29 May 2006

Of course children tell lies about sexual abuse – lying is unavoidably a part of human language

A technical explanation


By Marianne Haslev Skånland, professor
University of Bergen, Norway

* * * * *
This article is an extension of a talk given at the conference "För barnets bästa - ett kritiskt perspektiv" ("In the best interest of the child - a critical perspective"), arranged by witness psychologist Lena Hellblom Sjögren for Nordiskt Tvärfackligt Forum för Rättssäkerhet i Sexualbrottmål (Nordic Interdisciplinary Forum for the Rule of Law in Cases of Alleged Sexual Abuse). The conference took place at Skeppsholmen in Stockholm, Sweden, on the 24th and 25th August 1996. The article has previously been published in a slightly different form in the conference report which came out in September 1996. I has also been published, in its present form, on the 26 May 2006, by the r-b-v Forum in the
Section for English Speakers, listed under the title Children lie about sexual abuse.

Over the 10 years since 1996 there have been many impressive advances regarding theories of the origin and development of human language, which have, I believe, outdated central points of the paragraph about the evolution of language in the article. Cf the reference Bickerton
: Language and Species in the literature list, as well as Chris Knight, Michael Studdert-Kennedy, James R. Hurford (2000): The Evolutionary Emergence of Language. Social Function and the Origins of Linguistic Form. Cambridge: Cambridge University Press. ISBN 0-521-78696-7. I hope to follow up with an article about some aspects of this issue later.

* * * * *

Social workers and psychologists dealing with cases of alleged sexual abuse of children sometimes advance the view that children do not lie about sexual matters.

This opinion is of course quite sensational. Probably every human society up to now has known better. Simple observation, or thinking back to our own childhood with its often vivid fantasies of every kind and with plenty of secrets kept from adults, will normally be enough for us to assess the quality of any claim that children always tell the realistic truth about matters of emotional importance to them.

In addition, however, human language is known to possess certain properties which clearly support our scepticism, properties which can in fact explain why children as well as adults are fully capable of lying about just about anything, and why it is, in a sense, of advantage to our use and mastery of language to do so. In this talk I want to survey briefly a few of the most important such properties. My exposition may perhaps be felt to be somewhat dry in its technical parts but I hope it will still be fairly easy to follow. Philology, after all, is a long-established discipline which has engaged the interest of people working in many fields all through our documented history. Many characteristics of language were known and fairly well understood in Europe as early as in antiquity, in India probably even earlier. Certain aspects of language structure were intensively investigated and debated in the Middle Ages. Some have been rejected in certain periods and schools of thought, only to be rediscovered or reinstated later, sometimes in a more sophisticated form.


All communication takes place by means of signs, which are definable as combinations of expression and content. The expression is something perceivable through one of the senses, the content is the meaning which is conveyed from the sender to the recipient by means of the expression.

Signs can be analysed in several ways. Of particular interest in our present context is the following:

Signs fall, first of all, into two categories:
natural and conventional.

In natural signs the bond between expression and content is natural and there is no real dividing line between the two. Therefore, natural signs are not primarily signs at all, though they may be interpreted as such by an observer who understands the close relationship between expression and content.

As an example we can think of dark clouds signalling the probable coming of rain. The dark colour is a part of the chemical process that leads to the falling of rain; the colour is not there primarily for our benefit to remind us to bring our umbrellas, although we may utilise the colour of the clouds as a signal of what is coming. Likewise, running a temperature when we are ill is not primarily a signal to tell us, or the doctor, that we are ill, it is part of the illness and of the way the body reacts e.g. to an infection. When we observe this natural relationship and draw conclusions accordingly, we treat body temperature as a sign.

A conventional sign, on the other hand, is the bringing together, for the purpose of making a sign, of an expression and a content which are not intrinsically tied to each other. The bond between expression and content therefore has to be created by means of some convention, which can be explicit in the form of a formulated rule or law, or less clear-cut in the shape of custom or habitual use.

Next we must distinguish iconic signs from arbitrary signs. An iconic sign is a sign in which expression and content resemble each other (icon means "picture").

Since there is no proper dividing line between expression and content in a natural sign, such signs are always iconic - the expression resembles the content of which it is actually a natural part.

Some conventional signs are iconic too. A painting by Rembrandt resembles that which it is a picture of, and is therefore iconic. Other conventional signs are not iconic: A triangular traffic sign warning us that a crossing road has the right of way does not resemble what it means, although the use of red colour as a warning of danger/prohibition has at least an iconic basis, given that red is the colour of blood and that living beings tend to perceive blood as a sign of danger.

Signs which are not iconic are called arbitrary, meaning simply that there is no similarity between content and expression. Since there is such similarity in the case of natural signs, it follows that only conventional signs can be arbitrary.

Sign systems exist which combine iconic and arbitrary signification in quite intricate patterns.

Signs differ in how they are understood to be signs. Some natural signs are understood by humans through instincts. They therefore do not require learning and are no burden on our memory. For other natural signs, though, we have to learn what they are signs of. This leaning consists in understanding natural relationships and natural connections, and is independent of life in a community. Such leaning is still a relatively small burden on memory.

All conventional signs require learning in a community. For iconic conventional signs, learning is fairly easy and constitutes a relatively small burden on memory. For arbitrary signs the learning process is more difficult and demanding and is a considerable burden on our memory.

With the far greater difficulty of learning and remembering arbitrary signs, we might ask why one bothers using such signs at all. The answer lies in the different possibilities for communication which different types of signs allow: Natural signs which are understood through instincts can only be used to communicate by matters we can perceive through those same instincts, and only in situations in which the instinct actually triggers the reaction. Other natural signs can still only be used to communicate about natural relationships. Conventional signs give a richer repertoire of possibilities. Iconic conventional signs, though, can still only be used to communicate about matters which are amenable to iconic representation. Arbitrary signs have the tremendous advantage that they allow us to communicate about everything no matter what.

Human language

Human language is of course a system of communication. The signs of a language are primarily realised through speech, so their expression is sound produced by our vocal organs and perceived aurally. Any iconic signs must therefore have a content of audible sound.

The signs of language are always conventional and usually arbitrary. There is no advantage gained by - no functional reason why - the concept "horse" should be called
horse; the same concept is expressed equally well through Norwegian hest or Turkish at. All languages are learned without conscious effort and with equal ease by the children growing up in the speech community, although different parts of the language system may be acquired earlier in one language than comparable parts in another language. In fact, simple observation of the way in which different languages are equally capable of taking care of our needs for communication, totally speaking, is immediate proof of the conventional and arbitrary character of linguistic signs.

Most languages have some iconic signs, called onomatopoeia, e.g.
bang to denote an explosion, tick tock for the noise of a clock. However, they are in no way essential to the way language operates. A language can function in just the same way with arbitrary signs instead of onomatopoeia, but it could not function with only onomatopoeia and without arbitrary signs. Onomatopoeia are furthermore few and rather random, as again a comparison across languages will show. They are also subject to change through time - onomatopoetic words can undergo regular sound changes along with other vocabulary and thereby lose their onomatopoetic character. Furthermore, even within the area of clearly onomatopoetically based words, there is considerable arbitrariness. For example, the noise made by pigs is expressed in English as oink oink, in Norwegian as nøff nøff, and in Russian as chrjo chrjo.

The reason why onomatopoeia play such an insignificant part in language is clearly that they restrict us to talking about the comparatively few concrete things that make a noise, while we want and need to communicate about a vast number of other matters, some of them highly abstract.

There is even more arbitrariness connected with linguistic signs: that of the relationship between the sign and the extra-linguistic world.

Regarding the expression, we observe that while the types of sound which we produce when speaking, certainly utilise the available sound
substance – the kinds of sound which our speech organs are capable of articulating and which our ears are capable of perceiving – , the precise way in which these sounds are formed into a structured system is not wholly motivated by the sound substance, indeed not by anything outside of the system itself and its history (cf what I shall say later about phoneme categories in different languages).

Regarding the content of signs, its measure of independence of the outer world is even easier to demonstrate, by looking at the different ways in which different languages organise parts of the world into concepts. For example, Vietnamese has just one word meaning "mouse/rat", while our languages have two (incidentally, the Vietnamese classification, which mirrors a "folk biology", seems to coincide better with scientific biology on this point). What we consider to be different varieties of one type of fruit – melon – have different names and are considered as completely different fruits by several language communities in the Middle Eastern area. The colour concepts of Welsh differ from those of e.g. English and Scandinavian: Welsh "gwyrdd" corresponds to some nuances of our "green", but other shades of "green", plus everything we call "blue", plus some nuances of our "grey" are all considered to be "glas" in Welsh. The rest of our colour "grey" plus everything we call "brown" are called "llwyd" in Welsh. It is of course not the case that speakers of Welsh perceive colour nuances differently from Scandinavians in the physical sense, but their language groups that which is perceived, differently and thereby influences people's categorisation of what they perceive.

Generally speaking, lexical differentiation is at its greatest in semantic areas which are of culturally recognised importance in the community. Thus, the Lapps have a rich vocabulary for snow of different texture and condition, while societies in which snow is seen only on the top of mountains in the distance usually make do with one word for it. Differences not of degree of differentiation but of grouping into concepts are frequent in the anthropological field of kinship: While in my society my father's brother's children and my father's sister's children are grouped together as "cousins" and distinct from my own "brothers" and "sisters", in some societies father's brother's children may be considered the same kind of relatives as one's own brothers and sisters while father's sister's children are a different category altogether.

The advantage of having arbitrary signs clearly lies in the great power and flexibility this gives the communication system: anything - concrete or abstract, real or imaginary - can be formed into concepts and talked about. Arbitrariness means that linguistic signs are
abstractions from the world they refer to and from the spoken sound substance they utilise.

Arbitrary signs, however, require learning. In the case of language, the vocabulary of simple signs may run into several tens of thousands, and the complexity of the system by which we make composite signs (complex words, phrases, sentences) is considerable - a symptom of this is the fact that there is no general agreement among linguists about how a number of grammatical structures should be analysed.

Although human beings are endowed with the ability to learn language fast and doubtless with a strong drive to do so, the size and complexity of what is to be learned is such that mastery takes considerable time and effort. So does the acquiring of all the culturally stored knowledge and information which is transmitted to new generations by means of language. This again means that children normally need to spend quite a lot of time constructing, analysing, and to some degree practicing, utterances of varying complexity. A lot of this must needs be take place in situations to which the utterances have no direct and simple relationship. Children simply cannot wait around for a completely "natural" situation to turn up for everything linguistic and cultural they need to learn, and some things are better learned
without experience: We would want to understand the meaning of "Don't walk too close to the precipice" without any live demonstration of why.

Some further characteristics of the language system

Three properties are particularly important in enabling us to utilise language as a general communication system in human society:

Two-layered structure
A simple sign is one which cannot be divided into still smaller signs. In language, the expression part of simple signs is nevertheless constructed out of smaller units, units which are not signs. The term for such a sound-unit is phoneme. A phoneme can be defined as a class of sounds which can differentiate meaning, i.e. keep signs apart. Example: The possibility of making new words, with other meanings, as in a substitution set like Norwegian

[ l a t ]
[ l a k ]
[ l a m ]
[ l a p ] lapp
[ l a b ]
[ l a s ] lass

demonstrates that [t], [k], [m], [p], [b] and [s] all belong to different phonemes in Norwegian. (When we want to make clear that we are talking not about sound substance but about phonemes - classes of sounds - we write them in slants, for example: /t/, /k/, /m/, /p/, /b/, /s/ are phonemes of Norwegian.) In Polynesian languages, on the other hand, a substitution of [b] for [p] will never change the meaning of a word, not in any context at all; hence [p] and [b] belong to the same phoneme in these languages.

The letters of alphabetic writing systems are based on phonemes, although for various reasons the orthography in various languages is normally not perfectly phonemic.

The effect of having signs which are not opaque but are constructed out of phoneme combinations, is to increase greatly the number of possible simple signs. With only opaque signs, it is doubtful whether we could keep apart much more than a very few hundred signs. When signs have phoneme structure, and with no particular limit to their length, the combinatorial possibilities given by even a moderate phoneme inventory are far above what we need. For instance, Norwegian has above 5000 phonemically distinct one-syllable words. Longer words, even those that are simple signs, are several times as many, and there are still unused combinations available when new concepts demand new names. The phoneme systems of known languages of the world vary between 12 and about 82, and even 12 is enough to serve a perfectly normal language.

Productivity is the possibility of communicating something new by means of the old system. Human language is notoriously productive.

Although phoneme systems certainly change through history, they are usually fairly stable over several generations. The speaker who wants to say something new can therefore not achieve this by inventing new phonemes. New phoneme
combinations, on the other hand, do enter language quite often as new signs, mostly in the form of loan-words. I would like to emphasise that the process of borrowing vocabulary from other languages is in no way abnormal, though many linguists in the 19th century thought so and this romantic purism is still current in some schools of philological thought as well as among many laymen.

The fact that linguistic signs are arbitrary provides the explanation of why vocabulary enlargement and change through borrowing can take place with no ill effects on the language as a system: One sound-shape is as good an expression as another, functionally speaking, so long as the speakers keep unchanged a sufficiently large part of their language to be able to continue communicating in spite of the ongoing changes. Practically anything in language may change but comprehensive change must take place slowly.

But by far the most important kind of productivity is that which takes place in the production of complex utterances. We may truly say that almost every sentence we produce or hear is new to us. This is not because the same sentence may not have been uttered before - it may have - but because our mastery and understanding of it does not spring from the recall of an earlier occurrence, nor from memory in any other way, but from active operational construction and analysis of its grammatical structure. Whether or not we have heard a particular sentence before is irrelevant to our understanding of what it means. We have to remember the patterns of construction of the language and we have to remember the simple signs. In addition we often do remember some complex signs - composite words and short phrases - as wholes. But we do not clutter up our brains by storing long sentences as indivisible wholes. If we did, it would cancel out the richness and flexibility which arbitrariness and two-layered structure make possible.

A natural sign is bound to the so-called deictic situation, the situation here and now of which the sign is a part. A conventional sign, on the other hand, is not a symptom - a natural response - to anything, so not only can it be removed from the situation it refers to, but it can be transplanted to other situations, i.e. can refer to matters wholly unconnected with the situation in which the sign is used. When such removal takes place, the sign is displaced.

Many linguistic utterances of course bear some relation to the deictic situation. This often makes us overlook the degree of displacement present at the same time. Everything we say is displaceable, most of what we say is to some extent actually displaced. The same goes for the thinking we do in language form. This freedom to decide whether to speak and to decide what to say presents itself to us as an exercise of free will.

This again means that any attempt to describe or analyse language along simple, behaviouristic lines is doomed to failure. Even if the relationship between situations and the utterances produced in them were fundamentally deterministic, which is a matter of conviction to some, this relationship - of utterances, the system underlying them and the deliberations motivating them, on the one hand, and the situations in which speech occurs, on the other - would have to be considered so complex as to defy any cause-and-effect

In fact linguistic behaviourism, understood as the belief in a simple stimulus-response relationship between utterances and the situational context in which they are uttered, invaded American linguistics in the 1930s and onward, with devastating impoverishment of the whole discipline as a result. In spite of a would-be conscious revolt against it since the 1960s, its effects have still not been overcome, even by declared anti-behaviourists like Noam Chomsky. Behaviourism's most important legacy in American-style linguistics is perhaps the lack of any sign concept and only superficial understanding of arbitrariness. The consequence is that many scholars are bogged down in endless confusion of expression with content and of concrete with abstract.

Realistically considered, human society as it is could not exist without displaced language use.

Learning language

Babies go through a babbling stage in which they produce a rich variety of sounds. This is followed by a stage in which the variety of articulations goes down and instead categories of phonemes are constructed. The start is the formation of the class "vowel" as against the class "consonant". The first consonant is closed (no open passage through the mouth) while the first vowel is very open – [æ] or [a]. Early differentiation of consonant types establishes /p/ - /m/ - /t/. Early syllables are of the type CV (one consonant followed by one vowel). Further differentiation of consonants and vowels follows certain patterns which are acoustically, auditorily and articulatorily based, but also subject to variation depending on the phoneme system of the adult community.

From the time the child starts using sound categories in this structured way, it attaches its sound sequences to entities in its surroundings, i.e. it uses its sound productions as signs. Early signs refer to important features of the child's life, hence the familiar meanings of [mama], [papa] or [baba], [tata] or [dada] around the world (though which of these expressions is attached to which meaning is arbitrary, thus
mama means "father" in some languages, and dadda means "nanny" in Norwegian while daddy means "father" in English).

Subsequent development brings concept differentiation and enrichment. On the grammatical side there is at first simple juxtaposition of words whose relationship is additive or rather unstructured, this may be followed by a "starting again" when the child reverts to short combinations of two or three words but this time with a structured relationship between them (e.g. preposition plus governed noun). At this time, the child may discard its earlier longer utterances made by the additive principle. Its speech may therefore for a period appear to be regressing, but this is simply because the child is trying out new strategies, it is not a symptom of brain disturbance or psychological problems.

Now, the development of a language from the simple start to a fully developed adult version requires, as I mentioned a little while ago, considerable time and analysis without the support of any "natural" situation. This is not least true of the development of a phoneme system and a grammatical system, which are even more abstract and independent of the extra-linguistic world than are concepts.

Furthermore, since the use of human language is essentially displaceable and displaced, children cannot enter into the speech community unless they use language in the displaced manner of their interlocutors. But when an utterance is displaced from the immediate situation, its truth content cannot be checked by reference to the immediate situation. The truth or untruth of it therefore becomes a complex matter to assess. This is so even for very innocent pronouncements which we make all the time: "I thought John looked slightly ill today.", "Mary is very likely on her way here already.", "The weather this summer is nicer than two years ago." Only long term gathering and analysis of information enable us to guess sensibly about the probable truth of a statement like "I've baked some dandelions into a pie and put it in the fridge for your supper" uttered by our spouse. It all depends!

The daily use of language, by children and adults alike, therefore implies a continuous assessment of probabilities, and the necessity of living with as great uncertainties attached to the linguistic information we give and receive as those we experience in other parts of our lives.

The use of human language inevitably means displacement, and displacement leads away from the plain truth of mute actions and over into a more abstract existence, from which the continual possibility of untruthful utterances cannot be eradicated.

Lying, however, cannot be generalised without limit. Philosophers often investigate the ethics of an action, say theft, by asking what would happen if everybody were to steal continually. The answer is most likely that society would break down. We can ask an analogous question: What would happen if everyone lied in a similar way about something? Now bearing in mind that any arbitrary sign may be made to convey a particular meaning as well as any other, the answer is probably different: Society would not disintegrate, instead the lying utterances would change their meaning. We can see the essence of such change in what has happened to words like "people's democracy" in communist countries, to "child care" and "sexual abuse" and "familiehjem" in the community of social workers, to "culture" in the pop industry.

Lying is therefore parasitic on truth; it undermines truth but cannot exist independently of it. The stable, conventionalised meaning of the large majority of utterances is established through truthful use, use in which the meaning of the utterance is in harmony with the facts it refers to in the world outside of language.

Probably change of meaning through lying is not greatly different from change of meaning due to people gradually using words with a different content: Norwegian "saks" began as the term for "(piece of) rock", but has in the course of the technological change come to mean "pair of scissors". An "alternative" meant, to start with, "the other of two possibilities" but has for so long been used to mean "any other possibility" that this meaning must be accepted as the current one; hence we freely say e.g. "There are five alternatives".

Lying and other kinds of "untruths"

Displacement has as a consequence that true and untrue statements are linguistically equal. They are constructed in the same way, analysed and understood in the same way, pronounced in the same way. The difference lies in the relationship between the meaning of the utterance and the extra-linguistic matters they refer to, and this relationship is by no means always clear-cut.

Untrue statements are of at least three kinds, depending on the knowledge and attitude of the speaker to the truth of what he says:

Deliberate lies are uttered by a speaker who is aware of a mismatch between reality and what he says, and who wants his listener to believe in what the speaker says instead of in reality.

A variant of deliberate lying is that of telling something one believes to be a lie but which actually happens to be true, although one is not aware of it.

Saying something untrue in good faith is the communicating of information which we believe to be true but which in fact is not. This kind of falsehood is extremely common. Of particular interest to us at this seminar is the fact that it is very prominent in speculations of all kinds and even in science, where the subject matter is often so difficult to understand and check that certainty about the real state of affairs may be unavailable to us.

Irony is the telling of a deliberate lie but with the intention that the listener shall look through it and understand that it is a lie. Irony is sometimes signalled by a special tone of voice, by facial expression etc, but a very refined irony is that in which the speaker gives no such clue to his real meaning. He treads difficult ground, however, when balancing between giving information in this elegant but indirect way, and failing to get his meaning across.

But lying statements are by no means the only utterances that are not true. Large parts of language have no truth-value at all attached to them and are therefore not "true":

Questions are neither true nor untrue. "Did you go to Munich last week?" may elicit a true or a lying answer but is in itself neither. Nor is a command; if I tell you "Go and get me some coffee!" you cannot accuse me of lying but neither have you any guarantee that I really want the coffee. Nor does that alter with your response; the command is the same whether or not you choose to comply with it.

More surprising is perhaps the fact that even
some types of statement have no truth value in the usual sense.

This is so for statements which have a vague relationship to reality: "Well, perhaps the mail is delayed."

Another group are so-called
performative statements, statements which perform what they mean. An example is a baptismal: "I baptise this ship Victory". My act of baptising may be accepted and successful or not, but it cannot really be a lie (though it is perhaps unwarranted), and so we may also doubt whether it can be said to be true.

Yet another variant is found in sentences like "The present king of France is married." This is certainly not a true statement, but the "lying" consists not in saying something untrue about the king of France but rather in unfairly making it appear that you and your listener are in agreement that the topic of the sentence ("the present king of France") is accepted as a self-evident starting point.

Of even greater interest to philosophers through the ages, and in our times to computer scientists, who are very active in trying to understand phenomena like cognition, intelligence, consciousness and free will, are self-referential statements which may create vicious circles. The classical example is that in which I, a Norwegian, say "All Norwegians always lie." If this is true, I have uttered a true statement so the sentence is false. But if it is false, the statement was not true after all. The same kind of paradox has apparently non-linguistic varieties: A hair-dresser cuts the hair of everyone who does not cut their hair themselves. Does he cut his own hair?

The implications for logic and language of self-referential statements are interesting. I shall return to this question shortly.

Some evolutionary aspects

A substance – an item in the world – cannot be a lie; it simply exists. Only signs can lie, because only something taken as a sign can convey a message about something else, refer to something else.

How can falsehood arise in a sign system? How has it become part of human language?

There have been many fanciful speculations about the origin of human language, such as suggestions that all words have developed from a single word meaning "to beat" or "to eat", from the Hebrew word for "God", and so on. While such suggestions can be dismissed, positive evidence of early linguistic prehistory is harder to come by. The process of language change makes it impossible to reconstruct any language family or single language further back than, at a guess, about 4-6000 years, and then only with some help from written records. Such reconstruction only yields another language of exactly the same type as our well-known, documented ones. There are no techniques available for reconstructing concretely and in detail any development of human language from a different type of communicative system.

We can, however, propose reasonable hypotheses, on a more general level, about the kind of system this may have been and about how language could have changed from such another system type into what it is now, with its present-day characteristic features.

One suggestion which I should perhaps mention simply because it has a way of cropping up from time to time, is that language has evolved from onomatopoetic signs. It is certain that we can imitate many of the sounds we hear around us with some degree of success. Some such sound-imitation could bestow survival value, such as imitating the sound of a roaring lion to warn the group that a lion is near. However, the majority of sound-imitation that we may attempt: of birds singing, water falling, dogs barking, is not such as to benefit us greatly. Therefore there would probably be insufficient selectional pressure to favour clever imitators and develop simple imitation further. Nor do we observe in other primates any tendency to communicate in this imitative way.

Given the importance of communication with other members of the same species, human language is far more likely to have evolved from a simpler system of largely natural signs of the type that are in operation among the great apes today. (Studies of apes over the last decades have increasingly brought to light that apes may understand, and use, some of the arbitrary signs and grammatical structures of human language, although in their use of it, apes have to substitute our expression level of articulated sounds by visual means: they put tags of different colours and shapes up on boards, etc. Still, their own sign systems based on sound are presumably largely instinct-governed, i.e. natural.)

If this is so, then the acquiring of such properties as conventionality, arbitrariness, displacement, productivity, two-layered structure, and grammatical structure (beyond juxtaposition with additive meaning) must have been central for the development of language, since they are precisely the features that are characteristically lacking in a natural sign system.

From my discussion of them earlier it will be seen that several of these features are intimately related, one following from the other. Displacement implies arbitrary, conventional signs and may actually have established conventionality and arbitrariness historically. Thus, if we can provide a plausible explanation of how displacement could have arisen, we may not have to come up with separate explanations for arbitrariness and conventionality. I also believe productivity, two-layered structure and grammatical structure to be closely linked and perhaps of common origin.

In fact, although human language is far removed from an instinctual, here-and-now-bound system, the steps by which it has evolved out of such a system need not have been very complex.

Suppose, for instance, that a group of early hominids have in their "vocabulary" two signs which sound something like [wapo], meaning "food" and [kitsi], meaning "lion". (At this early stage they must be considered unitary signs. They may have something resembling a phonetic profile but in the absence of comparable signs which are partially similar, there is no way they could be segmented into anything like phonemes by the users. My spelling of them as a sequence of distinct sound shapes is therefore not really appropriate and should only be taken to roughly indicate their pronunciation.)

Then, if a hominid caught sight of a lovely source of food but at the same time spotted a lion nearby, he would have cause to signal both types of information to his mates: [wapokitsi] would be a way of saying "food and lion". At this stage [kitsiwapo] would be a way of saying "food and lion" too, the meaning of the compound signs being simply additive. The different linear order might imply slight differences in prominence, maybe, but nothing at all fixed.

But say the hominid, the excitement of his ambiguous situation making him hurry to say two things at once, mispronounces the signs and says [watsi] or [kipo] instead. We then have a blend, meaning "food and lion". Such a blend would, if it were understood by the others, be of value to them, conveying, in the same way as the longer [wapokitsi] and [kitsiwapo]: "There is a food source here but be careful, because there is a dangerous lion here too." Being of survival value if understood, the blend would have a chance of being used again and established.

Now say the blend [watsi] were to win through. At the same time [wapo] and [kitsi] continue to be used. In the interest of systemic economy there would probably be a comparison of the relationship between all three signs and a resulting reanalysis:

We now have [watsi] meaning "food and lion". Therefore [wapo] means "food and no lion", [kitsi] "lion and no food". Hence [wa] means "food", [tsi] "lion", [po] "no lion", [ki] "no food".

Then [watsi] would not necessarily be just a short variant of [wapokitsi], but would open the door to enlarging the vocabulary, because now it does contain signs partially the same, partially different. Such analysis of forms on the basis of a comparison with other forms is, not unexpectedly, called an

Analogy, then, can cause productivity. Now a small vocabulary enlarged by receiving items with precisely opposite meanings is still a far cry from a system capable of developing new shades of meaning as well as opening up entirely new areas of meaning. Nevertheless, the techniques of
blending, segmentation and making new combinations out of these segmented bits (this last activity we may perhaps call a form of editing), are precisely the way we handle human language in our daily lives. We arrive at partial meanings of parts of long utterances through comparing them tacitly with (parts of) other utterances that we have processed earlier; we piece together the bits we arrive at in new combinations, analogous to other combinations, to make up new meanings and new shades of meaning. New shades of meaning regularly develop when signs are placed in new combinations. Note e.g. the different varieties of meaning of "hus" in "mitt hus" (my house), "Kongehuset" (the royal house) and "Akershus" (fortress in Oslo).

Analysis by means of analogy is the dominant technique in our handling of our language. It would therefore be an attractive way of explaining the rise of productivity, since it does not require us to bring in features that are totally unknown to language (i.e. explain one unknown by another unknown) but only to examine more closely a mechanism which is present in language and fairly well understood already.

Another nice thing about this proposed explanation of productivity is that it is capable of throwing unexpected light on something else. That means: it can have useful by-products without being made more complicated. The "something else" concerns the development of grammar:

The blending-and-reanalysis process can explain how
linear order can come to be exploited grammatically. And linear order is in fact used as a grammatical means, a way of signalling differences of meaning, in all languages (so-called free word order in languages like Latin is always only partial). Simple examples abound:

horse bit the man.
man bit the horse.

Clever men love women.
Men love
clever women.

I do
n't like all this New Age superstition, do you?
I do like all this New Age superstition, do
n't you?

If we look again at the initial situation that might lead to a blend, we understand that although [watsi] and [kipo] are equally suitable blends, they could not both be regularly established. The reason is that whereas [watsi] for "food and lion" would lead to a reanalysis yielding

[ wa ] "food"
[ tsi ] "lion"
[ po ] "no lion"
[ ki ] "no food"

the blend [kipo] for "food and lion" would lead to other meanings being attached to the segmented parts:

[ kipo ] "food and lion"
[ wapo ] "food and no lion"
[ kitsi ] "lion and no food"


[ wa ] "no lion"
[ tsi ] "no food"
[ po ] "food"
[ ki ] "lion"

If the hominid group were to use both the system b) resulting from the blend [kipo], and a) resulting from the blend [watsi], it would obviously lead to confusion and be detrimental to the community. If blends are to be actually useful, they would very early have to give priority to
either [watsi] or [kipo] and discard the other. But giving priority to just one of them must imply that one of the linear orders for a combination of the full sign forms, [wapokitsi] or [kitsiwapo], has already, or at least at the same time, been preferred. In other words, a tendency to treat linear order as important must already be present in the communication system. We conclude from this that blending and reanalysis as a source of productivity also carries with it an important type of grammatical structuring.

The ongoing process of blending, reanalysis and varying meaning by placing signs in different combinations with other signs, also bears some relation to the way signs cease to be signs and become just phonemes or phoneme combinations. Examples: English /kap/ "cup" and /bo:d/ "board" are both signs. In "cupboard", however, phonetic change (the falling away of /p/ and the reduction of the vowel in "board") has broken the connection to "cup" and "board", with the result that "cupboard" had better be considered to be a simple sign. English /bousn/ "boatswain" is historically composed of "boat" and "swain" (= boy, man), but these are no longer recognisable as separate signs in /bousn/. Likewise Norwegian /fjø:s/ "fjøs" (cattle-house) is historically a compound of /fe:/ "fe" (cattle) and /hu:s/ "hus" (house), but again phonemic change has established "fjøs" as just one simple sign meaning.

The rise of phonemes - meaningless segments within the expression side of signs - in a system which did not previously have them, may perhaps require no other mechanisms for its explanation than the ones that are still present and operative in language today.

Tendencies to displacement are present even in communication systems which we consider largely natural. Any situation or event in the universe is unique, no two are exactly the same in every detail. If nothing else, time and/or place of two events must be different. Therefore, any perceiving organism, even if his reactions to a situation are instinct-regulated, must to some degree abstract away from the concrete deictic situation in order to understand what kind of situation he is experiencing. Perception is in itself categorising.

Now the more complex the situation to be assessed, the more the individual will be likely to benefit from an ability to delay his response while he takes in all aspects of it and weighs up all available information. If the communication system has already started to develop productivity, parts of it will need to be learned outside of any natural stimulus-response situation, and such learning requires a learning situation, which will favour individuals who are able to delay their impulses while they learn and handle displaced information.

It is therefore possible that full-fledged displacement can have developed as an extension of a general ability to abstract, and of a possibility of delaying our response to a situation which requires some degree of rational analysis.

If this attempt at reconstruction is even partly valid, one conclusion must be that abstraction stands even more firmly planted in the language system and its implementation than our first consideration of the individual features of language led us to believe. The very complex character of a system whose concrete, operative units and details are not biologically inherited, speaks of a lot of very abstract handling and learning going on behind the implementation of the system. And, as we have seen, abstraction ties in with displacement and opens up the possibility of lying.

The spontaneous acquiring of complex structure

At the beginning of this century, Russell and Whitehead, in their monumental Principia Mathematica, set out to prove that mathematics and logic were fully consistent. In 1931, however, an article by Kurt Gödel killed every hope of finding any such proof, by showing that if a sign system is sufficiently rich, it will of itself develop structures which are "undecidable" in the sense that they transcend a true-false classification.

Gödel showed this for the system of natural numbers: 1, 2, 3, ........ (0 is usually included for systemic reasons, although to our naive intuition it does not seem quite so "natural"). The number system is certainly a sign system. Numbers are not found in nature, they are abstractions for what collections of natural objects have in common: "five" is that property which is common to five cars, five books, five days, and so on.

Being signs, natural numbers can be used to refer to entities, i.e be given a defined meaning. What Gödel showed was that a number can be defined in such a way that it refers to itself and declares that such a number does not exist in the system. His conclusion about number statements resembles the conclusion we must draw about the following pair of utterances taken together:

"The next sentence is true."
"The last sentence is false."

So Gödel's proof is the mathematical version of the self-referential paradoxes I spoke of earlier, which cannot be fitted in among true or false statements.

The set of natural numbers can be generated in an extremely simple way, by "defining" a starting point and a "successor function" which we can think of as "the process of adding 1". An interesting implication of Gödel's proof is then that such a fundamentally simple collection of items cannot, when there are enough items, be prevented from developing the capacity to form very complex structures, which are fully regular occurrences within the system, but which as it were "take off" from their originally simple signifying relation and enter into unforseen structures of a kind which transcend the rules of the system itself.

Human language is not a well-defined system in the way of the natural numbers. But it is (slightly dependent on definition) probably equally rich, and of course we already know that it is rich enough for us to make self-referential statements in it.

What Gödel's result can tell us about language is that regardless of how it developed, there are
systemic reasons why a sufficiently rich system gets out of the control of any "censor" wanting to direct what statements may mean and wanting to constrain us to telling only what is true. Unless he can do away with human language, the way it is, and reduce us to the use of a much, much simpler system of communication, he will, like the sorcerer's apprentice, have to watch helplessly while the magic gadgets from time to time get out of hand.



Derek Bickerton (1981): Roots of Language. Ann Arbor: Karoma

Derek Bickerton (1990):
Language and species. Chicago: University of Chicago Press

Dwight Bolinger (1975):
Aspects of Language. 2nd edition. New York: Harcourt Brace Jovanovich

Noam Chomsky (1959): "A Review of B.F. Skinner's
Verbal Behavior". Language 35, no 1, pp 26-58. Reprinted in: Jerry A. Fodor & Jerrold J. Katz (eds) (1964): The Structure of Language. Readings in the Philosophy of Language. Englewood Cliffs, New Jersey: Prentice Hall

Kurt Gödel (1931): "Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I".
Monatshefte für Mathematik und Physik 38, pp 173-198

Louis Hjelmslev [1943]:
Omkring sprogteoriens grundlæggelse. København: Akademisk forlag (1966)

Charles F. Hockett (1960): "The Origin of Speech".
Scientific American, September 1960 vol 203, no 3, pp 88-96

Douglas R. Hofstadter [1979]:
Gödel, Escher, Bach: an Eternal Golden Braid. New York: Vintage Books (1980)

Roman Jakobson & Morris Halle (1971):
Fundamentals of Language. 2nd edition. The Hague: Mouton

Roman Jakobson (1972): "Motor Signs for 'Yes' and 'No' ".
Language in Society I, pp 91-96. London: Cambridge at the University Press

Otto Jespersen (1922):
Language. Its Nature, Development and Origin. London: George Allen & Unwin

Stephen Cole Kleene [1952]:
Introduction to Metamathematics. Groningen: Wolters-Noordhoff (1988)

Edward S. Klima & Ursula Bellugi (1979):
The Signs of Language. Cambridge, Mass: Harvard University Press

Bertil Malmberg (1973):
Teckenlära. Stockholm: Bokförlaget Aldus/Bonniers

Ladislav Matejka & Irwin R. Titunik (eds) (1976):
Semiotics of Art. Prague School Contributions. Cambridge, Mass: MIT Press

Ferdinand de Saussure [1916]:
Cours de linguistique générale. Paris: Payot (1965); Swedish edition: Kurs i allmän lingvistik. Stockholm: Bo Cavefors Bokförlag (1970)

Ronald A. Zirin (1980):
"Aristotle's Biology of Language". Transactions of the American Philological Association 110, pp 325-347.

Concerning especially children who lie or fantasise in sex abuse cases, see:

Children as the primary losers when cases of sexual abuse are falsely alleged: Two Swedish Cases
By Siv Westerberg

When the psychologist fuels a miscarriage of justice
By Knut Grepstad

Alexander Aminoff's language proficiency in childhood
By Marianne Haslev Skånland

Miscellaneous articles about sex-abuse cases

Articles archives 11: Incest and other sexual abuse

'Satanic Abuse' in Rochdale, Manchester