File clean-up May 24, 2022, 5:21 pm sdvig press Offprint from Writing in Focus (Coulmas/Ehlich, Eds.) © 1983 by Mouton Publishers, Division of Walter de Gruyter & Co., Berlin. * Slightly revised version of “Doppelte Artikulation in der Schrift”, Zeitschrift für Semiotik 2. 1980, Translation with the help of Donald F. Goodwin. Thanks are also due to the editors of the Zeitschrift für Semiotik for their permission to include the article in this volume.
4Orginally published in Konrad Ehlich & Florian Coulmas (eds), Writing in focus, The Hague, Mouton (1983): 45-62.
The anthropological relevance of double articulation
5A system of signs is called doubly articulated when it consists of two levels of organization, a first level of signs with a sense-determinative function (morphemes, words, sentences) and a second level with a sense-discriminative function (distinctive features, phonemes, syllables). Hjelmslev suggested that these be termed respectively plerematic and kenematic signs. The older generation of linguists considers this special kind of double articulation to be a peculiarity of human language (Benveniste 1952, 62; Hockett 1963, 12; Jakobson 1973, 44). Lévi-Strauss (1964, 37f.) treats double articulation as the criterion according to which a system of signs may be given the title “language”. When the discovery of the genetic code revealed a non-human sign system that undeniably had a double articulation — four nucleotide bases fulfilling an exclusively discriminative function and triplets of such nucleotide bases (“words”) along with hierarchically higher-order units (“sentences”) transmitting determinative chemical information — Jakobson (1973, 52), master of the search for invariants, promptly found a common denominator: The genetic code, the primary manifestation of life and language, the primary manifestation of humanity, are the two fundamental transmitters of information from generation to generation: the molecular and the verbal legacy are two indispensable presuppositions of cultural tradition.
6Younger semioticians (Prieto 1966; Eco 1972) contest an exclusive ascription of double articulation to language (in the narrow sense of the word). The cleavage between these standpoints makes a different orientation of the camps evident. Linguists of the older school take double articulation to be a criterion distinguishing human from animal languages; the younger semioticians, on the other hand, when they attack the linguistic “myth of double articulation” as being “obstructive” — which it is, in fact, on superficial analysis have other human sign systems in mind, which equally are featured by double articulation. This view can be accommodated without further ado by the linguists cited above, since they see in double articulation (disregarding the genetic code) a peculiarity not so much of language as of the human. Double articulation distinguishes human from all animal languages as far they are now known, and also from most, but not all other human sign systems.
7The creation and use of signs that are not intended to designate non-semiotic entities, but rather to produce other signs, presupposes a cognitive competence, the ability to produce tools to produce other tools, which anthropologists still consider as being a mark of the human (cf. Jakobson 1973, 58). The epistemological, not to mention anthropological implications of double articulation are an invitation to investigate non-linguistic sign-systems that exhibit double articulation, especially those that permit us to investigate double articulation in the nascent state, such as some recently developed writing systems. These newly developed systems stand in evocative contrast to those nonverbal sign-systems to which double articulation has been attributed in semiotics until now.
Functional conversion of sense-determinative into sense-discriminative signs
8Until now the debate has one-sidedly taken account of those secondary sign systems of which the second, or kenematic (quasi-phonematic) articulation occurs in sign units that in their original use had a sense-determinative rather than a sense-discriminative function. This is so of the examples adduced by Prieto and Eco, just as of Bühler’s old example of the sailor’s flag code (1934, 70f.). The elements of the traditional seaman’s code — round ball, triangular penion, and a quadrangular flag — represent geometrical figures. This original iconic meaning is dropped in the flag code. Neither the single elements nor subgroups of them have a constant and positive meaning. Their only function is to distinguish the complex signals from each other when they occur in different positions.
9In the case of six-digit telephone numbers, of which each pair of numerals (digits) designates successively a neighborhood, a street and a block of houses, the single numerals making up a sense-determinative pair have themselves only a sense-discriminative function. Likewise, each of the numerals of the two-digit numbers designating the various bus lines of a city functions only to distinguish meaning. Bus line 23 has nothing in common with other lines whose first numeral is 2, nor with those whose second is 3. The numbers do not mean that all buses with the same numeral in the same position run in the same direction, serve the same neighborhood, or belong to some special class for speed and comfort. Their sole function is rather to distinguish one bus line from all others with different numerals. The example of the bus numbers shows clearly that the usage of their components is an extended, metaphoric one. Signs that had a sense-determinative function in their original usage (as signs of natural numbers) come to have secondarily, in another sign system (that of public transport), a merely sense-discriminative function. They are now metaphorical signs, also in the sense that the echo of their original meaning is more or less latently heard in their new usage, according to the sensitiveness and attitude of the receiver. These examples of double articulation do not present a simple case of the production of tools for the production of other tools, as is the case in spoken language. Tools already available — with a sense-determinative function — are rather converted into tools with a new, now sense-discriminative function.
Structural and functional anisomorphism of speech and alphabetical writing
10Before going on to present newer writing systems as an illustration of non-linguistic sign systems with a second articulation of originally, not merely secondarily kenematic signs, it might be helpful to reflect on the rationale that deterred leading phonologists from describing writing analogously to speech as a doubly articulated system (in which the letters, styled as graphemes, might assume the part of the phonemes). Their reason was not the lack of structural isomorphism, nor the conspicuous fact that there is no one-to-one correspondence between letters and phonemes so that a grapheme often corresponds not to a phoneme, but to a morphophoneme, and conversely that a single phoneme often corresponds to a whole group of letters (Ger. sch, ch; Engl. th), nor that the internal structure of phoneme and letter are not congruent. In the case of the letters
| d | b |
| q | p |
11the binarily structurable constituent parts — a vertical line extended upwards or downwards; a semicircle open to the right or to the left — do not correlate to distinctive features of the corresponding phonemes (say to the oppositions acute grave and tense—lax). Their point was also not that alphabet systems are of non-uniform, heterogeneous origin, nor that the perception of letters is only in part digital, but predominantly holistic, by means of Gestalten: a heterogeneous origin does not preclude the selective coalescence into a unitarily ordered system; the perception of sounds, too, may be a mixed process, both digital and holistic. Furthermore, it is possible to analyze each Gestalt by means of a grid, so that in principle each figure can be registered either holistically or digitally, as the need may be. The need, and therewith also the factual extent of the hierarchically ordered digital analysis of written symbols that was originally holistically conceived, increases to all appearances with the increasing number and complexity of these signs.
12The decisive reason for the disjunction of speech and writing was rather the absence of a functional isomorphism. Whereas the phoneme m has a purely diacritical, relational, negative function, and evinces mere “other|ness”, the letter m has a determinate, constant and positive meaning: it is the sign of the phoneme m, which takes its place in the oral rendering of the text. Letters are, so to speak, metasigns, “signs of signs”, namely of sounds, whereas phonemes are special kinds of signs, sense-discriminative or diacritical “signs on signs”, namely on morphemes (Jakobson 1939, 296). The introduction of letters brings with it a different sort of con|version of signs, not a transmutation of signs with a sense-determinative function into signs with a sense-discriminative one, but rather a conver|sion from signs for units of sense (Sinneinheiten) into signs for units of sound (Lauteinheiten) and therewith from objective to metalinguistic signs. This modification is generally appreciated as such on the phylo|genetic level, where it manifests itself in the step from semasiography (in a wide sense including ideography, phraseography and logography) to phonography (in a narrow sense restricted to syllables and alphabetical writing). It is less well known that a similar step is just as manifest in the ontogenesis of writing. A child finds it much easier to correlate graphemic structures with senseful units than with elemental phonic units (Gleitman & Rozin 1977). A child learning to write has functional, not structural, difficulties. He has more trouble correlating written signs to sense-discriminative phonic signs instead of to sense-determinative signs than with the specific figuration of alphanumerical signs and the specific graphomotoric skills connected therewith.
13What distinguishes au fond speech and writing is not their differing physical media, an acoustic and an optical medium, respectively: that would be a too culture-specific, perhaps also a too species-specific view of the matter. Deaf-and-dumb languages are quite genuine languages, and meet all the specific criteria of languages (multimodality, grammaticality, productivity, semantic universality — and double articulation). Tele|graphic messages can be transmitted just as well acoustically (by key-sounds or drumbeats) as optically. On the other hand, there are signs in formalized languages for operations that are carried out primarily optically, and of which spoken designations have become secondary. Finally, there is the recent phenomenon of “noise-writing”, of the noise, specific even to the trade brand, that each key punch on a typewriter produces, and which is said to be used by intelligence sendees (at least in science fiction) to decode the letters typed.
14The difference between speech and writing is also not primarily structural. As regards structure, artificial writing systems are extremely flexible — in contradistinction to natural speech, they are capable of giving an exceedingly polymorphous performance. It is possible in principle to develop a writing system that duplicates the structure of the spoken language (see phonetic alphabets and spectrograms).
15What motivates and legitimates the distinction between speech and writing as two categorially different sign systems is the difference between their specific functions: speech is a sign system that immediately symbolizes meaning; writing is a sign system that symbolizes another sign system, and this either in another medium (thus our current visual writing system) or only in another structure (thus morse code, insofar as it is implemented by key punches). This specific, functional difference also immediately gives rise in normal cases so long as isomorphism is not the intended goal, as in the case of the phonetic alphabet to specific structural differences, which are common to writing systems and to other adaptations of language to various media of communication: for example, to pidginization, metricization (cf. Justeson 1976, 78f.) and, we can add, to its internalization, to the so-called “inner speech” (cf. Holenstein 1980, 126 ff). These transformations reveal regularities of simplification, for example in the reduction of marked phenomena in contrast to the conser|vation of unmarked phenomena, and also in the tendency to introduce similar operations to represent the same phenomena, for example redupli|cation to indicate extension (plurality, size, intensity, and so on).
16The functional disjunction of speech and writing does not preclude the possibility that a sign system that was originally introduced as a (secon|dary) writing system can assume the function of a (primary) sign system that immediately symbolizes senseful units, thus becoming, according to definition, a genuine language. In special cases (such as that of deaf and dumb), this change of function can be a complete transformation, in other cases (such as that of our normal usage of writing) only partial. Functional conversion and plurifunctionality are a universal possibility, not just semiotic, but also ontological (Holenstein 1979).
17In fact, the “mediation theory of word recognition”, the theory of the indirect access to the meaning of a word via its sound shape, has already become a classic topic in experimental reading research. Meaningless, but pronounceable, combinations of letters are not readable in cases of phonemic dyslexia. Rhymes based on homophones (hope–soap) are not recognized. Concrete or figurative nouns, on the other hand, are readable. Errors that do occur are semantically canal–river) and not phonically motivated. Here, clearly, the sequence of letters takes over the function of, or circumvents the sequence of, phonemes as the immediate carrier of the meaning of the word. (The pronunciation is derived from the rightly or wrongly apprehended meaning.) It may be assumed that, in such cases, the letters function as (sense-) discriminative signs and not, as is usual, as (sound-) determinative.
18The fundamental possibility that a normal, skilled reader has direct access to the meanings of words is now generally admitted — just as the fact that even fluent reading is never completely free of phonemic mediation. Research has gone on to clarify the factual extent, the condi|tions and the roles of the two modes of access. Direct access occurs mostly in the case of the more common words. It increases with the ability to read and, conversely, has the function of increasing precisely this ability along with the speed of reading; these two phenomena are best explained in relation to a context inferred by direct access and to the expectation of meaning that thereby arises. The complementary task of verifying the expected meaning “from the bottom up” devolves upon the mode of indirect access. The reader understandably takes recourse to this mode mostly when faced with unfamiliar expressions, but he also uses it to make precise and to confirm his direct understanding when this is required, when more than just the approximate meaning of a text is of interest, and when the exact sequence of words matters just as much as the literal meaning.
19Philosophers make a great fuss about the deficient categorial distinction in the designation of phenomena that belong to different categories, and view this deficiency as being responsible for the worst of all mistakes — the worst in a philosopher’s eye — the category mistake. The use of similar lexical categories to designate both (independent) substances and (dependent) properties in alternative sentences such as “The wall protected the inhabitants from the enemy missiles” and “The height of the wall protected the inhabitants from the enemy missiles” seems to imply that “wall” and “height” are things and causes in similar ways. But it should be philosophically just as fruitful not only to decry the disadvantages but also to consider the advantages of this peculiarity, characteristic of so many sign systems, of not systematically and pedantically rendering categorial differ|ences of the signified entities with exactly corresponding signifiers; and also to consider the ontological and cognitive conditions which keep the man of good common sense in general and the man of philosophical perspicacity in particular from being confused in his use of signs by the lack of categorial differentiation. The under determinateness of structure relative to function and interpretation should also be considered at this point, in view of it being, in addition to its role in banning monotony, the basis of linguistic creativity.
Functional conversion of graphemes from sound-determinative to sense-discriminative signs
20There are, however, words in which some graphemes have a sense-discriminative function, even though the decoding of the words concerned is phonemically based. These are phonemically ambiguous German words such as: Mal–Mahl, Lid–Lied; Stil–Stiel; Laib–Leib; Saite–Seite; Waise–Weise; Meer–mehr; du reist–du reihst. The range of examples in English (knight–night; know–no; pane–pain; lane–lain; totoo–two is so abundant that one wonders if perhaps English has made the turn back towards the logographic principle (cf. Gleitman & Rozin 1977, 21, 33, 35f.). The h in Ger. Mahl does not designate a sound that is to be pronounced in its place, and it is redundant as a sign for the lengthening of the preceding vowel. It does, however, immediately distinguish the meanings of Mal [mark] and Mahl [meal] just as the phoneme e in the place of a distinguishes between Mahl and Mehl [flour]. Here, too, the previous remarks about the examples of doubly articulated sign systems hold good: The discriminative signs were originally determinative signs that have been converted to fulfil a new function. It is sometimes even a case of a second functional conversion of these signs. Both letters h and e normally designate phonemes. In the course of the development of the German language they came to be used in certain contexts as signs of a distinctive feature, namely of the length of the preceding phoneme. Finally they came to be signs that immediately distinguish meaning.
21Features indicative of a tendency to give graphemes a discriminative function are also to be found in the use of groups of letters to designate a single phoneme, say, in the German usage of ch [x] and sch [/]. The single graphemes in such combinations no longer have a constant and positive meaning. They serve only to distinguish from each other groups of letters that designate different sounds.
22What is striking in such examples of a nascent double articulation is the anisomorphism of the signifier and the signified. Signifiers of one and the same category are used to render signified entities of widely different categories. The same grapheme e is not only used to designate two different sounds in elend, but also to designate a distinctive feature, the length of the vowel in siech [sick] (vs. sich [himself]), and to distinguish the meaning of Stiel [shaft] from that of Stil [style]. This neglect of categorial differences is even clearer in the cases of the morse alphabet and the genetic code. The delimitative devices or “punctuation marks” here have the same type of structure as the other plerematic signs. In morse code, in addition to the punctuation marks, the signs for a special category of words, the numerals, have the same type of structure as the signs of the single letters.
| J | · — — — |
| colon | — — — · · · |
| 2 | · · — — — |
Functional conversion of senseless into sense-discriminative elements
23The interpretation of letters as plerematic signs conforms with the fact that the majority of alphabetic characters can be most simply analyzed not directly into distinctive features (which do not have an independent existence), but rather into independent units, comparable to phonemes, which are then resolved into distinctive features. This conformity is merely a factual one, not a necessary one, which might be derived from the plerematic status of graphemes. Certainly a system of letters is possible of which the units determining sounds could be resolved directly into distinctive features: for example, a system consisting of lines drawn with differing length, color, brightness and position. The fact that the logically most simple realization was not taken up is in part grounded in the same way as redundancy in general (which is an adaptation to the needs of reception and production), and in part, as in printing, in the necessity of establishing a recognizable kinship to traditional alphanumeric systems.
24At least three starting points can be discerned in the decomposition of traditional alphanumeric writing systems into sense-discriminative features. The first was the upshot of the concern to standardize printing type, the second of the advent of illuminated writing, and the third of the development of electronic reading machines. The first of these tendencies reached both its technical and its ideological zenith in the Bauhaus between the two World Wars, and coincided with the world-wide establishment of phonology. The Bauhaus and the establishment of phonology had in common the aim of analyzing the respective phenomena on which they are based — spatial shapes and colors in the one case, speech sounds in the other — into abstract fundamental elements, and of combining these elements in complex structures, according to the nature of the material requirements, while preserving at the same time the “organic relations” between the elements (cf. Kandinsky’s lectures on form and color in the Bauhaus, Wingler 1969). The task of coping with the flood of vouchers for financial and goods transactions gave a much greater impetus to the third movement than did the humanitarian aim of providing reading machines to the blind. Waybills thus provided the basis for the Near-Eastern origins of writing, but also for the most advanced (electronic) mode of reading.
25The standardization of sign systems is made up of three processes:
- the decomposition of articulated as well as of inarticulated, continuous wholes into uniform constituent parts that may be artificial and abstract; the stencil lettering of Albers (Fig. 1) breaks up the vertical bars of b, d, p and q into three squares. In illuminated lettering (Fig. 4 & 5), the vertical bars of 1 and 7 are resolved into five or more dots.
- the reduction of allo-forms, that is of context-sensitive variants that exclude each other in any one context; hence the reduction of so-called complementary distributions; in illuminated lettering, slanting lines are rendered as vertical lines (for example, in the numerals 4 and 7 in Fig. 4), or slanting and vertical lines are both rendered with the same minimal slant (following the example of italic type). Similarly, round figures are recast as angular; or, as a compromise solution, the corners are left out, so that all figures are rendered essentially alike, with a slightly round looking form (see the numerals 5 and 8 in Fig. 4).
- the reduction of redundant signs; printing type omits as redundant the transverse bar in the middle of the numeral 7, which, together with a some¬what longer, horizontal rather than slanted Une at the top, distinguishes this numeral from the numeral 1. Zealots of unity among the reformers of German writing are also opposed to the superfluous use of capital letters to single out nouns: “we use only small letters, and thus save time” (Bauhaus maxim).
26In the stencil lettering of the Bauhaus collaborator Joseph Albers (Fig. 1 & 2), the letters are constructed out of basic geometrical shapes, in fact, of only three shapes (or more strictly speaking two): the square, the triangle (half a square), and the quarter circle. The same or almost similar shapes served more than ten thousand years ago as ideograms in the Near-Eastern precursors of writing. Later, they were used as sense-discriminative signs in an early modern non-phonic sign system, in the sailor’s flag code cited by Bidder. Now they turn up again in the construction of graphemes as sound-determinative signs. Recent research shows that not only concrete pictograms but also — even primarily — abstract geometric figures are to be considered as the earliest forerunners of abstract writing systems in the Near East. Aside from stylized animal shapes and, later, in the cities, also the forms of vessels, various abstract figures such as spheres, discs, cones, cylinders, triangles, rectangles and the like also functioned as counting tokens imprinted first on clay envelopes and later on earthenware tablets, and these made up the origin of the succeeding letters (cf. Schmandt-Besserat 1978; Fig. 3)
27Irrespective of the fact that the three fundamental elements of the stencil letters can be isolated as independent figures which are not inherently dependent components of the letters, it is also more economical to coordinate them in the analysis with the phonemes themselves, rather than with the distinctive features of the phonemes. If they themselves were treated as distinctive features, the quarter circle would have to be treated as two (unanalysed) complex distinctive features (a quarter circle rounded to the left vs. one rounded to the right; the curvature upwards and downwards is irrelevant because of complementary distribution in that system). Albers’ stencil lettering resolves accordingly into three graphemes analogous to the phonemes of the spoken language, and into a larger number of distinctive features. The position of the grapheme on the ground to be filled out by the characters, also functions as a distinctive feature in most visual writing systems; the ground to this end is usually imagined as being subdivided into cells. The use of the triangle may suffice here to illustrate the major principles of organization in the system of Albers. Beyond the position of the triangle in the figure, the position of the right angle and the doubling of triangles are distinctive features.
| c vs. e | - right angle top right vs. bottom right (this is no minimal distinction; the contrast between top vs. bottom, or moving the triangle half a space would suffice) |
| i vs. r | - triangle to the right vs. top |
| i &r vs. n & m | - a combination of bars composed of two squares with triangles vs. quarter circle(s) above square(s) |
| f vs. s vs. f | - two triangles vs. one triangle top right vs. one triangle middle right |
| ss vs. v & w vs. x | - top right: one triangle with the right angle top right vs. top left vs. two triangles; bottom: one vs two vs. no triangles |
| k vs. b | - two triangles vs two quarter circles |
| ß vs. z | - the triangles are redundant |
28The invention of electronic illuminated lettering made it possible to represent various letters without changing the basic hardware (say in the form of movable metal leaves, each imprinted with a different letter).
29The illuminated letters consisting of vertical and horizontal lines (Fig. 4), such as are found on pocket calculators, can be most easily analyzed as ten sense-determinative signs with seven distinctive features or elements lines distributed over seven cells or positions. The verticality or horizontality of their position is redundant in an analysis from the point of view of electronic production. The introduction of two independent kenemes (vertical and horizontal lines) with seven distinctive features, the seven cells or positions of the lines, however, meet better with the requirements of human perception. The discrimination is based on the presence or absence of the lines. Of course, the fact that they can be further analysed in terms of the antitheses right vs. left, top vs. bottom, inside vs. outside, opens up the possibility of alternate analyses. It is not exceptional, that distinctive features which occur as the final functional features can still be analyzed. The distinctive features of phonemes, too, can be subdivided into the properties of sonority, tonality and duration.
30The complementary distribution of vertical and horizontal lines can be reduced by the choice of dots (Fig. 5). This standardization requires, however, that the number of sense-discriminative elements be increased from seven to fifteen. The redundancy of qualitative differences (vertical vs. horizontal lines) is eliminated in favor of a uniform quantitative redundancy. The noteworthy feature of this model is that the atomic scanning of the fifteen discriminative elements is, not only for the human receptor but also for the computer, not the most economic method. The human receptor resolves the figures most simply into vertical and horizontal, partially overlapping lines, each consisting of three dots. The electronic use of luminous figures, too, can be simplified by the integration of single dots into groups of three, whereby the terminal dots can belong to two neighboring groups. Furthermore, some dots imply others: dot 1/2 does not light up unless the dots 1/1 and 1/3 light up, so that the operation of the dots 1/1 and 1/3 can be coupled with that of 1/2. The operation of all fifteen dots can be reduced to that of seven coupled with the rest, that is, the model represented in Figure 5 can be reduced with respect to production to the model of Figure 4.
31An increase in the number of cells from fifteen to thirty-five suffices to render the capital letters of the Latin alphabet as illuminated letters in a form that is easy for the human eye to recognize and discriminate (Fig. 6).
32Since the function of these systems of letters and numerals is not simply and directly to signalize speech sounds and numbers, but to do this while preserving a recognizable similarity to known alphanumeric characters, their shape is in great measure redundant. A comparison with telegraph codes (Fig. 7), which are based on the same dot system as illuminated writing, can illustrate this point. The international telegraph alphabet restricts itself to five cells, the Van Duuren Code to seven. The two-dimensional field of illuminated letters is here reduced to one dimension. The discovery of mistakes in the Van Duuren 7-Unit Code is made easier by an invariant property of the “letters” of this code; the regular manner in which they are composed out of three positively and four negatively (zero) marked compartments. The control function of redundancy now becomes a principle of organization. Four of the seven “letters” of the Van Duuren signals are zero-signs (blank compartments). On the level of distinctive features, the frequent presence of zerosigns is a characteristic of sign systems in general, and of technical sign systems in particular.
33In the case of printing and illuminated lettering, the development of a double articulation is motivated primarily by considerations of production. The decisive factors are technique and economy of production. The aesthetic aspect, concerning the reception of the signs, should of course not be suppressed. Unity in plurality and geometrical formation, too, are aesthetic moments. In
34the case of reading machines, however, the doubly articulated analysis is determined exclusively by considerations of reception. The writing systems, however constructed, whether holistically or digitally, exist. There is a machine that is projected and already in use (cf. Kurzweil 1978; Eckmiller 1969) that can decipher about three hundred current fonts and transpose them into another medium (either acoustic or haptic) which is accessible to the blind, that is, into speech or Braille. The success of such an “omni-font reading machine” is dependent upon the establishment of the invariant properties of each unit of the alphabet, as well as of the variations that are dependent upon factors such as the position and size of letters, continuous combinations of letters, ellipsis in the case of the fusion of two letters (cf. the ch in the case of Albers’ stencil lettering, Fig. 1), and the quality of the printing materials and the printing itself.
35Considering the range of possible variations discussed above, it is useless to fix on an absolute similarity of tokens of any one type. Where light and dark are distinctive features, it is misleading to orient oneself to an absolute range of lightness. In a grey and black composition the light part can be as dark as is the dark part in a white and grey composition. The figures for the number 7 in the Anglo-Saxon writing style and for the number 1 in the Continental-European are so similar as to cause confusion. Whether a given figure is a 1 or a 7 can be decided by the omni-font reading machine by no other means than those used by the human reader, that is, on the basis of the occurrence of the alternative figure in the context (say of a continental 7 with a transverse bar in the middle, or of an Anglo-Saxon 1 without a nearly horizontal bar at the top), or on the basis of some other kind of indication (say English text).
Concluding remarks
36Double articulation has, in the case of writing, a predominately economic rationale, to which ever-increasing importance has been attributed in the wake of mechanization. Aside from aesthetic considerations, which should not be underestimated, the clarity, and hence also the ease and reliability of discrimination are also important factors (seaman’s code, writing for the blind). With the increasing number of signs, and the diversity connected therewith, it becomes impossible to focus on a disordered multitude of holistic key features: an at least partial digital analysis, taking account of originally senseless elements, becomes indispensable (cf. the Far Eastern logograms and the omni-font reading machine).
37The sign systems adduced by opponents of the linguistic “myth of double articulation” are in many respects different from the writing systems presented here. Their examples show new sign systems consisting of composite signs constructed by making use of already given signs. The signs are converted from a sense-determinative to a sense-discriminative function. Signs that already have a current metaphorical function are preferred here, for example, numerals, playing cards (hearts, clubs, spades, diamonds), and the like, which, similarly, have been adduced as examples (Eco 1972, 239). In the case of the newer kinds of writing (printing and illuminated lettering), however, we are dealing with given sign systems, which in their traditional forms can be considered as being only in part doubly articulated, say in the case of single letter groups such as b, p, d, g. They are now broken down into discriminative elements which only partly and more or less by chance correspond to the given, normal shape properties that, in the traditional forms, function as the key features. Shapes are developed that are precise, though abstract, and are known not only in geometry but also for their diversified use in other sign systems. They stand, moreover, in a clear relationship to each other, in a relation|ship plausible also in terms of Gestalt theory. Three dots yield a continuous line in the morse alphabet, but a discontinuous line in the cited illuminated lettering.
38The seaman’s flag code cited by Bühler seems to hold an intermediate position. Despite the partial identity of its discriminative elements with those of Albers’ stencil lettering, it was classed with the first group. In this code, the geometrical figures all occur in isolation and in a prototypical position and shape, that is, they occur in such a way that they are generally familiar as iconic representations of themselves. In the stencil lettering, on the other hand, they occur in a non-prototypical position, in configura|tions that depict non-prototypical figures — overly long rectangles, for example — so that they are sometimes not recognizable as such at first glance.
39However much one is inclined to segment such squishes, so usual in semiotics, it should be beyond dispute that it is a matter not of a gradual, but, more radically, of a qualitative displacement: from sense-determinative to senseless units in order to obtain sense-discriminative units, from concretely iconic signs to more abstract (often geometrical) figures; and amongst these figures, from complex multidimensional shapes (sphere, square) to their elemental components (line, point); and this indeed both in the development of alternative sign systems (from the seaman’s flag code to the morse alphabet, from James Gills’ triangular modification of the ordinary alphabet for the use of the blind to Louis Braille’s writing for the blind), and also in the standardization of traditional alphanumeric systems (from stencil to illuminated lettering).
40Restricting the examination to the last two centuries, one could conjecture that universal regularities are at work in these prima facie plausible displacements, just as the development from semasiography to phonography is reflected also in the ontogenesis. But such a conjecture might be ethnocentric and rash. The counting tokens, which are the likely precursors of Near-Eastern writing systems, are not only, for the most part, abstract symbolic ideograms. As they increase in number and become more differentiated, it appears that there is an immediate emergence of abstract symbols, in addition to the concrete iconic signs on the tokens — dots and lines drawn in various directions, which function as sense-discriminative elements, that is, as signs lacking a constant and positive meaning. Abstract symbolization, such as the sense-discriminative function of signs, is not a prerogative accruing exclusively to the unconscious origin of natural languages, much less to the intense reflexion and formalization of the last two centuries. The Near-Eastern origin of writing also reflects and confirms the development from multidimensional to elemental forms.
