The rethinking of language

The rethinking of language

Probal Dasgupta

[The Visvabharati Quarterly 48:1-4.138-50. 1983.]

[An updated version in Bangla appeared in 2011 in connection with Tagore’s sesquicentenary. But this is the original 1983 English text.]

Today’s intellectual world has set itself, among other urgent tasks, the task of putting together a general, realistic and significant picture of language. To this end, scholars have been redirecting their language-studying energies. An attempt to place Rabindranath Tagore’s writings about language in this perspective may produce interesting results. His thinking changed between 1885 and 1938 in ways which form part of the wider rethinking and are usefully seen in relation to it.

I have mentioned three research goals. Let me go over them again.

Generality calls for theories that cover large amounts of data and express systematic similarities between facts.

Realism requires that linguistic analyses correspond as closely as possible to the complex existence of language in real life.

Significance, in this context, means success in unearthing simple and basic abstract principles of language which, acting in various combinations, have the effects which we see as complicated patterns in real languages.

The need for such mobilization of care and thought arises, of course, because language is rich in forms and patterns. That is to say, it varies. People say the same thing differently. Or they say different things. Some aspects of this variation let you tame them. Other aspects remain wild.

Taming variation or leaving it wild. What is this Taming supposed to mean? Two quite different answers sound right.

One answer occurs to modern linguists. “If you understand how much something can and does vary, you have tamed that variation.”

Another answer occurs to many people. “If a standard dialect replaces (completely or on a part-time basis) a group of dialects, some variation is tamed, and these languages turn into non-standard dialects.”

These answers, linguistics and standardization, though distinct, [p 139:] historically tend to link up in certain respects which we will look at later. For the moment, let us ignore the standardization answer. We ask how linguists have gone about taming variation.

Such taming takes place at three moments. Each moment of linguistics attends to particular sorts of variation, leaving others wild.

Moment One, the grammar-writing of India, Greece, Rome, Arabia and other classical civilizations, attains some mastery, one language at a time. This thinking identifies constant nouns and verbs and their modes of variation – tense, case, number and so forth. Indian thinking goes beyond this. It analyzes derived words into roots and affixes.

           

Moment Two, comparative philology, looks at systematic differences between sister languages, reduces many differences to “laws” of change such as Grimm’s Law, and establishes language families whose subgroups obey such laws of variation and which have developed as splinters (descendants) from original or “proto” languages, e.g. the Indo-Aryan languages from Sanskrit.

           

Moment Three, modern linguistics, works with the credo that the underlying nature of language must be such as to allow, and indeed to require, about as much diversity of languages, over time and space, as we see. The scales along which basic patterns of language can vary are fairly limited. The class of possible human languages is finite in principle. Modern research tries to chart this finite space, identifying the possibilities of motion in it, instead of just asking, as Moment Two does, exactly which of the available options have in fact been taken by languages or how they have moved into other options over the centuries.

           

Moment Two compares languages to see what comes from what. Moment Three surveys languages to see what can come at all, from anywhere. Such surveys indicate what language types can exist, and what linguistic features remain massively constant regardless of type. Moment Three also looks closely at individual languages to see how the character of a language hangs together or fails to.

           

 [P 140:] This gives you some idea of the modes of thinking that go into today’s linguistics. I oversimplify, but harmlessly. To see this, consider one thing I am glossing over. To the extent that language has universals (constant features), what is their form of existence? Chomsky says they belong to the human genotype. Piaget says they arise in each person through constructive interaction with the environment. Greenberg, who has led much of the survey-style work on universals, doesn’t say. But everybody agrees that an invariant core of universals forms part of the definition of human language. So I feel I can safely ignore the debate about the biology of this invariance.

           

Now that you see that this outline of linguistic concerns is not hopelessly oversimple, we can ask how the movement in Rabindranath’s thought from 1885 to 1938 participates in the transition from Two to Three, a transition still under way. It is time now to name our sources. Rabindranath’s early writings on language (1885-1905) collected in SObdotOtto (‘philology’; we will call it ST) and his 1938 monograph baNlabhaSa-poricOY (‘introduction to the Bangla language’; we will call it BBP) both appear in volume 14 of the centenary edition of his works (pp 3-114 and 435-506). All my page references are to the centenary edition. The transcription of Bangla used here is due to P.S. Ray, with E O standing for low or “open” vowels, Y W for mid semi-vowels, T D R for retroflex sounds, N for “ng”, S for “sh”, and aM oM etc. for nasalized vowels.

           

ST concerns itself a great deal with gazing at variations from one language to another in order to discover origins. BBP tries to see Bangla in its own terms, whatever they may turn out to be. This shift of emphasis immediately pushes meticulous historical pursuit of specific lexical items (word forms and meanings) into the background and plays up instead the systematic charting of grammatical elements which – e.g. emphasizers and question words – serve sentences rather than words. No word-histories turn up in BBP; no charting of grammatical items appears in ST. The shift is tho-[p 141:]-roughgoing. And it does not derive from other people’s ideas about moving into Moment Three. Suniti Kumar Chatterji and other linguists known to Rabindranath never left Moment Two. What led him to Three is a different matter. But we must recognize the fact that Rabindranath made it there self-propelled, independently of other thinkers.

In fact, his concern with syntax places him next to Otto Jespersen, among the true pioneers of Moment Three. When Rabindranath died in 1941, the work of Greenberg and Chomsky still lay in the future. Fragmentary though his remarks may be, they nonetheless deal with their subject matter more adequately than professional publications. This says something about how much we professionals have published about Bangla syntax. It would be a mistake to suppose that linguists can routinely collect his observations in obvious descriptive baskets. Let me mention two examples which pose problems for present-day theory.

Rabindranath (BBP 493) observes that “One can say poSur theke manuSer utpotti ‘humankind originates from animals’. But we don’t say manuS theke gOndho berocche ‘the person stinks’ (lit. ‘smell is coming off from the person’), but rather manuSer ga theke or kapoR theke, ‘from the person’s body’ or ‘clothes’. Instead of bipin theke Taka peechi ‘I’ve got money from Bipin’ we say bipiner kach thek Taka peechi ‘I’ve got money “from with” Bipin’. This is because the word theke ‘from’ directly deals only with the name of an inanimate entity. So it rains megh theke ‘from clouds’, but songs come out pakhir kOnTho theke ‘from the throat of a bird’, not pakhi theke ‘from a bird’.”

In modern jargon, the adposition theke ‘from’ is said to select an inanimate complement. This is the obvious way to pick up the observation made by Rabindranath. But, once we say that theke selects the inanimate complement megh ‘clouds’ in megh theke, we remember that a selector usually serves as the governor or head of the phrase in which selection takes place. So we expect theke to be [p 142:] the head of the phrase megh theke. In fact, assuming the rest of current theory, we feel like arguing that theke must head this phrase since it selects its complement.

However, it is unlikely on other grounds that theke heads megh theke. Consider pichone ‘behind’, which apparently does head megher pichone ‘behind the clouds’. You can conjoin it: megher Samne ba pichone ‘in front of or behind the clouds’. But theke stands alone. You can’t conjoin it with hoe ‘via’ and say protap dilli hoye ba theke aSbe ‘Protap will come via or from Delhi’. In this and other respects, theke fails to behave like clear cases of head adpositions. Either such failure means nothing and theke does head its phrase, or theke is a non-head and non-heads may select. Whoever draws the latter inference will need to produce a new theory of heads to match it. In either case, a problem arises. Rabindranath’s observation, made in 1938, is a time bomb, exploding now.

Another observation which modern research cannot incorporate without profoundly revising itself is the following (BBP 499): “In certain uses the pronoun je ‘which, who’ becomes a particle, as in hori je gElo na ‘that Hori didn’t go’. The word je determines the not going. In tini bollen je, aj-i taMke jete hObe ‘he (or she) said that this-very-day he has to go’, the word je virtually fences off the clause taMke jete hObe ‘he has to go’. It also determines facts, not just statements: in modhu je roj bikele bERate jaY ami jantum na ‘that Modhu goes for a walk every afternoon I didn’t know’, the word je attaches itself to the fact of Modhu’s going for a walk every afternoon.

One must read this passage within the Sanskrit-based grammatical tradition which takes it for granted that, in the third person, pronouns are also determiners. The je of je jaY lONkaY Se hOY rabon ‘who goes to Lanka he becomes Ravana’, a pronoun, and the determiner je of je dOrja diye EkTa beRal Dhoke SeTa die carTe-o Dhukte pare, literally ‘through which door one cat gets in, through it four can get in as well’ (in English order, ‘four cats can get in through [p 143:] a door through which one cat gets in’), are normally thought to be the same je. So Rabindranath’s point is that the pronoun-determiner je can play a particle role. A parallel observation occurs on page 489: “Where ki is a particle, it marks a question. It is also used as a determiner along with an understood noun. Thus, tumi ki korcho ‘what are you doing?’, i.e. ki kaj ‘what work’. (…) An example of the determiner function of ki: ki kaje lagbe jani ne ‘what purpose it will serve I don’t know’.”

So the pronoun-determiner ki also works as a particle – to mark, I add for non-Bengali readers, disjunctive or “yes-no” questions like hori ki jabe ‘will Hori go?’  And here Rabindranath goes out of his way to say how ki manages its pronoun-determiner function. When it is a pronoun, it is a determiner next to an omitted, understood noun. This analysis differs from today’s majority view that pronouns are nouns, and coincides with Michael Helke’s dissenting opinion.[1] I will not go into the many reasons for believing that Rabindranath and Helke are right and the majority wrong, as I have done it elsewhere.[2] Once theorists come round to the (correct) minority view of the matter, it will become necessary to recognize the determiner as the centre of reference, in the technical sense. That will undermine the standard logic of nouns due to Frege and Russell or at least its prevalent simplistic interpretation.

This is only the beginning. We still haven’t faced the observation that the determiners je and ki can also function as sentence-level particles, as in modhu je roj jaY ‘that Modhu goes every day’ and modhu ki roj jaY ‘does Modhu go every day?’ Such particles are now called complementizers, because some of them clearly “complementize” their sentence. Thus, in modhu je roj jaY ami Se kOtha jantam na ‘that Modhu goes every day I didn’t know that fact’, je ‘that’ turns its sentence modhu roj jaY ‘Modhu goes every day’ into a complement or subordinate sentence.

One question here is, what enables determiners like je or ki to [p 144:] accept complementizing as a second duty? Surely their ordinary work as determiners has something in common with complementizing. Otherwise we would be prepared to find languages where the word for “fish”, when added to a sentence, can turn it into a question, or where the word for “you” also has the meaning of “that” in “that Modhu will leave”! It is the linguist’s job to find out what the determiner je shares with the complementizer je, or the determiner ki with the complementizer ki.

We are happy to note that the kitchen of modern linguistics lets us concoct at least a fraction of an answer.

Consider the Bangla genitive ending r (similar to English ’s), as in modhur Taka ‘Modhu’s money’. It can parallel the complementizer je. In gitar bacca hOWar khObor jantam na ‘I didn’t know the news of a child being born to Gita’, the phrase gitar bacca hOWar ‘of a child being born to Gita’ behaves like the complement clause gitar je bacca hoyeche in gitar je bacca hoeche Se khObor jantam na, literally ‘I didn’t know the news that a child has been born to Gita’. The final r of gitar bacca hOWar parallels the je in gitar je bacca hoeche.

But the genitive ending also parallels the determiner je. Compare dOrja kholar lok, literally ‘door opening’s person’, that is ‘a person to open the door’, with je (lok) dOrja khulbe Sey lok ‘which (person) will open the door that person’ i.e. ‘the person who will open the door’. Here the phrase dOrja kholar, with r, matches the relative clause je (lok) dOrja khulbe, with the determiner je (which is followed by a noun, overt or understood).

Now, we think we know that the genitive ending r marks a variety of relatedness, which the Indian tradition calls sambandha. Thus, in dOrja kholar lok ‘a person to open the door’, r relates dOrja khola ‘opening the door’ and lok ‘a person’. In bacca hOWar khObor ‘the news of a child being born’, r relates bacca hOWa ‘a child being born’ and khObor ‘the news’. So it does not surprise us to hear that je dOrja khulbe ‘who will open the door’, an analogue to the sambandha phrase dOrja kholar, is called a relative clause, and its je a relative [p 145:] pronoun. Sambandha means relation. But we do need to add a story about the complement clause bacca je hoeche ‘that a child has been born’ and its relative complementizer je. It seems clear where the two sorts of je meet. They both relate a clause to something else. They diverge in ways which are difficult to understand in satisfactory detail. To extend the analysis to the ki pair is even more difficult.

I have attempted, elsewhere, to deal with these problems[3] and cannot yet significantly improve on these earlier efforts. Until someone can, the incorporation of Rabindranath’s remarks into official linguistics will remain incomplete. It should be obvious that a solution to these difficulties will have to invoke new and perhaps far-reaching principles, bringing about a modification of syntactic and logical theory. Just in case it isn’t all that obvious, let me state one aspect of the problem in contemporary terms.

Bresnan’s work on complementizers included the proposal, widely accepted in the seventies, that, in the complementizer-sentence combination, the sentence is the head which the complementizer specifies, just as, people believed, a tense marker specifies its verb phrase or a determiner its nominal phrase. That view uniformly placed grammatical deictics (a deictic element “points at” some place, time or thing) in “specifier” niches. Today, such innovators as Kayne and Chomsky propose instead that the head in the sentence-complementizer combination is the complementizer, and that inside the sentence the tense (and mood) element is the head. This hypothesis promotes two kinds of specifiers to headship.

The claim that Tense-mood heads its sentence clashes with Kayne’s 1980 reiteration of Dougherty’s 1970 reiteration of Chomsky’s 1965 point that Tense-mood is not an immediate constituent of its sentence at any level of structure.[4] And the Chomsky-Kayne claim which concerns us here, that complementizers are heads,[5] is implausible on other grounds which I leave out of the current discussion. Both claims however miss the earlier general- [p 146:] –ization about grammatical deictics. But they have gained some acceptance, for reasons unclear to me. If accepted, they destroy any hope one might have had of relating the function of je and ki in one specifier role (Determiner) to their function in what now ceases to be a second specifier role (Complementizer). Whoever gets us out of this mess will have to have clearer ideas than we do today.

And notice that neither Bresnan nor Chomsky nor Kayne help us understand how it comes about that a Bangla complementizer normally occurs inside, rather than next to, the clause it serves (whatever this service may be – headship or specifiership).

The matters I have been discussing belong to general linguistics as we know it, although they induce specific modifications of theory. But Rabindranath in his monograph BBP also stresses a theme which, if taken seriously, may affect the way linguistics approaches the domain of language. One might call this theme Morphological Gesture. It includes sound imagery which is often expressive rather than iconic – like dhu-dhu to express a flat, empty expanse, or cryptotypes such as Whorf noticed (as in slither, slide, slick). Morphological Gesture also subsumes other means of expression which feel “direct” and bypass the conventional route of referential, compositional meaning.

Rabindranath says (BBP 479-80): “Bangla has another sort of reduplication which carries a semblance of meaning, but which points at rather than refers to. Sanskrit says patanonmukha ‘about to fall’, Bangla says pORo-pORo (pOR ‘fall’). What is aasanna ‘imminent’ in Sanskrit is hObo-hObo (hObo ‘I will be’) in Bangla. Likewise: gElo-gElo, jaY-jaY ‘on the brink’ (gElo ‘went’, jaY ‘goes’). Sanskrit’s avaruddha-svare ‘in a choked voice’ is Bangla’s kaMdo-kaMdo (kaMd ‘cry’). Such Bangla expressions convey not just an idea but, as it were, a picture. (…) My book SObdotOtto, in its discussion of sound-effect words, elaborates on this use of obscure devices to make graphic descriptions clear.”

And the semantics of sentence particles (BBP 496-8/445-6) seems to Rabindranath to involve something similar. One might call it Syntactic Gesture, relating it to expressive use of word order, intonation (which linguists often call Vocal Gesture), and markers of focus or topic status like Japanese wa, mo, Bangla i, o (BBP 499): “There is another particle, i. While o unites, i isolates. Thus, tumi-o jabe ‘you too will go’; tumi-i jabe ‘just you, you’re the one who will go’; Se jabe-i Thik koreche ‘he has decided that he will go’, the going has been definitely decided on.”

Rabindranath repeatedly stresses the prominence of these informal, imprecise and somehow non-rational expressive devices in the ordinary Bangla style which, by the time he was writing BBP, he had come to regard as the only style destined to live, while the high style of his time was – he sensed – on its way out. The language that BBP is an introduction to is this newly self-conscious form of Bangla. Although he wrote the book for young students (BBP 438) as an afterthought to his biSSoporicOY ‘introduction to the universe’, he also saw it (439) as a first attempt to put together a standard picture of the emergent literary language, an attempt which would help give it a definite form. This brings us to an interesting topic.

I had promised to talk about how the two ways of taming variation – standardization and linguistics – tend to link up in certain respects. The crucial point is that the energy that people can use for investigating languages is limited. So they are going to lavish more attention on some languages than on others. And this will follow politically intelligible lines, with only the occasional major deviation for the sake of pure curiosity. Whether scientists like it or not, dialects which are (or stand a chance of becoming) standard will get undue amounts of scholarly attention. Indeed, codifying work is bound to form part of how they rise to and stay in power. Thus, one cannot strictly separate understanding and standardization. Many common goals tend to break such a separation down. Individuals who wish to understand the similarities and differences between certain dialects and those who create or maintain a stand- [p 148:] -ard literary elaboration of the common core of the dialect group are in part the same people doing the same work. Rabindranath is just a distinguished case in point.

I may be giving the impression that ST was all philology and that only late in life did Rabindranath become concerned about the present plight of his language and the need for a well-defined standard form of it, suitable for public (official) use. That would be wildly wrong.

Even the philology in ST was informed by an awareness of the difference between productive, living patterns and the unassimilated borrowings (whose patterns are, as linguists say, psychologically unreal): “Those suffixes which have come in with Sanskrit and foreign words and have no dealings with Bangla words cannot be regarded as Bangla suffixes.”

Here (ST 40) Rabindranath states a principle which modern linguistics rediscovered only in the seventies.

And, anyway, ST contains an essay, ‘Bhasha-bicched’ (79-82) explicitly discussing the need for a standard Bangla which Rabindranath then thought would be suitable for use in Assam and Orissa as well as Bengal (he wrote: “If Bangla were to become the written language of Assam and Orissa, this would be as good for Bangla literature as for those regions”, p. 49). We note with interest that this 1898 view gives way to a recognition, in BBP (1938), of the possibility that Hindi or Hindustani “may be accepted as a language for India’s national use. In other words, it is possible to adopt a language deliberately for some function, as we have adopted English. But there is a non-deliberate need for language: this need has to do with self-expression, not functionality. Official purposes must of course be served, but a more important purpose exists – to make the mind of our country responsive, fruitful and articulate. Only our own languages will serve this purpose.”

So Asamiya and Oriya have a right to grow along their own lines, apparently. But, dialects of Bangla? Here he draws the line (BBP 461): “The speech of the area around Calcutta has naturally [p 149:] been accepted as the language of all of Bengal. One should regard the general acceptance of this language as a good thing.” On this point 1938 echoes 1885 (ST 6): “The pronunciation in the Calcutta area is to be accepted as the standard. For Calcutta is the capital. It summarizes Bengal as a whole.”

The question of standardization of popular speech, the problem of morphological and syntactic gesture, and the well-defined puzzles posed by observations about adpositions and determiners contribute to modern linguistics in very different ways. We are far from being able to discuss standardization and gesture and syntax within an integrated work-space of “study of language as a whole”, given the present state of the linguistics profession. Perhaps working with and beyond the drift of Rabindranath’s later thinking will bring us closer to such a work-space. By way of conclusion, let me offer some remarks about working beyond what may look like a sprawling set of concerns.

Did Rabindranath say nothing substantial about language between 1905 and 1938? Well, he was then busy writing about metrics and the theory of literature. And BBP firmly embeds its linguistics in a literary matrix. One doesn’t need to read BBP carefully to see that literature is the question which brings standardization, gesture, and syntax together, and a lot else. Syntactic gesture is literary on a small scale, and the emergence of a literature is interwoven with the standardization of its language in ways which Prague school research clarified within the inter-war framework of linguistics. Perhaps that clarification no longer works. It has become necessary, in that case, to ask how the universality of literature across societies and the species-uniform growth of language in all individuals are to be related in a sound theory of both phenomena. Historically, a literature and its language take turns pushing each other forward. In an intuitive sense which theory cannot yet make precise, each puts its stamp on the other. This jointly formed stamp is the morphological gestural character or Sprachgeist of the speech com- [p 150:] -munity. Surely the set of possible human Sprachgeister, of possible marriages between expressive forms and colourful meanings, is qualitatively finite and obeys laws. And it is clear that normal children acquire Sprachgeist well enough to take part (more or less active) in the creation and recharging of expressive forms for everyday use. So you cannot separate this from the problem of how children learn vocabulary. Nor, if syntactic gesture turns out to be a viable notion, can the theory of developmental syntax stay away from the study of what we should really be calling style. But theoretical linguistics will, no doubt, take ages to get to such low-priority matters. In the case of Bangla, at least, there is a lot of basic syntactic research to do first. To put it concretely, Rabindranath may lead us to wonder at the fact that “while o unites, i isolates” (BBP 499, discussed above). But, before returning to this sense of wonder, we need first to puzzle out the difference between, say, kolkatar-i hoe ‘precisely on Calcutta’s behalf’, which can be said, and kolkata-i hoe ‘precisely via Calcutta, which cannot. With i at the very end, both become possible: kolkatar hoe-i, kolkata hoe-i (with the meanings indicated). What do you think is going on here?[6]

---

[1] ‘On the psychological reality of lexical items’, in Explorations in Linguistics, ed. G. Bedell, E. Kobayashi, and M. Muraki. Tokyo: Kenkyusha, 1979.

[2] Questions and Relative and Complement Clauses in a Bangla Grammar, 1980 Ph.D. dissertation, New York University.

[3] Jijnasa, vol. 1 (1980) and University of Melbourne Working Papers in Linguistics, vol. 5 (1979). See also note 2.

[4] Kayne’s ‘Unambiguous paths’ presented at GLOW, Dougherty’s work on coordination in Language, and Chomsky’s Aspects.

[5] Chomksy’s 1981 Lectures on Government and Binding.

[6] Throughout, translated passages from ST and BBP cite the examples and then gloss them. Needless to say, the originals did not do this. The translations are mine.