Diglossia, religion, and ideology: On the mystification of cross-cutting aspects of Bengali language variation

James M. Wilce
Northern Arizona University


This paper critiques several approaches to the Bengali language situation, approaches which touch on aspects of the language which I have not personally studied--particularly earlier forms of Bengali. It is a critique of sociolinguistics narrowly-defined, and more broadly of scholarly work which omits reflexive self-positioning. I hope experts in the Bengali language and its history will receive this paper as an attempt to shed light on problems and carry academic discussion into new ground.1

In particular, I juxtapose three discourses surrounding Bengali: historiography, sociolinguistics, and language-planning. Historians portray the Bengali language as a communal-political football and sociolinguists as one language riven by class and literacy. Language planners, viz. the East Bengal Language Committee, tried to shape the evolution of the Bengali language. Unfortunately, these planners were not alone in confusing facts with goals, history with hagiography, and the chimera of linguistic purity with the ravenous god of communal purity. It seems exceedingly difficult for any given scholar or bureaucrat who has weighed in on the Bengali linguistic situation to consider both religio-cultural and social-stratificational factors. Yet when the three traditions are examined together, insights arise which are missed by any one of them alone.

Historiography of the Bengali language: Narrating the rise of communal diversity in the competitive encounter of civilizations

The Bengali language, like most, reflects heterogeneous sources. In the interests of space I must begin my survey of this history with the fourteenth century, which was not, in fact, an era in which Bengali literature itself was very heterogeneous. Looking back on that era, the relatively uniform standard for Bengali literature--which transcended the boundaries between the medieval Muslim elite, which was foreign, and Hindu court authors--is more obvious than any heterogeneity of Bengali literary styles. Any author of Central/West Asian or indigenous ancestry who chose the Bengali medium--and the choice itself was controversial (Haq 1957: 115)--accepted the common literary standard.

Of course, this standard itself evolved, gradually including an increasing amount of Perso-Arabic influence at least at the lexical level. But it marked a new era of heterogeneity when dobha\s>i Bangla emerged. Ironically, according to Q.A. Mannan, it was fifteenth century Hindus at court who introduced a self-consciously mixed literary style, dobha\s>i Bangla.2 It was the coexistence of this dobha\s>i style with another, purportedly dominant and freer of Perso-Arabic admixture, which ended the era of uniformity. Again, by "uniformity" I mean only that before this point Muslims had joined with local poets in using a common standard and the possibility of a polyvocality or diversity based on differential degrees of borrowing from Perso-Arabic had not been realized. However, it is important to keep in mind that when dobha\s>i Bangla arose, it by no means served as an index of the author's own communal identity. Rather, writing in dobha\s>i might actually have served to mark a religiously Hindu identity embedded in a stance of loyalty to the Mughal state. Following Oberoi's new history of Sikhism (1994), we should be cautious about projecting backwards into the medieval era communal identity-boundaries (linguistic or otherwise indexed). If Hindus at court used a Persianized style at times, Muslims were also known to write Vais>nava poetry (Haq 1957: 51, Dimock 1967). In sum, no form of Bengali had communalist connotations, since communalism per se emerged later.

Complementary schismogenesis arising out of the colonial impact

The colonial impact was great and complex. The Orientalists at Ft. William College--Hastings, Jones, Gilchrist, and Hunter--at least paid lip service to the goal of reviving purer, older forms of India's religions. Analogously, their approach to the language probably combined some desire to Anglicize with stronger conflicting tendencies either to purify and standardize Bengali or to romanticize and glorify vernacular forms (Kopf 1969). At any rate, there arose at the College a class of pundits whose Bengali self-consciously eliminated borrowings from Islamicate languages.3 Sanskritization of Bengali proceeded apace. Not surprisingly, since this s;a\dhu bha\s>a\ differed from the common speech of Muslims in particular, there arose in reaction a Muslim form of speech--i.e. an form called "Musalmani Bangla" took increasingly distinct shape. So, Hindus under British influence took linguistically puristic steps, which provoked counter-steps by Muslims to "purify" Bengali of Sanskritic influence if that were possible. The anthropologist Gregory Bateson called this sort of mutual differentiation process "complementary schismogenesis," which is a process of differentiation in the norms of individual behaviour resulting from cumulative interaction between individuals [or groups]É [or the study of] the reactions of individuals to the reactions of other individuals" (1958: 175).4

The preceding discussion entails a composite historical picture summarizing studies by Kopf (1969), Mannan (1966), Haq (1957), and S.K. Chatterji (1934) at face value. What of their representations of history? From our vaunted position, of course, their interests or advocacy-stances are clear. Chatterji often railed against the 19th century "defiling" of Bengali in Musalmani literature; Haq simultaneously served as a Bengali apologist within Pakistan and an Islamist revisionist among historians of the language; and Kopf advocates the relative virtue, among colonialists, of the Ft. Williams Orientalists vis-ˆ-vis the later Macaulay.

For scholars to take such stances is natural, and only problematic to the degree that the reasons for such advocacy are opaque, arising out of covert ideological commitment. But there is another problem to the historiography as well-- its neglect of synchronic class-based or situational linguistic variation. Although the historian Benedict Anderson (1991) demonstrates the richness of a historiography attuned to synchronic linguistic diversity and the shaping role of "print capitalism" in the emergence of nationalism and other forms of identity, his work is exceptional and probably reflects the recent "linguistic turn" in history and the social sciences. Admittedly, historians of Bengal have not missed religious or ethnically-based variation in language. They have, on the other hand, missed the social facts which occupy sociolinguists, particularly class-based linguistic diversity. (Unfortunately, the inverse is largely true of those sociolinguists who have written on Bengali diglossia; their studies of the ways Bengali reflects differences of class and literacy have ignored religious factors. Moreover, Chatterjee's and Singh's papers are ahistorical.)

This paper can be read as an exploration of the ways in which an author's own class-status influence the shape of the research and writing. In this regard I repeat that such positionedness is only dangerous to the academic enterprise insofar as it is unacknowledged and exercises covert ideological influence. Here I offer only a foretaste of the illustrations provided below, the example of the historians of the Bengali language. Why would the Pakistani apologist Haq and the anti-Musalmani-Bangla scholar S. Kumar Chatterji contribute almost equally to the denigration of Musalmani Bangla literature.5 Might there, then, be more at stake here than communal identity?

Bengali diglossia: Class and literacy-based linguistic diversity

Another literature exists, complementary to the historical literature on Bengali. My sample consists only of three papers published in 1986 by Suhas Chatterjee, Dil, and Singh. All three describe the sociolinguistic phenomenon of "diglossia" in Bengali, taking a synchronic perspective on linguistic diversity. If the historians of language, in exploring competing religious tugs on the Bengali language, seem blind to situational and sociological variables shaping linguistic variation, these students of diglossia have omitted the diachronic perspective, ignoring evidence of conflict and communalism. These omissions are all the more noticeable when the sociolinguists' own data evince conflict along class and communal lines. A focus on historic struggles and competing ideologies must come together with the sociolinguists' discussion of synchronic variation. The sociolinguistic view adds specificity to the historical, and the historians place linguistic variation in larger contexts.


"Diglossia"--a conceptual overview

Every speech community includes variation yet contains it within common norms of use of its linguistic resources-- codes, dialects, registers, and styles. Some speech communities are characterized by a particularly stable relationship between variant forms of a language, one in which one variety adheres fairly closely to a literary tradition and derives prestige from it while another recognized spoken variant is considered inferior but intimate. Charles Ferguson (1959) called this hierarchical relationship between two variants within a speech community "diglossia." It entails a structural and functional gap between a Hi and a Lo variant, in which no one speaks the Hi natively. Diglossia, in his words, is:

a relatively stable language situation in which, in addition to the primary dialects of the language (which may include a standard or regional standards) there is a very divergent, highly codified (often grammatically more complex) superposed variety, the vehicle of a large and respected body of written literature, either of an earlier period or in another speech community, which is learned largely by formal education and is used for most written and formal spoken purposed but is not used by any sector of the community for ordinary conversation (1959: 336)

Used in official contexts-----Used in domestic contexts
Formal style-----Informal style
Must be taught in conjunction with literacy-----Acquired naturally by children in spoken form
No native speakers-----Many native speakers

Characterizing Bengali variation

How might the diglossia concept illumine the Bengali situation? This "etic" or analytic concept does capture something more than does the common "emic" (insiders') distinction between s;a\dhu bha\s>a\ and calit bha\s>a\ . Actually, as Tagore and Dimock pointed out in their times, even the distinction between s;a\dhu bha\s>a\ and calit bha\s>a\ is becoming blurred. "Diglossia" captures more of the dynamic relationships between class, literacy, and spoken language characteristic of Bengali at least since the turn of the century. The concept is thus more useful than other visions of the Bengali situation.

Still, whereas Ferguson's classic examples of diglossia were neatly divided into two "vertically" distinct forms as the term suggests,6 the Bengali situation is more complex. One problem with representations of Bengali in the sociolinguistic literature concerns the pervasiveness of certain s;a\dhu ("pedantic") features among all rural speakers, including the illiterate. Specifically, rural Bengali speech almost universally preserves the verbal stem suffix /-i/. Among the 1986 papers, only Suhas Chatterjee's does justice to this fact by categorizing certain feature sets as Low s;a\dhu bha\s>a\, LSB. LSB, he says, preserves the long verb stem but lacks Sanskritic lexical elements, substituting simple verbal nouns and adjectives for derived abstract forms. Thus Suhas Chatterjee's model--by adding a second, horizontal (historical-grammatical) dimension of contrast to the vertical (prestige)--more adequately captures the relation of Hi and Lo forms to calit and s;a\dhu bha\s>a\ :

Suhas Chatterjee's Model of Bengali Diglossia:

"Hi" and "Lo" forms of Bengali with subvariants

Diglossic status Bengali designation A Bengali designation B
Hi High s;a\dhu bha\s>a\ High calit bha\s>a\
Lo Low s;a\dhu bha\s>a\ Low calit bha\s>a\


High s;a\dhu bha\s>a\ ("pedantic language")-- HSB
Literary code, heavily Sanskrit in lexicon, employing processes of derivational morphology not used in other variants (Chatterjee 1986: 296), i.e. to create abstract and technical vocabulary suitable to Hindu philosophical discourse

Low calit bha\s>a\ ("current language")-- LCB
(Note: calit is commonly called calti )
"The Calcutta colloquial"; verbal forms shortened by deletion of the /-i/ suffix, lacking the lexicon produced by the derivational processes referred to above

Low s;a\dhu bha\s>a\ -- LSB
Identical to HSB in verbal forms (long), but lacking derived lexicon

High calit bha\s>a\ -- HCB
Combines shortened verbal forms with the derivational lexicon of HSB

Testable claims of diglossic variation in today's Bengali

From the viewpoints of the ethnography of communication or correlational sociolinguistics the 1986 studies have methodological problems. None of them used what those schools would recognize as an adequately empirical method, neither recording representative samples of speech in everyday contexts nor recording hundreds of participants whose speech attitudes and behaviors are elicited through the subtle methods of William Labov. Thus, I regard the 1986 claims as deserving of testing but by no means demonstrated. Specifically, the question of how Bengalis actually speak and write, and the extent to which those actual productions can be neatly characterized in either Ferguson's or Chatterjee's terms, remains unanswered. What follows, therefore, are claims which should be investigated. These claims roughly match my sense of the observable situation, and thus they are presented without quotation marks or other hedges, but my disclaimer should be kept in mind.

No one today speaks the Bengali of Carey's Bible or even of newspapers like the daily Ittefaq; if some can and do give such literate forms voice, the requisite skill is not learned at home but in school. Only in school do rural Bangladeshi students from illiterate families learn H varieties. In fact, even wealthy Dhaka students whose families speak a prestigious calit must learn s;a\dhu bha\s>a\ in school. Thus children acquire one form naturally and learn others in school. The gap between written and spoken Bengali far exceeds that in, say, American English. The gap between the "H" and "L" variants in a classically diglossic situation as described by Ferguson certainly obtains in Bengali. Chatterjee, citing Ferguson's nine parameters of a diglossic situation, finds that H and L codes in Bengali diverge in all parameters except phonology.7 At any rate, H and L Bengali show complementarity of function, a hierarchy of status ascribed even by those who do not command H, and literary heritage almost exclusively in H.8

Diglossic values enforced in school

The diglossia concept does not preclude the possibility of "code-switching" among diglossic variants, even within a given stretch of discourse or a single sentence. It does, however, predict that mixing will be judged negatively. That seems to be the case. Because the gap is great or at least ideologically salient, mixing H and L forms draws attention. Singh claims teachers mark down those students who mix calit and s;a\dhu bha\s>a in school writings (1986: 433). Thus teachers enforce a kind of linguistic purism in the interest of maintaining the stratification of forms which characterizes diglossia per se.

The linkages between diglossia and particular forms of social structure and practice make diglossia a heuristic notion, providing a perspective on Bengali culture complementary to that presented in the historical literature reviewed above. The notion of diglossia helps us reexamine the relation of language to religion and ideology, and the relationship of literacy to status. Chatterjee's two-dimensional model of Bengali "diglossia," in moving the concept to a higher level of complexity, problematizes its basic claims. Whereas Ferguson's model would predict that prestige should attach to archaic forms evocative of a literary tradition, Chatterjee rightly points out that Bengalis (often rural in origin) whose speech preserves some long s;a\dhu forms do not thereby gain prestige.

Chatterjee's analysis of Bengali diglossia is much more satisfying than a simplistic equating of s;a\dhu with H. Still, the actual expressions by which urban HCB speakers evaluate L varieties deserves more attention. Dhaka elites regard their own speech-- and Hi s;a\dhu writing-- as unmixed or pure. During my first days in Bangladesh, my Western-educated friend, Siraj, described the speech which Chatterjee calls LSB as a degraded mixture of archaic-literate grammar with a lexicon suited to the poverty of the speakers (my paraphrase). Siraj's objections point to the conflicting prestige values associated with the blended elements of rural speech-- H grammar and L lexicon. Diglossia is largely about language attitudes linked with class and literacy. Chatterjee's complication of the concept only adds to its usefulness in modeling dimensions of Bengali sociolinguistics. Still, at least in relation to Bengali, no model has been developed which stretches the notion of diglossia to incorporate its relation to religion and ideology.

The mystification of Bengali diglossia--Language ideologies in academia and language planning

The obfuscation of diglossia-religion links by 1986 sociolinguists


I have reviewed two of the literatures describing the Bengali language. Reading the Bengali diglossia literature after the history of Bengali language and religion is like being on the other side of a mirror. The opacity is striking. To anyone familiar with the history of the subcontinent, the relationship of communal-linguistic heritage to diglossia is plain. a) Religion and power are clearly linked, as are b) literacy and status; crucially, the two pairs (a) and (b) have also been linked in the history of Bengal. Yet even when data being cited are filled with communalist sentiments, the relationship of a literacy and class-based language gap to the competing civilizational forces at work in Bengal-- Hindu, Muslim, and Euro-colonial-- is obscured in the 1986 papers.

Is a Sanskritic lexicon a good index of H? All evaluations are positioned

A good example to start with is the place of Sanskrit ta\tsa\ma\s in Bengali varieties. Chatterjee and many others have seen Sanskritic lexemes as an index of the "pedantic," or H, s;a\dhu bha\s>a\. H's use of Sanskritic roots has been considered diagnostic even by those working before or apart from the diglossia concept. Yet, the equation of Sanskrit with prestige is problematic. Haq, in his role as revisionist historian and apologist for Bengali, was constrained to emphasize those figures in its history who most effectively used non-Sanskrit borrowings. "The Islamic atmosphere created by Nazrul in his poems, by the use of new words from Arabic and Persian, as well as the re-creation by him of Islamic ideas in Bengali and the use of similes and metaphors that remind of Muslim classics in other languages, are closer to their spirit, set in a new fashion in Bengali poetry" (Haq 1957: 213f). Haq and others advocated the self-conscious development of a standard which turned not to Sanskrit but to Islamicate languages for its prestigious importations.9

Here we see the blindness of these historians of Bengali to the elitism so intertwined with those imaginations of community which draw heavily on tongues distant in space or time (Anderson 1991). Religio-ideological motivations for importing obscure words into Bengali are not subjected to serious theorization in the diglossia literature. Yet, surely the play of various religious traditions is a key facet of the diglossia which Singh (1986: 431) finds in all the Indian languages. Given the importance of elite nurture of Sanskrit and Persian literary traditions in Bengal and much of the subcontinent, religiously motivated linguistic "reforms" have directly shaped diglossia.10 It is therefore remarkable that the tug-of-war between Sanskrit and Perso-Arabic should be neglected in the Bengali diglossia literature. Perhaps the thundering silence of recent sociolinguistic writings by Hindu and Muslim alike on this tug-of-war indicates the degree to which diglossia is a sensitive issue in a continent which has reason to fear ongoing communal violence. Fear and violence frighten away insight; thus fear fuels ideological distortion. For this or other reasons, academic voices fail to acknowledge that Bengali supersaturated with either Perso-Arabic or Sanskritic borrowings is not understood without specialized training. No one grows up speaking such H registers and few ever grow comfortable with them.11

Afia Dil on diglossia and language planning

What Afia Dil (1986) lacks in terms of an analytic model of Bengali diglossia comparable to Chatterjee's is compensated by her empirical data on diglossia in the spoken Bengali used in Bangladesh. Among the 1986 sociolinguists, Dil makes the most effort to situate diglossia within history. Her narration of the evolution of language practices and policy in the East Pakistan period evinces a perspective which adds greatly to the historical summary of Pakistan I presented above. Dil provides insights into the gap between the recommendations and actual practices of language policy makers in East Pakistan and later Bangladesh. Yet she fails to make explicit the ideological underpinnings of the historical actions she describes and set her own position apart from the communalist ideology of her bureaucratic subjects. Thus, despite her contributions, Dil exemplifies the mystification of history and ideology which I argue characterizes the 1986 papers.

After partition, the provincial government of East Pakistan appointed the East Bengal Language Committee with Mawlana Akram Khan as its Chair (Umar 1970: 275). Those were heady days, with Bengali Muslims feeling liberated from the superposed Calcutta standard in which they could never become fluent (Chowdhury 1960: 75). The policy goals which the Committee promulgated, summarized under the banner s;ahaj ba\m>la\ (Simple Bengali), were as follows: "i) that the Sanskritization of the language be avoided as far as possible by the use of simple phraseology and easy construction. . ., ii) that the expressions and sentiments of Muslim writers should strictly conform to the Islamic ideology [sic]; and iii) that the words, idioms and phrases in common use in East Bengal especially those in the Puthi and the popular literatures be introduced in the language more freely." [After this direct quote, Dil goes on to paraphrase other goals: (iv) to simplify the orthography and v) simplify the grammatical metalanguage, involving a decision paralleling #1 that] unintelligible technical terms of Sanskrit 'be substituted by [sic] the simple non-technical terms of Bengali language'" (Dil 1986: 454, citing and translating Chowdhury).

Many of the fascinating bits of discourse presented by Dil call for a level of analysis not provided in her article. By failing to analyze the ideologies underlying the discourse of the well-known language planners quoted, Dil not only leaves important stones unturned but gives the impression that she shares those ideologies. First, in regards to the third goal above, note the ambiguity surrounding the phrase "in common use." In the context of that goal as written, the "use" in mind is in "popular" literature. However, we might question the extent to which the "idioms and phrases" of the Puthi and Musalmani tracts were indeed "in common use" even in the speech of regional Muslims, let alone their non-Muslim neighbors.

The second example is the partially translated discourse of Principal Ibrahim Khan, a member of the East Bengal Language Committee.12 In the passage Dil cites, Khan echoes the EBLC's call for some sort of "simplification" of Bengali. But Khan and Dil collude in confusing linguistic realities and ideologies. Khan's use of "we" exemplifies the way pronouns facilitate the creation of imagined communities when he writes, "we shouldn't even say /khaibo/ we should say /khabo/, for that is what we say ordinarily. /khaibo/ is the s;a\dhu bha\s>a\ " (Dil 1986: 455).13 If by "we" Khan meant "we the elite who still speak Hi Calit Bengali (even after a political breakoff from its Calcutta source), then it is true that "we" do not "ordinarily" say /khaibo/. The implicit generalization that the stem suffix /i/ is absent in all "ordinary" speech, however, is not true of LSB-- the speech of most Bangladeshis. This confusion of facts clouds the claim of Khan to be resisting a "conspiracy of the educated class [which despises] 'the dialect of the common man'" (Dil 1986: 456; internal quote is Khan's).

Let me provide some background for the third example. In the Pakistan era, the Bengali Academy undertook intensive projects in dialectology and lexicography. One result was the three volume Dictionary of East Pakistani dialects (Shahidullah 1965). Visions for the replacement of the Calcutta standard were not homogeneous even among the projects' leaders. Shahidullah noted the seeming dissolution of the single standard after partition; he did not see the vacuum being filled by any one dialect, e.g. that of Dhaka. He and others involved in the project hoped "his dictionary might be instrumental in bringing about a national linguistic cohesion among the people with various dialectical backgrounds" (Dil 1986: 460). The communalist dualism of the director of the Bengali Academy, Syed Ali Ahsan, who wrote the preface to the dictionary, seems more strident than Shahidullah's. Ahsan's preface attacked earlier Bengali dictionaries which theirs was to displace at least in the dialectology of East Bengal. Dil's indirect citation of Ahsan shares his ideology in a way that seems dangerously unconscious or inexplicit, illustrating the tendency of all indirect discourse to blur the line between two authors:

[He objects that] dialectal wordsÉhad been left out as being asa\dhu 'incorrect.' Moreover, the dictionary writers had, without any twinge of conscience, not only left out the Perso-Arabic words used by the public but they had also entered many Sanskrit words in the dictionaryÉ [as an] attemptÉ to use 'Panditi Bangla' or scholarly language which was more often than not the Sanskritized Bangla (Dil 1986: 459).