Language and Linguistics
Ashok R. Kelkar




Writing, speech and Language:


      In ordinary parlance we speak as if there are two kinds of language, say, written English and spoken English, or written Urdu and spoken Urdu.  Indeed the so-called written Urdu alone is regarded as real Urdu.  So if somebody can converse in Urdu but cannot read or write in that language, he is apt to say that he does not know Urdu.  While we distinguish between Hindi and Urdu newspapers or books, we don’t quite clearly distinguish between Hindi and Urdu films.  Ordinary parlance, it must be pointed out, is misleading here.  We don’t have two kinds of language, rather we have the same English or Urdu and it is communicated either through speech or through writing.  Language is a system of meaningful symbols, while speech and writing manifest language through systems of communicative signals.  How are these two signaling systems, the speech system (phonology) and the writing system (graphology), hooked on to the symbolic system of language?


      Here, it must be admitted, students of linguistics fall into two schools.  Hjelmslev or Halliday, for example, think that speech and writing are on a par, simply two alternate modes of manifesting language in a parallel fashion. 







      Others, like Book field or Hackett or Bazell, think that language is essentially speech, and writing is an optional appendage to language.




      Let us call these the parallelism school and the appendage school respectively.  To my mind, evidence bears out that writings is an optional attachment all right, but that this attachment is not always made ‘after’ speech, that is, it is not always correct to say that writing is essentially a record, a rather imperfect record to be sure, of speech.  The relevant facts may be set out as follows: 


(1)         In the history of mankind, the beginning of writing cannot be earlier than 4000 B.C. while the beginning of speech is much older.  Indeed   it may coincide with the beginning of Homo sapiens alias ‘Homo   eloquent’.  In the history of any given language speech always antedates writing by a few centuries.  Indeed the large majority of language never got written at all.  (Speakers of such languages are not illiterate, they are simply pre-literate).  In the life-history of a single individual, speech always antedates writing.  Even with a literate language some individuals may remain illiterate.  In India illiterates 

          and pre-literates for outnumber literates.


      Adherents of the parallelism school may shrug off this argument as an example of the genetic fallacy.  But the temporal priority of speech over writing is too overwhelming to be lightly set aside.


(2)    Perhaps what is more to the point are the following facts pointing to the dependence of language as written on language as speech.


      While the child learns to speak and learns the symbolic system of language more or less simultaneously without any special coaching or training, the same cannot be said about the child’s learning to write.  Deafmute children find it difficult to learn language and writing in the absence of speech, while blind children have no difficulty in learning language and speech in the absence of writing.


      Similarly, the very system of language clearly shows that it is meant to be spoken.  Even language as written shows the imprint of speech—it is always linear.  It has to make up its mind as to the relative order, for example, of subject, verb, and object: it does not bypass the problem (as it could easily have done) by the placing the three in two-dimensional space as a triangle.  All writing systems (like all speech systems) are linear.  Again, no writing system makes use of explicit bracketing to resolve ambiguities as the one between—(old men) and (women) and (old men and women).


      Since English writing obliterates such things as accents and tones from speech it has to resort to such dis-ambiguations as—


women and old men   



as distinct from—


old men and old women.


      The speech system is more intimately connected with the language system than the writing system is connected.  This is shown by the fact that changes in phonology very often have important repercussions on grammar (thus coming comes from two distinct earlier forms ending in --d and – g) while changes in graphonomy have no such repercussions (thus the shift from the Arabic script to the Roman script did no bring about changes in the Turkish Language).


(3)         The passing mention of the obliteration of accents and tones in English writing leads us to the admission that when writing does attempts to record speech it does so only imperfectly.  When a child learns to read such writing it reads it ‘aloud’ and the vocalization is only gradually suppressed—sometimes only imperfectly.  One has seen lips silently moving presumably accompanied by inner vocalization at the time of silent reading, Indeed ‘adult’ fast reading presupposes the suppression even of this inner vocalization.


      But we must extract a further admission from the adherents of the appendage school.  While it is true that writing (and silent reading along with it) essentially depends on language as spoken not only temporarily but also structurally (our arguments 1 and 2), it is also true that writing may occasionally bypass phonology at the point where it is hooked on to language.  While speech helplessly accepts homonymy between bear and beer, writing accepts only the second harmony, but not the first: beat and beet stand clearly distinguished and have to be learned as such by the school child.  And what about a written signal like &? The ampersand does not clearly stand for the sequence and: we don’t write r&om for random.  It stands for the word and for the et in etcetera.  The word etcetera can also be written as etc. or & some writing systems, as we shall see later, this kind of bypassing in a much more thorough going manner.  More of this later. 


      In short, the writing system of a language is an optional appendage to that language so that the user of that language can communicate with the help of that language without restoring to speech and thus exploit the fact that while speech (and writing in the sky with smoke) is evanescent most writing is relatively permanent.  Indeed the earliest examples of writing are not speech substitutes like letter-writing (with the coming of the telephone letter-writing is loosing its importance) but records of commercial transactions, royal acts, and the like.  Though this optional appendage is heavily dependent on the spoken use of language in terms of life-history, it can and does bypass speech in recording a message in a language.


Writing, proto-writing and epi-writing:     


      We have earlier mentioned the road not taken by writing—namely, triangular or two-dimensional sentence structures or arithmetic like explicit bracketing.  But of course the road has been taken—not by writing systems but by other systems of visual communication.  This is true of visual systems older than and leading towards writing, namely, proto-writing as well as of visual systems younger than and leading away from writing, namely, epi-writing.


      An oft-cited example of proto-writing is that of certain American Indian ‘documents’ such as treaties between tribes speaking different languages.  Proto-writing may be a preparation that lead or that might have led to writing proper, but it is distinct from writing proper.  It is not only pictorial in symbolizing objects such as reverse and fish and through these objects even more abstract entities such tribal fishing rights, but also pictorial in symbolizing the inter-relationships between these concrete and abstract entities by freely using two-dimensional space.  Thus victorial proto-writing is distinct from pictorial writing, the latter is linearized.  Also, the latter corresponds to linguistic messages in a much more definite manner.


      Much more recently in human history, we see the rise of visual system that are dependent on writing.  Arithmetical notation with its bracketing is an obvious example of such epi-writing.  Other examples will be road signals, semaphore signals.  While with some kinds of epi-writing the dependence on writing is very heavy, witness, semaphore signals, Braille, Morse code or ciphers, with others the link with language becomes tenuous.  It is already difficult to read aloud tabulations, flow charts, and formulas in organic chemistry.  The visual notations of higher mathematics and symbolic logic leave language far behind in moving along certain directions.  But epi-writing is still anchored in writing proper, however tenuously.  Indeed it is very often interspersed with writing.


      Some visual systems may be wholly or almost wholly language independent—for example, some deafmute hand-signalling systems or the visual signs used at international airports or Olympic games.  Even such language-independent visual signs may incorporate language messages in writing—for examples verbal legends in mans, machine drawings, circuit diagrams, or musical staff notations.


      Much more rarely, spoken communication may be supplemented by writing proper.  A common sight in the city of Hong Kong is that of two Chinese talking and at the same time busily drawing characters in the air with a forefinger to overcome their dialectal differences.  Of course air-drawn characters are probably more accurately called epi-writing?  Not really, certainly not in the sense that Morse code is epi-writing dependent on writing.  Rather, we must say that characters are characters whether drawn in ink, paint, depressions in sand, smoke trials against the sky, finger movements in air, ink-stains left by stencils, or embroidered threads.  All this latter variety belongs to what one may with more justice call the technology of writing, which is, to a certain limited extent, comparable to the physiology of speech.  The technology of writing is only a appendage to the writing system proper, though it may occasionally govern the shapes of the written characters thus, the stylus strokes of Cuneiform writing the brush strokes of Chinese writing, and the rounded strokes of palmleaf writing seen in the Oriya script.  


Writing systems and types of Writing:


      A writing system is a system of signs that connects the visual signals to appropriate language messages for the reader and connects the language messages to appropriate visual signals for the writer.  The two ways of using the system, receptively or passively for reading and productively or actively for writing, must naturally match each other.  Thus, if we read a script from left to right, for example, we must also write correspondingly from left to right; or if we write the word and by the spelling &, we must also read & correspondingly as a word (and mot merely as a syllable).  This should be obvious enough.


      What is not so obvious is that the connection between the visual signals and the appropriate features of language messages such as syllables or words need not be a direct one.  In fact it is a medicated connection it has what linguists call a “double circulation”.  To clarify this point further, a writing system or graphonomy consists of two subsystems—a script system (or calligraphy) and a spelling system (or orthogaphy).  Thus, when we write Sanskrit in Devanagari and in Roman, we may be using two script systems but one and the same spelling system.  But when we write Hindustani in Devanagari and in perso-Arabic, not only are scripts different, but spellings also are somewhat different –thus the same word is spelt báta in Devanagari but bat in perso-Arabic, or (to cite another word) the same word is spelt taraha in Devanagari but tarh in Perso-Arabic (with a special t and a special h).  Consider also konkani as written in Roman—the same name will be written as X’anta by some (using the portuguese spelling system) and Shanta by others (using the English spelling system).


      The ‘lower’ subsystem or script system is directly hooked on to the visual signals but only indirectly hooked on to the language message.  The ‘higher’ subsystem or spelling system is directly hooked on to the visual signals.  Let us briefly look at these two subsystems.


The script system or calligraphy consists of the following:


 (i)    a list of graphemes (letters or characters as well as non-literal signs

         such as capitalness and punctuation marks).


(ii)    each grapheme has variants that are not distinctive (each Arabic

         grapheme has initial, medical, final, and absolute forms, small Greek

         sigma has a non-final and a final shape).


(iii)    each variant of each grapheme can be analyzed into strokes placed in a particular direction, (thus the Arabic characters go ‘from east to west in Arabic and Sindhi but ‘from north-east to south-west’ in Persian and Urdu), in a particular order (compare the Marathi and Hindi way of writing same Devanagari character ka), with a particular thickness (compare the Telugu and the Kannada way of writing what is essentially the same character ka), and within a particular enclosure (one can imagine any Chinese character enclosed in a square but not any small Roman character which may have a suspender or a flagpole).


(iv)   a scheme of linearization (left to right, right to left, top to bottom, or boustrophedon for graphemes in a line; top to bottom, right to left for lines in a page; left-sewn or top-sewn or right-sewn pages in a book or pages stacked top to bottom in a loose leaf portfolio or a codex; left to right volumes in a book rack but top to bottom in a binder).


(v)    accepted forms of layouts (justified or unjustified at the left or right or top or bottom margins; prose and verse alignments, pros paragraphs, verse paragraphs, and stanzas; centered titling; and so on).


      A respect of last three components of the calligraphy, the outline of the two-dimensional writing surface of course has to be predefined—the writer or the reader has to know which side is up, which down, and so on.  This applies even to cylindrical or spherical surfaces.  Consider the problem presented by characters embossed on a small circular disc—is this 6 or 9? N or Z?  Runic inscriptions do not presuppose a ‘flat’ surface, but an edge.


      The spelling system or orthography consists of the following:


(i)    a list of the smallest units of the language message recognized for spelling purposes (the ‘long u vowel’ seen in cute, few, pure, beauty,  feud, etc., but not in you, your, etc.; the words and, two, zed seen in &, 2, Z; the sentence type direct question seen in who’s there’s etc).


(ii)   a decision of the basic strategy and other details for the hook up with the lower level of the script system (in German, predicting pronunciations from the graphemic realizations of the spelling units is fairly straightforward, but predicting the grapheme sequence from the pronunciation is difficult in either direction—we have already seen the various graphemic manifestations of the ‘long u’ vowel, consider now the various orthographic sequence gh seen in ghost,  enough, fight, hiccough; in Roman, each character is enclosed in its own distinct plot of ground, ligatures like ae, ff, fi, ffi, in printing being purely accessory;  in Devanagari and in Arabic, ligaturing is the rule, not the exception).


(iii)  a decision of the basic strategy and other details for the hook up with  the units of the language message (in the English writing system, the  basic strategy is hook up from the grapheme or grapheme sequence) to the phoneme—in other words it has an alphabetic orthography; so has the Sanskrit writing system except that its true alphabetic character is obscured by the ligaturing of vowels to preceding consonants and of consonants to following consonants).


(iv)  accepted forms of layouts (matters such as paragraph division, dialogue layout, columns, chapter division and perhaps some kinds of epi-writing like tables, tree diagrams, flow charts, and the like could be considered here).


      It will be seen that the full description of a writing system involves the description of three hookups transitions : The transition from visual signal to calligraphy, the transition from orthography to language.  Each of these transition enables us to classify writing systems and their specific details into types.


      In view of the first transition (visual signal to calligraphy) we can recognize the following types:  The character orientation may be top-left to bottom-right (Roman, Devanagari) top-right to bottom-left (Arabic, Sindhi), top to bottom (Chinese, Japanese), and so on; the line orientation may be left-right ad top-bottom (Roman, Devanagari); top-bottom and right-left (Chinese, Japanese); left-right-left, to bottom (certain ancient Greek inscriptions in the so-called boustrophedon way of writing), and so on; the thickness may be uniform or variable (of course with a ballpoint pen one has no choice in the matter).


      In view of the second transition (calligraphy to orthography), we can recognize the following types:  The segmentation may be clear cut (Greek, Chinese, Japanese), ligatured (Proso-Arabic, Arabic, Devanagari) or conjunct (the characters ks and jn in Devanagari, some Chinese characters); the type identification may be one-one (Sanskrit, finnish, Czech), may-one (German), one-many (Marathi), many-many (English, French); the segment identification may be one-one (Czech, Finnish), may-one (sh in English, ch in French, sch in German for the hushing consonant), one-may (x in Greek, Latin, English, French or German, the shch letter in Cyrillic), and many-many (ei, ei in English kaleidoscope, fierce).


      In view of the third transition (orthography to language), we can recognize the following types:  The hook-up may feature-phonographic (the aspiration stroke in Kannada, the nasalization tilde in Portuguese, the fronting umlaut-sign in German), segment-phonographic (the so-called alphabets), syllable-phonographic (the Japanese and other syllabaries), lexis-logography (ampersand in English, French, German – respectively and, etc., unt, omkar characters in Marathi etc.,: present-day Chinese), grammar logographic (the interrogation mark in European punctuation systems), and semographic (mostly arbitrary, as the so-called Arabic numerals in European writing systems, iconic as in English U-tube, T-shirt, metonymic by the rebus principle as in English Xmas, Xt or by symbolism as with the obelisk against the name of a deal person in European writing systems).


      The complex nature of a writing system will now be appreciated.  This complexity gives rise to the large variety of types of writing systems.


      When a writing system changes through history, it may split itself into two or more writing systems.  Thus, the Brahmi script gave rise through successive stages to a large number of scripts found at present.  The two subsystems may change independently.  A given language may come to be written in different scripts—this may result in different script systems and a shared spelling system.  Problems of script reform and of spelling reform are thus different and two kinds of reform do not always go together.  Thus, for Marathi, Savarkar proposed a script reform and N.S. Phadke and N.C. Kelkar proposed  a spelling reform.


Written Languages       


      When the same script is used for the two languages, what may distinguish the two writing systems is not so much their script systems as their spelling systems.  The language is more likely to affect the spelling system than the script system is likely to affect language.  Can we think of influence in the reverse direction?  Can a script system or a spelling system be reflected in the language?  After we examine a message in language can we say that it is meant for or suitable for the written channel of communication than for the spoken channel of communication?


      To begin with there are the written ‘translations’ of certain specially spoken idioms—the foregoing and the following translate in writing as the above and the below, the former and the latter translate as the left-hand and the right-hand.  The hemmings and hawings cannot go into writing; on the other hand, idioms like if I may say so and how shall I put it? Which have similar functions can go into written communication.  The asides spoken in a low soft voice translate as parentheses and footnotes and marginal notes.


      Of course certain devices that belong to writing may have spoken translation—quotation marks translate as and I quote; headings and colophons translate as I am now going to talk about such and such and the like.


      Interesting as such observations are and useful too to a person composing the script of radio talks, film dialogues, or stage dialogues, they hardly touch the heart of the relationship between the spoken and the written se of language.  We have already alluded to the fact that in view of the non-availability in written use of accents and tones, facial expressions and gestures, feed-back from listener’s cross talk like and then?  And other responses like the skeptical headshake, the sender of the message, has to be extra careful in avoiding ambiguities, imprecision and the like – better say men and old women or old men and old women than risk the ambiguity of old men and women.  We can also point out that the availability of certain facilities in the written use should be fully made use of – facilities like cancellation, corrections, insertions, re-writings which are available to the writer and facilities like pausing to reflect, going over the text or any portion of it over again, skipping what one can afford to skip, jumping ahead and coming back, dipping and skimming, looking up from the index, scanning an alphabetized list which are available to the reader.


      We have also alluded to the fact that adult fast reading implies fuller suppression or articulate or inner vocalization of text.  In fairness we have to point out that there are written texts that demand some vocalization.  The play being enacted on the inner stage of the reader is an obvious case in point.  The same goes for a personal letter from a friend or an intimate.  The replacement of the elocutionary punctuation of Shakespeare by a ‘tidier’ logical-grammatical punctuation by his eighteenth century editors was thus  clearly a misguided ‘improvement’ on the bard.  And of course the playwright was a poet too.  Poetry also demands to be heard if only in an undertone by the reader to himself or silently by the reader’s inner ear.  But when due allowances are made for such spoken or silently spoken scripts, we have to recognize the larger body of written texts – most documents and prose fiction and discursive essays and scientific treatises, for example – that demand to be left alone with the silent reader without the intervention of either loud or silent vocalization.


      The coming of recorded and transmitted speech shows no signs of rendering obsolete of the hand-written or typewritten ‘manuscript’ and the multiplied and transmitted versions of it.  Occasions still arise when one would rather write a heart-to-heart letter then have that face-to-face talk or that telephone call.  Writing and reading permit certain special facilities as we have already seen.  Even the handicaps can be turned to advantage – the absence of supplementary vocal or bodily gestures encourages greater care in ‘composing’ a text ‘formulating’ the content.  Translations written at leisure have a better chance to be elegant and faithful than the necessarily rough-and-ready spoken interpretations done on the spot or written translations done in a newspaper office or a broad-casting studio.  All these point to the fact language in written use has some special assets – it can be cultivated, refined, elaborated, careful, detached, precise, unambiguous, exact, accurate rather than slapdash, banal, repetitive, hand to mouth, elliptical, commonplace, routine, run of the mill, current coin.  It can be, but it may not be.  Language in the written mode also has certain special liabilities – it is apt to be frozen, lugubrious, circumlocutionary, about-the-bush-beating, and the like rather than be fresh, involved, spontaneous, direct and the like.


      Spoken and written language modes are both here to stay.  Only the older tendency to a more rigid differentiation between the two is now giving place to a more flexible, give and take specialization and division of labour between the two.




      What I have been presenting so far has a rather modest aim, to bring together the best current thinking among linguists on writing and its relation to language before an audience of persons to whom these insights are likely to be of interest and use.  I am of course aware psychologists, ethnologists, and others have also made a study of writing.


      Thus language teachers will note perhaps the following:

(1)  In the elementary phase, writing should not loom large.

(2)  In the advanced phase, writing will play an important role.

(3)  In teaching the script system, eye-training for distinguishing resamblant

      letters and stroke analysis for hand-training should be emphasized.

(4)  In teaching the spelling system, visual recognition should be emphasized

(5)  In teaching language, the special features if any of the spoken and

      written modes should be brought to the notice of the learners.



     Those engaged in decipherment and analysis of ancient writings may note the following:


(1).  In tackling any data, the first task is to identify proto-writing, writing, and epi-writing.


(2).  In decipherment the three translations should be worked out independently from each other.  From the visual science to the scrip system, from the script system to the spelling system, and from the spelling system to the language.  At each point the varied types should  be identified.  A writing system usually displays a mixture of types but the dominance of one or two types out of the spectrum of possibilities listed earlier.


(3).  In analysis, segmentation and identification should be worked out jointly and graphemes (script-units) and ortho-graphemes (spelling-units) should be listed along with their non-distinctive variants.


      Those engaged in applied writing arts such as decorated and display calligraphy and typographic design may note the following :


(1)  A grapheme should remain recognizable, as also their sequence.  This especially applies to decorative directions.  Care should be taken that resembling characters will remain distinct – especially in the design of typefaces for printing, typewriting, stenciling, lighted screening and in  display calligraphy. (2)  While the relation between the script system and the language message is only indirect, there is good scope for exploiting the visual qualities of the graphemes in way of creating a style and making it associable with  the proposed message or message type by way of iconicity, metonymy, traditional association etc.,


      Finally, those interested in script reform and orthographic reform may well note the following:


(1)  The two reform proposal should be kept distinct.

(2)  A reform proposal should exploit the tendencies and possibilities of the

      existing writing system rather than bring in total innovations. 

(3)  A step-by-step staggered program and a wholesale shift should both be

       considered Habits from the existing writing system are bound to be

       carried over.  Better to move them than to fight them tooth and nail. 

(4)  Ease of reading, ease of writing, and ease of learning should be        

       considered – first separately and then jointly.  The demands of the three

       need not always be the same.

(5)  Ease of reading, writing, learning the handwritten and the printed forms

       of the script should be considered – first separately and then jointly. 

       The demands of printing and handwriting need not always be the same. 

       At the same time one should think twice before imposing the need for

       learning the script twice – first for handwriting and then for writing.




      This was published in Caltis-83 on the occasion of the calligraphy, lettering, and Typography of Indian scripts Seminar, Pune, January 1983.