Language Theory for the Computer: 

Monodimensional Semantics or
Multidimensional Semiotics?
Reflections on M. Th. Rolland's book "Sprachverarbeitung durch Logo-technik" (1)

ISKO Journal   Knowl.Org. 23 (1996) No.3, p.´147-156, 10 refs.

Computer linguistics continues to be in need of an integrative language-theory model. Maria Theresia Rolland proposes such a model in her book "Sprachverarbeitung durch
Logotechnik" (1994). Relying upon the language theory of Leo Weisgerber, she pursues a pure "content oriented" approach, by which she understands an approach in terms of the
semantics of words. Starting from the "implications" of word-contents, she attempts to construct a complete grammar of the German language. The reviewer begins his comments with an
immanent critique, calling attention to a number of serious contradictions in Rolland's concept, among them, her refusal to take syntax into account despite its undeniably real presence.
In the second part of his comments, the reviewer then takes up his own semiotic language theory published in 1981, showing that semantics is but one of four semiotic dimensions of
language, the other dimensions being the sigmatic, the pragmatic and the syntactic. Without taking all four dimensions into account, no theory can offer an adequate integrative language
model. Indeed, without all four dimensions, one cannot even develop an adequate grammar of German sentence construction. The fourfold semiotic model discloses as well the universally
valid structures of language as the intersubjective expression of human self-awareness. Only on the basis of these universal structures, it is argued, is it possible to identify the specific
structures of a native-language, and that on all four levels. This position has important consequences for the problems of computer translation and the comparative study and use of


1. Major lines of Rolland's approach to language

For the diverse forms of machine-based language processing, in particular for the natural language dialog between man and
computer, we still have no generally accepted model (despite intensive research) with which to describe the characteristics of a
particular language. The lack of an appropriate model becomes painfully apparent in conjunction with the task of translating
between different languages. Maria Theresia Rolland's monumental study proports to solve this problem, first for the German
language, but indirectly also for other languages. Indeed, she claims that her study is not simply a partial contribution, but in fact
the decisive break-through, a solution almost exclusively in terms of pure semantics, i.e. based entirely on immanent
word-content, but claiming to cover every aspect of language. 

Rolland develops her proposals on the basis of the language-content research of the Bonner linguist Leo Weisgerber, who sees
himself in the tradition of Wilhelm von Humboldt when he asserts, that the reality accessible to a human being is constituted
solely in and by his language, his or her world-view being constructed essentially along the lines of the "inner form" of his or her
native-language. Rolland's claims for her position are set out in the following. 

"As the following elaborations will show, we have succeeded, on the basis of Weisgerber's (1962a: 13 ff) content-oriented approach (i.e. direct reference to the semantics of the
language) in identifying the rules and regularities and their underlying principles, which are constitutive for the German language. Furthermore, it is shown, that these principles hold
good for the structure of every other language, whereby, naturally, the specific concretization will vary from one language to the other, since each language has its own specific way of
grasping reality" (p. 41).

Oddly enough, despite her recognition of such "principles" of language structure, Rolland repeatedly and vigorously rejects the existence of universal language structures
(linguistic universals) or of universal grammar (p.10, 20f., 31, 257, 551f.).

After two introductory chapters with the titles "I. Topic of research" and "II. Viewpoints", which do nothing more than to outline the valence theory of the verb and the theory
of cases, the reader comes to the central chapter of the book "III. Logo-technique". The first part of this chapter concerns itself with word-classes (= "Wortarten") and
sentence-members (= "Satzglieder"). Here Rolland asserts, that the basic semantic rules, in accordance with which words function as the reality-defining carriers of meaning, are
fixed definitively in a small and thus manageable number of word-classes. She identifies six such classes: verbs, substantives, adjectives, adverbs, prepositions and
conjunctions. Each class, she claims, has its own typical inflection - an astonishing assertion for the grammarian. Thus the "inflection" of the preposition is the case of the
substantive it governs, the "inflection" of the conjunction is that of the verbal or nominative sentence-members, and for the "inflection" of the adverb the comparative form of
the so-called "adjective-adverbs" or of a few innate adverbs, e.g. "gern", "lieber", "am liebsten", is called into service (p.101, 173).

The word-class to which a word belongs determines its abilities to function in a distinct way as part of a sentence. Rolland distinguishes six types of functional
sentence-members (= "Satzglied"): predicate, subject, object, circumstantial determination, attribute, conjunctive determination. Although not identical with the word-classes,
these sentence-members are set in relation to them by the authoress, though not according to any recognizable principle (see the survey on p.343). Two word-classes have only
one such function: verbs function only as predicates, adjectives only as attributes. The other classes can exercise more than one function as sentence-members.

The remainder of this chapter "Logo-technique" (III, 5-11) is devoted to the structure of word-contents (= "Wortinhalt"), moving on then to the structures of syntagmas and
sentences (III, 12-13). What the authoress has in mind in this second part of the chapter is summed up in the following quotation.

"Within the patterns typical of the particular word-class, the word-content (= "Wortinhalt") is constituted. Word-content is a composite of special content, i.e. the meaning proper to this
particular word alone, and of general content, the meaning shared with other words of its class; such general content being subdivided along two general lines. Access to the meaning of
a word is provided by its implication (= "Implikat"), i.e. the word as viewed in the context of the structure of its surroundings in the sentence, the whole of which reflects the complex
specific meaning of the word. Corresponding to the two types of partial meaning, special and general, two types of partial implication are to be distinguished, complement and
supplement. Complement, corresponding to the special content, is to be found in the structural constellation of dependent introductory words and predicates and in the connections of
substantives and adverbs. The supplement, corresponding to general content, reveals itself according to the two lines of general meaning, the one, corresponding to the word's
inflectional patterns, constituting the inflexional group, and the other, corresponding to its constructional patterns, constituting the constructional group" (p.343).

Rolland distinguishes further the word-classes in accordance with "agreement in at least partial aspects of the complement" and "agreement in the inflexional group". All in all,
the authoress's intention is to show that every word contains within itself (implicitly) its own individual construction plan, according to whose rules and patterns it is to be
used in sentences or syntagmas. Thus verbs imply and condition sentential structures, words of the other classes imply and condition syntagmic structures.

The following chapter "IV Speech processing" operationalizes the foregoing theory in the form of computerized lists of word meanings (p.346-549). There is no point in
discussing these applications before arriving at an understanding of and agreement about the theoretical foundations.

What emerges from this wondersome construct of symmetrical relationships between word-contents, is in fact, though not perhaps in the intention of the authoress, a kind of
comprehensive semantic lexicon of the German language, listing all the possible usages of a word and thus making possible "knowledge interrogation" (= "Wissensabfrage")
aimed at disclosing all the possible elements of information stored in the German language. 

Rolland speaks of the "construction of a fully automated, natural language dialog-system" (p.552) and sums up: "Thus the understanding of the structure of language proves to
be the basis for solving problems in a wide variety of applications, from the practical uses of computer science, through the interpretation and use of language as such, on to the
theoretical explanation of thinking itself (p.556).

2. Immanent Criticism: Contradictions in Rolland's Theory and Practice

The authoress herself has not hesitated to describe her efforts as epoch-making or to extoll their virtues in the best advertizing style. The reviewer, by contrast, is obligated to
greater modesty. Right at the outset, let it be granted that a well worked out semantic lexicon of the German language could indeed be a meaningful and useful contribution.
Such a lexicon may not confine itself to only a few, often merely idiomatic uses of a word, as is the practice in most existing dictionaries, in particular in bilingual ones.
Instead, it must explicate the full panoply of possible meanings of each word, especially of the verbs. On the other hand, it does not lie in my competence to judge, whether
the projected utility of such a lexicon would in fact be sufficient to justify a publicly funded effort to achieve it. Such an evaluation must also take into account a comparison
with English language databases. 

Quite different questions, however, must also be answered. Would such a database for the semantics of the German language be sufficiently free of dogma to be practically
useful? Would such a database reflect the real structure of language? Is its grasp of linguistic structure sufficient to provide a solid foundation for the functions it proposes to
fulfill, e.g. translation, linguistic comparison, stylistic analysis etc. Responding to these questions, I propose to begin with an immanent criticism of Rolland's work,
measuring the claims she makes in terms of her own standards, in particular her claims to be contradiction free, solidly grounded and intersubjectively comprehensible. Only
when this immanent critique is finished, will I proceed to introduce what I believe to be a more comprehensive, more coherent, and more internally consistent theory of
language. However, I repeat, for emphasis: the immanent critique in the first part of this review of Rolland's work is entirely independent of the concept developed in the
second part of this paper. Moreover, the accent throughout is on the theory of language as such; questions of computer programming or the utility of "machine-based" language
processing will only be touched on by way of future prospects.

a) Contradiction between sophisticated conceptual systematics and inadequate conceptual definitions

The long quotation from Rolland's text reproduced in the first section of this article may well have impressed the reader as sounding quite meaningful; in fact, however, on
closer attention it proves to be considerably less intelligible. Personally, I must confess openly, that, despite my most intensive efforts of interpretation, the whole conceptual
system of the authoress remains incomprehensible. One reason for this is that the concepts she uses are almost never defined. Instead they are introduced suggestively in the
course of developing a thought, thus evoking the impression of having something definite to say. In fact, however, it is this very definiteness which is lacking. What is an
"implication" distinguished according to "special content" and "general content"? One might expect that the lengthy glossary at the end of the book would help the confused
reader out of his/her predicament. Let us see! In the glossary, "complement" is defined as "that part of the implication, with the help of which the special content can be
identified" (p.575). This definition refers us to "special content". The glossary defines "content" as "the intellectual side of linguistic instruments" (p.574). This is at least a
rough identification, though it is problematical, since the pragmatic element implicit in the concept "linguistic instruments", in short the "intention", is also an intellectual
element. The frequent explanation of "content" (= "Inhalt") through "intellectual component" (= "Geistiges") is not only homely and old-fashioned, but also lop-sided and
directly false. Later in this paper, I will come back to the equally "intellectual" side of the speech act, which is studied by linguistic pragmatics; at the moment, however, it is
the authoress's notion of "special content", which is under discussion. This is defined as "a partial content, which signifies the particular component of the word's content"
(p.583). "Particular" (= "eigenständig") is here contrasted to "general content", i.e. "that part of the word's content which the word shares with other members of its class"
(p.573). This much one might have figured out oneself. But has one come any closer to clarifying the notion of "complement"? And what of its pendant "supplement", defined
as "that part of the implication, with the help of which the general content can be identified" (p.584)? Let us turn then to the crucial notion of "implication" (= "Implikat"):
"the structure underlying a particular word in a particular language, see also complement, supplement" (p.574). This is going in circles, and one may be forgiven for thinking
that the circles are vicious. When such a suspicion is substantiated in numerous, clearly defined examples and concepts, then the whole system with its often astonishing
symmetries proves to be little more than a house-of-cards drawn from a deck of the authoress's own making. Such a construction can hardly be regarded as containing real
knowledge about linguistic structures.

As further examples of Rolland's circular reasoning, I shall next take up her definition of "substantive" and its corresponding sentence member and then her definitions of
"adjective" and "attribute". With these examples, we find ourselves on a more concrete level of linguistic phenomena than that of the above constructs.

b) Contradiction between the claim to pure semantics and the actually syntactical definition of word classes

One could perhaps dismiss the preceding criticism as a formalistic critique based on divergent notions of conceptual and definitory clarity. With the notion of word-classes, by
contrast, we are dealing with one of the main supporting pillars of Rolland's whole system: words belong to classes. How are "word-classes" (= "Wortarten") defined? She
writes: "The decisive and universal criterion for the distinction of the word-classes among themselves is the function which the words of a particular class exercise as members
of a sentence" (p.56). Later, we find the following definition completing the circle: "The characteristic of the substantive is its ability to function as a direct member of a
sentence, either as a direct subject, a direct object or a direct circumstance. The term 'direct' means derived from a substantive..." (p.83).

Equally circular is the definition of "adjective" in terms of "attribute" and "attribute" in terms of "adjective": "The characteristic of the word-class adjective is its ability to
function as a special sentence-member, namely as an attribute" (p.94). With this claim, the authoress has already per definitionem (falsam) excluded predicate adjectives and
predicate nouns from the class of adjectives, treating them as adverbs, because, in German (by contrast to Latin), they happen not to be declined. In doing this, she ignores
what Hans Glinz, another Weisgerber pupil (not mentioned by Rolland!) already in the 50's had claimed to identify as a characteristic of the "inner form of the German
language"1 (2). Furthermore, the circular definition of the adjective in terms of the attributive function and then of the attribute in terms of the adjectival word-class (p.95) leads
the authoress to transform the genitive attribute, e.g. "Das Buch des Lehrers" (= "the teacher's book") into a genitive object, because this interpretation better fits the symmetry
of her tables of word-classes and sentence-members. 

This example reveals a whole list of typical errors: e.g. she defines the word-class "adjective" in a way doing violence to linguistic usage and anything but "semantically", she
defines the sentence-member "attribute" in a false and equally un-semantical manner (in fact, sentence-members can only be defined in terms of the whole of the syntactic
sentence complex); in an illogical, circular manner, she defines one in terms of the other; in general, she defines word-classes through their function as sentence-members. But if
there is anything which qualifies as "syntactic", it is the definition and function of the sentence-members. Here in fully unjustified manner, they are treated as word-class
implications of supposedly pure semantics. 

As long as the theoretical basis of Rolland's analyses remains unexplained and ungrounded, any further study of her numerous "beautiful" tables is pointless; these surveys are
mere houses-of-cards constructed by arranging time and again the same self-made playing-cards in ever new artificial patterns arbitrarily postulated rather than empirically

Here and there, one encounters suggestions of how word-classes can be defined semantically; indeed, in one case rather late in the book, a whole list of semantic definitions of
word-classes is offered: verbs signify a "process" (= "Process"), substantives a "something" (= "Gegenstand"), adjectives a "condition" (= "Zustand"), adverbs a "circumstance"
(= "Umstand"), prepositions a "relationship to" (= "Beziehung zu"), conjunctions a "connection between" (= "Verbindung zwischen") (p.257). These more or less correct
semantic definitions of word-classes would, however, naturally lead to quite different conclusions regarding the word-class definitions of the sentence-members. Pars pro toto,
this can be illustrated by the example of the predicate adjective or more specifically the predicate noun. Contrary to Rolland's opinion, there really exists an adjectival,
non-adverbial attribute to the predicate, as is illustrated by the assertion "Ich trinke den Kaffee schwarz" (= "I drink coffee black") or by the familiar hymn-verse "Der Wald
steht schwarz und schweiget" (i.e. "the forest stands black and keeps silent"). In Rolland's system, such constructions have no place; they are treated as an adverb of manner, as
though the forest really "stands" in a "black manner" or the drinking occurs in a black manner. Such predicate attributes, like their cousins the predicate nouns, do not cease to
be adjectives simply because their position in a sentence is not that of the usual substantive attribute. That such adjectival sentence-members remain adjectivistic is clearly
demonstrated by the ease with which they can be transformed: the demonstrative or explanative statement "a beautiful tree!" can easily be transformed into "This tree is
beautiful!" Such transformation through transposition is not possible, where a member of one word-class is to be replaced by a member of a different word-class.

Rolland shows no awareness of the problems with her definitions of word-class and sentence-member. As a consequence, she is unable to communicate clear understanding,
contenting herself with arbitrary constructions in the guise of apparently symmetrical tables. In short, her subdivision of "word-classes" and "word-types" leads only to an
unpalatable mixture of semantic and syntactic viewpoints contradicting her own claim to pure semantics.

Had Rolland clearly and consistently distinguished between word-class, which is defined semantically, and sentence-member, which must be defined syntactically, she need
not, for example, have subsumed the article, the pronoun and the name under the class of substantives. Here again one sees her penchant for house-of-cards symmetries: she
speaks of "processual" "conditional", and "copulative" substantives (= "Verlaufs-", "Zustands-" und "Kopula"-Substantive), because allegedly there exist parallel distinctions
with verbs and adjectives, indeed with all six word-classes. Behind all this terminological nonsense one recognizes the problem of the relationship between semantics and
syntax, a problem which the authoress herself refuses to face, because supposedly, i.e. according to Weisgerber, everything can be explained alone in terms of the
"word-content" (= "Wortinhalt") and its implications. This leads to the strange phenomenon, that Rolland attempts to construct a complete German grammar without taking
account of syntax, indeed without even defining the difference between semantics and syntax. When, however, these two linguistic dimensions are not distinguished, semantics
itself suffers, since semantics represents the conceptual logic of the linguistic units (words, in particular). The specific idiom of the native-language may well play with this
logic, but it by no means replaces or destroys it.

c) The contradiction between the claim to holistic perspective and the denial of the "pragmatic"

When one is forced to speak of a denial and repression of the syntactic dimension in this book, the next question is, how does the book treat the theme which has been in the
forefront of attention since the 60's and 70's, the topic of linguistic pragmatics, in short the "speech act" theory? While it is true, that in language systems everything is
definable in terms of relations, our authoress recognizes only dual oppositions. Thus she reduces the original spectrum of speech acts to but two, declarative and interrogative
sentences (p.291-). Where do expletives, wishes, self-portrayals, and the so-called perlocative speech acts like promises or such performative expressions of an executive
character as e.g. nomination, baptism etc.) fit into the authoress's scheme? A theory of language claiming to be holistic - and this Rolland aims quite emphatically - cannot fail
to take account of linguistic pragmatics. "Dialog" with a computer, which is unable to understand typical interpersonal figures of speech such as threats and dissembling
expressions like irony is in principle reduced one-sidedly to mere data-bank functions, i.e. to mere normative speech. Perhaps such a reduction may be necessary at the outset,
but then one must openly avow the conscious character of such a restriction, all the more when one claims to follow Humboldt's energeia conception of language, as the
authoress does repeatedly.

When Humboldt namely, as Rolland cites him, insists that language is not a ready-made "ergon" but rather a continually active "energeia" (cf. p.41f.), then this implies the
Kantian shift from the object to the transcendental, i.e. the practice oriented conditions for the very possiblity of objectifying information2. Even when one understands
language in the sense of "langue" as an intellectual intermediary world (= "Zwischenwelt"), as an intermediating reality (= "mediale Wirklichkeit") - and this the authoress
does with full right - the fact remains, that both the construction of this intermediary world and its receptive reconstruction in the course of using language are in fact actions,
i.e. acts in the broad sense of the cognitive processes involved in "acts of understanding" as Kant would say. With Humboldt, many ideas remain at the level of the initial
empirical differentiation of linguistic plurality and, on the theoretical side, on the level of intuitive programmatics. Thus his expressions are particularly suited to ceremonious
but vague quotation. Nevertheless, his emphasis on the energeia-character of language demands, without doubt, in the spirit of Kant, that language be conceived as a system of
activities. The reduction of language to semantics, i.e. the objectivising, or better the already objectified dimension of language in fact directly contradicts Humboldt's own
energeia-postulate. This critique holds already for Weisgerber's position. Regretably, the early transcendental philosophers, the German idealists, had themselves failed to
analyze language sufficiently from the point of view of action theory (= "handlungstheoretisch"); this is due principally to their failure to recognize the plurality of semiotic

From this wide notion of pragmatics in the sense of action theory a narrower notion in the sense of interpersonal action through language must be distinguished. For in fact,
only in the interpersonal dimension does language become immediately practical. Again, I repeat, this practice is not reducible to the simple opposition between declarative
and interrogative sentences. Where such reduction is made, one should not speak of linguistic computers with "dialog abilities" or "computer dialog". As long as this
pragmatic or dialogical dimension of language is not taken into account, such a manner of speaking is not, in more senses than one, "linguistically conscious"3.

In an aside, it should be noted here, that Rolland in no way takes into account the metaphorical, artistic meta-linguistic4 usage of language, which is rooted in day-to-day
language-games. Instead, she postulates apodictic rules of correctness, after the fashion "this is possible, that is not". Such rules tend to sound like carping criticism, e.g.
"They conversed for hours" is admissable, but "they conversed for years" should not be (p.211). Personally, I find the second sentence much more interesting. Admittedly,
such a sentence may presuppose a prior, "normal" manner of speaking; different levels of speaking must no doubt be distinguished, but there are no grounds for setting up
prohibitions or for programming the computer as dialog partner to admit the one sentence and reject the other.

d) Contradiction between a specific native-speech notion of semantics and a broader logical notion

The deceivingly simple and self-evident notion of "semantics" is nowhere explicated by Rolland. Is the "semantic" identical with the "intellectual" (= "geistige") content of a
word, a syntagma or a sentence in the logical sense? Rolland starts from a supposed "unity" or "wholeness" (= "Einheit" or "Ganzheit") of sound and meaning, sensuality and
sense, and criticizes quite correctly those who (in positivistic or behavioristic manner) postulate an immediate relationship between the sound of a word and the non-linguistic
world (p.51). On the other hand, she puts such emphasis on the "intellectual" in the sense of the logical/conceptual, that she does not hesitate to treat even slightly divergent
usages of one and the same word as pure and simple homonyms, i.e. as different words unrelated except in their chance like-sounding pronunciation. Here a sample in the
original German text with translation:

"So gibt es u.a. viele Präpositionen 'aus':

Er ging aus dem Haus (von welchem Ort?)

Er trank aus der Tasse (woraus?)

Er stammt aus dem Ruhrgebiet (woher?)

Er handelte aus Verzweiflung (aus welchem Grund?)

Ein Buch aus dem vorigen Jahrhundert (aus welcher Zeit?)

Ein Tisch aus Holz (aus welchem Material?)

Ein Bild aus dem Nachlaß (aus welchem Besitz?) usw."


"Thus there are many different 'from' prepositions:

He went from the house (from what place?)

He drank from the cup (from of what object?)

He stems from the Ruhr (from what provenience?)

He acted from dispair (from what basis?)

The book from the last century (from what time?)

The table from wood (from what material?)

The picture from the estate (from what possession?)


The logician will be delighted by the way the authoress here differentiates the diverse meanings of the word "aus", treating them as intellectually unrelated homonyms. On the
other hand, he must call her attention to the fact that, in order to explain these ostensibly separate words, she falls back on the very word she seeks to explain. This is a clear
case of circular reasoning, for which the authoress has an obvious penchant. (Such circular reasoning reveals itself as well in the intricate network of cross-references from the
present to later discussions and from later to earlier discussions. For the reasoning behind a particular statement, she almost invariably refers the reader to later discussions.
Then, in the later discussion, the topic is said to have been explained in the earlier passages.)

When, in terms of "pure" logic, a plurality of meanings are treated as entirely separate concepts, though subsumed in a particular language under one and the same
like-sounding term, this should be a clear warning that "native-speech" makes use of analogical thinking which is logically anything but "pure", giving place to both
similarity and dissimilarity. The question is, which semantics should we use, that of pure logic or that of analogical thinking. The latter is the semantics of "native-speech".
Rolland, a self-styled advocate of native-speech in the school of Weisgerber, here does violence to that very native-speech, when she treats analogically related meanings of one
and the same sounding word as though they were entirely separate words, and only by chance homonyms.

Herein lies the complete inconsistency of Rolland's postulated "unity and wholeness" (= "Einheit" und "Ganzheit") linking sound and meaning! Her failure to bring together
her own ideas has serious consequences and raises a host of questions: to what extent is semantics to be understood as a linguistic interpretation of words at the level of
native-language? To what extent is it a logical interpretation of concepts independant of their particular expression in a native-language? What is to be said of the proported
complete dependance of thinking on language?

Could it be that thinking - even granting that it is normally articulated in interpersonal linguistic form - transcends the unity of sound and meaning after all, in perception, in
feeling as self-perception, in thinking as such, to say nothing of intuition? Could it be that the once fashionable professorial thesis about the complete immanence of thinking
in language and the so-called "Unhintergehbarkeit der Sprache" - a thesis often posed on the authority of Humboldt - is obsolete after all? Had one really done language and
language analysis a service by making or trying to make it the quintessence of thinking in all its forms? But Rolland herself by no means consistently follows the ideology of
the linguistic immanence of thinking which she articulates (p.53). In the end, it is by no means clear, what her "interrogation per computer-'Dialog'" has to do with the
German language. The suspicion arises, that this "natural language" interrogation is in fact little more than an aid for users insufficiently in command of the English language.
In any case, the attempt to clarify the specific native-language grasp of reality is quite incompatible with the homonomy passion which manifests itself throughout the book.

e) Contradiction between the authoress's fundamental claims and her unphilosophical denial of linguistic universals

The insufficiently explained relationship between the logical/conceptual element and the native-language element in the authoress's notion of semantics is closely related to the
question of the unity underlying the diversity of mankind's native-languages. Already at the beginning of this article, I called attention to Rolland's denial of universal
linguistic structures and to her claim that the "principles" of her own version of language theory are valid "evidently for the structure of every language, ... albeit naturally in
different concretisations proper to each specific language" (p.41). "Naturally" indeed! What else is meant by linguistic universals than just such principles? If the authoress is
not in a position herself to work out such principles, this by no means gives her the logical licence, in contradiction to her own propositions, to deny their universal-linguistic
character. What forces her into such self-contradiction? Certainly not her positive, native-language program. The answer to this question lies in the historical severance between
language research, later linguistics, and philosophy. Much could be said at this point, but I will confine myself to one point alone: it is absolutely impossible to develop a
"holistic" theory of language without doing philosophy, because, first of all, language is the privileged instrument of human encounter with reality and of human thinking as a
whole, inasmuch as thinking articulates and communicates itself intersubjectively, and, secondly, precisely because language is in fact energeia and not ergon. For this reason,
the role of language must be interpreted by reconstructing its underlying structures in terms of action theory (= "handlungstheoretisch"). For the same reason, language theory
must address itself in philosophical terms to the full complex of the phenomenon meaning. Philosophy is, after all, to use the words of Kant, the universally oriented "art of
the concepts" or the science of meaning (= die aufs Ganze gehende 'Kunst der Begriffe' oder auch die Wissenschaft vom Sinn). This means, of course, on the opposite side, that
the "philosophers" must descend from their ivory towers, or, to put it better, that those who intend to philosophize must undertake to reconstruct language on the basis of the
principles of consciousness. Then they will no longer be tempted to withdraw into the notorious ivory tower of the "guardians of being", where holistic and concrete knowing
never did take place. (Much the same must be said for the reflection on social structure and the relationship of philosophy to sociology.)

3. Critique from the standpoint of a reflective language theory of the semiotic dimensions

In 1981, the author of this review himself published a philosophical theory of language as the second part of his study "Reflexionstheoretische Semiotik" (6). The first part of
such a philosophical semiotic, a "study of the process of meaning" (= "Sinnprozeßlehre") consists of an action theory (= "Handlungstheorie"), i.e. a study of the semantics of
human actions according to their respectively constitutive intentionality (6). In this context, language is interpreted semiotically, i.e. in accord with a theory of signs, as a
meta-action, characterized by the fact that it regulates itself in the course of performance by means of its own syntactic meta-symbols. This theory of language, which is at one
and the same time semiotic and philosophical-holistic, takes up the distinction between diverse semiotic dimensions as elaborated by Charles Morris; these are the syntactic,
the semantic and the pragmatic dimensions of language (7). In addition, however, a fourth dimension of language is distinguished as elaborated by the former GDR philosopher
and semiotics-expert Georg Klaus; this is the sigmatic dimension, i.e. the realm of denotation or object-relatedness of the signs (8). In my study, these four semiotic
dimensions of language are explicated according to a principle relating them one to another and defining them in terms of increasing cumulative reflexivity:

     1. the sigmatic or denotative dimension: the original relationship of the speaker to non-linguistic reality

     2. the semantic or meaning-dimension: the relationship of the speaker to an already established intermediate reality of conceptual content (presupposing the first

     3. the pragmatic or intersubjective action dimension: the relationship between speakers by means of semantics

     4. the syntactic or connectional dimension; the relationship between linguistic signs, presupposing and reflecting the three previous dimensions.

Although the sequence, or better the hierarchical order of these four dimensions is grounded in the increasing levels of reflexivity, the sequence can be reversed when it is a
matter of practice. In fact, the hierarchical viewpoint is fully compatible with a circular viewpoint as the following diagram shows:


2. 4.


What is at stake here is nothing less than the recognition and elaboration of the thesis that the principle of human self-awareness, i.e. self-reflexivity, is likewise the
foundational and constructive principle of language: Language is the intersubjective self-expression of human self-awareness. It is self-evident, that the basic linguistic
structures must be just as universal as general human self-awareness itself is universal. Nevertheless, such universal linguistic basic structures are realized only contingently in
the diverse concrete native-languages. In this view, the supposed opposition between native languages and universal linguistic structures is rejected as undialectical and abstract
and un-thought-out. (That human beings of different races share the same basic anatomy, is in fact more astonishing than the fact that their languages, despite their obvious
diversity, manifest the same principles and fundamental structures.) Each native-language is a unique, contingent "incarnation" or application of the universal linguistic
structures. There are no fundamental difficulties with the distinction between the universal, generally human and logically necessary level on the one hand and the contingent,
individual linguistic level on the other.

Everything depends upon the reciprocal "inter-penetration" of the four semiotic levels. Iin the light of theoretical reflection, it can be shown that there are no more and no less
than four such dimensions. This goal is served by the method of dialectical subsumption. In contrast to the usual formal subsumption of the individual under the general,
"dialectical subsumption" designates the ordering of the general or comprehensive under its particular determinations, in such a way that the particular determinations or
subordinate distinctions themselves are further differentiated according to the principal criteria of differentiation. At stake here is the "harmonic" or "holographic" principle of the
reflection of the whole in its individual constituents, a principle not unknown in intellectual history. The dialectical subsumption of the language dimensions within each
other, sometimes in repeated further sub-differentiation, can here only be sketched schematically and in terms of a single step of sub-differentiation.

     1. The sigmatic dimension (denominative dimension)

     1.1 sigmatic sigmatics (the perceptibility of the sign-bearer)

     1.2 semantic sigmatics ([the character of] the sign-bearer as the bearer of meaning)

     1.3 pragmatic sigmatics (the localization of the linguistic sign in the context of action)

     1.4 syntactic sigmatics (the determination of the linguistic sign within its system).

Under the heading 1.3, an important problem is treated, which particularly occupied L. Wittgenstein, although he mixed it up with other "pragmatic" inquiries under the rather
indefinite title "usage"; this is the question: How do the linguistic signs originally acquire their meaning as reference, i.e. as a relationship to the non-linguistic reality or at
least, even when it is a borderline case of self-referring relationship to linguistic reality, to an intended reality other than itself? The four forms of acquiring reference (1.3.1
through object-related; 1.3.2 through subject-related; 1.3.3 through socially-related and 1.3.4 through auto-referential language games) cannot be further discussed here. The
point is to illustrate the uniform, though by no means schematic-formalistic principle that reigns in language as a developed, dynamic system of action and reflection.

In connection with Rolland's proported purely semantic project, the semantic dimension of language deserves special attention. Within it, the following subdivisions arise in
virtue of the application of dialectical subsumption.

     2. The semantic dimension

     2.1 Sigmatic semantics: identifiers (pronouns and names)

     2.2 Semantic semantics: descriptors (word-classes)

     2.3 Pragmatic semantics: logical predicate-classes

     2.4 Syntactic semantics: combined predication (the logic of conditional sentences)

The attentive reader will immediately recognize the fundamental differences between my approach and Rolland's "semantics". Here the word-classes really are introduced
semantically (without borrowing from an otherwise dissavowed syntax). Not through mere empirical fact gathering, but rather through logical reconstruction of the empirical
evidence, two generic groups of word types are identified. The first is the sigmatic-deictic group (2.1 in the above table), composed of pronouns and proper names, whose
function is to point to or to stand for objects. The second, the semantic group (2.2) is properly "descriptive" in character and includes four word-classes with their own proper
semantics, namely, substantives, adjectives, verbs, and situators with adverbs, prepositions and conjunctions as sub-classes. What real semantics of word-classes implies,
illustrating as well how universal logic and native-language particularity interpenetrate, manifests itself in the further subdivision of the descriptorial word-classes (see (6) Pt.2,
p.114-167). For example, relying entirely on word-content and not, as with Rolland having recourse to word-structure - the detailed demonstration would take us too far afield
here - it is possible to subdivide the class of substantives into object-substantives, characteristics-substantives, process-substantives and idea-substantives.

The next group (2.3) consists not of isolated words but rather of types of predication understood as the semantic synthesis of word-contents. This is a properly logical problem
and is identical with Kant's theory of the categories. The fourth group (2.4) corresponds to syntactical semantics and includes the logical possibilities of combining
predications, i.e. the logic of conjunctional sentences. This is the theme of modern junctor-logic. Note that all this is only remotely related to specifically linguistic, i.e.
native-language syntax; here we are dealing "only" with the general logical foundations of such native-language syntax. On the other hand, the further subdivision of the
different word-classes leads to a sifting of native-language vocabulary. And the comparative study of diverse languages in terms of such a general standard of comparison will
reveal the significant differences in the world-views corresponding to different native languages.

In order to further clarify the relationship between semantics and syntax, let us skip over, for the time being, the pragmatic dimension, which is number 3 in the four-fold
scheme above, and go directly to number 4 in that scheme, the syntactic dimension.

     4. The syntactic dimension

     4.1 sigmatic syntax: principles of morphology

     4.2 semantic syntax: principles of sentence construction

     4.3 pragmatic syntax: principles of text composition

     4.4 syntactic syntax: principles of style (rhetorical figures)

The principles of sentence construction elaborated in semantic syntax (4.2) likewise contain a logical, universal linguistic scheme of potential sentence parts. Here too, the way
this universal logic is realized concretely in a particular native-language remains quite open. Thus, in the syntactically constructed sentences in any language whatsoever, one
universally finds a subject-predicate core, which may be combined with one or more of the four primary sentence components: objects, adverbials, identifiers, and modifiers.
Here too, the way particular languages realize these possibilities of logical syntax can vary considerably and is by no means predetermined. Among the instruments used are:
inflexion, rules of congruence, rules for word order, etc.

In principle, word-classes and their syntactic functions as sentence-members are variable with respect to each other, i.e. in principle, each word-class can fulfill the function of
any sentence-member, even though individual word-classes may well have preferential functions (both generally and in specific languages). This virtually boundless variability
between word-classes and sentence-members is well illustrated by the German language: "Geben ist besser als nehmen" (= "To give is better than to receive"), "Für ist besser
als gegen" (= "For is better than against"). The functions of the subject and the equation-member in such sentences can be filled not only by substantives and adjectives, but
also by verbs and even prepositions. (Facts like this do not fit into Rolland's apodictic rules based on an insufficiently understood relationship between word-class and

Which of the many possible forms of expression for the universal grammatical structures are actually used by a particular language is a matter for the study of the individual
language and can at best be generalized in terms of empirical language typology. On the other hand, without recourse to universal linguistic structures, it is impossible to
explain native-language syntax satisfactorily or to demonstrate the simplicity of its structures, aside from the particularities of morphology. I am well aware of the
methodological breadth and critical implications of this claim with respect to existing grammar studies and am prepared to deliver corresponding proofs, on the basis of the
theory of language, which this article can only sketch. Thus any German sentence, no matter how complicated, can, provided it is understandable and grammatically correct. be
interpreted and expressed graphically as a combination of simple syntactic basic diagrams and can be represented optically in accordance with the additional distinction of
primary, secondary and tertiary sentence parts which can only be mentioned here.

Granted, that what is here postulated, on the basis of philosophical, universally human structures of meaning, must be verified in detail for each individual concrete language.
However, the prospects for such validation are at least as good or better than those for the validation of anatomical correspondences between Australian aborigines and human
beings of European extraction. In an age of world-wide communication within the one human race, it should be more than legitimate to call attention to such universal
grammatical structures underlying the wide spectrum of the native-language variety of the language of one mankind.

The demonstration of a universal linguistic grammar raises a monumental claim and opens a much wider perspective than that offered by mere native-language semantics. This
is the claim to facilitate machine-based translation from one native-language into another on the basis of common (universal) syntactic fundamental structures, in short,
computerized translation. Clearly, such a procedure must take into account the specific semantics of the languages in question, but in basing itself on a common underlying
"depth grammar", it goes well beyond the mere consideration of word-fields and "word-implications"6.

"Depth grammar" takes on the meaning - comparable to N. Chomsky's7 use of the term - of a universal, logically grounded grammatical structure, in relation to which the
native language formulations constitute a surface level of expression. To designate this level, the term "expression level" (= "Ausdrucksebene") seems to me to be most
appropriate, since this is the real level of language as contrasted with the underlying level of logic. In this "colorful reflection" of a connective logical deep structure we find the
real vitality of native-language. However, this distinction between universal logical deep structure and contingent linguistic expressive structure must in fact be drawn
repeatedly for each of the semiotic dimensions described above. Every native-language, or better every native-language family, has its own way of expressing the universal logic
inherent in the denotative, the conceptual, the interactive and the syntactic-connectual dimensions. Precisely the last of these, the syntactic dimension can be called the formal
or expressive dimension of language par excellence. At the same time, it represents the specific systematizing dimension of language, in which the language reflects back upon
itself and in stylized play with itself becomes artistic language.

It was a major error of the pragmatics boom of the Sixties and Seventies to exault language pragmatics as though it were the all-comprehensive dimension of language, an error
into which both Morris and Klaus fell. Behind this error lay a confusion of the two meanings of "pragmatics" distinguished above, pragmatics as linguistic action-theory in the
comprehensive sense (= "handlungstheoretisch überhaupt") and a more specific meaning focusing on the "social" or interpersonal side of language. One of the sources of this
confusion was the demand raised by the 1968-Movement, to view language like every other phenomena in its social and political context.

Understood in this narrower sense of the social or interpersonal pragmatic dimension, the following forms must be distinguished:

     3. The pragmatic dimension

     3.1 sigmatic information pragmatics

     3.2 semantic expression pragmatics

     3.3 pragmatic reception pragmatics

     3.4 syntactic role pragmatics

These forms are reflected back in language as a self-referent syntactic system, i.e. as style. The one-sided theorists failed to explain - had they even recognized the problem -
why the rules of grammar and meta-grammatical style are not the rules-of-the-game for social action as such. The reason for this failure lay in their failure to distinguish
between social interaction as such and linguistic interaction as a meta-activity with respect to such action.

In the present review of Rolland's work, it is not the one-sided pragmatic-political understanding of language which is in the forefront, but rather the one-sided semantic
understanding (which is coupled with Rolland's failure even to recognize the existence of the other dimensions). It is not the intention of this paper to set off against such
approaches an equally one-sided syntactic approach. What is here at stake is a holistic-semiotic approach, which does justice to all four primary dimensions of language:

     (1) the deictic relationship to reality as such;

     (2) the semantic relationsip to the intermediate reality of meaning;

     (3) the relationship of social interaction; and lastly

     (4) the relationship to a self-referent system of grammar and style.

Within such a holistic view, the demands placed upon computer linguistics are significantly higher. At the same time it must be said, that, except in the case of such highly
specialized tasks such as "informational interrogation" (= "Wissensabfrage"), only such a holistic approach is in a position to fulfill such expectations. Till now, the realization
of this program has failed, because such a theory of language sounds too philosophical for the linguists and too linguistic for the philosophers. No wonder, then, that such a
program has failed to come to the attention of the computer linguists, just as it failed to catch the attention of the authoress of "Sprachverarbeitung durch Logotechnik". Perhaps
her "logo-technology" could serve as a first step, the semantic step so to speak, towards a holistic system of language processing, but to do this, it must be freed from its
problematical trimmings and reduced to its valid core, the lexicon of word-usage.

The decisive step, the first step needed to open the way to such a holistic view of language, will be the use of computers to model language as a multilevel reflexive system, as
far as this may prove possible. Cybernetic auto-reflexivity (11) will never be in position to achieve, much less replace, the four reciprocal dimensions of self-reflexion, which
constitute human self-consciousness. Nevertheless, for those who understand, what is involved here is not the creation of computers which can think and speak as human
beings do, but rather the development of computers which can process language in a way that is comparable to meaningful human speech. Such computers need not operate on
all levels simultaneously, a rapid sequential shift from one level to the other according to the logical progression from object-relation, self-relation, social-relation and
system-relation through language will suffice. When will the encounter between linguistics and the revolutionary information technology of our time finally bear fruit? One
thing is sure, only then, when linguistics and the philosophical theory of meaning once again are reunited9.


1 The evident falsity of Rolland's position becomes obvious when she compares the Latin adverb "pulchriter" in the construction "pulchriter cantavit", in German "er sang schön" (= "he
sang beautifully") with the German predicate-noun "schön sein" (= "to be beautiful"), concluding that the German predicate noun is in fact an adverb. This goes far beyond the assertion it
was meant to support. Thus the authoress makes the claim, that in the German sentence "Das schöne Buch liegt auf dem Tisch" (= "The beautiful book lies on the table"), the word "schön"
(= "beautiful") is indeed an adjective, but that in the sentence "Das Buch ist schön" (= "The book is beautiful"), the word "schön" (= "beautiful") functions as an adverb.

2 Humboldt's contemporary J. G. Fichte called "objectifying" the fundamental function of all language, whereby he himself was primarily interested in the transcendental conditions of
objectification. See my contribution in (3).

3 See my contribution in (4).

4 For the notion of art as meta-language and language as meta-action, see my study in (5).

5 For more on this point, see my paper under (9).

6 Prof. Dr. Heinz Hamm of the Sophia University in Tokio has taken up the problem of the translation of reflexive-theoretical language theory; he is evidently convinced of the utility of
this theory for trans-linguistic and trans-cultural translation.

7 At first sight, Noam Chomsky’s "generative transformation grammar", with its dichotomous (dualistic) branching structures, appears to corresopond well with the binary principle of
the computer. At closer sight, however, it becomes clear that speech, as the expression of human self-awareness, is - except for certain special aspects like phonetics - by no means
constructed according to a dichotomous binary logic. Instead, speech follows the four-value logic of human self-reflexivity. The challenge of constructing a mathematical formalism to
represent the logic of reflection was recognized and taken up, albeit still inadequately by the logician Gotthard Günther (see (10)) and some of his students (see (11)). On the other hand,
the processual reflective logic of language by no means requires logical formalization to be reconstructable by the computer. Technical operationalizing does not necessarily entail
logical formalizing. Thank goodness! Otherwise we would probably have to wait even longer to reap the fruits of the difficult re-encounter between logic and philosophy.
Independently of such considerations, the reconstructability of speech processes in terms of information-technology is grounded, I believe, in another aspect of computer technology,
namely the cybernetic aspect, which is analogous to the principle of reflexivity (see Footnote 12) and independent of the binary principle. The four-fold structure of human self-reflexion
outlined, though not extensively demonstrated in the course of this paper (see Footnote 9 and the work there cited), can very likely be simulated by the quasi-reflexivity of a cybernetic
information hierarchy. Viewed from this angle, the computer would appreat to relate more closely to the model of language in terms of reflection-logic than it related to Chomsky’s
dichotomous binary model, which in fact has failed as a theory of language. 

8 The parallel between cybernetics and the problem of transcendental philosophical reflection was first thematized by Gotthard Günther in (10).

9 N. Luhmann has spoken repeatedly of "reflexive mechanisms", but has failed to grasp what makes self-reflection unique, namely the identity of the two related beings and the
relationship joining them (the knower, the known and the act of knowing). With this failure, however, he is in good company of the philosophers D. Henrich, M. Frank and their
disciples, who believe they have refuted the reflection-theory of self-awareness. When the social sciences show a deficit with respect to a "concept-culture", this is for the most part due to
deficits in the foundational discipline philosophy (Cf. Dahlberg, I. in (13)).

10 Once again, I repeat, meaning is not just a matter of language meaning. One does language a disservice, when one tries to make it the quintessence of everything knowable or
thinkable. The "linguistic turn" can only succeed, when it is seen as an expression of and partial realization of the Kantian "transcendental turn". The mistaken attempt to view it as the
replacement of a philosophy of consciousness by a philosophy of language has time and again found advocates from Humboldt via Wittgenstein and his followers in analytical
philosophy to Habermas, but inevitably proves to be unfruitful. Rather than reducing thinking and consciousness to language, the real task is to conceive language as related in
thinking to the structures of consciousness.


     (1) Rolland, Maria Theresia: Sprachverarbeitung durch Logotechnik. Sprachtheorie - Methodik - Anwendungen, Bonn, F. Dümmler,1994, 497p.

     (2) Glinz, H.: Die innere Form des Deutschen, Düsseldorf, Schwann, 1959.

     (3) Heinrichs, J.: Nationalsprache und Sprachnation. In: Fichte-Studien 2 (Kosmopolitanismus und Nationalidee). ed. K. Hammacher u.d., Amsterdam, Rodopi, 1990, p. 51-73.

     (4) Heinrichs, J.: Dialog über Dialoganalyse. Ernst W.B. Hess-Lüttichs 'Grundlagen der Dialoglinguistik' in kritischer Diskussion. Kodikas/Code 6(1983)p. 369-385.

     (5) Heinrichs, J.: Handlung - Sprache - Kunst - Mystik. Skizze ihres Zusammenhangs in einer reflexions-theoretischen Semiotik. Kodikas/Code 6(1983)p.245-62.

     (6) Heinrichs, J.: Reflexionstheoretische Semiotik. Teil 1: Handlungstheorie, Bonn, Bouvier, 1980, 192p. and Teil 2: Sprachtheorie, Philosophische Grammatik der
     semiotischen Dimensionen. Bonn: Bouvier 1981. 490 p.

     (7) Morris, C.: Signs, Language and Behavior, New York, Braziller, 1955.

     (8) Klaus, G.: Die Macht des Wortes. Ein erkenntnistheoretisch-pragmatisches Traktat, Berlin-Ost, 1974.

     (9) Heinrichs, J.: Die Logik der Vernunftkritik. Kants Kategorienlehre in ihrer aktuellen Bedeutung, Tübingen: UTB Franke 1986. 286p.

     (10) Günther, G.: Beiträge zur Grundlegung einer operationsfähigen Dialektik. 3 vols. Hamburg: Meiner 1976-1979-1980.

     (11) Kotzmann, E. (Ed.): Gotthard Günther - Technique, Logic, Technology. München-Wien: Profil 1994.

     (12) Günther, G.: Das Bewußtsein der Maschinen. Krefeld und Baden-Baden: Agis 1963.

     (13) Dahlberg, I.: Zur Begriffskultur in den Sozialwissenschaften: Lassen sich ihre Probleme lösen? Ethik u. Sozialwissenschaften 7(1996)No.1, p.3-91

PD Dr. Johannes Heinrichs (b.1942) studied philosophy, theology, and the German Language in Munich, Bochum, Frankfurt, Paris. His doctoral dissertation (1972, Bonn) was the
Hegel-Study "Die Logik der Phänomenologie des Geistes" and his habilitation in philosophy (1975). He teaches philosophy and social ecology at several universities. He published
10 books and about 50 papers in journals and serials, also in language theory. 
Zurück zu den Publikationen Zurück zu home