Tamil food' by S. Namasivayam points to another source of somewhat later references, writing,

பல கல்வெட்டுக்கள்‌ தோசை அக்காலத்தில்‌ இருந்தமையை சுட்டிக்‌ காட்டுகின்‌ றன.
pala kalveṭṭukkaḷ tōcai akkālattil iruntamaiyai cuṭṭik kāṭṭukiṉ ṟaṉa.
'Many inscriptions indicate the existence of dosa at that time.'

and giving references to a few such. Finding more like that is easy using catalogs now scanned online. For example, search in காஞ்சிபுரம் மாவட்டக் கல்வெட்டுகள் kāñcipuram māvaṭṭak kalveṭṭukaḷ 'Kanchipuram District Inscriptions' for தோசை அரிசி உளுந்து எண்ணை tōcai arici uḷuntu eṇṇai 'dosa rice gram oil'. Narasivayam's observation,

அக்காலக்‌ கல்வெட்டுக்கள்‌ தோசை சய்யும்‌ முறையை நமக்குக்‌ கற்றுத்தந்தன என்று கூறினால்‌ அது மிகையாகாது.
akkālak kalveṭṭukkaḷ tōcai cayyum muṟaiyai namakkuk kaṟṟuttantaṉa eṉṟu kūṟiṉāl atu mikaiyākātu.
'It would not be an exaggeration to say that the inscriptions of that time taught us the method of making dosa.'

probably really is an exaggeration, I'm afraid.

It seems that the English dosa form comes through Tulu or Kannada dose, so from Karnataka. The large crispy form of dosa found in restaurants is generally supposed to originate in Udupi specifically. At the very least, masala dosa must follow the introduction of potatoes (and chili peppers) by the Portuguese.

Looking around the internet, one may encounter an alternate explanation for the early source of idlis. This is that Arab traders brought them to South India when they settled down and married local women, because, being concerned about Muslim dietary restrictions, they found it easiest to just eat rice balls with coconut paste. Various specific authorities are given for this:

  • modern food historians such as Lizzie Collingham, Kristen Gremillion, Raymond Grew, Makhdoom Al-Salaqi (Syria), Zahiruddin Afiyaab (Lebanon)
  • references available at the Al-Azhar University Library in Cairo
  • Encyclopaedia of Food History, edited by Collingham and Gordon Ramsay of Britain, Oxford University Press
  • Seed to Civilisation, The Story of Food, by Heiser Charles B, Harvard University Press, 1990

This very strange. Some of those historians only seem to be associated with food history on pages making this claim. Collinham has indeed written a number of books about Indian food history, as both Lizzie and E. M. Collingham. Topics include food of the British Empire and entertaining at Rashtrapati Bhavan. Her Curry: A Tale of Cooks and Conquerors tells just such stories about interactions between cultures, resulting in dishes like Biryani. But here Arab traders teach their Indian wives to make seafood pilaus. And idlis only get passing mention as already-established South Indian rice breads, contrasting for the Portuguese with leavened wheat breads, or as later modern diaspora dishes. (As is common here, Lieut. Richard F. Burton makes a brief appearance, praising the leavened bread in Goa. Later in that same work, he complains of its lack and of the substitute, “unleavened wafers, called aps,” viz., appams.) Moreover, there does not seem to be any encyclopedia of food history from OUP or anyone else. And I doubt Ramsay would be associated with anything other than television. Prof. Heiser's book does exist. His works have focused more on food plants and that one on seed-like ones in particular. It does have chapters on grasses (rice), legumes (dal), and coconut, but that seems to be as close as it gets. I could almost imagine that this whole explanation is an AI hallucination, like if somenoe had asked one where idli come from. It fits the pattern of a detailed answer, plausibly sourced, even if not academically rigorously cited. But the earliest occurrence of this claim I can find it from Feb 2015, just before Idli Day was founded, which seems a little too early for LLMs to have done it. It could be some elaborate trolling. Or it could be legit, no funny epistemology; but the genuine new facts are just hard to find as given. Still, rice balls seem like less of an innovation than fermentation, and it isn't clear how the lack of rice in early idlis is resolved. (If someone understands what is going on, I will, of course, amend this post.)

The OED has a headword for idli, added in 1976 and not revised since, and ones for uthappam and appam added in 2006. It does not have dosa, although that word is used in quotations for uthappam, rice-pancake, chutney, and sambar. The source of uthappam is multiple nearly identical cognates, “Partly < Tamil ūttappam, and partly < Malayalam ūttappam,” and similarly for appam.

I would have thought there would be some casual British mentions of these dishes in the 19th century. Of course, they do appear in dictionaries. The DDSA Kannada ಇಡ್ಡಳಿಗೆ iḍḍaḷige entry listed above is the same as in Kittel's 1894 printed version, and that is, in turn, taken verbatim from Reeve's 1858 “Canarese” dictionary. Rottler's 1834 Tamil dictionary has entries for இட்டலி iṭṭali 'a kind of cake' and தோசை tōcai 'rice-flour pancakes'. Beyond that, there is a mention of appas or hoppers (another almost English word not in the OED) in Ceylon in 1859. And, from 1909, of “cakes, called dosai,” among the Nagarathar caste.

The OED's earliest idli quotation is from the 1958 novel The Guide by R. K. Narayan. Idlis are not surprising as a way to recreate the atmosphere of Malgudi. The two 1972 quotations, cited only as New Yorker are, in fact, both from a short story “Naga,” published there and also by Narayan. Keeping with his work, the earlier (1955) Waiting for Mahatma has “Oh, how long it was since he had eaten anything like idli, those white sensitive things made by his granny on most Sundays.” and “The thought of idli, soft and light, and of dosai, was aluring. It seemed as if he had tasted them in a previous birth.” Before that, the author seems to have avoided foreign names; for example, Mr. Sampath (1949) described eating rice, or rice and curd, or rice and buttermilk. Autobiographically, a passage in My Dateless Diary describes a dinner at the Indian Consulate in New York with an exiled Maharaja.

The dinner was a triumph, establishing once for all that supremacy of tranquilizing qualities of South Indian food—Rasam, Sambhar, Masala Dosai, pickles and so forth. I'm more than ever convinced that the South Indian diet marks the peak in the evolution of culinary art and that the South Indian, however well he may be received, will never feel really at home anywhere in the world unless he can have his spices too within reach. My regard for His Highness went up when I found him uttering little cries of joy at the sight of Sambhar and Dosai. I knew then that the man could do no wrong.

Another place that Narayan's stories appeared was the London-based quarterly Indian Writing, published irregularly from 1940 until ended by the War in 1945, and edited by Ahmed Ali and Iqbal Singh. At least by Ali's telling, those two did all the work and the other nominal editors were either for show, or, in the case of Alagu Subramaniam, not very useful. Nevertheless, the third number (March 1941) published a short story by Subramaniam titled, “The Flood,” containing the following.

Her speciality was a sort of pancake called thosai. Flour, mashed lentils and coconut milk were mixed together to form a liquid paste. Two or three spoonfuls of this were transferred to a griddle and baked. She was famous for these pancakes, and men, women and children went joyfully to her shop.

A 1944 Government planning document for food rationing categorized, “Restaurants, Puri Parotawalla, Idli Dosa shops.”

In 1953, Adlai Stevenson, having lost the Presidential election to Eisenhower, embarked on a world tour of the Middle East and Asia. The June 15, 1953 issue of The New Republic has “Stevenson In India” on the front page with this rather cringeworthy photo and inside a piece titled, “Sir, you are in a Solid Democratic Territory,” starting with a play on the Governor's first name.

“Idli” is the name South Indians give to the steamed rice-powder cake that forms their favorite breakfast dish. And “Idli” was naturally the name they bestowed on their favorite visitor from America. They greeted Stevenson with garlands and he, in turn, split many a fresh coconut with them in the name of One World.
Of the coconuts of Kerala and the dark beauties of matriarchal Malabar even Baudelaire has sung. But it was left to Stevenson to declare:
“My visit to Travancore-Cochin has fulfilled a life-long desire and ambition to see the Malabar coast. Here I found the most luxurious and beautiful fragments of the entire globe, inhabited by a proud industrious and ancient people.”
The only aspect of Stevenson's visit that the proud, industrious and ancient people regretted was its shortness. He passed through Hyderabad on the Deccan plateau on May 7, spent a day in Madras on the East coast, flew over to Trivandrum, near the southern tip of India and was back in Bombay on May 10.

The byline for this piece is “Oliver Pirie,” explained in a note as a pseudonym for the combined separate accounts by the five journalists who followed Stevenson in India. The rest of the article somewhat betrays this, including discussions of politicians and crowds he met, and giving questions and answers from a press conference, which center on what might be expected, such as Korea and the French in Indochina (this was around the time of the Franco-Lao Treaty). There is also repeated reference to Communist challengers to the Indian Congress Party, a theme also taken up by one of the pieces that Stevenson was commissioned to write for Look from his trip, “Will India Turn Communist?

I admit that, seventy years on, the notion that “South India” would immediately suggest to a reporter or editor «À une Malabaraise» does not quite feel right. Though it may be more wholesome than all of Asia suggesting a Red Menace.

There is little certainty in undirected searches in scanned books. Typos are introduced in manually transcribing MARC metadata from the OPAC, or it may be wrong there, or confusing, as in the case of periodicals. OCR results are full of errors, particularly with italics, which would still commonly be used for these new words. Much better is to find something to go after. For example, in 1978, the New York Botanical Garden published a Plant Bibliography specifically for Vegetable Cookery (to introduce vegetarian meals). It refers to a 1973 The Art of South Indian Cooking, which does, indeed, have recipes for dosa and idli.

Back in the present, when making our own dosas, we sometimes use an obvious local filling, avocado. Although California restaurants don't seem to make this, it does seem to be a real thing elsewhere, as ಬೆಣ್ಣೆಹಣ್ಣಿನ beṇṇehaṇṇina 'butter fruit'. (Alternatively, the phonetic ಆವಕಾಡೊ āvakāḍo.) There are also recipes that call for mixing it into the batter itself, which we haven't tried. Achaya has a paragraph on avocado / butternut in the New World to India section, but does not mention any particular culinary uses there or include it in the Glossary. Ever the oilseed chemist, he mostly promotes unsaturated fats in the Indian diet, as in another paper giving an evaluation of the vegetarian diet. Though, since his works are informationally quite dense, any attempt at summary risks distortion, in a blog post or in a newpaper article.


Anand said...

Thank you for such a deeply researched post! And in particular, for delving into the primary sources rather than just taking Achaya's word for it, which is all that most Indian food writers care to do.

I just wanted to make one minor correction though: I think ūttappam is not from ūtu- 'blow' but from ūṟṟu- 'pour' (which is pronounced ūttu in modern spoken Tamil). This is often used metonymously for the entire act of cooking a dosa from the batter.

MMcM said...

Thanks for the suggestion. I have updated the post to not just assert the 'blow' etymology, but source it to the OED. Pavanar has the 'pour' etymology, too, with exactly the phonetic development you describe. And, then, of the Tamil dictionaries in the Chicago DDSA, it seems, reading between the lines, that Cre-A has 'pour' and Madras 'blow'.

Perhaps the argument is that all appams batters are poured and ūttappam is different in being thicker on account of the leavening. Though that doesn't really feel decisive and I'm not qualified to judge.