Introduction

To the island, that we know as śrī laṃkā today have been allocated many toponyms. One of them is siṃhala/sīhaḷa being Sanskrit/Pāli. One of the earliest references to these toponyms is in a Chinese record that goes back to the end of the 3^rd century.1 Another toponym is īḻam, being Tamiḻ. The earliest references go back to the 2^nd/3^rd centuries AD.2

 

Let us look at the now generally accepted statement īḻam<sīhaḷa. It shall be read as īḻam is ‶derived" from sīhaḷa by means of dropping the initial dental sibilant and by means of additional phonological processes to which I shall come later. The statement goes back to Robert Caldwell from the 1840 s and 1850 s.3 He had many successors. Therefore, I speak of the Caldwell school.

 

In this paper I shall try to show that the formula īḻam<sīhaḷa is questionable and that it should be replaced by the formula īḻam͠sīhaḷa, which is read as ‶īḻam alternates with sīhaḷa". Using my own terms, I rephrase the Caldwell school´s position in the following three statements, which also indicate a critical evaluation.

 

1. According to the Caldwell school, the etymon of the word īḻam is allegedly that of the Pāli word sīhaḷa. We could also say that according to the Caldwell school īḻam is synonym with sīhaḷa, albeit their not being homonyms and homographs, but what the two toponyms mean is never stated. Although synonymy is projected, the quest for the etymon through morphological analysis is suspended. The Caldwell school´s exclusive handling of the words as phonemes, onomastica, and toponymical distinguishers only, discarding their status as morphemes, has had a decisive negative consequence on our understanding of the two toponyms.

 

2. īḻam is allegedly, according to the Caldwell school, a loan in Tamiḻ from an Indo-Aryan word that has undergone radical sound change in Tamiḻ. If we apply Sanskrit and Tamiḻ grammatical terms on Caldwell school´s interpretation, it can be said that this school treats īḻam not as a tatsama/taṟcamam, which is evident, but as a tadbhava/taṟpavam.

 

By taṟcamam, ‶sameness", is meant a word that is a Tamiḻ near-homograph, near-homonym and a synonym of another word from another language, in this case from an Indo-Aryan language. A taṟcamam as graph should be almost identical with the original word. If I spell in Tamiḻ script Sanskrit hiṃśa, ‶violence", with the help of kirantam (Grantha) as hiṃśa, then it is a taṟcamam.

 

A taṟpavam, Sanskrit tadbhava, ‶thus-becoming", is also an Indo-Aryan word by origin, but it has gone through radical changes when used by Tamiḻar. Some scholars would speak of a tamilisation of Indo-Aryans words, which again has a correspondence in a similar Tamiḻ grammatical concept known as vaṭamoḻiyākkam, ‶development of the Northern language" in the Naṉṉūl. If I spell in Tamiḻ the Sanskrit word hiṃśa, ‶violence", without the help of kirantam as iṅkicai, then it is a taṟpavam.

 

By tradition, Sanskrit is not the only language identified as source for a taṟcamam and a taṟpavam. Prakrit and Pāli, in general Indo-Aryan languages, are included. Modern grammarians even include English. In āspatri<hospital the former is classified as taṟpavam.4

 

The phoneme īḻam also is a taṟpavam— if, and only if, we accept that the Caldwell-school has hit the point with the formula īḻam<sīhaḷa. Here the graphical change is radical. I think the Caldwell-school´s statement is questionable and I shall try to show why.

 

Another such controversial taṟpavam having allegedly gone through a radical change, a taṟpavam that has been spread by Robert Caldwell also, is tiru in the formula tiru<śrī.5 tiru is classified as taṟpavam.6 We can see in both cases that homonymy and homography is absent, but that they in spite of this are regarded as synonyms meaning both ‶auspicious". A taṟpavam formation can be radical, but it would be wrong to call it distortion. The change follows phonetic rules. In the Tamiḻ grammar Naṉṉūl from the 12^th century AD., these rules are formulated that regulate the formation of a taṟpavam. Some of these rules have been implemented by the Caldwell school in the formula īḻam<sīhaḷa. Other ‶rules" have been invented for the occasion. Let us look at one of the favourites of the Caldwell school, the above-mentioned formula tiru<śrī.

 

Caldwell himself stated that the palatal ś was treated in analogy to a dental s that regularly was replaced by t. We get śrī>sri >t (i) ru.7 It is, however, unclear why the palatal ś should be treated in analogy with a dental s in this word only and not in other words with initial palatal ś. Moreover, how to explain the shift from long ī to short? Furthermore, how to explain the final -u? There are underlying difficulties that have been neglected. I do not exclude that Caldwell is right, but his explanation is not yet convincing. There is always a risk that rules are applied deductively. The scholar knows a stipulated start, here Sanskrit śrī, and he knows a stipulated goal, here Tamiḻ tiru. He selects rules to get from start to goal. If it does not work, he has to manipulate or even invent ‶rules". I think this is what happened when the formula īḻam<sīhaḷa was created. Before Caldwell, this formula did not exist neither in Tamiḻ or Siṃhala indigenous lexicography. It is a typical Orientalist conceptualisation by mainly two Western scholars and missionaries, by Robert Caldwell and Herbert Gundert. There were extra linguistic, ideological, motivations to do this, which I, however, shall not identify in this paper.

 

3. According to the Caldwell school, the only referent of the word īḻam is allegedly the island known as sīhaḷa. The fact that this word is multireferential and polysemous is neglected. This had serious consequences for the plausibility for the formation of the formula īḻam<sīhaḷa.

 

The Caldwell school´s interpretation of īḻam<sīhaḷa shortly presented above in three points can be regarded as established by many scholars in Dravidian historical linguistics—but not by the couple Burrow-Emeneau. Caldwell´s missionary colleague Herbert Gundert accepted this derivation and put it in his Malayālam-English dictionary from 1875. This was an important move because in Keraḷām was and still is a large contingent of īḻavar. Īḻavar is a caste name of toddy-tapers. The formula was used to support the legend that they had originally came from Īḻam [=sīhaḷa>Ceylon].

 

The Caldwell´s school´s extended interpretation was popularised without reference to Robert Caldwell in the West and East by spreading the famous dictionary Hobson-Jobson in 1886, that we find even on Internet today.8 It was also taken over by the compilers of the important Tamiḻ Lexicon in the 1920 s9 also—again without any reference to Caldwell. The compilers of this Tamiḻ Lexicon did not acknowledge the authorship of this very important derivation that has made history and still influences the consciousness of many intellectuals today. I give here the Tamiḻ Lexicon´s entry:

 

‶īḻam, n. < Pāli, Sīhala. Siṃhala. 1. Ceylon; ciṅkaḷam. (tivā.). 2. Gold; poṉ. (iraku. nakara. 68.) 3. Toddy, arrack; kaḷ. (cūṭā.). 4. Spurge, Euphorbia; kaḷḷi. (malai.)".10

 

The first part of the entry copies Caldwell, (but the following polysemous presentation is the work of the compilers).

 

The Caldwellian interpretation was made use of among several others, by for example the Lankan scholar R A H L Gunawardena in the 1980 s and 1990 s. He published in 1984 an often quoted and stimulating paper called ‶The People of the Lion. The Siṃhala Identity and Ideology in History and Historiography".11 Even if the word sīhaḷa occurred in the Dīpavaṃsa only from the 4^th century AD, R A H L Gunawardena took up the well-known [Caldwellian] interpretation that the Tamiḻ word īḻam is allegedly derived from sīhaḷa. So, whenever and wherever we find the word īḻam, we can conclude that this word has been preceded by the word sīhaḷa/siṃhala . As the word īḻam first appears in the Drāviḍī inscriptions of South India from the 1-2^nd centuries AD, he concluded that the term sīhaḷa was used at this time.12

 

I mention R A H L Gunawardena´s interpretation here to show that Caldwell´s interpretation is still instrumental and to show how a now active professional historian at Peradeniya University makes use of Caldwell´s historical-linguistic interpretation in an attempt to establish that the word sīhaḷa was earlier than the word īḻam. I regret that Caldwell´s name is absent in this presentation also. It refers to the Tamil Lexicon that had suppressed the reference to Caldwell.

 

Another modern important case is the historian Kārtikkēcu Intirapālā´s work. He tried to refine the Caldwellian thesis as late as in 1965, but again without mentioning Caldwell. His concern was to convince his readers that īḻam means ‶sīhaḷa".13 On his work the present Siṃhala ethnonationalist slogan ‶īḻam means sīhaḷa" was based. The formula īḻam<sīhaḷa has been exploited for political ends. Some persons still today use it to show that sīhaḷa/siṃhala must be older than īḻam, because the latter is allegedly ‶derived" from the former. They imply also that a priority of ethnonym reflects a priority of ethnie. They preclude that the signified is attached somehow to the signifier. They promoted the slogan that īḻam means the same as sīhaḷa. Therefore, when īḻam appears in a historical source, it does not refer to Tamiḻs, but to Sinhalas. Caldwell has been made use of in a political and pseudo-linguistic debate that concerns the rise of Siṃhala assertiveness as against the rise of Tamiḻ assertiveness. It is an irony of history that Siṃhala ethnonationalism, which is radically anti-Western, uses an Orientalist concept as pillar in its Siṃhala ethnonationalist ideology. It is called Siṃhalatva by its promotors to approach it to Hindutva.

īḻam — A Corrupt Form?

Robert Caldwell was speaking of a situation when Tamiḻ speaking people adopted loanwords from Indo-Aryan languages. What happens with the word siṃhala/sīhaḷa when used by Tamiḻar? He said that īḻam is a ‶corrupted" form of siṃhala or rather sīhaḷa.14 He also said that īḻam ‶comes from" sīhaḷa,15 which is another way of saying that īḻam is derived from sīhaḷa. I do not think, however, that Robert Caldwell meant ‶derivation" in the technical sense of affixation or umlaut. He did not intend to say that the relationship between the two words is of the type ‶kind<kindness" or ‶sleep<sleepy" where the derived part appears as a new and different word by suffixing, (prefixing or by umlaut). He used the word non-technically because of a vague use in an older tradition of historical linguistics. All kinds of word change over time were loosely explained as ‶derivation". Caldwell´s interpretation is not a ‶derivation" in the technical sense (but īḻa-m and īḻa-v-ar are genuine derivations from īḻa-). What he means is that the word sīhaḷa was corrupted when pronounced by Tamils, who said īḻam for sīhaḷa. Still, he did not take the opposite side, saying that this sound change was completely arbitrary. No, he tried to show that it followed some rules of sound change. The rules he identified were not wrong, but his way of applying them was questionable.

 

If I rephrase Caldwell´s interpretation of īḻam in modern terms, we could say that he presents it as a blend that has resulted in a combinative sound change combining loss of initial s with contraction. The loan effect is of course not an ‶assimilation" other in the wide sense of an ‶adoption". In his interpretation is no semantic change implied. īḻam means ‶sīhaḷa". They are synonyms, but not homonyms. Again, his own word ‶corruption" characterises his stand. It refers to phonetic changes that do not influence on the semantic side.

 

My own view is that īḻam is a Tamiḻ word referring to toddy or gold, etc. In contrast to this statement, we have to look at my other statement also. What happens when Tamiḻ adopts siṃhal (ḷ) a is that the word is modified into ciṃkaḷa (m). This adaptive process, indeed, is an example of a genuine blend. Its loan-effects are regular modifications of phonemes. ciṃkaḷa (m) is a blend, not īḻam.

 

The word siṃhala and its modified blend ciṃkala (m) are Indo-Aryan words meaning ‶lion-like" or ‶small lion".16 siṃhala/sīhaḷa and īḻam are not cognate and congruent. The words īḻam and sīhaḷa have been connected by Caldwell and by his followers based on sounding similarity (when allegedly the initial s has been dropped). He and his followers interpreted partial homophony, i.e. partial agreement of pronunciation, as synonymy. Caldwell´s and his followers´ interpretations are not even based on complete homographs, i.e. agreement of spelling, because of his neglect of the retroflex approximant ḻ in īḻam. In their presentation of (s) īlam, we find an alveolar l or retroflex ḷ. In my view, the two interpretations can be made plausible only by conjuring away these and other facts.

 

Robert Caldwell stated in his monumental A Comparative Grammar of the Dravidian Languages from 1856 that the word īḻam seems to have been corrupted from sīhaḷa, by the omission of the inititial s.17 He left this statement unchanged even in the second edition of his grammar from 1875. It is important to see his own wording. He wrote: ‶… Îṛam [= īḻam], Ceylon, a word which appears to have been corrupted from the Sanskrit Simhalam [sic], or rather from the Pali Sihalam [sic], by the omission of the initial s… "18

 

In this statement were implied some other statements, but he exposed only one. He said simply that initial s- is dropped. As the formulation stands in the form of a general rule, it is rather odd. Neither the Tolkāppiyam or the Naṉṉūl or Vīracōḻiyam has such a rule. His statement gives, however, meaning if we connect it with an earlier correct statement in the same work, namely that if (initial) s is the first consonant of a Sanskrit derivative, it is sometimes omitted altogether.19 Here we have two new elements. The statement is now only sometimes valid and is valid only for Sanskrit words, which are derivatives. Here ‶derivate" is correctly used for words that have a Sanskrit base like siṃha-la/sīhaḷa.

 

Therefore, there is no rule and the range is limited to some Sanskrit words. This makes sense, which he demonstrates convincingly by giving some examples of how Tamiḻ transforms Sanskrit words with inititial s: sandhyā>anti, ‶evening "; sthānam>tānam, ‶place".20 We can increase the list with one further example (that is not given by Caldwell): Prakrit samaṇa is sometimes reproduced in Tamiḻ as amaṇ (ṇ) ar.

 

Thomas Burrow published in 1947 a paper where he took up the elimination of initial of all three Sanskrit sibilants. He gave many more examples.21 According to him, the elimination of initial sibilant is frequent in Tamiḻ and Malayāḷam, but more rarely in Teluṃkam and Kaṉṉaṭam.22 He also points out that there exist parallel forms that have not dropped the initial sibilant and that those who have dropped it get it reintroduced from Sanskrit at a later stage.23

 

In 1988, M. B. Emeneau took up Thomas Burrow´s thread again and made additions to the theme ‶sporadic development of c/s to zero". The emphasis is only ‶sporadic", which has to contrast against ‶regular". Sporadic change happens occasionally in a seemingly arbitrary manner. Emeneau, however, accepted sporadic sound change as against those who insisted on strict regularity. Therefore, let us take c/s>zero, for what it is, a sporadic development.

 

There is another development also, the replacement of c/s with t and k.24 So, there is no rule that says that initial c/s should be dropped. It is regularly kept, and sporadically dropped or replaced by t and k. So far, there is nothing controversial about stating that in some Sanskrit loanwords initial s is dropped sporadically when taken up as taṟpavam in Tamiḻ.

 

We have to notice also that the occasional drop of initial only occurs for the cases of palatal ś and dental s. It does not happen with cerebral ṣ in Sanskrit words. ṣaṣṭi does not become *aṭṭi, but caṭṭi. ṣaḍaṃga becomes caṭaṅku. ṣaṇmukha becomes caṇmukaṉ, etc.

 

Moreover, those words that have dropped inititial dental s or palatal ś have sometimes double forms: capai, avai<sabhā; canti, anti<sandhyā; camayam, amayam< samaya; cintu, intu<sindhu; cūci, ūci<sūci, etc. This also shows that the sporadic dropping of initial s or ś was counterbalanced by rules of regularity for the formation of a taṟpavam.

 

Let us participate in Caldwell´s language game. Now, we have dropped the initial s in sīhaḷa and we have -īhaḷa. What happens now? We have a long way to go. We have to arrive at īḻam. The problem is that Caldwell in his Grammar… does not guide us properly. We have ended up in a blind alley. Caldwell has, however, not abandoned us. He has written another work called A History of Tinnevely, where he instructed us how to proceed. He wrote: ‶Sihalam… is the name by which it [the island Lanka] was called by the later Buddhistic writers, from which came in regular succession the forms Sihalam, Sīlam, Selen-dip, Sereendib, Zeelan, Ceylan, and Ceylon… From the form Sīlam comes the Tamiḻ īḻam".25

 

From this statement, we can extract the following. He regards the name laṃkā to be older than siṃhala, which according to our present knowledge is wrong. This is, however, another story, not to be told here. When he spoke about a ‶regular succession", it is not clear whether he meant that, the word came in the order he has given or whether they also are supposed to be derivations. As he says ‶came… from" the latter seems to be the case. When the second ‶derivation" sīlam was attained and was adopted by Tamiḻ, the Tamiḻ speaker ‶corrupted" this Indo-Aryan word by dropping the initial s. The word īḻam was finally ‶derived" from sīlam. So, the form *īhalam is no problem, because it did not exist. The base from which we have to start is *sīlam.

A Critical Examination

I comment now critically on this derivation by Caldwell in 12 points.

 

1. The form *sīlam does not exist anywhere. It is hypothetically constructed by Caldwell. When constructing such a word it must be shown that the construction follows phonological rules. In this case, Caldwell moved in a circle. He knew that initial s is (sometimes) dropped in Indo-Aryan loan words when adopted by Tamiḻ. So, he constructed *s-īlam and let the initial s- be dropped. He got what he wanted: -īlam. The result was included in the premises. Such a way of reasoning is not convincing.

 

2. Furthermore, he does not explain the steps sīhaḷa>*sīlam.

 

3. If we apply the rule for the formation of a taṟpavam strictly, we get the following result. His starting point was sīhaḷa which allegedly was followed by *sīlam. The phoneme sīhaḷa would, in the mouth of a Tamiḻ, have resulted in *cīkaḷa (m) or *cīyaḷa (m) and not in *sīlam. Medial h can be changed into y. Furthermore, when the Pāli word sīhaḷa is transliterated into modern Tamiḻ, there are two possibilities, to change the ha to ka or to use a kirantam ha for ha. The initial s is not dropped, but changed into a Tamiḻ akkuru c or into a kirantam s-. Therefore, following the rules for taṟpavam formation, sīhaḷa is expected to appear as *cīkaḷa(m) or *cīyaḷa(m).

 

There was a point for Robert Caldwell to avoid to end up in *cīkaḷa (m) or *cīyaḷa (m). It would not have brought him to the goal, to īḻam. So, he invented a member of the chain of derivation, namely *sīlam. Now things became easy. Just drop the initial s- and you get īlam. His problem is that *sīlam is technically a taṟcamam/taṟpavam. Here, we could end the whole exercise, because initial c/s in an already formed taṟcamam/taṟpavam is not dropped.26 Let us speculate what would happen if we would accept the case of a sporadic loss of initial s- in sīhaḷa in a correct formation of a taṟpavam. The result would be *-īkaḷa. This does not bring us anywhere.

 

4. In the case of the word siṃha, there is an established tradition not to eliminate the initial s, but to replace it with c-. siṃha is regularly Tamiḻised as ciṅkam or cīyam. There is no form *iṅkam. We do not have the drop of the initial s- as rule, but its transformation into c-is regular. We find many examples.27ciṅkam or cīyam is only one of them.

 

It has been stated by Anavaratarinayakam Pillai that ciṅkaḷam is a modern taṟpavam.28 This statement implies that in the pre-colonial period it was not there. Then, we allegedly only find forms that have dropped initial dental s. ciṅkaḷam is, however, not a modern taṟpavam. We find it in one of the oldest nikaṇṭu-works, the Tivākaram, available from the 8^th century AD., and even earlier, we find this word. It could, however, hardly be expected much earlier as the form sīhaḷa, having the island as referent. It is from the 3^rd-4^th century AD.

 

My interpretation of ciṅkaḷam being a taṟpavam deviates from a ‶Tamiḻ" interpretation once launched by a Tamiḻ scholar, by Ma. Cirinivāca Aiyaṅkar (M. Srinivasa Aiyankar). He thought that ciṅkaḷam was not a taṟpavam, but a Tamiḻ word: ‶There can be no necessity for us dragging a Sanskrit word through many stages, when there is already in the Tamiḻ language the simple word Singalam".29 This is an expression of Tamiḻ linguistic nationalism, not of historical linguistics.

 

5. Robert Caldwell neglected the Sanskrit form siṃhala and focused the Pāli form sīhaḷa. He does not explain why. This creates an extra problem for him because the first appearances of the toponym īḻam are from the pre-Pallava period in Tamilakam where we find no trace of Pāli.

 

It is evident to me why he chose Pāli. With the Pāli form, he does not have to explain what has happened to the anusvāra in the phoneme siṃhala. Caldwell should have at least chosen the classification Prakrit to make his analysis more plausible from a historical point of view, but at his time, inscriptional and literary Prakrit in Tamiḻakam, and tamilised inscriptional Prakrit in Drāviḍī was not yet known. He had to appeal to the reader´s imagination to place Pāli somehow to Tamiḻakam before the time when īḻam was formed. Buddhaghosa´s translation team, bringing Pāli to Tamiḻakam, did not arrive before the 5^th century AD.

 

Robert Caldwell ended up in Pāli. The focusing on sīhaḷa instead of siṃhala is, it seems to me, not based on a finding of the word sīhaḷa in this empirical world in or near the language world of the Tamiḻar, but on the wish to avoid a phonological problem that were created by the anusvāra in Sanskrit.

 

In reconstructing Caldwell´s thinking, we can see that it is deductive all the line, but that he also did not end up where he actually wanted to end up, in the phoneme īḻam. By neglecting rules for the formation of a taṟpavam, he ended up in the phoneme īlam having an alveolar l, not even a retroflex ḷ, and still less a retroflex approximant ḻ, which makes an enormous difference. īlam, with an alveolar l, having the island as referent, does not exist in the world of Tamiḻ or any other language world. īlam is corrupted, indeed, by Caldwell, but not by Tamiḻar.

 

6. Where do we find in Tamiḻakam from the 1^st or 2^nd century AD., or even earlier, any reference to a Pāli form sīhaḷa? The Caldwell school has to show that the word sīhaḷa existed in this world whenever and wherever the word īḻam first appeared. This cannot be demonstrated for the case of Tamiḻakam and īḻam. Furthermore, the transformation from sīhaḷa to īḻam is conceptualised as a process of a long time, may be of hundreds of years. When Caldwell wrote, he did not know about the pre-Pallava occurrence of the word īḻam, nor did he know about the Pallava and Cōḻa occurrences. His view was that the īḻavar, bringing the term īḻam, were brought to Tamiḻakam by the first Syrian Christians in the first centuries AD. This implies that when the word īḻam was finalised, it had a long development behind it. Caldwell´s theory precludes that the word sīhaḷa stands at the beginning of a chain of transformations ending up in the first centuries AD. with īḻam. Therefore, sīhaḷa can be said to be very much older than īḻam. This is, however, not the case. Our present stand of knowledge tells us that īḻam appeared somewhat earlier than sīhaḷa.

 

7. Caldwell used diacritical signs. Therefore, we can take the spelling of sīlam with an alveolar l as a conscious choice. This created another difficulty, namely to explain how an alveolar l can become a retroflex approximant ḻ that has proto-Dravidan origin. He never made himself aware of this problem. To convince, the Caldwell school had to show a rule of transformation from alveolar or retroflex l to retroflex approximant ḻ in loanwords from Indo-Aryan languages. I do not deny that the development ḻ>ḷ>l can be demonstrated in some cases, which are not mere misspellings, but I do deny that its reversal can be demonstrated as a rule. It is this rule based reversal that Caldwell and his school precludes having as a foundation. The existence of free variation between ḻ and ḷ, like in pavaḻam, ‶coral", and pavaḷam is of course not the same as a diachronic transition from the former to the latter.

 

A Tamiḻ scholar has made an attempt after, but still in the sprit of Caldwell, to explain what Caldwell did not explain regarding the transition from sīhaḷa to sīlam and from there to īḻam. In this attempt other phonemes are invented: *sīlam is written as *sīḷam and said to have been preceded by *sīyaḷam… So we get sīhaḷa >*sīyaḷam >*sīḷam>īḻam. The form *sīyaḷam is a taṟpavam that is constructed only seemingly following rules for the formation of a taṟpavam, namely the rule that medial h can be changed to ya. The formation should, however, be cīyaḷam, if we follow rules of taṟpavam formation strictly. Tamiḻ words, including a taṟpavam cannot start with dental s. Tamiḻar transform the Indo-Aryan s or ś into c (or t) or drop it sporadically. Therefore, in a taṟpavam we can expect either zero or c as initials.

 

We follow the exercise to the end. We have, however, already reached the end, because the sporadic drop of s or ś does not occur after a taṟpavam already has been formed. Let us take the Prakrit form samaṇa. In the mouth of a Tamiḻ, it turns out as either amaṇa or camaṇa. There is a possibility that by force of analogy amaṇa is taking the form camaṇa and that camaṇa by the force of analogy takes the form amaṇa, but this is something else than saying that camaṇa is exposed to the loss of initial s or ś. There is no initial s or ś. Furthermore, drop of an initial caused by complicated sound collisions is something else than a sound change caused by force of analogy. Both Caldwell and Pillai have made the same act: they let a taṟpavam—*cīlam and *cīyalam respectively —undergo the (sporadic) drop of initial s. As there was no s, they had to make their analysis plausible by endowing their taṟpavam with an initial s. It is not a taṟpavam, but it is the original Indo-Aryan word that may be exposed to sporadic drop of initial s or ś.

 

This ends the exercise, but we hang on. Now, another ‶rule" is invoked according to which the stress on the first syllable has led to the dropping of ya after it. Therefore, we get allegedly siyaḷam>*sīḷam from which by dropping the initial s, the word īḷam is obtained. This is a little better than Caldwell´s sīlam, because it has retroflex ḷ, but still, it has no ḻ.

 

There is, however, no such ‶rule" for Tamiḻ—it is different for Teluṃkam— according to which the stress on the first syllable has led to the dropping of ya after it. What about cīyakkāy, cīyaku, cīyaṅkal, cīyaṇar, cīyar, cīyaṉ, cīyāṉ, etc.? They should all have dropped the syllable ya, or alternative forms without ya should have developed, if the ‶rule" has any application. Not least, we have the phoneme cīyam from Prakrit/Pāli sīha. It should have taken the (alternative) form *cīm or even *īm, if we follow the scholar´s own homemade rule. I am aware that in Teluṃkam there is a drop of ya, like in sīkāya—Tamiḻ gives it as cīyakkāy— but we cannot argue for Tamiḻ by just projecting one rule valid for Teluṃkam into Tamiḻ. It seems that Teluṃkam also has inspired the scholar to write a dental s in *sīyaḷam. Teluṅkam has a dental s, Tamiḻ has not.

 

Finally, the scholar appeals to another observation, namely that some Tamiḻar are not inclined to attach much value to the change of ḷ to ḻ, as often the Tamiḻar in the so-called pure-Tamiḻ districts, Tinnevēli and Maturai, make no difference between them.30 This statement is true in modern spoken Tamiḻ, and even in modern written Tamiḻ, but not in classical written Tamiḻ, if we neglect mere misspellings. The object of our study is classical Tamiḻ. I am aware, that there is a change from ḻ to ḷ, but to find cases for the reversed change from ḷ to ḻ, changes which are not mere misspellings and which are applications of a rule, are nonexistent. There is no such rule.

 

8. Caldwell´s argumentation excludes by implication that īḻam is a genuine Tamiḻ word, but we do not get any arguments for this exclusion. The identity marker ḻ is one indication that the word is not a loan word, but is a Tamiḻ word. This marker cannot be just neglected.

 

9. Caldwell´s statement seems to be deduced by inspiration of the other examples of Indo-Aryan loanwords given above which drop initial s. His is not an inductive study, but a deductive one, based on the force of an analogy like īyam<sīsa. Again, the general statistical rule is that initial s is not dropped, but is replaced by c, like siṃha>ciṅka (m) or cīyam.

 

If we compare īḻam with the near-homophone and near-homograph īyam, ‶laid", we find that they have the same root-type, V: C, but in the case of īyam, we can be sure to encounter a genuine taṟpavam from Sanskrit sīsa. The initial s has been dropped and the medial s has changed into ya. They are synonyms, both meaning ‶laid". Both have the same referent, laid. A case like īyam<sīsa may have been in the mind of Caldwell when he presented īḻam<sīhaḷa.

 

An untrained mind in the art of taṟpavam formation may not even be aware that the two are cognates; they look so differently. He may take īyam as a Tamiḻ word and invent a story about Tamiḻar being first to have produced laid. I, however, do not take it for a Tamil word, because I know that it is a taṟpavam. I can identify synonymy in connection with a regularity of sound change. These two become arguments for classifying them as cognates. If I apply the same procedure to the formula īḻam<sīhaḷa, I find that that there is neither synonymy nor regularity of sound change in any direction. I can safely say that there is no homology between the two derivations. Therefore, in my mind, īyam<sīsa has no persuasive force on my classification of īḻam<sīhaḷa.

 

10. Caldwell did not care for the polysemy of the word īḻam and of its several referents. To īḻam, he ascribed only one referent, the island sīhaḷa. Consequently, an īḻavar was a ‶Singhalese". The absurd consequence was that millions of Singhalese populated South India. His pupils took up his monosemic and monoreferential interpretation.

 

Robert Caldwell did not care about the referents toddy, gold or of spurge plant for īḻam. He must have known them, but he just did not care. He had access to the nikaṇṭu tradition, to the Tranqebar Dictionary, and to the dictionary of his missionary colleague Winslow that pointed at the polysemic character of the word. His approach was to strictly follow some selected arbitrary rules of derivation to present īḻam as a cognate of sīhaḷa. His acting was a performance restricted to phonology excluding morphology and semantics. Having accomplished this, he said that īḻavaṉ means ‶Singhalese", but ‶Singhalese" was just a phoneme of the onomastic type for him. He did not see it as a junction of morphemes and as a semene.

 

If he had been aware of what ‶Singhalese" means analytically and if he had been aware of the polysemic nature of the word īḻam, he would have realised that his derivation must be provided with a question mark. Sīhaḷa means ‶lion-like (person)" or ‶small lion".31 In consequence of Caldwell´s making īḻam a derivation of sīhaḷa, īḻam should also mean something like ‶lionlike" or ‶small lion". It does of course not. Caldwell´s īḻam>sīhaḷa is an interesting case of what may happens to a pure phonological analysis when morphology and semantics are consciously neglected.

 

11. Baffling is that Robert Caldwell, one of the founding fathers of comparative Dravidan linguistics, did not apply his knowledge about Dravidian languages to the word īḻam. If he had done so, he would have found that the word is spread in its basic form to several Dravidian languages, which points at a common origin in proto-Dravidian. This proto-Dravidian is beyond any possible influence from the word sīhaḷa. Even if the spread of the signifier īḻam is a result of diffusion to other Dravidan languages from one Dravidian source language, it remains to show when and where sīhaḷa could have influenced this source.

 

12. There is a puzzling statement by Caldwell in his History of Tinnevelli, which can be interpreted as an awareness of the associative and the analytical meaning of siṃhala. He wrote: ‶Simha means a lion, Siṃhala the lion country, that is, either the country of the lion-slayers or more probably the country of the lion-like men. "32 The reference to lion-slayer appeals to the traditional morphological analysis of sīhaḷa as consisting of allegedly two unbound morphemes sīha and la>lā, a noun and verb, constituting allegedly the meaning ‶lion-slaying". This analysis is explicit in the source itself and therefore it is not astonishing that Caldwell knows it.

 

I prefer the morphological analysis of sīhaḷa as consisting of an unbound morpheme sīha and a bound morpheme, a nominal suffix -ḷa, constituting the meaning ‶lion-like". Whatever may be the truth, Caldwell was aware of the associative etymology of sīhaḷa. If he had made a simple morphological analysis of the word īḻam, considering the polysemic and multireferential character of this word, he would have realised that the two are not synonyms.

 

I conclude from points 1-12 that Caldwell and his school, have not (yet) shown that īḻam is a taṟpavam of sīhaḷa.

Morphology of īḻam

I divide here īḻam into its morphemes. The free morpheme īḻa- belongs to a declension ending in -m in the ‶nominative" case, like mara-m, ‶tree". -m is a bound morpheme.

 

We can describe īḻam also in terms of classical Tamiḻ grammar going back to one of the oldest Tamiḻ grammars, to Tolkāppiyam, Collatikāram, section 7 on iṭai-y-iyal, and section 8 on uriyiyal, and say that īḻa- is an uriccol, ‶prominent word". uriccol and Sanskrit dhātu, ‶base" are sometimes connected. -m is an iṭaiccol, ‶affix", or —as other later grammarians have said—pakkacol, ‶word on the side".

 

In the other ‶cases" than the nominative another inflectional base, īḻa-tt-, is used. It is an empty morph by which a stem is expanded.

 

The morphemes īḻa- or īḻattu- are inflectional bases, but īḻa- (or īḍi)- is also a root. It belongs to the fourth root (of six Dravidian roots) of the type V: C (= long vowel + consonant). Here, we have to be especially careful not to go into a trap. A taṟpavam may have the same root form V: C, like for example īyam, ‶laid".

 

The nominal bases īḻa- and īḻattu- can also be used attributively as adjectives, for example in compounds.

 

To be hypercorrect, īḻa- per se refers to toddy, gold or a spurge plant and, but īḻa- per se is of course not in language performance. In use is only īḻa-m and īḻa-ttu- that refers to a neuter ‶thing" distinguished from a female and male ‶object". This neuter ‶thing" is the island, toddy, and gold or spurge plant. This ‶thing" can be classified in terms of classical Tamiḻ grammar as aḵṟiṇai, ‶non-class", a class of non-persons.

 

The Dravidian Etymological Dictionary (DED) distinguishes between īḻam the island [=550] and īḻam toddy [= 549]. It presents them as homonyms, and consequently as two separate entries. I think this is correct. Making two entries is not only a pedagogical device.33 It has a justification in the fact that although the two entries are absolute homonyms, they are not synonyms and there is no relatedness between them.

 

The Sanskrit and Tamiḻ nikaṇṭu-tradition does not make our modern distinction between analytical and associative etymologies, but Tolkāppiyam, collatikāram 397 has a fourfold classification of words with special regard to the proviniens of words:

1. iyaṟ-col, ‶natural words" [=common native Tamiḻ words]. Tolkāppiyam, Collatikāram, cūttiram 394, says: ‶… of them [=of the four words] iyaṟcol is used in the centamiḻ area and elsewhere without change in meaning. "34
   In other words, an iyaṟcol is an indigenous Tamiḻ word. In my interpretation, īḻam is an iyaṟ-col.
2. tiri-col, ‶varying words" [=literary native Tamiḻ words]. Cūttiram 399, speaks of two kinds of variations, which later have been exemplified by commentators. The variations refer to sound changes that in one case do not change the meaning but in another case does. The two variations have been exemplified with kiḷḷai, ‶parrot ">kiḷi, ‶parrot", and maññai>mayil, ‶peacock". Sīhaḷa>īḻam are of course no tiri-col which appear only in the context literary Tamiḻ words.
3. ticai-c-col, ‶words of the directions", more precisely of the twelve neighbouring countries. The commentator Nacciṉārkiṉiyar of the Tolkāppiyam, Collatikāram 397, has specified the twelve. The first is ciṅkaḷam35 that we can take as an example of a ticai-c-col. This classification of ciṅkaḷam is acceptable to me also, even if I would like to go one additional step and classify it as vaṭacol (see below): ciṅkaḷam is a taṟcamam of Sanskrit siṃhala. Īḻam is no ticai-c-col in traditional classifications, but there is no principal objection to classify it as such…
4. vaṭacol, ‶northern words" [=Sanskrit].36 The commentator gives vāri, mēru, kuṅkumam, maṇi, māṉam, mīṉam, vīram, etc, as examples.37 As these words are almost unaltered, they are classified as taṟcamam by the Naṉṉūl. The commentator Teyvaccilaiyār also includes Prakrit in the category vaṭacol. In my interpreatation, īḻam is not a vaṭacol, but ciṅkaḷam is.

Unfortunately, īḻam has never been taken up in this fourfold classification by commentators. Evidently, there is nothing remarkable about this word. īḻam is just one among hundreds normal words of the class iyaṟcol, I presume.

 

A person thinking along the lines of Robert Caldwell might expect Tamiḻ paṇṭitar to classify īḻam as vaṭacol. There is no indication that this ever has been done. Moreover, there is no way either to relate this formula to traditional Eḷu– Siṃhala lexicography before Caldwell. These facts show that the derivation īḻam<sīhaḷa is not supported by traditional learning in Tamiḻ and Siṃhala lexicography. One of the most important sources is the Tivākaram that teaches īḻam͠sīhaḷa.38 It is not plausible to say that the paṇṭitar of old could not have found a precursor to this formula īḻam<sīhaḷa because their knowledge in linguistics was undeveloped. The old Tamiḻ grammarians and lexicographers had enough knowledge to anticipate parts of Caldwell´s formula īḻam<sīhaḷa. I refer to the fact that these grammarians knew fundamental rules of etymology and they could build on a long Sanskrit tradition. They had the means, but they did not take up īḻam within the fourfold classification of the Tolkāppiyam. My tentative explanation is that they regarded īḻam under the category of iyaṟ-col, ‶natural words" [=common native Tamiḻ words]. Therefore the precondition for creating a proto-Caldwellian formula—the classification of īḻam as vaṭacol, — was non-existent. The Tamiḻ paṇṭitar of old took up, however, the word īḻam in relation to ciṅkaḷam, and in such a way that it becomes clear, that they did not defend a derivational relation, but an alternating one.

 

Furthermore, it is possible to treat the word īḻam as a Tamiḻ word without ending up in anomalies. Therefore, we have no reason to question that īḻam is a Tamiḻ word.

To sum up, the word īḻa-m has:

• Two inflectional bases, īḻa- and īḻa-t-tu. Only the latter is also used in the oblique form.
• Three for us relevant derivations (=affixations), īḻa-m, īḻa-va-r and īḻatt (u)-āṉ (see section 2.3.1).
• Two meanings, which represent an etymon, ‶juice" and ‶metal".
• Four for us relevant referents, toddy, gold/metal, spurge plant, and the island known also as laṃkā.
• A spelling with the retroflex approximant ḻ that we only find in Dravidian languages.
• A long root vowel that can be traced to proto-Dravidian origin.
• A root, that belongs to the fourth root (of six Dravidian bound roots) of the type V: C (= long vowel + consonant).
• Congruents in other Dravidian languages.
• A negative, but important instance, is that traditional learning among Tamiḻ paṇṭitar has not classified īḻam as vaṭa col and that Siṃhala intellectuals before Caldwell have not promoted the formula īḻam<sīhaḷa.

Alternation and Convertibility instead of Derivation

There is a beautiful example of īḻam͠ciṅkaḷam͠ilaṅkai in a Tamiḻ Cōḻa inscription where all three, īḻam, ciṅkaḷam and ilaṅkai, more precisely derivations of them, are used at the same time and place, in the same passage by the same writer of the same inscription.39 In this very case, we could also write īḻam --> ciṅkaḷam --> ilaṅkai. Alternation and convertibility shake hands.

 

I focus the Cōḻaṉ Rājādhirāja (1018-1054) who left a Tamiḻ inscription40 [mixed with kirantam] dated to 1046. In this inscription, he talks about political conditions in the island from a South Indian perspective. It was during the period of Cōḻa occupation of the island in the 11^th century. He mentions four insular Kings that are all known in the latter part of the Mahāvaṃsa [=Cūḷavaṃsa] under similar or other alternative names. These Kings all tried to oppose to Cōḻa rule from Rohaṇa, but all failed. Some were corrupted and had achieved the shores of South India. The Mahāvaṃsa has little good to say about them,41 probably because they failed in rejecting Cōḻa rule and because they were morally corrupt. I neglect here the complexity of politics and take only what is necessary for my purpose to show the alternation and convertibility of Tamiḻ toponyms.

1. vikkiramabāhu [Vikkamabāhu I =Kassapa VI, 1029-1041, 1029-1040] is [ironically] said to be ilaṅkaiyar ko[m]āṉ vikkiramabāhu,42 ‶King Vikkamabāhu of the Lankans". Here ilaṅkaiyar refers to the whole population or to demos. At this stage the concept of ilaṅkaiyar had fully developed.
2. vikkirampāṇṭiyaṉ [Vikkamapaṇḍu, 1044-1047, 1042-1043] is said to have entered īḻam having lost his possessions in teṇṭamiḻmaṇṭalam,43 ‶Southern [part of the] Tamiḻ hemisphere". This part is important because it reveals a consciousness of three separate regions. Southern Tamiḻmaṇṭalam implies a consciousness of Northern Tamiḻmaṇṭalam. To this we can now add a third region, īḻam.
3. vīracalāmekaṉ [Jagtīpāla, 1047-1051, 1043-1046] is said to believe that īḻam is superior to the area that he formerly had occupied in South India. He is called [ironically] ciṅkaḷattaracaṉ,44 ‶King of siṃhala". Here we have the case of īḻam~ciṅkaḷam, īḻam --> ciṅkaḷam in the same paragraph.
4. cirvallava mataṉarājaṉ [Parakkamapaṇḍu I, 1051-1053, 1046-1048]45 is [ironically] called īḻattairācaṉ, ‶king of īḻam".

 

The use of ilaṅkai, ciṅkaḷam and īḻam in the same inscription reflects the pluralism of alternating signifiers in use. This pluralism —limited to ciṅkaḷam ͠īḻam —was already codified in a normative Tamiḻ glossary like the Tivākaram from the 8^th century. Later, all three are used as alternations. The references ‶king of ciṅkaḷam" and ‶king of īḻam" are used interchangeably. We can just add in our mind ‶king of ilaṅkai" without breaking any convention of language use. ilaṅkai aracaṉ appears for example in the kamparāmāyaṇam, pālakāṇṭam 365.

Conclusion

I have concluded that īḻam is a proper Dravidian word. It is an iyaṟ-col. This analysis is incompatible with the formula īḻam<sīhaḷa. I have also concluded by examining literary texts and inscriptions that this formula should be replaced by the formula īḻam~sīhaḷa.