1.0. Introduction

The present book is a humble attempt to treat Bangla phonology in the metrical framework. As far as the classical (segmental) generative phonology of Bangla is concerned, Paul (1985: 1) observes correctly that it is a 'virtually unexplored' field with only more or less half a dozen published papers. While this situation may cause concern, the later development called metrical phonology has not yet touched the field of Bangla linguistics at all. The field of Bangla metrical phonology is absolutely 'unexplored'. And this very fact itself is justification sufficient for the present study.

As the pioneer work in an unexplored field the present book, however, is obliged to perform the necessary groundwork by working out the steps that lead from our present (segmental generative) understanding of Bangla phonology to the new approach. As a result, the present study, with its limited scope, is bound to be less ambitious in terms of a wider application of the methods and tools of metrical phonology to Bangla.

As far as the groundwork is concerned, I have tried to deal with at least three aspects of language relevant for a discussion of metrical phonology of Bangla in chapters 2, 3, and 4. The first and the second of these three aspects are related to the analysis of Bangla speech, while the last one is related to the field of Bangla writing.

Chapter 2 aims to identify the characteristic features of one cardinal concept of metrical phonology, viz. the syllable, as they are available in the existing literature of Bangla linguistics.

Chapter 3 provides a segmental treatment of a representative domain of Bangla phonology, viz. verb morphology, in the classical generative phonological framework. The limitations of this very framework gave rise to the theories of metrical phonology.

Chapter 4 presents a full fledged indigenous theory of Bangla metrical phonology first propounded by Prabodh Chandra Sen, a systematic thinker, in the period from 1922 to 1986 and available in the research tradition of Bangla literary studies.

Chapter 5 goes beyond the initial groundwork and sketches a treatment of Bangla syllables in a metrical framework.

In the present chapter section 1 will introduce the language, viz. Bangla; section 2 will comment on the model used; section 3 will deal with the transcription employed here and lead up to the concluding section.

1.1. The language

Bangla, also known as Bengali, belongs to the eastern branch of Indo-Aryan. The language, as is shown in the 1991 Census, the most recent one of India, is spoken by 58,541,519 speakers in West Bengal (88,752 sq km) plus 11,054,219 other Indian speakers. Apart from the Indian data it is also spoken by an estimated 107 million in Bangladesh (143,998 sq km) in 1991. As per the present language situation Dasgupta (2003) comments that apart from the diasporic speakers for whom no systematic figures are available Bangla has a total of 177 million subcontinental speakers in 1991.

As per the language status, in Bangladesh, Bangla is the sole national language; whereas in India, it is one out of eighteen national languages.

Two very prominent characteristics of the language are as follows:

Firstly, Bangla cuts across the political boundary between India and Bangladesh, formerly known as East Pakistan.

Secondly, Bangla cuts across the religious boundary between Hindus and Muslims. In a secular country like India a language cutting across several religious divisions is a quite common phenomenon. But in this regard Bangla presents a slightly different picture. In the case of Bangla, unlike other languages, in India the majority of Bangla speakers is Hindu, who for religious purposes uses Bangla only as subsidiary to Sanskrit; whereas in Bangladesh the majority of Bangla speakers is Muslim, who for religious purposes uses Bangla only as subsidiary to Arabic.

In both the countries, however, Bangla has a good literary tradition and the speakers are deeply attached to their language.

As a result of all these reasons, viz. a wide geographical distribution, political boundary, religious division, national language status, deep attachment, a rich literary tradition, and the speaker-strength, some sort of parallel standardization, which may be termed as Kolkata Standard Bangla and Dhaka standard Bangla, came into being in the language situation. Ray et al. (1966) recognize these two parallel standard dialects, viz. the standard colloquial Bangla of Calcutta, presently known as Kolkata, and that of Dacca, currently spelt as Dhaka, the capital of East Pakistan. Now with the politically independent existence of an almost twenty-five year old Bangladesh the Dhaka standard colloquial has grown as powerful and rich in every respect as the Kolkata standard colloquial.

The Kolkata Standard Colloquial Bangla, hereafter KSCB, is the main concern of the present study.

At this point, however, I would like to raise a question regarding the identity of this KSCB. In this first decade of the 21st century what do we understand by KSCB? Or more precisely, what is KSCB composed of?

The scope of the present book does not provide any suitable place for an answer to this question; but the question strikes me as a major one for the following reasons:

Firstly, those of us who live in Kolkata are very much aware of the two types of migration, viz. intra-country and inter-country, into this very city. Projects on different aspects of such overwhelming and continuous migration are taken up. But an investigation of the linguistic impact of such migration is yet to be taken up.

Secondly, within Kolkata itself remarkable variations are evident in the intuitive judgements about Bangla forms.

Thirdly, these variations of KSCB have not yet been captured systematically in any socio-linguistic research project.

In the present study, however, I neither have an answer to this question nor do I wish to find out one. But as I feel this question to be a relevant one, I wish to unfold the question itself a little more.

In Bangla, a language with a diglossic situation, Chalit is the name for the spoken standard colloquial or the low variety; whereas Sadhu indicates the high variety. According to Ray et al. (1966: 2) "in the Chalit, the main historical contributor has been the speech of Hooghly and Krishnagore, small towns along the river somewhat north of Calcutta."

And (1966: 3) "other languages also spoken in the area are chiefly Hindi-Urdu and English, each spoken almost wholly in urban areas, where each is also just as adequate for most purposes of life as Bengali is."

Apart from the historically inherited variability and the inter-language influences, which are a natural concomitant of the circumstances described above by Ray et al. (1966: 3), the present day KSCB has experienced many other linguistic and non-linguistic forms of acculturation. I feel that the strongest factor at work has been migration, which happens due to two main reasons, viz. economic, resulting in intra-country (or more specifically rural-to-urban) migration within West Bengal, and political, resulting in inter-country or Bangladesh-to-India migration.

I shall assume the inter-country migration to be more relevant than the other type for the following reasons:

Firstly, some sort of record of this migration is available.

Secondly, these migrants used to consider the Dhaka dialect as their standard variety, a fact that has socio-linguistic relevance.

Thirdly, the contribution of this migrant group is more than the other in different domains of this state, be it the West Bengal politics, or the literature, or the elite society culture. This, however, is a subjective opinion.

Chakrabarti (1990), a study of the refugees and the left political syndrome in West Bengal, provides the government records of this migration available till 1952. According to Chakrabarti (1990: 1) this migration "started before partition with the Noakhali riots of 1946 and continues down to this day. The migrants came in waves. At times the migration hit West Bengal in the shape of formidable breakers and often subsided into a trickle. But except for a very brief period, it never stopped."

Chakrabarti (1990) indicates that the first phase of migration took place between 1946 and 1949; the second phase between 1950 and 1951; the third phase during 1952; and the last phase began after 1960-61 and continues till date.

On the basis of the records of the rate of admission to Government camps Chakrabarti gives an account of at least 2.8 million migrants from Bangladesh to India during the period between June 1948 and 1971. Out of these 2.8 million at least 1.3 million entered West Bengal during 1961 -1971 and of the remaining 1.5 million also a high proportion entered West Bengal.

Apart from the recorded migrants unrecorded migration too is there. According to Chakrabarti (1990: 3) "the refugees who sought shelter in Government camps represented only a small fraction of the total influx."

Even after this study, viz. Chakrabarti (1990), since 1991 till the present day, with the changes in the political scenario of Bangladesh, unrecorded migration into India is taking place.

In West Bengal, Kolkata, the centre of the KSCB area, always received the largest proportions of the migrants. These millions of migrants, speaking the same language but different dialects, obviously contributed strongly in shaping the present day KSCB. This contribution is not yet quantified. But it is necessary to undertake such research. It is high time we raised the question of the identity of the present-day KSCB, and to describe the variability within the norm.

I, however, leave the problem of variable norms open for future researchers and prefer to work with non-controversial data from KSCB as much as possible.

1.2. The framework

As the title of the book suggests the present study is done in the metrical framework. But such a broad statement excludes many details of the framework aspect of this study. In fact, the framework of the present study may be discussed with due consideration under two subsections, viz. the phonological framework and the framework of the indigenous research work.

1.2.1. The phonological framework

In the wake of Chomsky and Halle's (1968) The Sound Pattern of English, henceforth SPE, till now the most prominent landmark in the field of phonology, the field has witnessed a variety of competing frameworks. A diagrammatic representation of the relationship of a few of these models is given below; an expository outline follows; this material is based on Durand's (1990) account:

Phonology
__________________________

structuralist / generative phonology
classical phonemics ___________________________

linear phonology non-linear phonology

(SPE; Concrete; (Autosegmental;
Underspecification Metrical; CV
theory; Lexical phonology;
phonology etc.) Dependency phonology
etc.)

In classical phonemics the phonemes of a language are identified in terms of more or less four central criteria: opposition, complementary distribution, phonetic similarity, and free variation. The structuralist works mostly give a list of allophones beside each phoneme and use the term phone for a sound unit whose status is not yet decided.

By contrast, generative phonology, according to Durand (1990: 5), "is committed to a programme where the primitive terms and the rules for their combination and transformation must be couched in symbolic notation and formally defined. Thereby, an important step is taken towards making the system easier to fault, or to refute, and our description should be highly valued according to a common view of what defines a scientific enterprise."

Within the generative tradition the earlier models depicted the phonological representation as a linear arrangement of sound-segments on which the phonological rules operate. Such models may be termed as the linear phonology.

The linear generative tradition was initiated by SPE, the monumental work by Chomsky and Halle (1968). In brief, SPE offered, in the words of Durand (1990: 1), "a theory of internal structure of sound-segments, a theory of levels and derivation, a theory of link up between syntax and phonology." Afterwards, however, many of the fundamental aspects of SPE was questioned in a variety of models, e.g. the internal structure of sound-segments, especially the feature-specification is questioned in the underspecification theory; the abstractness in the underlying level is questioned in the Natural Generative Phonological theory and in the Concrete theory; the role of the different types of boundaries and the morphological brackets involved in the link-up between syntax and phonology is questioned in the Lexical Phonological model and so on.

The latter models mainly questioned the different aspects of the linear representation of phonological strings. According to Durand (1990: 2) "they all share the belief that phonological representations need to be much more articulated than traditionally assumed and that a number of phenomena (particularly, but not exclusively, stress and tone contours) cannot be appropriately accounted for if phonological representations are limited to string-like arrangements of segments and boundaries." Non-linear phonology, a cover term for all such models, is quite popularly used these days.

Among the non-linear models the autosegmental framework introduces the notion of CV tier between segments and the suprasegmental hierarchy. In fact, the most convincing examples of the autosegmental model come from the tone languages.

The metrical framework goes beyond segments and recognizes the syllable as a unit and studies the syllable-constituents and the syllable as a constituent of higher-order units.

The dependency phonology, following Durand (1990: 277), recognizes "not only the syllable as a structured unit ….but also relationships between syllables."

In the current study, the segmental treatment of Bangla verb morphology, a representative domain, is done in accordance with one of the linear generative models, viz. the concrete model, proposed by Kiparsky (1968b) and applied to Bangla by Paul (1985). The concrete model, as it is designated in Paul (1985), is a modified version of the standard theory of SPE in the direction of restricting abstractness of the underlying representations. The framework will be more fully described in chapter 3.

A non-segmental treatment of some aspects of Bangla phonology is done in terms of a non-linear model, viz. the metrical model. The metrical framework, mainly questions the segmental treatment of English stress as proposed in SPE. SPE assumes stress to be a property of an individual segment and assigns stress in terms of sequences of particular segments.

The SPE based segmental approach to stress is first questioned in the paper named On stress and linguistic rhythm by Liberman, M. and A. Prince (1977), hereafter LP, published in Linguistic Inquiry 8. LP (1977) pick out some seven properties of the SPE-based theory of stress because of which, on the one hand, it deviates from the rest of the generative phonological theory, and on the other hand, it prepares a field for a new theory. LP propose the theory of metrical phonology that handles English stress in accordance with the relative prominence of syntactic and morphological constituents in terms of a hierarchical organization of phonological strings based on syllables. Beyond the syllable the bigger units in the hierarchy are foot, mot, and word.

LP (1977) is the first major work in the field of metrical phonology. From 1977 onwards, mainly in the pages of Linguistic Inquiry the present-day metrical theory developed gradually on the basis of a good number of constructive criticisms of the various aspects of the theory proposed by LP (1977). Among the chief contributors to this theory mention may be made of Kiparsky, Clements, Feinstein, McCarthy, Nespor, Schane, Selkirk, Ingria, Leben, Prince, Hayes and so on.

Afterwards the themes of this well-developed metrical framework are summarized in a textbook fashion in Hogg and McCully (1987).

Following Hogg and McCully (1987) the present book analyses the various aspects of Bangla syllables. It studies the internal hierarchical structure of Bangla syllables and the conditions thereof in terms of the concepts of sonority hierarchy and syllable template; problems of syllable boundary in terms of the concept of ambisyllabicity; and the predictable syllable stress in terms of metrical grid respectively in chapters 2 and 5. The present study restricts itself to the level of syllable.

1.2.2. Indigenous framework

The research tradition of Bangla literary studies presents a good deal of metrical analysis of the Bangla verse language. Among the contributors to this field mention may be made of Satyendranath Dutta, Rabindranath Tagore, Dwijendralal Roy, Dilip Kumar Roy, Prabodh Chandra Sen, and Amulyadhan Mukhopadhyay of the initial stage. Among them some are poet metricists while the others are grammarian metricists. Both the types of metricists, however, share a common feeling that the grammar of Bangla metrics should be formalized independently of the Sanskrit tradition because of the simple reason that the Bangla sounds do not behave like those of Sanskrit, which are often regarded as their historical counterparts. Among the metricists Prabodh Chandra Sen is the first person to develop a systematic metrical theory of Bangla verse language on the basis of his own extensive research work extending from 1922 to 1986 independently of both the Sanskrit and the Western traditions.

Sen's metrical theory consists of a two-level analysis, viz. (in his own terminology) the phonological level and the morphological level. The phonological level treats the syllable as a unit and studies its internal structure and methods of quantification. The morphological level studies the various combinations of syllables resulting in a higher level unit, viz. foot. In other words, the phonological and morphological levels may be termed as the levels of syllable formation and foot formation respectively.

The present study limits itself to the findings of the phonological level of Sen's theory which are more fully illustrated in chapter 4.

1.3. Transcription

In this book the transcription convention of Ray, Hai, and Ray (1966) have been followed. Some of the key symbols are described as follows:

T= voiceless unaspirated retroflex stop
D=voiced unaspirated retroflex stop
R= retroflex flap
E= front mid-low vowel
O= back mid-low vowel
e= front mid-high vowel
o= back mid-high vowel
Y= front mid-high semivowel
W= back mid-high semivowel
y= front high semivowel
w= back high semivowel
N= velar nasal
S= palato-alveolar sibilant
M= nasalizes the nucleus preceding it

Single quotes are used for real glosses and double quotes are for literal renderings of morphemes.

The rest of the symbols are not listed as they agree with the IPA conventions followed by Ray et al.

1.4. Conclusion

I wish to conclude the present chapter by saying a few words about the motivation behind the present study. The motivation is supplied chiefly by the frameworks used here. To be more specific, the dissimilarity and similarity between the two independently developed metrical models motivated this book to take shape.

The similarity lies in the methodology of analysis of these two models. Both of these, viz. the current theory of metrical phonology and the indigenous metrical theory, offer a syllable-based phonological analysis capable of handling suprasegmental features like stress, pause etc.

The dissimilarity comes from the temporal point and the social background of these two models. The indigenous Bangla metrical theory with an underprivileged social background of a third world country is chronologically much ahead of the current theory of metrical phonology with a western social background as the former started taking shape since 1922 while the latter in the late seventies.

These similarity and dissimilarity, however, served as justification enough for undertaking the present project.