I consider Pure Meaning as the basis of language – meaning, which becomes embodied in form, that is, in structured sequences of acoustic waves. It should be noted that vibrations of the air, that is form, fundamentally differ from meaning in a very precise understanding: there is no way of deriving any meaning from mere vibrations of the air or from their representation as graphic symbols (letters, words, sentences) on a sheet of paper. Acoustic waves or their graphic representations are merely assignedto meaning so as to evoke it. It was the basic error of Chomsky’s so-called Generative Grammar to have missed this essential point.
On the other hand, it is perfectly possible to construct machines that totally ignore meaning. They transform the vibrations of air belonging to some language A into acoustic waves correlated with a language B (or do the same for their respective graphic representations). In the case of basic items like words this transformation is laid down in dictionaries where German ‘Baum’ is identified with English ‘tree’, without the dictionary being aware of any meaning. In the case of larger formal units like sentences, translation machines proceed in a similar manner. In many cases, the transformations of larger units like sentences do, however, lead to spurious results. In order to exclude such errors, reference to the linguistic environment is mostly sufficient. This means that a broader formal environment determines a narrower one, so that such reference mostly produces appropriate translations. As the broader environment still exclusively consists of other formal elements (acoustic waves or their graphic representations), translation machines may rely exclusively on formand be highly reliable (which is already true of the most developed among them).
Such basic formal procedure was initiated by Distributional Analysis, which has thus made an immense contribution to the pragmatic purpose of translating languages. But translation machines do not further our understanding of language – understanding is therefore quite a different matter. As Distributional Analysis could keep meaning strictly out of its way, it has on the contrary be an obstacle to such understanding. The reason should be perfectly clear. Meaning represents the very fundament of language.
Chomsky must have felt the shortcomings of his theory, as he tried to amplify it by means of a concept he called ‘deep structure’ – opposed to a surface-structure. If the first was to have any sense at all, it should refer to what lies at the bottom of form, namely meaning. But meaning could not be arrived at by purely formal Distributional Analysis, so Chomsky got stuck and soon abandoned Linguistics altogether – and it so happened that after him General Linguistics got stuck as well.
The impasse was obvious, for Chomsky was right and wrong at the same time. Meaning was of no use to the flourishing new science of computerized translation, so he could feel assured that he need not bother with it. On the other hand, it was Chomsky’s avowed and ultimate aim to explain man’s linguistic generativeness – and he was totally wrong when he believed that this goal could be reached without reference to what is the very core of language. In order to lead to a true theory of generativeness, the deep-structure would have to embrace what I call ‘Pure Meaning’, while the surface-structure would be pure form, that is sequences of acoustic waves (or their graphic representations) described by means of Distributional Analysis.
Let me illustrate this basic point right at the beginning.
First, historically (or phylogenetically): Animals already conceive reality and act accordingly, even if they do not translate these conceptions into auditory or any other signs and signals.
Second, ontogenetically: Infants add auditory signals (‘signifiants’ or words) only to concepts already present in their mind, otherwise they would pronounce empty sounds. Mere crying is, of course, meaningful too as it usually is the outward expression of pain.
Third, pathologically or ‚a contrario’: Deaf-mute persons (like Helen Keller) are equipped with ‚meaning’ independently of its formal realization in sound structures. The fact is proven by their ability to replace sounds with a sign language consisting of gestures.
Fourth, pragmatically through translation: When translating an English sentence into Chinese, I must, first, go back to its meaning before, subsequently, applying the specific rules governing its formal realization in Chinese (as mentioned above, computerized translation-machines proceed without reference to meaning).
And fifth, methodologically: The preceding considerations acquire their most general significance as soon as we switch to comparative linguistics. There are but two tertia comparationis between any too randomly chosen languages. These lie at the bottom of any specific translation as well as of any general statement about comparative linguistics and linguistic laws. The first tertium comparationis consists in Pure Meaning (that is meaning apart from and prior to any realization in form by sounds, letters, gestures etc.). The second tertium comparationis are the Formal Means at the disposition of human beings.
What such an approach aims at should be evident: It endeavors to explain, first, the particular generativeness of the single speaker of some given language, say English. What is it that enables him to create an infinite set of sentences after having been acquainted with only a finite number of such – mostly when being a child? Second, this approach wants to explain general generativeness. What are the necessary conditions and the constraints that govern the human capacity for creating any natural language whatsoever, that is those which have already been created and those he may still create?
Somebody has termed my approach ‘pragmatic’ – a misnomer. Computerized translations that transform a formal sequence of some language A into the formal sequence of some language B deserve such a characterization as they may be put to pragmatic use. The approach expounded in this book is not useful in this sense. If it has any merits at all, it is its capacity to make us understand the true nature of meaning and its relation to form in natural languages. Noam Chomsky had conceived the grandiose idea that a speaker must dispose of a set of rules if we want to explain his particular generativeness. Chomsky even hinted to general generativeness when referring to innate ideas at the base of human linguistic capacity. But Chomsky was unable to prove his point. Generative grammar based on distributionalism, that is on a purely formal procedure (like machine translation, cf. A, c, e, 4), is by its very method barred from proving what Chomsky wanted to prove.
Syntheses – the basic Units of Meaning
If the preceding assumptions are correct, the first task of comparative linguistics must consist in a definition of Pure Meaning, that is, its units and subunits. Obviously lexical semantic items can only be considered subunits as they fall short of the requirements of information. Nobody converses in the following way:
Cold, Peter, volcanoes, serenity, green etc.
Hence the real units of Pure Meaning are what I call ‚Types of Synthesis’, for example the Action-Synthesis /S, a/ or /S1, S2, S3, a/ or the Quality-Synthesis /S, q/ where the first two are made of substances modified by an action while the latter refers to a substance modified by a quality.
In English, these are formally realized in sentences like ‘Peter jumps’, ‘He gives the ball to Mary’ or ‘The wall is green’.
Thus, comparative linguistics starts with the following scheme consisting of three basic parts:
1) /Meaning/, which is ‘formally realized’ by means of:
2) definite formal units (words etc.) in:
3) structured formal sequences, the most elementary units of which we name ‘sentences’, as these normally represent what in the sphere of meaning are the basic ‘Types of Synthesis’.
What I call definite formal units are (a) mono- or polysyllabic sounds, (b) tones, (c) intonation and (d) position.
In natural languages semantic lexical items are for the most part formally realized by mono- or polysyllabic sounds called ‘words’. Tones may, however, substantially reduce the number of sound units used for lexical items, as happens for instance in Vietnamese or Chinese.
Intonation is often used to distinguish the functional appearance of a synthesis as assertion or question, assertion or doubt. According to how I pronounce it in German, the Action-Synthesis íS, aýfunctionally appearing as either /S, a/!or /S, a/?may be formally realized by two different intonations. According to intonation, the German sentence ‘Er kommt’ (He is coming) is then understood either as an assertion or a question. The same effect may be produced by means of position, for instance ‘Er kommt’ (He is coming) versus ‘Kommt er?’ (Is he coming?) or finally by means of a specific sound particle as in Japanese: ‘kuru’ versus ‘kuru ka?’ the first meaning “(he) comes”, the second ‘does (he) come?’
Position may thus be used as alternative formal device to express semantic relations like Agent and Patient or Assertion and Question but it may not be used in natural language to express different lexical items in the manner of artificial languages. Digital computer language expresses allpossible semantic differences by different sequences (positions) of just two signs + and -. What I call the ‘Differentiation Value’ (Dif-Val) of position is thus quite different in natural as compared to artificial languages. In the first instance this value is quite low while in the second it is almost infinite. The reverse holds true for sound units (words). In natural languages their Dif-Val may theoretically be limitless while it reaches its minimum in digital ones (+/-). The Dif-Val of Tones seems to reach a maximum of six in natural languages. There are, for instance, five tones in Chinese (high: ma1, rising ma2, falling-rising: ma3, falling: ma4and neutral: ma). Tones substantially reduce the number of elementary sound units (words appearing as syllables, in this instance). Indeed, Chinese only uses a fraction of those needed by languages without tones.
The Dif-Val of formal elements used in natural languages is responsible for the constraints operating in the formal realization of any possible language.
In other words, it delimits the boundaries of possible linguistic variety and thus allows us to formulate basic laws governing the formal realization of meaning.
Let me once again repeat what I assume to be the foundation of truly comparative linguistics:
1) /Meaning/, which is ‘formally realized’ by means of:
2) definite formal units (words, tones etc.) in:
3) structured formal sequences, the most elementary units of which we name ‘sentences’, as these normally represent what in the sphere of meaning are the basic ‘types of synthesis’.
This leads to the natural conclusion that truly comparative linguistics is necessarily made of two different compartments namely ‘Analytic Linguistics’ and its counterpart ‘Constructive Linguistics’.
The first deals with natural languages still in use or found in historical documents. It explains the particulargenerativeness of a speaker of English, Chinese or any other definite language. The second describes generalgenerativeness, namely how natural languages are developed by human beings when – obeying to the constraints of formal realization – they create any new idiom.
Its task is to describe how specific languages like Chinese or English realize Pure Meaning. But Pure Meaning is not a monolithic entity. It is rather composed of two quite distinct parts, one of which I want to call the ‘Logical Structure of Meaning’, while the other represents its informational aspect: the ‘Informational structure of Meaning’. These two taken together constitute the entire structure of meaning.
Let’s take the example of a simple Action Synthesis /S, a/. Like any other synthesis it may appear in two entirely different semantic shapes: either as ‘information’ or ‘non-information’ like in the two English sentences ‘Men eat rice’ and ‘Man eating rice (are usually tall)’. In the first instance, some fictitious man of the moon may just be explaining the eating habits of terrestrials. The speaker provides true information as he supposes to say something new to the listener. In both cases, the logical structure of meaning remains identical, it is the Action Synthesis, but its function as a piece of information is altogether different.
The second sentence ‘Man eating rice (are usually tall)’ contains two types of synthesis, first, an Action-Synthesis /S1, S2, a/NI, more specifically /men1, rice2, eat/NI, where ‘NI’ means ‘non-information-synthesis’; and, second, a Quality-Synthesis /S, q/Ithat is /men, tall/I, where ‘I’ means ‘information-synthesis’. The Action-Synthesis does not convey any information as the listener is supposed to know that there are men, who eat rice. Relevant information is only provided by the Quality-Synthesis, which states that these men are invariably tall.
Let me use a more convenient way to distinguish both functional types of synthesis by calling /synthesis/I a ‘free’ and /synthesis/NIa ‘bound’ synthesis, as it must always be part of another synthesis that conveys information. Further informational variants are statement versus question, topic versus novum (comment), the varieties of semantic effacement and so on.
The Formal Realization of Meaning
Meaning is the foundation of language, which form is meant to ‘realize’, that is to transform in material signs susceptible of being exchanged between the members of a linguistic community. Meaning as such – i.e. mental images formed in the heads of speakers and listeners – is totally distinct from form, just as form – acoustic waves or written letters – is from meaning. An interesting and intriguing aspect of natural languages is to be found in the fact that the formal realization of meaning may proceed in quite different ways. On the one hand, identicalunits of meaning may be formally realized in differentways, while, vice versa, identical formal means may embody more than one meaning. For instance, the above conjunction of two syntheses may be rendered in English in two alternative ways. ‘Men eating rice are usually tall’ or ‘Men, who eat rice, are usually tall’. In Chinese only the first of these formal alternatives is admitted leading to a sequence like ‘Eating rice men usually tall’. This is an instance of different formal realization of identical meaning, in this case the bound synthesis.
The case of one and the same formal pattern expressing more than one meaning is, for instance, to be found in the so-called English ‘relative clause’. The latter may express either an ‘information’ or a ‘non-information synthesis’. Take for instance ‘Peter, who (by the way) is a fantastic young lad, has my special approval’. Though identical in formal appearance to a bound synthesis, we are in fact faced with a parenthesis, that is an information- or free synthesis. The speaker wants to expressly inform the listener that he believes Peter to be a fantastic young man. He could have chosen the more usual formal realization: ‘Peter is a fantastic young man. He has my special approval.
Syntax versus Paratax
The most basic operation of formal realization does, however, not concern rules of syntax, that is the manner how a specific language formally embodies semantic relations like in an Action-Synthesis of the type /S1, S2, S3, a/I, where it gives rise to an English sentence like ‘He gave his bicycle to his brother’.
Much more elementary is the realization of Formal Paratax. In order to illustrate what I mean, let’s take three basic types of synthesis, the Action, the Quality and the Psychic-State-Synthesis (the latter being represented in English by sentences like “I feel”, “he believes”, “they are happy” etc.). In the notation of Pure Meaning let these be rendered as /S, a/I,/S, q/Iand /S, p/Irespectively.
Now, it is a peculiarity of English and a number of certain other but by no means of all languages that entirely different semantic classes like /S/, /a/, /q/ and /p/ may be formally realized either as eNouns or eAdjectives:
‘The womanis crying’, ‘the light is bright’ or ‘she is happy’
These represent the normal case. But all three may likewise fall into the same formal classof English nouns = eNouns. As witnessed by the following examples:
‘The woman is crying’, ‘Her crying is hard to bear’, ‘The brightness dazzles us’ and ‘Happiness makes you feel good’
In other words, entirely different semantic contents like Actions, Qualities and Psychic States may be grouped into one and the same formal class of the English noun as defined by preceding elements such as ‘the’, ‘a’, ‘some’ (or in a more general way by the method of ‘distributional analysis’).
/S/, /a/, /q/, /p/ ®eNoun
/woman/, /cry/, /bright/, /happy/ ®eNoun
Paratactic classification is quite different in a language like Chinese.
Paratax as the way of grouping different semantic classes in formal ones is part and parcel of formal realization at large – or rather it constitutes its very basis. And it is specific for each language. It will be seen later that differences in paratactic order are largely caused by ‘rank-lifting’.
Paratax is the logical counterpart of syntax, as the elements combined in the latter are made of classes. Paratactic formal classes are language-specific, so there can be no nouns as suchbut only English, Chinese, Japanese nouns, adjectives, verbs etc. This is an item of utmost importance, when it comes to evaluating the possible performance of Chomsky’s Generative Grammar. Only if the semantic contents grouped in paratactic classes were identical in all human languages would Generative Grammar, as conceived by Chomsky, would make a sense. If not, Chomsky’s Generative Grammar, instead of explaining the variety of languages, explains it away. Generally speaking, paratax as a basic and distinct procedure in every language has been all but overlooked in Traditional Grammar with the exception of ‘Distributional Analysis’, which, however, discarded meaning so that it cannot produce any results in the field of comparative linguistics.
Its scope is much more ambitious, though as a matter of fact it does but reproduce on the higher level of scientific theory the mental operations at work in any society where human beings create their own language out of Pure Meaning and Formal Means (doing so under the sway of certain definite constraints). Constructive Linguistics wants, first, to describe the field of arbitrariness in natural languages, while, in a further step, it aims at specifying the limitsof arbitrary variety, that is, the constraints of formal realization.
Taken together, Analytic and Constructive Linguistics cover the entire field of General Linguistics, which I alternatively call ‘General Grammar’ or ‘Semo-Formal Grammar’. In order to be complete, the above scheme describing my procedure has still to be widened so as to include paratax.
1) /Meaning/, which is ‘formally realized’ by means of:
2) definite formal units (words, tones etc.) in:
3) syntactically structured formal sequences called ‘sentences’, which consist of subunits that paratactically organize semantic classes into formal ones (traditionally called English, Chinese, Japanese etc. nouns, verbs etc.) in a language-specific way. These subunits are syntactically connected in a language-specific manner (so that German syntax is quite different from its Chinese counterpart even when both realize identical structures of meaning).
Let me conclude this basic outline of my theory with a further remark on Chomsky’s Generative Grammar. The latter may be thought to represent a powerful tool of Linguistics as it describes the rules governing the ‘generative’ process of some particular language. An English child having listened to a finite setof sentences uttered by his parents intuitively extracts the rules of their formation so that after some time it is capable of independently creating an infinite numberof sentences on its own – all based on the rules which Chomsky tried to make explicit. The same happens when a Japanese, a Spanish, a German child extracts the respective rules of their own languages.
Now, the shortcoming of Chomskyan theory lies in a rather basic omission. The child does not only learn the rules of syntax but those of paratax as well. In other words, it intuitively performs ‘distributional analysis’ learning this way what semantic classes may or may not be included in the formal class of, say, the English noun or the English verb. To my knowledge, Chomsky happened to be well versed only in the English and the Spanish languages. Had he known Chinese, he would have learned that in this language paratactic ordering of semantic classes into formal ones proceeds in an altogether different manner. Chomsky’s grammar, though representing a certain advance in comparison to previous traditional ones, is far from being an adequate tool even as an instrument of Analytic Linguistics. When we confront its Constructive counterpart, it must be criticized as a serious impediment on the path to linguistic progress. The concepts of nouns, verbs etc. being language-specific cannot possibly serve as Tertia Comparationis for Comparative Linguistics. It is inadmissible on logical grounds to consider them linguistic universals.
This is the introduction to ‚Principles of Language, revised 2017’ (http://gerojenner.com/mfilesm/Principles_revised_2017.pdf).