The Romanisation of Indic Script Used in Ancient Indonesia

The paper calls the attention of Javanists, and Indonesianists at large, to theoretical as well as practical issues connected with conventions for the Romanisation of Old Javanese, Old Malay, Old Sundanese and other Indonesian languages which were traditionally written in Indic scripts, and underscores the difference between transliteration, on the one hand, and transcription or orthography (‘spelling’) on the other. In doing so, it replies to the points of criticism raised by Dick van der Meij in his review (2012) of From Laṅkā Eastwards, edited by Acri, Creese & Griffiths (2011).

A review by Dick van der Meij of the collective volume From Laṅkā Eastwards; The Rāmāyaṇa in the Literature and Visual Arts of Indonesia (kitlv 2011), coedited by the present writers with Helen Creese, recently appeared in Bijdragen tot de Taal-, Land-en Volkenkunde 168.2-3 (2012), pp. 337-40. Despite a measure of criticism concerning some of the individual contributions, Van der Meij concedes that our volume is 'provocative and original, and shows that interpreting old textual and visual materials is a richly challenging endeavour that's reward-ing, if inherently fraught with uncertainty' (p. 339). We thank Van der Meij for his on the whole balanced review, which addresses both the strengths and weak points of the volume. Nevertheless, we feel compelled to reply to the final section of Van der Meij's review, under the heading 'Spelling' . The section reads as follows: My main critique is a practical one and concerns the spelling of Old Javanese which is inconsistent throughout the book. For no reasons provided, Acri and Griffiths deviate from the spelling Zoetmulder (1982) uses in his Old Javanese-English dictionary. Griffiths proposes a spelling that adheres 'strictly to international norms for the transliteration of Indic script types' (p. 133, note 1) without explaining what these norms are and he does not adhere to his own rules. Both Acri and Griffiths fail to provide reasons why they should deviate from Zoetmulder's spelling. I see no reasons why students of Indonesian literatures and manuscripts need to turn to India for inspiration for transliteration systems of Indonesian scripts. Now, of all things we see the word kakavin (Acri) where in Bali and Java the spelling has been kakawin for ages. There is no /v/ in Old Javanese, or in Balinese for that matter! In my view, these spelling alterations are ludicrous and rather than being an expression of sound scholarship, degrade the book. More importantly, I fear that these spelling changes herald a return to the deplorable situation where Indonesian cultural phenomena are not considered in their own rights but rather in those of a so-called 'Sanskrit Cosmopolis' . […] It is to be hoped that these scholars continue their work but please, do not complicate matters unnecessarily. It is difficult enough.
A reply to Van der Meij's points of criticism will provide us with an opportunity to give the desired arguments in favour of the Romanisation conventions implemented in our contributions to the edited volume, as well as in other works by the present writers, published or in-progress. Our aim is to call the attention of Javanists, and Indonesianists at large, to theoretical as well as practical issues connected with conventions for the Romanisation of Old Javanese, Old Malay, Old Sundanese and other Indonesian languages which were traditionally written in Indic scripts, and to underscore the difference between transliteration, on the one hand, and transcription or orthography ('spelling') on the other. Our reply is meant not so much as a rejoinder to Dick van der Meij but rather as the rekindling of scholarly reflection about Romanisation conventions.
Bijdragen tot de Taal-, Land-en Volkenkunde 170 (2014) [365][366][367][368][369][370][371][372][373][374][375][376][377][378] We may begin by pointing the interested reader to the very thorough discussion of the general problem of the conversion of scripts provided in the monograph of that title by Hans Wellisch (1978), and think it useful to reproduce here the following general observation and three definitions (pp. 17-18, italics original, bold ours): All scripts, particularly alphabetic ones, can be used for at least three different purposes which, though interrelated in various ways, must nevertheless be clearly distinguished and which have to be considered on their own merits and for their individual objectives if confusion is to be avoided. As we shall see later, it is the failure to make such clear-cut distinctions between the purposes of using a script that has led to lengthy and largely fruitless discussions about the value and usefulness of different schemes for conversion of scripts for various applications.
The first and foremost use of script is in the orthography of a language.1 Although, in theory, the letters and letter combinations of an alphabetic script ought to express as unambiguously as possible the phonemes of the language for which the script was designed, this has probably never been the case even in the distant past when various alphabets were invented. […] The second use of script is in transcription, when the phonemes of a source language written in a dissimilar script (or not written at all) are represented more or less faithfully by the characters (letters and other graphic signs) of a dominant script. […] The third use of script is in transliteration, when the graphemes of a source script are converted into graphemes of a target script without any regard to pronunciation and also, at least in the strictest sense, without either adding or deleting any graphemes that are not present in the source script.
Throughout this paper, we use the term Romanisation to denote any form of conversion of a source script to the Roman alphabet, whether orthographic, transcription or transliteration. We also note that the term 'spelling' is not a basic part of the conceptual framework we borrow from Wellisch.2 After these introductory remarks, we will now proceed by addressing specific statements of the reviewer (in italic and within quotation marks) one by one.
'The spelling of Old Javanese […] is inconsistent throughout the book' . Van der Meij's complaint about 'spelling inconsistencies' is understandable to some extent; however, it must be admitted that Acri has already discussed this aspect of the book in his Introduction (p. xiv), justifying the editors' choice with the following rationale: […] we had to face the practical consequences of differing scholarly practices in dealing with matters of transliteration, transcription, and spelling. We have decided to give virtually free rein to the authors' individual preferences, the resulting variability of usage being an eloquent reflection of the diversity of perspectives and scholarly backgrounds which it has been the purpose of this volume to give open forum.
We had hoped that this editorial acknowledgement of the points where we felt unable to impose uniformity, and why, would be enough to dispel any objection as to 'spelling inconsistencies' throughout the book, as long as the individual papers would adhere to internally uniform policies. Regrettably, this has turned out not to be the case.
'For no reasons provided, Acri and Griffiths deviate from the spelling Zoetmulder (1982) uses in his Old Javanese-English dictionary. […] Both Acri and Griffiths fail to provide reasons why they should deviate from Zoetmulder's spelling' . The main problem with Van der Meij's choice of words is that it misrepresents the state of affairs, giving the false impression that the 'spelling' implemented by Zoetmulder in his dictionary of 1982 constitutes some kind of generally accepted scholarly norm for Romanisation, and that it does not take into account the different applications which may justify different Romanisation schemes. Even if we limit ourselves here to the domain of scholarly editing of texts in Old Javanese (which was far from the only scholarly approach represented in our volume), as far as we are aware, there are no clear, consistently applied norms currently agreed upon by a majority of scholars in the field, which can be deduced from the editions of Old Javanese texts published over the last decades-that is, after 1982. There is no observable scholarly standard even if one looks only at kitlv publications, and certainly none if one includes (as one should) publications in Old Javanese epigraphy. In the Introduction to his edition of the Old Javanese Śaiva scripture Dharma Pātañjala, in the subsection 'Notes on Conventions' , Acri (2011b:xiv) discusses this matter as follows: No consensus has been reached yet among scholars about the adoption of a standard orthographic system for the roman transliteration of the varieties of in origin Indic scripts in which Old Javanese texts are written. Previous generations of editors have used different systems, either adopting the conventions used by early Dutch scholars, by Zoetmulder's monumental Old Javanese-English Dictionary (ojed), or introducing their own codifications-often with little or no success in drawing further followers.
In hindsight, it is regrettable that the concepts of orthography, transliteration, and transcription were not kept thoroughly distinct here. Anyhow, in footnote 2 Acri describes, by way of example, the system used by Willem van der Molen to transliterate the three Javanese codices of the prose Kuñjarakarṇa (1983), which was subsequently used by Wiryamartana (1990) and by Sedyawati, Van der Molen & Wiryamartana (2002). In spite of being by far the most analytic system proposed so far-to the extent of introducing specific diacritics in order to avoid the use of more than one Roman letter to transliterate, for example, the 'aspirated' and 'unaspirated' stops (p/ph becomes p/p)-it has the drawback of being very rich in diacritics and using them in a manner inconsistent with all previous systems, so much so that it is not immediately intelligible even to the specialist in Old Javanese (let alone related fields of study), requiring instead a significant amount of familiarisation. Moreover, the system is not adapted to the representation of Indic syllabaries with the four-way distinction (d/dh/ḍ/ḍh) proper to Indic writing as it was received in Indonesia in the middle of the first and used well into the second millennium ce. In any case, to return to Van der Meij's objection, this system too yields a very different woordbeeld from Zoetmulder's (dharma becomes ḍaṙma). We will come below to the justification of our own choices in matters of Romanisation.
'Griffiths proposes a spelling that "adheres strictly to international norms for the transliteration of Indic script types" (p. 133 note 1) without explaining what these norms are and he does not adhere to his own rules' . As to the objection that Griffiths (and Acri)3 did not explain the spelling norms adhered to, both of us took for granted that scholars involved in the study of Old Javanesea language whose lexicon consists for no less than forty percent of Sanskrit loan-words, and whose literary forms are thoroughly rooted in the Sanskrit tradition-must be aware of the transliteration system sanctioned by the International Conference of Orientalists of 1894 (Senart & Plunkett 1894), commonly referred to nowadays as International Alphabet of Sanskrit Transliteration (iast). Moreover (as we ourselves have realised only after our publications referred to here), the iast system has been fine-tuned and further codified since 2001 in the form of a real international standard: 'iso 15919:2001 Information and documentation-Transliteration of Devanagari and related Indic scripts into Latin characters' .4 Although the published iso standard ignores Indic scripts from Indonesia, there is no clear reason why it should do so, and no reason not to apply it wherever possible to the transliteration of Indonesian documents in Indic writing. Anyhow, iast and iso 15919 are very similar, and it is almost exactly this system that was also used by Zoetmulder in his Old Javanese-English Dictionary, except that Zoetmulder uses the signs ĕ, ö, w and ŋ which are not provided for in iast or iso 15919.
Griffiths outlined his choices of Romanisation as follows (p. 133 n. 1): The transliteration used in this contribution adheres strictly to international norms for the transliteration of Indic script types. This means that I use v (not w) and that anusvāra/cecak is ṃ irrespective of its pronunciation. The only additions to the internationally standard repertoire of signs are the raised circle (°) which precedes 'independent vowels' (namely vowels which form a separate akṣara) and the median dot (·) which represents virāma/paten. Since some (sequences of) phonemes can be spelt in more than one way, there is occasionally need to work with a normalized transcription. Van der Meij's objection is a text-book example of the 'failure to make such clear-cut distinctions between the purposes of using a script' (Wellisch 1978, quoted above). His comment that Griffiths 'does not adhere to his own rules' is obscure, but this much is clear: he has ignored the difference that is being made between transliteration and transcription. Van der Meij could have gained a grasp of the linguistic, philological, and cultural-historical issues at stake here by going through Griffiths' (2010) review of Kozok et al. (2006), also published in the Bijdragen, where, after highlighting some positive features of the highly idiosyncratic Romanisation conventions adopted by Kozok et al. in their work on the Nītisārasamuccaya (an Old Malay text preserved in a codex from the Kerinci highlands of Sumatra), Griffiths concludes as follows (p. 136): All the other innovations, including the use of ŋ for the 'velar nasal' of the Indic alphabet (conventionally transliterated as ṅ), seem to me in needless contravention of existing international conventions. As such, they hinder the comparability of this codex with other documents pertaining to the Sanskrit cosmopolis and the multifarious vernacular literary traditions it has spawned from Afghanistan to Bali.

'I see no reasons why students of Indonesian literatures and manuscripts need to turn to India for inspiration for transliterations systems of Indonesian scripts.
[…] More importantly, I fear that these spelling changes herald a return to the deplorable situation where Indonesian cultural phenomena are not considered in their own rights but rather in those of a so-called "Sanskrit Cosmopolis"' . Van der Meij seems to forget that the majority of Old Javanese Romanisation systems, notably, as we have pointed out above, those implemented in Zoetmulder's Old Javanese-English Dictionary as well as in several philological works that appeared after its publication, are fundamentally inspired by the conventions, such as iast, developed in the course of the nineteenth century by scholars in Indian studies to transliterate a plethora of Indic scripts, starting with but not limited to the problem of how to transliterate Sanskrit. The deviations from this 'Indian' model, including those observed in the Old Javanese-English Dictionary (cf. pp. xiv-xv), are actually rather few and not tremendously significant: -One finds in Old Javanese studies no less than three renderings for the velar nasal grapheme (i.e. the fifth consonant sign of the Indic syllabary). Besides the sign ṅ sanctioned by iso 15919, one finds also ng and ŋ, while one or the other of these three signs will also be found to represent multipurpose nasal anusvāra (or cecak in Modern Javanese), which in iso 15919 should be rendered as ṁ.6 -The visarga sign rendered by ḥ in iso 15919 is often represented as h, just like the akṣara h. -The twenty-ninth consonant of the Indic syllabary representing the phoneme /w/ of Old Javanese is romanised with w or, less frequently, with the v sanctioned by iso 15919. -The vocalic r̥ (or ṛ) and l ̥ (or ḷ) tend to be transcribed, rather than transliterated, by their Old Javanese phonetic counterparts, viz. the clusters rĕ (r + schwa) and lĕ (l + schwa).
Our choice not to adopt any such 'deviations' from the internationally recognised norms of transliteration of Indic scripts in our transliteration scheme, and to adopt them only partially whenever we use a transcription, is motivated by the consideration that maximal adherence to international standards, as well as maximal correspondence between transliteration and transcription, maximally facilitates comparative research, without hindering research exclusively focused on Indonesia.
That the dominant systems of writing used in Indonesia before the age of colonialism-those known to contemporary Indonesians as 'ha-na-ca-ra-ka' or 'ka-ga-nga' systems-belong to the Indic family of writing systems is an undeniable paleographic fact, and the nearly seamless applicability to Indonesian documents of the iso standard for the Romanisation of Indic script is likewise undeniable.7 In other words, Indic writing in ancient Indonesia is only a branch of a single family of closely related writing systems applied to a multitude of Asian languages; the adaptations in use of the system that we see in different South and Southeast Asian regions do not alter its fundamental unity. This is why it is possible to transliterate a Sanskrit manuscript from the Indian Subcontinent, an Old Javanese manuscript whether from West Java, from the Merapi-6 As noted by Acri (2011b:xv, note 4), 'It is likely that the two graphemes represented in Old Javanese one and the same (velar nasal) phoneme, a fact that can be inferred from the 'reinforcement' of the ṃ into ṃṅ in intervocalic position, and also from the outcome of ṃ as ṅ in intervocalic position (e.g. saṃ hyaṃ bhaṭāra vs. saṃ hyaṅ ātmā)' . It may be noted that we ourselves so far used the sign ṃ preferred in iast instead of the transliteration ṁ prescribed by iso 15919. Henceforward we intend to make our own usage conform with the iso standard. 7 Phenomena that pose challenges to the use of iso 15919 for Indonesian textual material arise only relatively late in the history of writing in Indonesia, in the second millennium, notably in Old Sundanese manuscript tradition, but are never challenging enough, in our judgment, to justify refusal to use the iso standard.
Bijdragen tot de Taal-, Land-en Volkenkunde 170 (2014) 365-378 Merbabu scriptoria, or from Bali, and a Sanskrit/Khmer inscription from Cambodia according to the same-or in any event consistent-principles. In this situation, a choice to apply the iso standard to Old Javanese (and other ancient Indonesian languages) is at least one reasonable choice among others, if not actually, as we personally hold, the single most reasonable choice to make. However that may be, none of this has anything to do with a return to a situation where 'Indonesian' textual facts are considered through an exclusively 'Indian' lens (if such an imagined scholarly past has ever existed). As explained by Acri (2011b:xvi) in his Introduction to the Dharma Pātañjala: The adherence to the Indic (i.e. Sanskrit) system of transliteration implemented in this book does not aim at 'Sanskritizing' the Old Javanese but rather to rationalise and simplify the present situation by favouring a system that both Sanskritists and Old Javanists are acquainted with. The Sanskrit system has also the obvious advantages of being fully standardised and internationally established, and of being in use to transliterate a variety of Indic (languages and) scripts.8 Why Van der Meij regards as 'deplorable' the study of Indonesian cultural phenomena (we would rather say: premodern cultural phenomena of maritime Southeast Asia) in the light of the notion of a 'Sanskrit Cosmopolis' elaborated by Pollock (1996Pollock ( , 1998Pollock ( , 2006 is, frankly speaking, beyond our grasp. To us, Pollock's concept of a 'Sanskrit Cosmopolis' and the closely related idea (ignored by Van der Meij) of a 'Vernacular Millennium' that followed the cosmopolitan age constitute fundamental theoretical advancements that greatly further our understanding of the cultural dynamics at work in the premodern and early modern sphere of South and Southeast Asia. They have significantly con-8 In fact we are aware of other editors who have previously called for a reconsideration of the Old Javanese spelling system in a way that more closely conforms to the Sanskrit standard. One is Haryati Soebadio (1971:67), whose attempt remained a mere declaration of intents without materializing into real practice for, in spite of her claim to 'have chosen to transliterate the Old Javanese according to the Sanskrit system' in her edition of the Jñānasiddhānta, no real correspondence is found apart from the rendering of w as v. On the other hand, a completely consistent Sanskrit-inspired Romanisation (including the rendering of cecak as ṃ) was adopted by French scholar V. Sukanda-Tessier (1977) to render (Old) Sundanese and (Old/Modern) Javanese words throughout her book (even including the modern placenames, the titles of texts, etc.). In several publications involving textual material from Bali, C. Hooykaas (1964Hooykaas ( , 1966Hooykaas ( , 1973 has likewise chosen to adhere to international forms for the transliteration of Indic scripts, using v instead of w and ṅ instead of ŋ or ng. tributed to rendering obsolete the politically-charged notion of 'Greater India' , on the one hand, and the equally political principle that anything Indonesian should be studied from a purely local perspective. This last attitude, palpable in Van der Meij's statements, is that of the generation of scholars trained in the era of compartmentalisation of area studies that developed after decolonisation, and the rise of national(ist) historiography in the various countries of South and Southeast Asia. It seems to us that Van der Meij is running into the intellectual walls erected along the boundaries of contemporary nation-states, which stand in the way of recognizing the connected histories of different parts of South and Southeast Asia, and of studying the supralocal aspects of many cultural phenomena. These considerations, among others, were already presented in the Introduction to our volume (p. xiv). 'Now, of all things we see the word kakavin (Acri) where in Bali and Java the spelling has been kakawin for ages. There is no /v/ in Old Javanese, or in Balinese for that matter!' This statement reflects in a nutshell the reviewer's confusion about the matters under discussion. First, the 'spelling' kakawin has not been in use 'for ages' , as claimed by Van der Meij, but is merely one possible transcription/transliteration on an indigenous term, and it is that term itself, in its original spoken/written form, that has been in use for a considerable amount of time. The use of one Romanisation or the other is totally irrelevant. The reviewer apparently fails to grasp the fundamental distinction between different levels of analysis, namely transliteration, occurring at a graphemic level, and transcription, occurring at a phonemic/phonetic level. Neither Sanskrit nor Old Javanese knows a phonemic distinction between /v/ and /w/; neither in ancient India nor in ancient Indonesia does Indic script know more than a single corresponding grapheme, namely the twenty-ninth consonantal akṣara of the Indic syllabary. The phonetic realisation of the phoneme and its written representation is known to have differed from area to area, both within India and within Indonesia, and for purpose of a unitary transliteration it is not more arbitrary to favour v than to favour w.
An exhaustive discussion of this specific transliteration problem was already offered by Louis-Charles Damais decades ago (1958:10 n. 3 and 1970:19 n. 1), and his choice for w rather than v is immaterial to our argument, for it was based on different priorities than ours.9 The fervent attachment of Van der Meij and many other contemporary Indonesianists to the use of w rather than v is never Bijdragen tot de Taal-, Land-en Volkenkunde 170 (2014) 365-378 defended with anything comparable to the sophistication of Damais' discussion. It is quite evident that the distinction made in standard Latin orthographies of modern Indonesian languages between v and w, paired with the arising of a marginal phonemic distinction between /v/ and /w/ in Bahasa Indonesia, tends to confuse scholars about the phonemic and graphemic interpretation of written data of the pre-Latin past, when no such distinctions existed.10 This is all the more noticeable in the regrettably common case of scholars who have only a weak understanding of basic analytic concepts such as 'phoneme' and 'grapheme' .
Let us give one concrete example that is relevant to the present discussion. This is the problem inherent in editing Old Javanese texts that also contain Sanskrit verses, such as Parvas or Tuturs. We note that there has been a tendency to Romanise the two languages with different systems (concretely, using w for Old Javanese and v for Sanskrit). This convention conceals the fact that the script used in the manuscripts does not make any distinction as to the graphic form of the characters in question. Old Javanese and Sanskrit were written in a single script, and this is, in our view, a fact of great significance in cultural history that needs to be accommodated, not obfuscated, in our choice of Romanisation conventions.
'In my view, these spelling alterations are ludicrous and rather than being an expression of sound scholarship, degrade the book' . In view of the arguments we have advanced above in favour of conformity with existing scholarly standards, we cannot but consider the reviewer's choice of words ungrounded and disproportional. Given the current practices in the majority of other South and Southeast Asian philologies, where compliance with Indological norms tends to be considered a virtue, one might argue that the burden to provide arguments is on those who would want to differentiate Romanisation conventions used for Indonesian textual material from those in use for other South and Southeast Asian areas.
'It is to be hoped that these scholars continue their work but please, do not complicate matters unnecessarily. It is difficult enough' . Whether our adherence to the international standard for a thoroughly supralocal issue such as the standard transliteration of Indic scripts rather than any assumed phonemic reality behind its Indonesian pronunciation(s). On the other hand, we accept Damais' (1970:20) persuasive arguments against the convention to transliterate ĕ/ ö for short and long pepet, and follow him in rendering them as ə/ə̄instead. 10 See Lombard (1995: 10) on the 'graphic breaking off' in Southeast Asian countries such as Indonesia, which has 'led in a few generations to an alienation with considerable effects of which people speak too little' .
Romanisation of Indic script complicates things or conversely makes matters simpler, is obviously a question of perspective, but we really do not think there is anything difficult about it. The case of Romanisation of ancient Indonesian scripts in fact seems mercifully simple, by contrast for instance with such related cases as Burmese or Khmer.11 In conclusion, we do not think that the actual outcome of reflection about these matters-that is, the concrete Romanisation policy arrived at-is the fundamental issue at stake. Rather, it is the process of reflection itself and the grounding of Romanisation choices on reasonable, conscious arguments, whatever the outcome of the decision, that we consider crucial. We hope that our arguments advanced here will stimulate such reflection, and will help create a situation where scholars become more conscious of the issues involved in dealing with pre-modern Indonesian scripts and languages, and more mindful of the distinctions between orthography, transliteration and transcription. It is high time that the study of ancient Indonesia once again dares to be relevant to Asian studies at large, and for this it must learn to go beyond the simplistic paradigm of a radical opposition between Sanskrit and the 'Indian'/'Indic' , on the one hand, and the 'Indonesian'/'local' , on the other.