I hope I have, in the foregoing discussion, succeeded in isolating diglossia as an issue that is of interest to the educationist. I have not yet given the reasons for my choice of the Kannada, Sinhalese and Tamil communities for my research; this I shall do at this point, before entering into a discussion of specific characteristics of literacy in these communities.

I have already shown that there are two kinds of situation, which are diglossic. Firstly, there are those situations where 'prestige' is religiously or culturally defined; in such circumstances the linguistic implications of diglossia would include the choice of the prestigious norm for the written usage as well as for the formal spoken usage. There, the written and formally spoken languages share a fair degree of similarity, at least in grammar; they are governed by norms specified identically for both. Secondly, there are those situations in which diglossia is a product of some type of puristic elitism which is not conditioned by religious or broadly cultural factors; it is a product more of insecurity and insensitiveness to the nature of language as a tool, a means of communication. As for the linguistic usages in these circumstances, the norms specified via classical models are only employed for written purposes. In these situations, the prestigious form is virtually the written about the first type of diglossia. The second type has been dealt with much less; Krishnamurti (1975, 1976), De Silva (1967, 1974a) are perhaps the only available material. In order to show the similarity between the Sinhalese and Telugu situations I shall present in a subsequent chapter a thumb-main sketch of Sinhalese diglossia; in it I shall also attempt to illustrate the role of purism in the evolution of diglossic communities.

Arabic and Tamil are examples of the first type of diglossia described above; Sinhalese and Telugu belong to the other type. (The Telugu situation is somewhat different from the Sinhalese one in that the role of purism has been different in the two: I shall refer to this eventually; see also Krishnamurti 1976). In this way, the choice of Tamil and Sinhalese has enabled me to take account of both types of diglossia in my study.

Kannada, strictly speaking, belongs with Tamil. However, the Kannada attitude towards the maintenance of the linguistic duality is much less rigid than the Tamil attitude and, as a result, modern Kannada literature tolerates a fair amount of colloquial lexis, etc. Still, it is the case, however, that most writers adhere to, or profess to adhere to, a variety of language clearly distinct from the colloquial languages, not only in the lexis but also in grammatical matters.

Although at the beginning I chose the Tamil and Sinhalese situations as distinct types of diglossia and included the Kannada community in my study because of its declaredly more relaxed adherence to a Tamil-like diglossia, I have discovered since I began my survey that Kannada presents a further type in the typological array of diglossic communities. It is a 'different type' not so much because it is different from the Tamil type in the separation of the high and low varieties as because of the choice of an apparent social dialect for high or formal use. I have not gone into this matter fully enough to make any bold pronouncements about it, but I have learnt, on best authority, that the formal language of many learned Kannada speakers is similar to the home language of the Brahmin community of Karnataka. The status of caste as a dominant variable in linguistic diversification in India has already been questioned (Pattanayak 1975), and so any mention of caste here or elsewhere in this essay should not be interpreted beyond its face value. Be that as it may, if the language of a particular segment of a community is chosen by the rest of the community for prestigious usage, the diglossia symptomatic of such a situation would perhaps be based on the recognition of that chosen segment as in some sense synonymous with the heritage which initially motivated or was chosen to symbolize the prestigious norm. Also, in such a situation, the prestigious segment would not be a diglossic community, for whatever they did would be prestigious any way; it is the rest of the population that would form a diglossic community. The model of excellence does, in this sense, come from avowedly the same society, all members of which claim adherence to the same culture and profess to speak the 'same language'. In the case of Tamil and Sinhalese, all speakers who claim to speak the 'same language' use are acknowledge linguistic symbols of diglossia in appropriate circumstances. What societies like the Kannada one shows us is that a firm belief in a 'same language' concept does not necessarily make all members of such macro-communities diglossic or non-diglossic. Values such has prestige any symbols of such values segment such communities into smaller units. Linguistic communities are obviously, not entities, which transgress functional diversification.

I am assuming all along that what I have learnt from educated Kannada speakers about this Brahmin versus non-Brahmin situation has some truth in it. Because of the way it seems to work, the Kannada community might even be found to be similar to the English one. In the English community, too, a model of linguistic excellence is often sought in the realms of educated speech in the South Eastern parts of Britain: these educated speech in the South-easterners are, then, the British Brahmins. Looking at it from this point of view, one diglossic at all but operates, like the community of English speakers, within a standard versus non-standard distinction. Such an interpretation must, however, be received with caution, for, as I have said before, English speakers do not make a rigid distinction with regard to the functions of the two varieties while, from all account, the non-Brahmin Kannadigas do. At least for the present I would, therefore, treat the Kannada situation as a diglossic one. (See also Nayak 1967.)

The three communities I have chosen do, in this way, represent diglossic behaviour in all its ramifications. Adjacent geographical areas were chosen primarily for convenience; Sinhalese also happens to be my own first language. This choice has, however, proved to be a productive one rather than merely a convenient one, for the attitudes towards and some difficulties in the acquisition of literacy can now be traced directly to the nature of diglossic behaviour and people's disposition towards diglossia. We are in the same cultural area and are, therefore, able to reduce to a minimum the various cultural variables, which, if the situations were chosen from different parts of the world, might have been too numerous to control.

In all three communities, and indeed in the whole of South Asia, book learning is rated very high. Even five and six year old children do a fair amount of homework in reading and arithmetic each day and the parents are forced by the system to help the children with their school work every day. Some children know the first primer by heart before going to school although they are not necessarily capable of identifying individual words or pronouncing them in isolation. The teacher often interprets this as 'knowing the test'; the repercussions of this on the child's learning are obvious. The entire phenomenon is, of course, circular. The parents expect the children to be taught book knowledge at school and the teachers set the class work in order to achieve this parental objective. The children are not always encouraged to discover, make and enquire; this is especially so in rural schools, and memorizing forms an integral part of leaning even at the university level. A concomitant factor of this slavishness and allegiance to the written word from the first day at school is that at school every school child is required to read everything that is written in his reading exercises. He is required to pronounce every single letter including those that are superfluous from the standpoint of the phonemic structure of his language but are there because of its diglossic character. The grammar, i.e. inflections and rules of concord etc., being different in the written language, the child is neither able nor encouraged to read the sentences with an intonation that would be appropriate for their spoken counterparts that are familiar to him. Many parents make their children memorize whole reading passages which at school they rattle off parrot-fashion and also extremely fast. The written language being distinct from his own spoken usage, for which he naturally has a good ear, the teacher is often unable to identify the pronunciation errors in these fast readings; my experience has shown that in a number of instances where the teacher has confessed to having heard the distinct but related pronunciations prescribed for different letters the pupil has never made such a distinction in his pronunciation: I refer particularly to the various sibilants, the nasal consonants and, in Tamil, the r-type sounds. These features I have described in this paragraph are, probably, symptomatic of all diglossic communities; they must not be interpreted as an implied criticism of teaching techniques in general.

I have given this preamble in order to focus on the important place reading has in the Indian child's early learning. The only language that the child is made to read and write is the literary variety. (Some sentences in the Kindergarten readers may contain a few sentences, which resemble the colloquial variety, but they are only an extremely small proportion.) During their thirteen or fourteen years at school before going to university and then for three or four years at the University pupils are exposed to a great deal of high language. They read it and, when they write in their mother tongue, they write it. The irony, however, is that, in spite of this extensive training, it has been found that these people are not always able to perform in full accordance with the norms prescribed for the literary usage. In Sinhalese this has been noticed very frequently, and purists have criticized the universities for not giving their undergraduates sufficient training in normative usage. When I was a lecturer at the University of Ceylon (now Sri Lanka) the Head of the Department of Sinhalese required us to take this as a serious criticism. As I have shown in De Silva (1974a), there are also some famous Sinhalese writers whose command of the literary norms occasionally falters. As I have said before, the Kannada situation is a slightly relaxed one and one could, in such a context legitimately expect a writer to represent the colloquial grammar and phonology in his writing. In Tamil, however, the cleavage is very rigidly upheld and it is generally believed that secondary school children and university students do not violate the normative literary rules. Despite this belief experiments have shown that even Tamil language graduates violate specified norms and in fact Shanmugam Pillai (1965) shows tendencies toward a merger between the two varieties in certain circumstances.

What all this means is that the maintenance of the linguistic duality involved in these instances is an extra burden on the learner. Notwithstanding this, many native speakers of these languages want to restrict the term 'literate' only to refer to persons who are able to read and write correctly. By correctness they mean the ability to observe the normative literary rules with regard to spelling and grammar. The opinions, however, vary in the three communities in a significant way: this variation seems to be a result of the differences in the definition of prestige in the three communities (De Silva 1974b) and has an influence on the degree of 'correctness' that people are able to observe in their reading and writing. In a survey conducted to study the social expectations of literacy (and thereby the social definition) I have obtained some revealing results which I summarise below. The persons chosen for this survey had al completed their secondary school education and ranged form the B.A. first year (in India, the so-called pre-university first year) students to professionals holding doctoral degrees, including some Ph.D.s in the language in question. They belonged to different occupational categories such has clerks, teachers, lecturers, businessmen, civil servants and university students. They had all gone through the machinery for acquiring sufficient literacy as required by their communities. Between sixty and one hundred were questioned in each community. Illiterate persons and school children were not included in the survey: among other things the illiterates are not able to pass judgment on matters like literacy and would generally be happy to be guided by established opinion; school children would not always be able to give dispassionate answers because they live in the midst of prejudices for or against matters pertaining to their school curriculum. The percentage of results with regard to correctness were as follows:

	Tamil	Kannada	Inhalese
'Literacy' is the ability to read and write correctly	100	50	30
'Correctness' means thecorrect rules of spelling and grammar	99	45	15
'Literacy' is the mere ability to read and write	0	50	50
Unable to decide between 1 and 3	0	0	20

As the above figures show, 100% of the people questioned in Tamilnadu wanted to define literacy as the ability to read and write correctly: correctness was defined by 99% as correct adherence to the rules of spelling and grammar as specified for the high variety. (The one per cent that differed was concerned with correct spelling only). Ideas were less settled in the Kannada community. In Karnataka, 50% wanted to define literacy as the ability to read and write correctly, defining correctness as the correct adherence to the normative literary rules. An equal proportion was, however, happy to settle for the mere ability to read and write the letters of the alphabet as the definition of literacy. Among those who defined literacy as the ability to read and write correctly, the majority (45% of all those who were questioned) wanted the literates to observe correct rules of spelling as well as grammar while 5% was happy to restrict the meaning of correctness to spelling only. This divergence of opinion corresponds to the more relaxed attitude to diglossia among the Kannada speakers. In spite of this attitude to correctness, the overwhelming majority wants to retain the teaching of the high variety in schools. Contrasting with both Tamil and Kannada the Sinhalese response was divided significantly in favour of the mere ability to read and write rather than the ability to read and write correctly: while only 30% wanted no more than the ability to read and write as the attribute of a literate individual. Even among the conservative 30% there was no unanimity on the definition of correctness. One half of them implied by correctness 'correct Sinhalese usage' (not specifying it as literary or colloquial) which is not adulterated by foreign language influence. They were particularly concerned with the influence of the English idiom on Sinhalese usage through the Sinhalese writings (and the Sinhalese speech) of the English- educated persons. All in all, then only 15% wanted the literates in the community to be proficient in the normative literary usage. It is significant that 20& felt that, in the climate of present day polemics on the subject, a decisive answer was not possible.

The reader might find this vast difference between the Tamil and Sinhalese attitudes surprising, considering that high Sinhalese is a specifically written variety, which has, therefore, to be learned with particular care for that purpose only. In the case of Tamil, the high variety is spoken on formal occasions and, while the learning of it is reinforced by this oral-aural use, the divergence between the two (see Shanmugam Pillai 1965, 1975) might have forced people to give their answers in favour of communicability rather than the preservation of the duality. To a person who is aware of the different ways in which linguistic prestige is defined in these communities, however, these results would not come as a shock.

In Tamil, the ability to use the high variety is associated with social superiority and intimacy with the Tamil heritage. Formerly, the high variety of Tamil was almost identical with the formal, educated speech which showed much Sanskrit influence. The Brahmin speech has also been defined as highly Sanskritized. Since the ascendance of the Dravida Munnetra Kasagam (DMK, the ruling party in Tamilnadu) the Brahmins have been forced out of their prestigious place in the society and, along with this the high language has been stripped of some of its Sanskritic lexis; however, this has not made the language any closer to the colloquial speech, for more archaic features have been introduced into the language by the DMK philosophy in its quest of pure Tamil culture and language. The diglossic character has, therefore, not been affected by the state's political changes, and the high variety is still looked up to with the same respect and awe. (As I have reported before from the experience of two persons, even the uneducated people would not take people seriously when they attempt to address them in public using the colloquial language known to the audience.)

The Sinhalese practice is different from this. Not being a spoken language, literary Sinhalese has no class or caste or any other similar overtones. Although historically speaking, literary Sinhalese is a resurrected variety of language that belonged to the 'purity' of Sinhalese have ensured that the literary variety is not widely recognized as the symbol of the Sinhalese heritage. In the circumstances, the ability to use literary Sinhalese correctly only reflects the person's erudition in the Sinhalese language. Among the Sinhalese educated élite there are at least three points of view on the character of literary Sinhalese. There are the purists who clamour for the use of 'pure Sinhalese' phonology and lexis in the framework of literary grammar (for the implications of the term 'pure Sinhalese' see De Silva 1967 and chapter 5); while the purists want the Sanskritic and other foreign elements out, there are others who have no aversion to Sanskritic lexis and phonological representation, and would in fact encourage them, but within the accepted normative literary grammar; there is, in addition, a third group that regards the special literary features as irrelevant and cumbersome and, therefore, campaigns for the abolition of the present linguistic duality. Unlike in the case of Tamil, the ability to use the literary variety correctly is not associated with an intimate knowledge of, or a passionate belonging to, the glorious past culture of the people. The indecision reflected in the answers given by 20% is a direct result of this lack of unanimity among the Sinhalese élite on this matter.

I have said earlier on that a situation akin to the Sinhalese on existed in Telugu a little while ago. In Telugu, the opponents of the classical or literary grammar have won their battle, at the recommendation of a Telugu Language Committee the use of the classical language for teaching and examination has been relaxed in the universities and the secondary schools. The Telugu Language Committee Report (1973) discusses the history of Telugu diglossia and the circumstances that culminated in the Committee's recommendations. As in Telugu, the only prestige that literary Sinhalese enjoys is that it merely symbolizes a particular type of learning, namely the learning of the classical usage. Before I leave the Telugu Language Committee Report, I must hasten to add that the main reasons that seem to have influenced its recommendations are the absence of any substantial social prestige associated with the literary language and the difficulties in the teaching and the learning of an extra grammatical complex for a restricted purpose. (For more details see Krishnamurti 1976.) In the Sinhalese community there is, as I have mentioned before, a campaign against literary grammar, motivated by the same reasons. Although this campaign has not succeeded in abolishing the existing linguistic duality, the statements that have been made in answer to my questionnaire reflect somewhat the present-day thinking on the subject.

My questionnaires were in the respective languages. Among other things I was interested in finding out to what extent the people's definition of literacy was compatible with their own performance in the literary variety (e.g. did those people who expected the literates to know the 'correct' usage and professed to know it themselves actually perform 'correctly'). With this in view, I designed some questions to elicit long answers (of the type 'why do you think so', 'explain why', 'what is your own opinion', etc.). The answers I received to such questions contained two to five sentences each and the total number of sentences formed a fair quantity on which the subjects' capabilities in the language could be assessed. What these sentences revealed was that people's expectations of literacy and their educational qualifications were not clear predictors of their own proficiency in the literary language. A person may declare that literacy is the ability to read and write correctly according to specified norms; he may say, with some pride, that he experienced no difficulty in mastering the literary grammar at school in a short span of, say, two years; he may possess a B.A. degree in the respective language; he may be one who has little patience with those who violate the normative rules of literary usage: yet he can make an average of one grammatical mistake for every two sentences and a spelling mistake for every three sentences. The following is a summary of such errors (the mistakes I have counted are those identified by groups of teachers who served as assessors during my survey).

Tamil:
100% wanted the literates to know the correct usage.
Total number of sentences written in answers: 170
Average of mistakes - grammar: 1 for every sentence
- spelling: 1 for every 3 sentences

Kannada:
50% wanted the literates to know the correct usage.
Total number of sentences written in answers: 121
Average of mistakes - grammar: 1 for every 2 sentences
- spelling: 1 for every 3 sentences

(There were 8 instances which appeared to be spelling mistakes but might also have been inadvertent slips of pen due to the similarity in the shape of some letters of the alphabet. I have not counted these.)

50% did not want correctness to be part of the definition
Total number of sentences written in answers: 108
Average of mistakes - grammar: 1 for every 2 sentences
- spelling: 1 per sentences

Sinhalese:

30% wanted the literates to know the correct usage.
Total number of sentences written in answers: 88
Average mistakes - grammar: 1 per sentence
- spelling: 1 per sentence

70% did not want correctness to be part of the definition.
Total number of sentences written in answers: 102
Average of mistakes - grammar: 3 for every sentence
- spelling: 1 per sentence

I shall postpone the discussion of the theoretical implications of these errors till the next chapter. What is significant in these figures is that even among those who profess to know the correct usage and to follow it there is a fair number in whose writing various types of mistake recurs. It is necessary to enquire into the reasons for this. The primary school teachers in these languages are not unanimous on the time a child usually takes to master the literary grammar and spelling, but they all agree that, by the time the child is ready to leave the primary school (i.e. age 11), he will have mastered the rules of grammar and spelling, leaving only the lexicon to be acquired gradually as he goes along. The above figures based on the performance of adults contradict this belief. In the primary school there are five standards or grades. Some teachers believe that the child is capable of manipulating the literary variety in the third standard, some the fourth standard and some the fifth. These statements are very much tied to whether the teaching of the language in the primary school is in the hands of one teacher or many teachers. If the responsibility is in the hands of one teacher, he or she usually plays safe by raising the time to the fifth standard. Where the responsibility is shared by many, the customary thing is to pass the buck: statements like 'it will be done next year', 'it should have been done last year: it is too late now', etc. have often been made. I do not think that these attitudes necessarily reflect the teacher's incompetence or indecision; the teaching of the high variety has been done for decades and the textbooks and the teaching methods have envolved alongside this experience. The problem, as I shall attempt to show in my analysis elsewhere, is rather due to the relationship of the literary variety to the child's spoken usage. I do not wish in any way to put the entire blame for all problems of literacy on the diglossic character of these languages, but I am convinced that diglossia makes the acquicision of literacy more difficult than it ought to be and causes indecisions and diffidence in the use of language which inhibit the child's free expression and creativity.

I have attempted to assess the primary school children's proficiency in the literary usage as well as the rate of increase in their proficiency by giving them five tasks, namely, reading of known texts, taking dictation of unseen texts and writing of free compositions. These tests showed not only that the children were not anywhere near full proficiency in the literary language by the time they were 11 or 12, but also that their rate of progress was slow in comparison to their rate of progress (that was more easily assessable), say, in mathematics. As an illustration of the errors the fourth and fifth standard children made in these tests, I have taken, at random, ten one-page free compositions written by them in each language and tabulated the errors as follows:

Tamil:
79 errors
Wrong spelling :40
Wrong inflection :15
Wrong word: 6
Wrong grammar or colloquial grammar: 6
Colloquial words:12

Kannada:
156 errors
Wrong spelling: 85
Wrong inflection :15
Wrong grammar or colloquial grammar :23
Colloquial words :24
Wrong words :6

Sinhalese:
197 errors
Wrong spellingn : 92
Wrong inflection or wrong postposition :26
Wrong word :2
Wrong grammar or colloquial grammar : 49
Colloquial words :28

Before I leave the subject of free composition, I must define what free composition means in these school situations. Free composition here does not mean writings of children's own choice and imagination which would give vent to their innovative nature. In my experience, the teacher first sets one or more essay topics (e.g. the coconut palm, the happiest day in my life); then he usually writes on the blackboard some useful (literary) expressions that may be used and also some notes on concord, etc. Some teachers give a skeleton of the essay on the blackboard. The entire operation appears not so much an exercise in creativity as an exercise in literary usage. The mistakes I have counted as above were made in spite of this training.

If, after thirty or forty hours a week of exposure to a variety of language for over four years, the average performance of a child is no better than I have found and described; if the social expectations of literacy are not altogether compatible with the performance of the high literacy ideals; if a social attitude enmeshed in such incompatibility creates a state of tension (De Silva 1974a; 1974b; Shanmugam Pillai 1972) that is not in the best interests of creativity; then the system that gives rise to these circumstances might not be the ideal one to perpetuate for the accomplishment of literacy as 'an ability to express oneself articulately for a variety of purposes, socially, intellectually and vocationally both in speech and writing; to command a capacity to read for information, enjoyment and enrichment and to respond sensitively and intelligently to what is said as well as to what is written' (Goddard 1974, p. 21). The teaching methods in these communities might need revision, but it would not be realistic to apportion full blame on the teaching methods or, indeed, the syllabuses. It is my contention that the circumstances and conditions of usage are not favourable enough for most persons to internalize the literary language (in any sense of the term 'internalize') to the point of having a full command of it. Where a form of language has a restricted function in the society, leaving little opportunity for the learner to use it fully, frequently and in a wide variety of circumstances, that language will always be comparatively alien to him; his personal spoken language which he constantly uses in every normal circumstance will exercise its powers more to block the individual's familiarization with this alien form than to facilitate it. Colloquialisms that have infiltrated the writings of many authors and the hybridisms created by many in their writings (De Silva 1974a; Shanmugam Pillai 1965, 1972) are far more numerous than the literary influences on the speech habits of any section of the population; any influence of the literary language on speech is primarily lexical and is the result of extensive reading.

The history of the diglossic situations under survey shows that the high varieties in all three are not natural to the communities but have been imposed upon their normal linguistic habits by revivalists who have, in the name of cultural renaissance and standardization, resurrected linguistic forms which are several centuries older and whose overall structure is outside the linguistic competence that the individuals display in their speech behaviour as grown up members of the community (De Silva 1967). Bright (1970) proposes that the literary form should be stated as the underlying structure of the language from which the colloquial form may be conveniently derived by deletion rules. It would certainly be neat and convenient to do this on paper; however, if a form of language (in this case the literary language) is the basis or the 'underlying structure' of all speech acts and is therefore within the competence of the speakers, why is it that young individuals (who in similar circumstances can master aspects of a second language with two or three years intensive learning) find it difficult to operate it in full accordance with the norms despite their exposure to it for thirty or forty hours a week over a period of four or five years?

This takes us on to the subject of competence about which we hear a lot in linguistics and which has been differently defined by different persons for different purposes (Chomsky 1096; Campbell and Wales 1970; Hymes, mimeo; Le Page 1976; etc). I propose, therefore, to devote the next chapter to a discussion of linguistic homogeneity and the notion of competence in some detail, but with particular reference to diglossic situations.