Standardization or Restandardization: the case for `Standard' Spoken Tamil

Harold F. Schiffman
Professor of Dravidian Linguistics and Culture
South Asian Regional Studies
Williams Hall Box 6305 University of Pennsylvania
Philadelphia, PA 19104-6305

The Tamil language has had its current standard written form since the thirteenth century, but due to an increasing diglossia, spoken Tamil dialects have now diverged so radically from earlier norms, including the written standard (LT, or Literary Tamil) that no spoken dialect (regional or social) can function as the koiné or lingua franca. Since LT is never used for authentic informal oral communication between live speakers, there has always been a need for some sort of spoken `standard' koiné for inter-dialect communication. Aside from interpersonal communication, one hears this inter-dialect koiné most clearly in the so-called ``social" film, which arose out of its antecedent, the popular or ``social" drama. Conversational portions of novels and short-stories also exhibit spoken forms, though not always as clearly `phonetic' as a phonetician might expect. The goal of this paper will be to examine the concept of `language standardization' as it has been applied to other languages, focussing on the role of literacy and writing on this process; then evidence for, as well as the sources of, koinéization of `Standard Spoken Tamil' will be presented; then we will determine whether the thesis that SST is an emergent standard is in fact sustainable, given the challenges of literacy and writing.

(Indexing: standardization, Tamil, diglossia, linguae francae, koinés) 1


The Tamil language has the second-longest history of standardization in South Asia, having been codified by Tolkappiyanar in the early centuries of the Common Era; it has changed radically over time and subsequent standard written forms have evolved, the most recent being the codification by the grammarian Pavanandi in the thirteenth century. Due to an increasing diglossia, spoken Tamil dialects have now diverged so radically from each other, and from the written standard (LT, or Literary Tamil) that there are problems of mutual intelligibility between many Tamil dialects.2 We will not even begin here to deal with the lack of intelligibility between Indian Tamil and the extremely divergent Sri Lanka Tamil dialects.3 Since LT is never used for informal oral communication between live speakers4, some sort of spoken koiné has filled this gap, and been in use for centuries, though it is not always clear retrospectively what the linguistic features of this koiné were or have been.5 It used to be the case that the Brahman dialect of Tamil was once the koiné used for inter-caste and inter-regional communication, but in this century this dialect has been replaced by another, non-Brahman dialect. The domain most clearly dominated by this koiné is the so-called ``social" film, which arose out of another domain, the popular or ``social" drama. Conversational portions of novels and short-stories also exhibit spoken language forms, though these are not always as clearly representative of a basilectal phonetic spoken Tamil of the non-Brahman koiné as a phonetician might expect.6 The goal of this paper will be to examine the concept of `language standardization' as it has been applied to other languages, examine the conditions under which a fairly uniform Tamil spoken koiné (SST) evolved, then present evidence for standardization features of SST, and then determine whether the thesis that SST is an emergent standard is in fact viable.

Review of the Literature: Standard Spoken Tamil.

For a review of the rather extensive literature on the question of the existence of a Standard Spoken Tamil, see Andronov (1962, 1975), Asher (1982), Bloch (1910), Britto (1986), Annamalai (n.d.), Shanmugam Pillai (1960), Vasantha Kumari (n.d.), and Schiffman (1999). Various studies by Zvelebil (1950-65) depict a range of dialect variation among the modern non-standard, essentially regional varieties of Tamil.

During a recent research visit in Singapore, the question was often put to me whether there is such a thing as ``Standard Spoken Tamil" (SST), and if so, what its main features might be, and how Singapore Spoken Tamil would compare with it. Since I have actually spent a significant part of my academic career investigating the question of SST, and I have evidence there in fact is such a thing, I have decided to address this issue directly in this paper. My belief that there is such as thing as SST is bolstered by my Singapore data, where, if anything, a wider use of SST koiné is evidenced than in general in India.7

SST, as I and others believe it to exist, is based on the everyday speech of educated non-Brahman Tamils, and its most obvious public domain, as noted above, is the Tamil movies of the so-called ``social" type, other modern ``social" stage dramas, some radio broadcasting (radio plays etc.) and to some extent in television.8 This language is not the same as regional or social varieties of Tamil (Trichy-Tanjore Non-Brahman, or Mysore Brahman, or Ramnad Adi-Dravida varieties, etc.). In many ways it is closer to literary Tamil, though nowhere near identical to it. It is not invariant, i.e. there is some variation in it, but natural languages are variable, and given the absolute lack of formal standardizing pressures put on SST (it is subject to no Academy, no school system, no literary society's strictures) it is remarkably uniform.9 It is spoken by educated people of various castes and regions to each other, and people learn it by listening to the dialogues of plays and films, working on communicating with each other in college hostels, and other places where educated people come together and try to communicate in Tamil. Of course it is used mostly for informal purposes, but one also hears it used by educated people for certain high-register purposes, where LT has no functional register: for example, Tamil linguists trained in western linguistics will often begin a discussion in English, and gradually code-switch into Tamil with English loans. I have heard entire university level lectures on Tamil syntax presented this way, with sentences like


``If we admit a morpheme boundary in this position, then your morphophonemic rules and your constituent structure will both be affected."

Many teachers, moreover, whether they teach in Literary Tamil or in English, use SST to paraphrase what they say, since students will not otherwise always understand them. In Singapore schools that I have visited, Tamil classes are regularly conducted with SST as the language of `explanation' (in fact, the medium of instruction!) though SST is never the object of instruction. The assumption is that students actually understand SST (which may not always be the case in Singapore), complicating the pedagogical problematicity of this issue. In mainland (Indian) Tamil schools, children do already speak some kind of spoken Tamil from their home environment, though it is rarely the `standard' koiné. How students `acquire' standardized ST is a question that has not been examined or reported on except anecdotally, since most teachers do not recognize the extent to which they actually use SST in school settings, and because there is no prevalent notion that some forms of it might be more acceptable than others. That is, teachers do not `correct' children for `erroneous' use of spoken Tamil since they possess no overt knowledge of what that might be. In fact children probably acquire a regional koiné in elementary school, and then acquire the kind of SST forms I am examining when they come to live in college hostels (dormitories) as young adults.

The role of literacy in standardization.

Another prickly issue facing us here is that of the role of writing in the development of and transmission of great traditions. For years the existence of orally-transmitted bodies of literature in the Indian subcontinent has vexed westerners (though not South Asians) because it appears that ancient Indic texts were codified and transmitted without overt evidence of writing having been involved. By `writing' I (and others) mean specifically marks made on paper, leather, clay, stone, wood, palmleaves, or other materials, using styli, pens, or other markers, such that a visible record, however perishable, is left. Western scholars such as Goody (1986) have taken the position that codification and transmission of such great works as the Vedas simply could not have been possible without writing, so evidence to the contrary, such as the elaborate and complicated systems of memorization observable in traditional Sanskritic colleges, is dismissed. I have discussed this issue in a recent work (Schiffman 1996:171-72) and must concur with Staal (1986:27) that the Goody hypothesis is contradicted by the Indic evidence.

What may be the problematical issue here is that Goody and others make a distinction between writing/literacy on the one hand, and orality on the other, whereas the real distinction may be between writing (marks on surfaces) on the one hand, and literacy (including oral literacy)10 on the other. Whatever we are able to conclude, in India what seems to be thought to be necessary for standardization, or invariant rule-observation to occur, is that it be codified, by which is meant that eventually the `grammar' is recorded, in rhyming sutras, and memorized. In modern times these grammars were also written down, and are now found in `books'. The notion that a language might be codified without having been committed to memory in rhymed sutras is not a prevalent one, or perhaps even an acceptable one, in modern India, but the fact that a grammar may not in fact be written (i.e. marked on surfaces) is an ancient and acceptable state of affairs. In the case of Tamil, for example, the idea is that the grammatical rules existed a priori, and were taught to the Vedic sage Agastya by Murukan, the son of the god Siva who then taught `divine Tamil' to his disciples (Schiffman 1996:175). This accords nicely with the modern linguistic notion that structure is in the language, and must be discovered by the linguist (though the idea that Murukan might have some new ideas about SST, and would want us all to rethink his earlier lessons, is not so likely.)

Definitions of Standardization.

When the question arises as to whether whether SST is standardized, we must have an idea of what constitutes a general definition of standard language, or failing that, what constitutes standardization in a particular language. We have evidence in many languages of both conscious, planned standardization (via language academies, dictionary- writers, printers and proofreaders) and of the somewhat haphazard choice of a particular dialect of some city or ruler (Madrid, Paris) and standardization via use in official texts (the Bible, the Koran etc.) followed perhaps by royal fiat (e.g., the Ordonnance de Villers-Cotterêt, promulgated by the French king François I in 1539, which established that only the French language should be used for record-keeping throughout France.)

Since the crux of the question is whether iron-clad definitions of standardization exist, and whether SST meets those conditions, i.e. is standardized or perhaps on the way to being standardized (some writers refer to emergent standards) we need to review some working definitions of what standardization might entail. The best recent review of this issue is probably Joseph 1987, though his treatments focus either on well-known western languages (English, French) or on non-literary languages such as Inupiaq/Inuit. The kinds of problems that face non-western but long-standardized languages such as those of India or other parts of Asia have not been the focus of Joseph's work, nor in fact of most work on the subject. Acutely diglossic languages such as Tamil and Arabic constitute perhaps a completely different kind of case, one in which restandardization (Joseph 1987:174) seems to be what is happening, i.e. a newer version of the language, with its own spoken form, is emerging to challenge and attempt to capture some of the domains of an older, highly prestigious literary language that has ceased to be a vehicle of oral communication. As Joseph points out, however, restandardization will never totally replace the older standard language (LT); it (the older norm) will simply be elevated to a `classical' status that it will continue to inhabit, but no-one will try to emulate it except for a few archaizing die-hard purists, or, in the case of liturgical languages, priests and pundits.11

Some useful early work on the issue of standardization emerged from the Prague School of linguistics, and has been summarized by Garvin (1964); his key concepts are urbanization, flexible stability, and intellectualization. Many definitions of standardization (codification, etc.) involve official choices being made about the corpus of a language, but as we have mentioned and will see in detail below, we are talking about the development of a spoken standard, which may involve other kinds of decision-making.12 Arabic, also acutely diglossic, is faced with a need to develop a koiné dialect that would be usable throughout the Arab world but would be closer to spoken dialects than classical Arabic. The result so far is the emergence of so-called MSA (Modern Standard Arabic), the features of which can be readily described (it is even what is taught to foreigners) but is still in the process of evolution (Mitchell 1985, 1986, Walters 1996). Another study puts it like this:

A new system of Arabic language varieties is developing which includes the emergence of a new international koiné which is rapidly overtaking classical language prescriptivism and which is compatible with emerging national or subregional dialects of what will remain one Arabic. Within each nationally controlled educational system, the massive growth in educational participation by people from all walks of life, and the penetration of mass media of multiple linguistic origins into all homes, together draw on an inevitable medley of vernacular and grammatical sources, from the highly deliberate to the necessarily unconscious, to bring about stylistic differentiation of Arabic to suit today's communicative needs. A strong force in this restructuring of the Arabic language system is cross-communal, fed by intense exchange of people, goods, messages, and ideas--and simply intent--between all Arabic language communities, toward a higher degree of mutual accommodation (Jernudd and Ibrahim, 1986:6).

The main problem with the comparisons with MSA, and indeed the whole dynamic of its evolution, is that it is not used as a spoken language by native speakers of Arabic, though they do write it and use some variety of it in, e.g. schools. Foreigners who learn it (e.g. in university settings) subsequently have to learn a local spoken variety such as Cairene or Lebanese colloquial.

Status planning and corpus planning.

Though a distinction is often made between status planning and corpus planning13, in fact corpus planning may also be viewed as a collection of decisions about the status of individual elements of the corpus of the language: this pronunciation is preferred over that; this spelling is correct and that is not; this plural-marker or past-tense form is preferred over that; this syntactic construction is ``valorized" and that is ``stigmatized". When all these status decisions14 have been made, the corpus has been ``standardized". It may then be disseminated through printing (the Bible, the Quran), through its use in royal or other administrative edicts (Charlemagne's grandsons' Strasbourg Oaths, the edicts of Asoka) or nowadays, as the form of language taught in schools (Malaysia, Norway). The set of decisions may sometimes be summarized in the form of a (prescriptive) grammar.15 As Garvin (1964) points out, decisions about standardization may get made, and perhaps even published, but dissemination of the results may fail; i.e. the standard may fail to be implemented, and implementation may in fact be the Achilles heel of most language planning. Garvin's requirement that there be flexible stability means that there should be some stability, usually through printing of a dictionary, spelling book, or reference grammar. But it must also be flexible, allowing for eventual revisions, addition of vocabulary, and adaptation to more modern technology. Garvin also posited four functions of a standard language:

The unifying, i.e. the ability to unite several dialect areas into a single standard-language community; this function is largely symbolic, since it renders subjective value to notions of what kind of linguistic community the speakers inhabit. For Tamil, this is in general not a problem, since SST is accepted and spoken widely in Malaysia and Singapore as well as in Tamil Nadu.16
The separatist, whose function is to set the group off from others, and establish boundaries; again, this is largely symbolic. Since Tamil is already recognized as a separate literary language, separate from its neighbors Telugu, Kannada and Malayalam, the problem of deciding what is or is not Tamil does not arise.

The prestige function, i.e. the prestige of possessing a standard language; this function is also largely symbolic, i.e. it has abstract meaning for the speakers, i.e. gives them pride in the ability of their language to `hold its own' among other written languages.

Tamil already has a prestigeous literary language; this is thus not an issue here; rather, capturing some of the prestige for the spoken language is a problem.

The frame-of-reference function, which Garvin refers to as providing objectivity; this is the ability to serve as a frame of reference or an objective standard for correctness and for the perception and evaluation of poetic speech.

As far as this affects Spoken Tamil, one needs an objective standard for what would or not be considered `correct', but it is not necessary for poetry, since the older norms dominate the domain of poetry.

As is obvious, some of these apply to the development of SST and some do not; since Tamil already has a written standard (LT) some of these do not apply and will not unless SST captures domains currently dominated by LT. It must also be noted that LT is not a unitary norm; there are many varieties of LT, some extremely conservative or archaicizing, but since Tamil culture conceives of the language as being only one (rather than multiple stages or varieties) taking refuge in the archaic style is often the strongest defense of the recalcitrant resisters to modernization: they can so easily demonstrate how modern spoken forms are totally inappropriate for something like religious usage.17

Other definitions of standardization.

It may be useful to review some other attempts to define language standardization; as I have tried to indicate, much of the debate on this issue has to do more with English or other western languages and may not be germane for Tamil or Arabic.

Pakir 1994: Discusses what she calls ``unplanned language planning", or ``invisible" language policies.

Kachru (1985): Proposes four types of codification:
Authoritative or mandated (by academies etc.)
Sociological or attitudinal codification: social and attitudinal preference for certain varieties, accents.
Educational codification: dictionaries, media, teacher training, standardization of textbooks, school grammars, etc.
Psychological codification: constraints on, e.g. Sanskrit.

Milroy and Milroy (1985): ``In the strictest sense, no spoken language can ever be fully standardized." Writing and spelling are easily standardized; spoken standardization is an ``ideology", an idea, not a reality. If languages were not standardized, they would break up into regional spoken dialects and end in mutual unintelligibility.

Haugen (1972): proposes that linguistic cultures are `intolerant' of optional variability in language. There must be selection, diffusion, maintenance, elaboration of function.

Joseph: 1987: Joseph's work is mainly useful for showing how one highly standardized language, French, managed to rise from the position of an L variety to that of an H variety, and to displace Latin, the previous H variety. In doing so, it had to prove that French had classical features as valid as those of Latin and Greek, and once French had done this, other European languages were able to follow suit and expel Latin from H-status, e.g. in University education.


The Milroy and Milroy hypothesis (i.e. that there is a Standard Language Ideology (SLI)) seems to be predicated on the notion that all languages are in the same kind of sociolinguistic situation, and go through the same kinds of stages of standardization. This is surely an unexamined and unprovable hypothesis, but serves the ideology that standardization not only cannot be shown to exist, (i.e. standardization is a figment of someone's imagination, a mere social construct) but that the ideology that fosters standardization is hegemonistic, imperialistic, and hurtful. Not much evidence is given for the universal application of these two claims.18

Since the SLI is an unproven hypothesis, we may treat it as itself an ideology, the SLI Ideology. It views standardization as hegemonistic just because English is a language spoken beyond its borders, and because exonormic19 standards of English pronunciation etc. are demanded of speakers who will never be able to meet the demands of the norm, mostly because the evaluators will constantly (and unfairly) shift the criteria to make attainment impossible. But there are differences between standardization of a language like Tamil and languages of wider communication like English. For one thing, Tamil is not a LOWC (Language of Wider Communication), so the notion of hegemony over other languages does not arise. Secondly, Tamil already has a standard literary language; with the focus on standardizing the spoken language, different issues come to the fore.

Tamil already has a strictly codified written norm (Literary Tamil), used and accepted by all Tamils (in Tamil Nadu, Sri Lanka, Malaysia, and Singapore), and it is LT that is the exclusive and excluding language. Mastery of its correct forms is difficult, and illiteracy is high. Ability to speak SST is less difficult to acquire than ability to master LT. If there is any notion of hegemony or inequality it applies to LT.
Tamil is not a language used widely beyond the membership of its mother tongue community; there are very few non-native speakers, and the question of even teaching the spoken language to non-native speakers did not arise until fairly recently.

Because of Tamil's extreme diglossia, only LT was widely understood, but only by educated or semi-educated speakers. Its spoken dialects (regional, social) vary widely; some dialects (e.g. Sri Lanka) are mutually unintelligible with Tamil Nadu dialects.

LT as a panlectal standard had become archaic and problematical. The goal was to find a hyperlect or edulect acceptable to all, not marked by region or caste features. SST is the result.

Decision-making Bodies.

Decisions about language standardization may be made by a body, or perhaps even by an individual (Panini, Tolkaappiyanaar, Martin Luther, Ben Yehuda); if it is a body, it may have as its immediate task, not the codification of the language, but the officialization or standardization of some text, e.g. the English Bible of 1611. Perhaps the work of one or more writers becomes the model for what is acceptable and what is not (Shakespeare, Goethe, Cervantes, Pushkin, Tagore). Large bodies, however, have more trouble coming to a decision than would small bodies or an individual; the decision-making process is simply too complex for any large group to be able to do effectively. Therefore the Academy model, though perhaps politically necessary, is in actual practice very ineffective; it must delegate decision-making to subcommittees, and once the body is established, becomes a force for conservativism, blocking even then most trivial reforms.20 Purism or some other cultural agenda may hold sway, with passionate denunciations of the most innocent suggestions or proposals .21 Sometimes academies, though given the ultimate authority, keep their finger on the pulse of the linguistic community, and ask for suggestions and/or non-binding approval of any changes they may suggest, from their users. On the other hand, hyper-democracy in the language standardization process is usually counter-productive, and may lead to the troubles experienced by Norway, where floor-fights in the national legislature over trivia such as the gender of nouns were once common.

Informal Consensus.

Another possible model for language standardization is an informal consensus model, where a small but influential body of people (poets, intellectuals, writers) come to agree on the choice of a norm without any formal decision-making whatsoever. This kind of linguistic decision-making is less well-understood, because it is only noticed retrospectively, after it happens. The participants may not be conscious of what they are doing, but if we follow the accommodation theory of Giles et al. (1991), we can see this as a kind of accommodation going on--people are making adjustments in their habits and tailoring their linguistic production to their perceptions of what their hearers/interlocutors want to hear. This kind of standardization is more likely to be what happens in the choice of spoken norms, than in the choice of written norms. It occurs to me that it is what is responsible for the choice of the spoken norm known as RP, the British Received Pronunciation, (also known as RSE, or Received Standard English) since that apparently emerged in the British Public (i.e. private) schools in the 18th and 19th centuries as those schools came to prominence.22 Generations of British leaders were trained in those schools, and there was remarkable consensus about what the RP norm was like; yet no-one had to issue edicts or officially declare any standards about pronunciation. There was already an agreed-upon grammatical and syntactic system for standard English, but how this was pronounced was not, in the early days, explicitly standardized. Gradually, RP became to some extent a standardized pronunciation, though many experts now disagree about how extensive this was.

Similarly, in America a grammatical/syntactic system of English quite similar to that used in Britain continued to be agreed upon after the American revolution (probably because of the ``standardization" of the English Bible) with spellings influenced by Webster's dictionary, and disseminated by McGuffy's Readers. By the end of the 19th century a pronunciation norm for public speaking (preaching, oratory) held sway, based on the speech of products of prestigious eastern seminaries and colleges (Harvard, Princeton, Yale) and the New York stage; and, as far as the pronunciation of final r's is concerned, was quite similar to the southern British norm. In the early twentieth century, however, this norm gave way, and sometime between the two World Wars another model emerged, this time a rhotic (r-pronouncing) one. This model was without much doubt disseminated by radio, and within a generation, was also the norm in talking movies and television. It is known as `broadcast standard' and its best representation is the speech of news presenters on national networks, especially when reading from texts (i.e. not speaking extemporaneously.)23 Commercial radio broadcasting in the US never set any standards for its announcers; there was no central ownership, no state-owned broadcasting system; there was never a school, a rule-book, nor a pronunciation guide (unlike there is for the announcers of the BBC, the CBC, the NHK etc.) The Broadcast Standard, because of its rhoticism, sounds more like `mid-western' styles of speech, though this seems so more to eastern-seaboard speakers than to midwesterners. It is probably closest to the educated speech of Americans from large northeastern cities other than New York, Boston and other non-rhotic areas. Together with the evolution of RP, American Broadcast Standard evolved without conscious control, yet both display remarkable uniformity.24

I claim that Standard Spoken Tamil (SST) also emerged via an informal decision-making process, similar to the way British RP and American Broadcast Standard evolved, but included in it was decision-making about the grammar and syntax as well, not just the pronunciation. After a certain consensus was reached on the broad features of SST, it could become the natural choice for use in the ``social" film, and was thus disseminated widely to all Tamil speakers everywhere, serving both as a model of ``correct" speech (spoken by the central characters, the hero and heroine) while character actors cast as buffoons and rustics provided models of ``incorrect" speech; the ``Jerry Lewis" character Nagesh was famous for this in the Tamil film; other linguistic cultures have their own equivalents.

These days it is fashionable in many circles in the west to deny both the existence and the legitimacy of standard English or other standard languages, because standards have often been used capriciously, and maliciously, to deny non-standard speakers access to power etc. Therefore we now hear and see a great deal about hegemony, power imbalance, linguistic prejudice, maintenance/denial of privilege, empowerment, and many other descriptions of ideological control of language. And indeed, much wider tolerance is now permitted in how standard Englishes, whether American, British or other varieties, are pronounced, although there seems to be less tolerance in news broadcasting, for example, for non- standard grammatical forms such as negative concord (known popularly as `double negatives'). And of course in broadcasting, different levels are recognized for news readers, sports announcers, talk-show hosts, cartoon characters, and other informal usage.

As anyone who has ever had to teach a language knows, however, choices have to be made as to which forms to teach; pedagogically it is simply unworkable to accept any and all utterances students produce, so teachers, especially language teachers, find it essential to adhere more or less strictly to one set of forms rather than allow variation in students' writing and speech. 25

New ways probably need to be devised to broaden the concept of standardization, to allow for variation, perhaps in register and domain, without giving up the whole notion of having a form of language of widest communication, or the utility of some kinds of agreed-upon understandings. Too often, standard grammars are in fact norms for written language, but this gets forgotten when spoken language is taught, as it is today. 26 Computerization alone will demand various things; just try your spell checker (which also checks your grammar) and see if you agree with the kinds of decisions it makes about your usage. The fact is that when all is said and done, speakers of all natural languages make judgments about different kinds of speech and writing that they hear and see samples of, and some of those judgments are, like it or not, hierarchical social judgments. There seem to exist understandings, a whole network of understandings of what is appropriate speech/writing, and what is not. Another way of putting it is that there can exist forms of speech and writing which evoke no particularism; they do not remind us of any region or social class, and they do not immediately mark their user as a member of any particular class, caste, or ethnic group (other than the class of educated speakers). They convey content without calling attention to the form. Understandings exist as to which form does this `best'; understandings can, of course, also break down. In order to get a grasp on whether my own students have any consensus of what a non-particular form (``standard") might be, I have undertaken informal surveys about their linguistic preferences. I find that students who attack the notion of standard English do so mostly for their own convenience, not for the supposed benefit of subaltern non-standard speakers of the English language. They wish to be able to speak and write any way they please, but conversely they also wish to receive written and spoken English communication in a standard form, as I have ascertained by testing their tolerance for messages (e.g. telephone information messages, pharmaceutical labels on medication, airline emergency evacuation announcements, etc.) delivered in non-standard forms.27

In any event, the issue of standardization has become highly politicized in this day and age, perhaps more so than in some other periods, but perhaps not. The fact that in practically no society do people actually use language according to the rules that have been devised, rules which often date from a previous era, is given as evidence on the one hand for either:

Total decay and ruination of the language, leading to illogical thinking, moral turpitude, and the decline of civilization; or

Evidence that there really is no such thing as a standard language, the notion being kept alive only to benefit an elite ruling class, a small coterie of mandarins, or whatever, who in any event behind closed doors don't use the standard language eith er.

In the Tamil context, both of these arguments are used; the first to validate the notion that the spoken varieties of Tamil are corrupt, decadent, and worthless, and the second (partially) to challenge the idea that there might be an alternative to the rigid literary standard. The essential thing to consider about standard language is that all of the above can be true, and that there is still something useful to be said for it. Standards do change; words, phrases, spellings that were highly stigmatized when I was a child have now become commonplace, and phrases I never thought I would utter I now hear coming from my own son's mouth. People now regularly split their infinitives, dangle their participles, and end sentences with a preposition. Most speakers of American English now use the form `you guys' as the plural of `you', a form that was highly stigmatized when I was a child, and which I must remind my son not to use with his grandparents. We must remember, however, to distinguish between style and grammar; much of what is considered ungrammatical is actually different in style, since grammar by definition is the structure inherent in a language. If people use their language and are understood every day without miscommunicating, they are speaking grammatically.

Some three or four decades ago an American cigarette commercial used the phrase ``Winston tastes good, like a cigarette should." English teachers were up in arms about this ``error": one was supposed to say " a cigarette should," etc. Later the Winston people capitalized on the furor by airing another commercial: ``What do you want, good grammar or good taste?" Some people may not like the use of `like' for `as' but it is hard to call this a grammatical error. 27 In other words, what may have once been considered ungrammatical may in a latter day have to be called a stylistic difference. And no matter whether the word `standard' has become the whipping-boy of post-modernist culture-critics, there are nevertheless `understandings' that people in various societies have about what kind of language is acceptable, and what is unacceptable, and what different kinds of language are for.

What teachers need to have is a framework to adhere to, so that they can be fair in their determination of what is acceptable and what is not; otherwise grading, promotion, everything they do will be capricious. But they also need to know the difference between style, register, and grammar, and be able to teach it. They need to distinguish between formal styles of a language and informal varieties, between expository writing and creative writing, and be able to convey this to their students.28

Standard Spoken Tamil: What is standard and what is not?

In the situation as it applies to Tamil, similar constraints apply. I can state that in most Tamil dialects as well as in SST there is very little variation in past tense formation of verbs, for example. Most verbs form past tenses as in LT, but in verb stems that end in a final -i sound (e.g. teri- know, utai `break') the past tense markers nt, tt that are typical of verb classes II, VI, and VII (Fabricius' and Dr. Graul's classification (Fabricius 197229) undergo palatalization to nj, cc in spoken Tamil. This is consistent in all dialects that I am familiar with. It is therefore possible to state that this is a standard feature of SST, even though it is not the same process as is found in LT. There is evidence of palatalization having begun earlier, in LT, and therefore incorporated into the orthography.30

Another feature of SST that is quite regular, though different from LT is the use of what used to be considered to be an `emphatic' marker, the clitic form -ee as in naanee vandeen `I (emphatic) came' which contrasts with emphatic taan e.g. naan-daan vandeen `I alone came; only I came.' This `emphatic' marker is semantically complex and difficult to describe31 but it can be used expressively in many ways in both LT and SST. One way that is new, and is in fact a semantic change, is its use as a redundant marker of location. That is to say, wherever LT has forms that indicate location, such as the locative case il, the `deictic' adverbs inku, anku, enku, the points of the compass, postpositions such as meel `on, above', kiir `below, down, under', SST has these forms plus -ee, i.e. viitt(u)lee, in the house', ingee, angee, engee `here, there, where' meekke, tekke, vadakke, kirakke `west, south, north, east', meelee `on', etc. That this ee cannot be analyzed as simply an emphatic marker is shown by the fact that when emphasis is required, emphatic ee is added to forms already marked with ee: viittukkulee-yee `right in the house', angeeyee `right there', etc.

In fact I would argue that this ee is perhaps not semantically new, but maybe in fact old; in Old Kannada, to take the example of another Dravidian language, ee often functions as a locative marker. Whatever the case may be, the addition of ee to locatives and semantically-locative phrases in SST is quite regular, and moreover, semantically different from LT in this regard. (Here we run into another problematical area, that of instances where the grammar, syntax and/or semantics of SST differs from LT. To LT purists, there can be no such thing as SST having a different grammar or different syntactic rules, because this might lead to the notion that such a different system were somehow legitimate.)

There are many other examples of systematic regularities in SST that could be adduced; 32 the point of this paper is not to enumerate them all, but rather to try to show what kind of a system or systems SST displays. We will often find one-to-one correspondences, that is where LT has x, SST has y; but we will also find cases where something found in LT is not found in SST (such as plural marking in neuters, like avai `those things')33 or the use of the aspectual verb vaa to indicate a historical or narrative past. On the other hand, we will also find examples of constructions that SST has that are not found in LT, or do not have direct equivalents. Such constructions as avan solraaple `as he says' , which seems to be derived from some form of a verb plus poola `like, as' has to go back to LT avan solrapatiyee `ibid.' but how forms like solraaple, irukkraaple, vandaaple, en kaale vaaraaple irukku arise has not been explained. SST has kitte as a postposition meaning `near, on the person of' e.g. avan-kitte (often reduced to avan-tte by deletion of ki, a regular process) but for this in LT we can only use avanitam.

Those who would require a standardized language to exhibit no variation whatsoever will quickly point out that there are areas of great variation in SST, and these can also be described quite accurately. One is in the use of kinship terms, which vary tremendously from caste to caste. In order to avoid caste-marked forms, Tamil speakers often have to resort to English, or to euphemisms, e.g. for `wife' there is no caste-neutral form, so we may hear people saying oyfu (<wife) or viittule `in the house' for `wife'. The same goes for many other terms, which are also used as address forms, e.g. annaacci `elder brother' may be used as an address form to give mild respect to a younger man, but it is not the `standard' form for `elder brother', which I would say is annaa or annan. The former is originally a `vocative' form of the latter, but many address forms (vocative) have become terms of reference also, e.g. ammaa has replaced taay for `mother', maamaa has replaced maaman as `uncle' (mother's brother). This form has now passed into Singapore English to refer to the kind of general store run by an person of Indian descent, selling newspapers and other sundries, i.e. the maamaa kade.

Other areas of variation are found in the following:

Plural marking. Though plural-marking is not obligatory for nouns in SST, pronouns must be, and animate nouns tend to be, and some other nouns may also be optionally marked for plurality. The LT plural marker kal is rarely if at all realized in SST, instead intervocalic k is softened to phonetic [h] to begin with and is often deleted, especially in extended syllables (e.g. pookalaam may be shortened to [poo'laam] `let's go'). Beyond that, there seems to be a perception in some dialects that the most common plural marker is - ngal rather than kal, as in maram marankal, pronounced [maranga] so this -nga is extended to other nouns as well. This is reinforced by the fact that -nga is also (perceived to be) the plural of some pronouns, e.g. nii,/niinga; avan /avanga, naama/naanga. Therefore the commonest plural form we now see in SST is nga, e.g. korande `child' korandenga `children', tambi `younger brother', tambinga `younger brothers'. But note that the underlying form of nga should actually be nga(l) because when any other morpheme follows, e.g. the question marker -aa the retroflex lateral appears: korandengalaa? `children?' niingalaa? `you?' So we would want to retain a slightly more abstract form of this than the phonetic spoken form nga since it would simplify our grammar to do so. There are also some `irregular' plural forms like pasanga the plural of payyan `boy' that are difficult to explain according to the rules of LT, but nevertheless are just there. I believe the [s] of this form to be an old alternation with [y], going back to Proto-South Dravidian; i.e it is not just a modern `corruption'. In fact, many forms found in ST are old, but have existed in the nether regions of the language, never sanctioned by the grammars, but still ancient forms. Some other dialects use the ka(l) form as their plural marker, which comes out as nga(l) after words that end in nasals, and ha in other positions, e.g. korandeha `children'. But since intervocalic k (realized as [h]) is usually deleted in SST (e.g. in pookalaam becoming [poo'laam] `let's go', and all present tense markers of weak verbs have the ki deleted) this is not very stable as a plural marker. Some dialects are known to use English plurals in some (foreign) words, e.g. muslims `Muslims' or even both English s and a Tamil plural: muslims-nga. This would depend on whether the word was native or borrowed. Other speakers tend to substitute the quantifier ellaam `all' for a plural marker: andak kanakk-ellaam `(all) those bills' books-ellaam `(all) the books'. There is a tendency for some reduplication to be used where English would have plural marking: nyaayittukerame nyaayittukerame varraanga `they come on Sundays' but this may be more to express repetitive or distributive notions. But the fact is, perhaps because of the collapse of some distinctions in the pronominal system (pronouns like niir `you sg. polite' is now archaic) nga(l) has emerged as the strongest plural marker, much as English s emerged as one among many plurals (oxen, children, kine, geese) and has become the most productive marker in modern English. But because of the optionality of plural marking in nouns, and because plural marking in animate nouns (obligatory) often involves kinship terms (variation!), SST has not been able to settle on this totally as yet. We would need to survey speakers' ideas of what they consider preferable, and why they would prefer one form over another; if we did, we would often find that justifications based on LT would be given, since LT is perceived as the `real' language, and SST is not. Thus, ideas about the proper plural morphology would be derived from LT, which is of course a different system with different rules about plural formation.

Past Neuters. Another area of variation is that of the past neuter forms of verbs. In LT this is simply -adu as in vandadu `it came' for all verbs except poo `go' and aaku `become' which have the forms pooyirru `it went' and aayirru `it became'. There are also some verbs, those classified as III in the Fabricius/Graul scheme, which optionally may have this last ending, e.g. tuunkirru `it slept', but may also have the `regular' ending tuunkinadu or tuunkiyadu `ibid.' In many forms of SST, however, especially those used in the Trichy-Tanjore area, the rru form, which palatalizes in SST to [ccu] (or [cci]) has spread from its restricted use with poo and aahu and as an optional marker of class III verbs, and is now used as a past neuter form of verbs with all forms: not just pooccu and aaccu, found in all dialects, but vanduccu, aruducci, saappiducci, and so on. This tendency is too strong to be excluded from notice; usually in SST when there is a proliferation of forms the `standard' choice has been to pick those that are closest to LT, but in this case the change cannot be ignored. Here we must admit variation. It may receive some stigmatization, but I have not tested this to find out. 34

Aspectual Verbs. The area of Tamil syntax known as aspectual verbs vinai nookku is an area where SST has changed, expanding the aspectual system of LT in its inventory, in the level of grammaticalization of the system, and in the pragmatic use of the system. The system thus varies in a number of ways:

Dialectally: Different spoken dialects use different `main' verbs as aspectual verbs; the set is finite but somewhat open-ended;

Pragmatically: The way aspectual verbs are used, and for what purposes, i.e. what their illocutionary force might be, varies tremendously; the fact that aspectual verbs are not usually negated when declarative, but may be when imperative, or are more likely to be used in declarative sentences than in interrogative sentences, all make it difficult to write grammatical or syntactic rules for this part of the language.

Grammatically: this is the question of to what extent aspect is now an internal morphological device (probably derivational rather than inflectional), and no longer a purely syntactic phenomenon. Evidence for variable grammaticalization of aspect exists, but phonological variation indicates it is not complete.

The aspectual system is therefore a very difficult system to describe and to master, especially for non- native-speakers, but it is one of the more interesting and creative parts of the language, and is vastly more complicated than in LT, partly because LT is not used pragmatically for communication, or for negotiating meanings.

Stigmatized forms. It may come as a surprise to some speakers that SST, which is thought of as a variety without prestige, with no rules and regulations, may not admit certain forms because it considers them too low. That is, there is also agreement in SST as to what is not allowed, or at least are `going a little too far'.

One of these is the tendency to round vowels when preceded by a labial consonant and followed (usually) by a retroflex consonant, as in poTTi (<peTTi) . What is interesting is that some forms with [o] are acceptable, e.g. poNNu (from LT peN) but some others are not, e.g. (v)uuDu `house' is not acceptable to some but (v)uuTTule `in the house' is acceptable. Studies need to be done of this phenomenon to determine how acceptable the rounding actually is.
Another tendency is the lenition of sequences of nasal plus consonant; in particular the sequence NT in some words but not in others. LT veeNTum regularly comes out as veeNum but the negative form veeNDaam is not regularly realized as veeNaam except when a brusque reply is required. That is, veeNDaam `means' '(I) don't want (something)' but veeNaam is more casual or more emphatic; `I don't want any! Get lost!' This lenition of NT also occurs in some other places, e.g. koNDaandu (from koNDuvandu `having brought') may be laxed to koNNaandu as in ellaatteyum koNNaandiTTrundeen `I brought all the stuff.' Interestingly, this lenition has long since been completed in other consonant-plus -nasal sequences in SST: [mp] can be laxed to [mm] or [m] as in kaaNpikkoLLaveeNTum `I want to show' which comes out in SST as [kaamikkaNum ]. The alveolar sequence nr has long since become [NN] in Non-Brahman dialects (and [nn] in Br. dialects). The tendency is for this all to happen after long vowels, but not always. This process is complete in Tamil's sister language, Malayalam, but has not been discussed in print regarding Tamil. The same is true for sequences of nk --> "eng-eng", where [g] becomes a nasal.

Case and Postpositions. A third area where there is great variation in SST is that of case-markers and postpositions. Though there is overall agreement as to what the traditional case system (the seven-cases plus vocative system of Tolkaappiyam) is now like, and what the forms are, it is not possible to hold to this eight-case system in SST (it probably isn't in LT, either, and was problematical even for Tolkaappiyanaar) where we must admit that the boundaries between what is a case marker (e.g. iliruntu though clearly made up of two pieces, il `locative' and iruntu `past participle of iru `be'' seems to be a case because Sanskrit had an ablative.) But there are many other situations .) But there are many other instances where postpositions can substitute for case markers (viiTTu-meele means something different from viiTTukku meele), or we get a case marker followed by a case-marked postposition (e.g. viiTT-ukku pakkattu-le `near the house'). These kinds of variation are not a problem for SST per se, but they are perceived as a problem by those who see grammar as invariant and/or as governed by rules set down in the 13th century. In fact the seven-case system, as already mentioned, was problematical for Tolkaappiyanaar, who had to admit two markers in the instrumental, aal as instrumental, and uDan and ooTu, the latter being `sociative' because there was no place for them both in a seven-case system. Had Tolkaappiyanaar not bought the Sanskrit seven-case system lock stock and barrel, but opened up the Tamil system to let it fit the Tamil facts, we could now allow other facts of Tamil to be accounted for, and the case-postposition system is one of these. I note that the new syllabus for Tamil in Singapore schools, to be used from next year, no longer talks about seven (or eight) veerrumai, but treats each case marker as a kind of veerrumai. This is a welcome change; the system that has been used is unworkable, mainly because it has been kept in the seven case box for two long.

What decision-making process has been involved in the evolution of SST?

I have mentioned that I believe SST to be already highly uniform, and that this uniformity has somehow been involved with the spread of mass-media forms that use it, such as radio and film. Impressionistic accounts35 attribute the development of this inter-caste, interregional form as taking place in college hostels when young educated people from all over TN come together and must negotiate some form of communication. The inter-caste inter-regional form used to be the Brahman dialect, but this is no longer the case; now even Brahmans use NBr. Tamil, and clearly SST has evolved out of this panlectal NBR dialect soup.

I have also used terms like ``SST does not allow form x" or ``when in doubt, SST prefers forms close to LT" and with such locutions I have been speaking as if SST were a person or a decision-making body. In fact the decision that went into the choice of this form or that form are covert, i.e. they are not available for observation, but anecdotal reports from speakers who have learned their SST in college hostels confirms that a kind of decision-making process goes on. Certain forms are stigmatized, e.g. Brahman forms, so Brahman speakers quickly learn not to use their home dialect, if they have not already figured this out.36 Other speakers may bring regional or caste forms to the process, only to have them stigmatized through ridicule and other forms of overt comment; they quickly learn to not use these forms again. If this business sounds familiar, it is probably because a similar process seems to have evolved in English public schools in the 18th and 19th century, whence the ``standard" Received Pronunciation (RP) evolved.

The interesting thing in this decision-making about what is an acceptable SST form and what is not, is that it is not governed by rules set by an Academy, by lexicographers, by eminent writers, or any of the other elite language control boards found in many societies, e.g. the French Academy, the Duden Gesellschaft (for German), etc. Yet college students are an elite, and they have in common that they are or were (I do not have a date for the evolution of this NBr. SST) educated, either in English or in Tamil37 Originally then the body of people who made the decisions were most likely to have been male, of higher non-Brahman castes, and from families wealthy enough to afford higher education of the western type. This is of course, not at all unlike the situation applying in the British RP model.

In the mid-twentieth century, it is without question that the chief disseminator of this SST has been the modern Tamil ``social" film. There is remarkable uniformity of SST irrespective of whether the studios were DMK-dominated or Congress-dominated, i.e. MGR films vs. Sivaji Ganesan films, to take only two examples. Despite the DMK's special ideas about Tamil, their films used SST that varied hardly at all from the kind found in other studios' films, except when the hero feels the overwhelming urge to expatiate in the special DMK-preferred alliterative style. This variety is also found in the stage dramas of the social variety that in fact have a symbiotic relationship with the Tamil film industry, and is also used in radio plays, and to a lesser extent in television. Another place where some kind of SST is also used, but with less consistency, is in the so-called ``social" novel and short story. Here writers are involved, but not as prime movers in the decision-making process.

Beginning with the advent of novel and short-story writing in Tamil, there evolved a kind of writing that was concerned with social problems, moral uplift, the independence movement, and other social issues brought on by the collision of colonialism with traditional India. This kind of prose-writing did not actually exist before, nor did almost any kind of prose--everything in Indic languages tends to be in rhymed sutras, more suitable for memorization. In order to make these writings appear to reflect the lives of real people, writers began to use some spoken styles in the dialogues of their writings. Never, to my knowledge, or perhaps very rarely, was a novel/short story written totally in a spoken style. The narrative and descriptive portions of the novel are always written in a form that I would call modern Literary Tamil, which does not admit most of the spoken changes that have occurred since the 13th century, but is more relaxed about, e.g. sandhi rules, than would be older forms.

However the spoken styles are not perfect examples of spoken Tamil, i.e. we cannot use them as true phonetic renditions of how people actually spoke, because there are a number of inconsistencies in this use.

Writers vary; some use SST in their dialogues; some don't. The well-known modern writer Mu. Varadarajan did not use much spoken dialogue in his writings, though he often wrote about it.
Some use it everywhere in spoken style; some use it only for effect. Sometimes there is a switch from SST to LT for a certain kind of effect; or a switch from LT to SST for another kind of effect.

Some writers place SST only in the mouths of rustic or comic characters, reserving a more LT style for the `heroic' or main characters. This is also true in western writing as well; Shakespeare has his `high' characters speak `standard' English but his buffoons and grave-diggers and `low' characters speak dialect, and speak it in prose, rather than verse.

Even when consistency is attempted, we are more likely to find SST forms in the verbal element of the sentence than in the nominal portions. I have discussed the reasons for this in an earlier article (Schiffman and Arokianathan 1978)--issues of recognizability, of position in the sentence, and other non-linguistic reasons for lack of phoneticity can be adduced.

Another may simply be that writers may not consider it important to be consistent, since there are no rules and therefore nobody will be offended if in one part of the sentence one writes mudalilee and in another sentence one writes modallee. And it is just a fact that things written in a non-standard form of language are difficult to decipher, even if they are phonetically correct. English speakers are used to seeing the word `once' spelled that way, and would find it strange to see it spelled <wunts> even though the latter is phonetically closer to what most people say. In English, phonetic spellings are sometimes used in cartoons and other non-standard writing (advertising is another genre) but certain words are never tampered with: know is always spelled with a /k/ and a /w/ even though phonetically it would just be [no]. Obviously comprehensibility would be affected if one went totally phonetic, even in writing dialect. Anyway, a writer's goal is not to report phonetic `field work' but to communicate something, in this case something `social' about certain characters in the story, and since non-standard language may often be associated with certain social characteristics, perhaps stereotypically, in the minds of the linguistic community, a phonetic rendering may tell something more succinctly than a detailed description of, e.g. the character's rusticity and bumptiousness, or whatever. If an American southern writer has a character say ``It idn't none of your bidness" instead of ``It ain't none of your bidness" or ``he dudn't know the answer" (instead of `he don't (doesn't?) know the answer') this is done for a particular effect and with some particular intention on the author's part, and with an expectation that there exist certain understandings in the minds of the readers about what these details mean. Both of these alternatives are non-standard, but they are non-standard in different ways, and used with different effect.

Written and Unwritten Standards.

Though we may speak of an informal standardizing process taking place, and we may recognize the kinds of standardization and regularity that exist in SST, some people are still loath to admit that a speech-form can be standard(ized) unless it has a written grammar, i.e. a book between two covers. This is because of the idea that grammars are imposed or bestowed upon languages, not that languages have grammars, i.e. have rules and regularities that people can discover, and organize into a book. I would like to introduce another notion here, and that is that though Spoken Tamil may not be completely standardized, i.e. there are areas of variability, it is in a position where standardization could in fact be brought about. That is, the potential for standardizing the language is there, and were certain conditions to be met, the process could be complete.

The conditions necessary would, in my opinion, be:

In the areas where inconsistency and variability exist, where people do not agree on the form to choose (such as past neuters) attitude surveys and usage surveys could be undertaken. The areas of variability could be subject to questionnaires in order to determine what people think of, e.g. the cci past neuter, and whether it should be made the standard, allowed as a variant, or excluded.
Educated Tamilians from various communities and regions could make decisions, formal decisions, about the areas where doubts exist. This would be a revolutionary idea, but it could be done.

Consensus could be reached on having flexible stability rather than rigid stability, in order to allow for the possibility that certain sound changes, such as the rounding of front vowels between initial labials and before retroflex consonants, might complete themselves.

A newer, more open way to obtain recommendations for development of new vocabulary could be agreed upon, allowing for flexibility and opening the language to natural and folk resources, rather than only classical resources.

The Confessions of a Standardizer.

Since I have actually written a book called A Grammar of Spoken Tamil it might be of some use to reveal what things were in the back of my head when I did it. That is, did I simply record what I had found, or did I make decisions in favor of one of two competing alternatives in a capricious and prejudicial way? And did the decisions I made help to perpetuate linguistic inequality and the hegemony and privilege of a particular class of Tamil speakers? Or did I simply ratify the existing situation, namely, that standardization has already taken place, and all I have done is to describe it?

Regularity. Most of the time, my concern was with regularity, with finding what patterns repeated themselves in the grammar of SST, and stating this. When competing alternatives existed, I chose the one that seemed to be close to what I in general heard non-Brahmans use. Thus a time-expression marker such as Verb+ pootu meaning `when (X) verb-s' can occur either as irukkr-appa, irukkr-appo, or irukkr-appam where the -am portion is actually a nasalized [õ]. The LT form of this would be more likely irukkum pootu and some speakers would use this to disambiguate their speech, if some other speaker didn't understand irukkr-appa, irukkr-appo, or irukkr-appam. In fact, this rule of thumb (``Disambiguate by using LT, or something close to it"), is a fairly useful one; all literate Tamilians use it, although some feel that they should use it all the time. Thus the LT-like irukkum poodu, with a nasalized [u] to tell us it's spoken, not LT, is also a feature of SST; it is not a totally `natural' thing for the spoken language, but it is used. In fact other things like it are also used, which means that it is a kind of `elaborated style' in the Basil Bernsteinian sense of the word; it is neither SST nor LT. It is perhaps akin to what speakers of French sometimes do, i.e. spell out a word, to disambiguate it: il lui faut la mer, m-e-r. `He needs the sea (not his mother)'; Chinese speakers are also known to finger-spell Chinese characters in the air or on their hands to disambiguate certain things.
Pedagogy. Since my goal with my grammar was to provide something that was pedagogically useful for second-language learners of Tamil, and since literate Tamilians invoke literary forms in case of doubt, I also chose to do so when it made certain things more simple pedagogically. For example, the present-tense marker for all `strong' verbs in Tamil is kkr and for weak verbs it is r: paar+kkr+een `', i.e. I see' poo-r-een `go-pres-1st' `I go'. This is so for all verbs except the copula iru, which has a present-tense marker kk only iru-kk-een `be-pres-1st' I am'. However, since this verb has a LT form with LT kkir like any other verb, I chose to not make an issue of this point except as a footnote, when discussing iru in particular. That is, no one will fail to understand irukkreen when they themselves say irukkeen, and students are burdened with one less rule.

On the other hand, iru also has an ``irregular" neuter singular form; the LT form is irukkiratu but the spoken form is just irukku. Some speakers do produce a form irukkutu [irukkudu] but a more LT-like irukkradu would be a step beyond that. Since irukku is the most common, I chose to list it as such; if they encounter irukkutu they will understand it. Here frequency of use took precedence over regularity or whatever. For literate Tamilians, the LT variety always comes first, and they can always consult the grammar of LT in their heads if they have questions; for them, SST should always defer to LT, and be based on it. Such a viewpoint does not allow for the possibility that a foreigner might not have a grammar of LT in her head to consult in moments of doubt. On the other hand, the second-language learner of Tamil must at some point confront the fact of the grammar of LT, since it is culturally expected and is sometimes useful to know; in any event all reference works, practically, are written about LT. A person writing a grammar of Spoken Tamil cannot ignore the existence of LT, even though American linguistics may tell him that literary languages are irrelevant. One must find a golden mean, between the structure of the spoken language, and the structure, some of it quite identical and useful, of LT. Tamil is a diglossic language (Britto 1986), and this fact must be acknowledged; what linguistics, structural or theoretical or whatever, must also admit is that in a diglossic language, the spoken variety is strongly influenced by the literary variety. This fact is inescapable; it is well-nigh oppressive. But the spoken language also has life and juice and zing that the literary variety does not possess, i.e. it has an authentic vitality, a life of its own that is often lacking in the stultified norms of LT.

Simplicity. Another criterion we value in linguistics is that of simplicity; we are taught that linguistic descriptions are best if simple, and that simplicity, when found in a language, should be preserved. When one is forced to have to learn both the rules of LT and SST, it is simplest if one can learn them in a way that proceeds simply from one set to the other. This criterion underlies some of the decisions made above, such as the choice of kkr as the present-tense marker for iru even though it is more commonly kk. At an earlier period I would have castigated myself for giving in to the rules of LT, but since life is short, whatever is simplest is best.

Stigmatized forms. As mentioned above, SST, though itself stigmatized from above by LT, also has forms it does not accept. Those that are neither vulgar nor obscene, but still need to be dealt with, are things like the rounding of e and i between an initial labial consonant and a following retroflex (usually) consonant already mentioned, or the deocclusion of NC to [nn] in some places. In such cases, I am forced to choose whichever form is most acceptable, but which leads to a funny kind of irregularity already noted: poNNu `girl' is acceptable, but (v)uuDu `house' is not; (v)uuTTle `in the house' however is. This is no problem for the student who does not know LT; for students knowing LT it seems strange that the rounding has proceeded irregularly through the lexicon. To the variationist it is simply evidence of a sound-change in process.

Abstractness. Another decision already alluded to is that one of the differences between LT and ST is the apparently deletion of certain final consonants in SST, such as final laterals and rhotics, and the nasalization of final vowels. If one were to be perfectly true and phonetic about this, one would state that SST and LT differ in the presence and or absence of these sounds, but this is over simplifying. In fact, most of these laterals are only absent when the word/morpheme in question occurs before pause; but if anything is added, such as case, or clitics, or interrogative markers, the sound is not deleted. Thus the LT form vantaal `if X comes' has a spoken analog vandaa. but if anything is added, such as um `even', then both forms are identical: vantaal-um and vandaal-um `even if X comes'. Thus the `underlying' or `base' forms of many morphemes in SST will be the same as in LT; what will differ is that in final position, before pause, certain consonants can (and will) be deleted.

Thus I have, in the interests of simplicity and regularity and the other criteria linguists generally use, made decisions about what form or other is `standard', even in cases where variation in the language may exist. This is, I think, no violation of anybody's rights, nor does it do any injustice to the language. No Tamilian ever tells me they can't understand the forms I use; I never have trouble making myself understood on the telephone (where people can't see my face and therefore don't expect me to be speaking English); all the forms I have chosen are in fact used by somebody though there is perhaps no one individual who speaks exactly the way I have described the language. Thus it is perhaps the case that there is no native speaker of SST yet, and everybody still speaks their local dialect most of the time, reserving SST for inter-caste, inter-regional communication. As long as people are closely bound in kinship systems, this will certainly mean that special caste-related kin terms will be used that cannot be used by all castes (there is no word for `wife' that is not caste-bound.) Perhaps then only foreigners will speak SST, or Singaporeans.

This brings me to my last point; I find that in general, Singapore spoken Tamil (at least that variety still learned at home as a native language, i.e. not the variety learned only at school) is more or less congruent with SST; the few things I notice that are different are a tendency to use more LT-like forms than TN Tamils would use. Singaporeans say perroorkal for `parents', whereas Tamilians in TN would say appaa-mmaa; Singaporeans say muunru for `three' instead of SST muuNu and there are some other hypercareful forms I hear; but aside from the occasional Malay word, I do not notice great differences, i.e. I cannot tell from a Singaporean's Tamil (unless he says the two words above) whether s/he is Singapore-born or TN born, but I can tell from his/her English.


Phonological ``Rules" that have applied in SST
These are phonological rules that a linguist would characterize as having been added to the grammar of LT, the application of which gives SST as the output.

Palatalization: tt, nt [cc, nj] after high front vowels ([i, ai]) as in paDitteen [paDicceen] `I studied'. Found in all dialects, perfectly regular.
Doubling of sonorants in CVC where V is short: kal kallu `stone' etc.

CVC where V is long, add u; but not in all dialects: naaLu `day' naa `day'.

l in plurals, pronouns, verbs, deleted in final position, otherwise present.

Nasalization of final [Vm, Vn] (but not [Vn])

Lowering of [i, u] to [e, o] in sequence (C)...Ca, as in iDam [eDõ] `place', utkkaaru
deletion of l, r, l before stops, internally: uTkkaarndeen okkaandeen `I sat' etc.

general cluster reduction: uTkaaru okkaaru `sit'; keeTkireen keekkreen `I ask'.

Monophthongisation of ai to e: in accusatives, noun endings, internally. Exception: in monosyllables, vai vayyi `put, place'.

Intervocalic v and k deletion. Complicated, but examples: paarkkavillai [paakkale] `didn't see'; pookaveentum [pookaNum] `must go', pookireen [pooreen] `I go'; koNTuvaa [koNDaa] `bring', etc.

Intervocalic deletion of R, the retroflex frictionless continuant. This sound often is merged with [L], but in some cases it is deleted instead, resulting in compensatory lengthing: poRudu [poodu] `time'; mukaRndadu [moondadu].

Rounding: [i, e] [u, o] between Labial ...Retroflex consonant: poNNu `girl', poTTi `box', (v)uuDu `house', puDi `like', but also: mida [moda] `float', piranteen [porandeen] `I was born'. Some of these may be more acceptable than others.

Optional or Stigmatized Developments.

Already described: NC cluster reduction is variable, and somewhat stigmatized veeNum `want, need' is okay, but *veeNaam `don't want, not needed' is not; koNDaandeen (koNDuvandeen) `I brought' is okay, *koNNaandeen `ibid.' is not.
Rounding of vowels before retroflex /L/ in Br. dialect: niingo `you pl.', avango `they'.

Metathesis: Br dialect: enakku [neekku] `to me'; unakku [onakku] [nookku] `to you' etc.

Other dialects: merge [L], [D], [R]; some merge [n] and [N]. Retroflexion is lost except in [t] (vs. [T])

Unrounding of [u] when nasalized: Coimbatore: veenum `want, need' [veenim ], etc.


Andronov, M. (1962) Razgovorny tamil'ski yazyk i ego dialekty. [Spoken Tamil Language and its Dialects.] Moscow: Nauka.

Andronov, M. (1975) ``Problems of the National Language in Tamilnad." Anthropos Vol. 70:180-193.

Annamalai, E. (n.d.) `Notes and Discussions; The Standard Spoken Tamil--some Observations.' (ms.)

Arden, A. H. 1954 A Progressive Grammar of Common Tamil. Madras: Christian Literature Society.

Arokianathan, S. (1981). Tamil Clitics. Trivandrum: Dravidian Linguistics Association.

Asher, R. E. Tamil. LINGUA DESCRIPTIVE STUDIES, Vol. 7, (1982). Amsterdam: North-Holland.

Britto, F. (1986) Diglossia: A study of the Theory with Application to Tamil. Washington, D.C.: Georgetown U. Press.

Crowley, Tony. 1989. Standard English and the Politics of Language. Urbana and Chicago: University of Illinois Press.

Eastman, Carol. (1983) Language Planning, an Introduction. San Francisco: Chandler and Sharp.

Fabricius, Johann Philip. 1910, repr. 1972. A Dictionary, Tamil and English. Tranquebar: Evangelical Lutheran Mission Publishing House.

Ferguson, Charles. (1988). ``Standardization as a form of Language Spread." In Language Spread and Language Policy, Peter Lowenberg (ed.), Georgetown University Press. Pp. 119-132.

Garvin, Paul. (1964) ``The Standard Language Problem-Concepts and Methods" In. Hymes (ed.) (1964), pp. 521-526.

Giles, Howard and Philip Smith. 1979. ``Accomodation Theory: Optimal Levels of Convergence." In Giles, H. and Robert St. Clair (eds.), Language and Social Psychology. Pp. 45-65.

Haugen, Einar. (1972) ``Language Planning in Modern Norway." in Dil, A. (ed.) The Ecology of Language: Essays by Einar Haugen. Stanford: Stanford University Press. Pp. 133-147.

Goody, Jack. (1986) The Interface between the Written and the Oral. Oxford University Press, New York.

Hymes, Dell (ed.) (1964) Language in Culture and Society. New York: Harper & Row.

Jernudd, Björn and M. H. Ibrahim. (1986). ``Introduction." ASPECTS OF ARABIC SOCIOLINGUISTICS, Vol. 61 of International Journal of the Sociology of Language p. 5-6.

Joseph, John Earl. (1987) Eloquence and Power: The Rise of Language Standards and Standard Languages. New York: Basil Blackwell.

Kachru, Braj. (1985). ``Standards, codification and sociolinguistic realism: The English language in the Outer Circle." In English in the World Randolph Quirk and H. G. Widdowson, ( eds.) Cambridge: Cambridge University Press. Pp. 11-30.

Mitchell, Terence F. (1985). ``Sociolinguistic and Stylistic Dimensions of the Educated Spoken Arabic of Egypt and the Levant." In J. Douglas Woods (ed.), Language Standards and their Codification: Process and Application. University of Exeter: Vol. 9, Exeter Linguistic Studies. Pp. 42-57.

Mitchell, Terence F. (1986). ``What is Educated Spoken Arabic?" International Journal of the Sociology of Language 61:7-32.

Milroy, James and Lesley Milroy. (1991). Authority in language. Investigating language prescription and standardisation. London: Routledge and Kegan Paul.

Pakir, Anne. (1994). ``Education and Invisible Language Planning: The Case of English in Singapore". In English and Language Planning: a Southeast Asian Contribution. Thiru Kandiah and John Kwan-Terry, (Eds.) Singapore: Centre for Advanced Studies, National University of Singapore and Times Academic Press. Pp. 158-181.

Pakir, Anne. (1991). ``The Status of English and the Question of `Standard' in Singapore: A Sociolinguistic Perspective." In Tickoo, Makhan L., Languages & Standards: Issues, Attitudes, Case Studies. Singapore: SEAMEO

Phillipson, Robert. (1992). Linguistic Imperialism. Oxford: Oxford University Press.

Pillai, Shanmugam M. (1960). ``Tamil, Literary and Colloquial." In Ferguson and Gumperz (eds.) LINGUISTIC DIVERSITY IN SOUTH ASIA. International Journal of American Linguistics 26(3):27-42.

Schiffman, Harold. (1999). A Reference Grammar of Spoken Tamil. Cambridge: Cambridge University Press.

Schiffman, Harold. (1991) ``Swiss-German Diglossia." In A. Hudson (ed.) STUDIES IN DIGLOSSIA. Southwest Journal of Linguistics, Vol. 10(1) pgs. 173-188.

Schiffman, Harold. (1996). Linguistic Culture and Language Policy. New York and London: Routledge.

Schiffman, Harold and S. Arokianathan. ``Diglossic Variation in Tamil Film and Fiction." in Krishnamurti, Masica and de Silva (eds.) Look up the Title of this.

Staal, Frits. (1986). The Fidelity of Oral Tradition and the Origins of Science. vol. 49(8) of Mededelingen der koniklijke Nederlandse Akademie van Wetenschappen, afd. Letterkunde, Nieuwe Reeks. North-Holland Publishing Company, Amsterdam.

Tickoo, Makhan L. (1991). Languages and Standards: Issues, Attitudes, Case Studies. Singapore: SEAMEO.

Vasantha Kumari, T. (n.d.), ``The Standard Spoken Tamil." (ms.)

Walters, Keith. (1996) ``Gender, identity,and the political economy of language: Anglophone wives in Tunisia." Language in Society 25:4, pp. 515-555.

Harold Schiffman