The Sources of the Written Language Bias in Linguistics

Traditionally linguists have been occupied with written language to a much greater extent than with spoken language. Even today, when much lip service is being paid to the need to study spoken language, we approach language with a conceptual apparatus which shows numerous signs of being derived from this tradition. This I will demonstrate later on (sections 5-11). At the outset, it would not be out of place, however, to ask the question how and why linguistics became biased towards written language in the first place. I am of course far from being able to account for the entire history here, but I will nevertheless suggest four factors which it would seem necessary to consider in this context.

The four points, which, by the way, are not mutually independent, are these:

a) the technological evolution of the medium of written language as such, which has determined important aspects of our conception of what language in general is or should be,

b) the motives and goals behind the activities of linguists in the past. These goals are no longer so dominant today, but the traditional directions that linguistics took as a consequence of them still characterize our theories and practice.

c) the high status of written language in almost all societies and therefore also among linguists belonging to those societies,

d) the role of written language as the metalanguage of linguistic description. Thus, written language was not only the subject of scientific analysis, it was also (and still is) the medium in which the products of the analysis, our various theories and metalinguistic statements, were stated.

The first two points merely follow the general tendencies of sciences to adjust to technological and practical-political needs and possibilities. The latter points, (c-d) are more specific to linguistics, although they too have repercussions on all scientific and other activities which are dependent in some way or another on the existence of writing.

3.1 Technology

It is a commonplace that technology always tends to have an impact on what kinds of theories scientists develop and what kinds of goals and problems they set up in their enterprises. Technology leads to new cultural advances. No wonder then that technological devices and their preconditions and inherent possibilities are regarded as particularly important and interesting by society in general, and hence also by scientists. Perhaps the gradual evolution of writing represents the most important technological advance of all. It makes civilization and more advanced culture possible, and therefore it is in fact a precondition for all other technological advances so far made. One is nevertheless justified in saying that contemporary linguists do not in general have any deeper appreciation of the influence that the medium of writing has had on their own science. Yet, what has happened is that, for the reasons I have just outlined, linguists have been occupied with explicating certain properties and inherent possibilities of written language, and these properties have then very often rather naively – been assumed to be characteristic of language in general.

There are at least three important stages in the development of writing: a) the first picture-based writing systems, b) the syllabaries and alphabetic writing systems which were based on the phonetic form of the corresponding spoken words and utterances, and c) printing, which made large-scale distribution of written works possible. As a fourth milestone of similar importance one may consider today’s rapid development of computers. Basically what these perform are very sophisticated manipulations of symbols which are or can be written (written out).

Olson (1977) has described the development of writing techniques as a gradual transition from the context-boundness of spoken language to successively higher levels of autonomy of written messages. An oral message is bound to the time and place where it is uttered. The outer form, the speech behavior, is a very ephemeral phenomenon, and the semantic interpretation is crucially dependent on various factors in the communication situation, factors which may be considered extrinsic to language as such. Thus, oral language is far from being autonomous (II.1).

When man began to write down symbols of various types, this meant that the products of the sender’s activities were made permanent, and this, Olson argues, is the first step towards an autonomy of the text.

However, as long as the written symbols were basically pictures of some kind, there were heavy restrictions on the kinds of messages that could be written down. In practice, a great number of abstract meanings and meaning elements could not be symbolized. However, when the writing systems gradually developed from being more or less completely content-based to being expression based instead (signs or letters were used to indicate the sounds of speech quantized as syllables, vowels and consonants), these restrictions were removed. In principle, the segmental aspects of any string of spoken behavior could be satisfactorily symbolized in writing, and this established a “formal autonomy” of the written text.

But even after the invention of the Greek and Latin alphabets there were quite strong restrictions on both the form and content of written texts. One reason for this was that much of what was written down in the beginning consisted of important parts of the already existing oral tradition: epics, verse, song, orations used as laws, folk tales, legends as well as proverbs, adages, aphorisms, riddles, etc. These texts were biased both in form and context to fit the requirements of oral communication and auditory memory. Thus, they often had a formally bound structure (e.g. verse), they were typically elliptical and left a lot of the interpretation implicit. For their interpretation, they were heavily dependent on the interpreter’s background knowledge of the culture and of the referents involved, and the functions of the various texts in the entire cultural context, etc. Thus, they tended not to say exactly what they meant; they were not “semantically autonomous” in Olson’s terminology.

Before printing was invented, every text had to be laboriously copied by hand each time it was to be duplicated. Of course, t also put the writer under strong pressure to keep texts short, and hence the texts were not very explicit. The situation was radically altered, when printing made it possible to spread a very large number of copies of the same text. It was by this means possible to increase the length of texts and to write in a much more explicit manner. Many more background assumptions could be expressed if necessary, and this, according to Olson, made the written text “semantically autonomous”.

Olson points out that soon after the advent of printing, Martin Luther pronounced his opinion that the meaning of the Scripture depended, not upon the dogmas of the church, but upon a deeper reading of the text. This is, in other words, the theory that the meaning of the text is in the text itself. Olson traces this theory back to Luther and to the British essayists of the seventeenth century, who were among the first to try to pursue the ideal in writing of explicitly formulating all the steps in the writer’s argumentations without any reliance on implicit presuppositions or personal interpretations. This ideal later on motivated the rectification of natural language into a language of logic, which was supposed to meet these requirements better than normal prose. Olson seems to accept the thesis of the semantic autonomy of (certain) written texts; meaning is said to be “intrinsic to language” in a written message but “extrinsic” in spoken communication. Such a view is hardly acceptable, however. The interpretation and comprehension also of a written text crucially involves background knowledge which is not contained in the text as such (VII.7) . Of course, it remains true that written text is in general more explicit and more autonomous than a spoken message (II).

With the invention of writing it became possible to retain messages over centuries, something which had an enormous impact on the development of human cultures through the accumulation and maintenance of the insights of earlier generations. However, literacy had many other important psychological and social consequences for individuals and social groups 1 some of which will be briefly reviewed here (see Goody 1977). (1)

Many of the very earliest written documents were of an administrative or economic nature; inventory lists, catalogues, and lists of taxes, credits and debts etc. Other types of texts that appeared in early times were laws, decrees and instructions of various kinds. The use of such specializations of writing as lists, tables, diagrams and formulas is particularly interesting in that some of them exploit the spatial dimension of the written surface in a way which has no counterpart in speech behavior. The user thus gets an opportunity to manipulate his subject matter in new and important ways. Logical and mathematical calculations become possible. Such arithmetic operations as multiplication and, in particular, division are hard, if not impossible, to carry out without written notations. Moreover, could the position system of arithmetics have been invented without writing?

A written text is independent of its producer and interpreter in a way that has no counterpart in the case of spoken messages. This enables the user to keep a certain distance to the contents, as well as to his communication partners. Speech communication, on the other hand, is firmly anchored in the immediate environment, the speech situation which is embedded in the everyday culture, all of which exerts a strongly normalizing pressure on both speaker and listener. They have to follow the established conventions of their culture, unless they deliberately accept confrontation with the ensuing negative reactions from the social environment, thus taking the risk of being expelled from the community in the long run. The written medium, however, greatly extends and enhances the user’s chances of taking an independent personal stand with respect both to the contents of messages and to his fellow human beings. This will have several important consequences.

In a literate culture attitudes towards reality are changed. Some features of the down-traded picture of everyday reality can be questioned, alternatives can be considered and compared, and rationality, skepticism, logics and science developed – all of which is commonly associated with Western culture. Knowledge tends to be more abstract and general, and less tied to the contexts of everyday reality. The development of logic, for example, has to do with the relative freedom to manipulate the elements of written texts; the written medium allows us to separate words, which means that their internal ordering can be more easily manipulated, and that syllogistic reasonings can be developed. The use of tables makes it easier to construct classificatory systems with hierarchically organized categories and subcategories. Writing allows the user to work out and survey all of the theoretically possible alternatives.

Communication by means of written texts puts the emphasis on the intellectual, cognitive, descriptive, and argumentative functions of language, whereas the emotive and social aspects are kept firmly in the background. After all, most of these latter aspects are conveyed in speech by means of prosodic phenomena, accents and intonations etc, and various non-verbal signals and gestures, and this is all very poorly represented in writing. McLuhan (1965:86) has argued that the use of alphabetic writing as a medium of communication has made Western civilized man into an intellectualized individual who represses his feelings and his commitment.

Writing helps its user to sharpen his thinking; language is thereby given a more individualized, monologic function. This in turn facilitates the development of intellectual independence on the part of individual persons. While deviant opinions and attitudes are difficult to convey and maintain in a society where communication is based on direct face-to-face interaction, the literate culture enhances individualism; heretics, free-thinkers and philosophers are given a certain amount of elbow-room. Single historical individuals can now make their own original contributions to the development of knowledge and culture.

In this context we must reiterate that writing implies that knowledge can be accumulated over a period of time. There is no longer any need for each generation to repeat the same process of trial and error over and over again, since one is no longer forced to rely only on the oral tradition, which is, after all, limited in scope.

Another consequence of the rise of literate culture is the possibility for men in power to control secondary groups. By issuing written decrees and edicts the desires and orders of a central power can be distributed to large groups of people living at different places. This presupposes and carries with it the development of an impersonal bureaucracy. The possibilities of governing and dominating people with the help of written messages are of course particularly great in societies, where literacy is restricted to a small minority, and this is precisely the way in which almost all societies have been governed up to our own time. This also means that those who are competent to use the written language enjoy and the written language itself is normally regarded as the correct language (III.3). Thus, language in a literate society, or rather a society with varying degrees of literacy, becomes a potent means of social stratification. The “restricted codes” of spoken language are insufficient as a basis for societal advancement and career. Washabaugh argues:

“Non-autonomous, restricted codes are poor social integrators since they point speakers to no authoritative group or tradition (Mueller 1971:111). Non-autonomous, restricted codes increase a people’s commitment, to, and dependence on, the social order and lead them to support a repressive social order all because such a language form decreases a people’s ability to reflect on their condition and to generate privatized’ alternatives to the social order (Mueller 1971:106).
(Washabaugh 1980:211)

It has also been pointed out that literacy makes the writing of a (relatively) objective history possible. Non-literate cultures are characterized by the existence and vitality of myths, both religious and secular. Every culture feels a need of explanations of why nature and society are the way they are. Myths explicate and legitimize the existing societal conditions, and they normally do so in a way that satisfies the interests of those in power. Though myths generally build upon a kernel of historical truths, their contents constantly change through oral tradition as society develops and as society’s demand for new types of explanations and legitimizations are altered. When written records are available and can be preserved over a period of time, one of the preconditions for making objective history possible has been created.

Finally, the invention of writing has naturally had important consequences for the linguistic code itself. In most literate societies we will, sooner or later, be faced with two codes or variants – one is tempted to say two languages – rather than just one. The written language is different from the spoken language not only as regards the medium of representation graphic as opposed to phonetic. Its grammar, particularly its syntax, is much more constrained and prescriptively controlled, and its lexicon is greatly expanded. When written dictionaries and encyclopedia are being developed and used, the size of the total vocabulary of the language involved is very much increased, and as a consequence the delicacy and precision of word meanings are also enhanced. Of course, these gains achieved through written language will have important feedback effects on the spoken language too. (It is hardly a coincidence that sign languages (i.e., those languages used by the deaf) have considerably smaller vocabularies than most spoken languages. There may be several reasons for this state of affairs, but the fact that spoken languages are supported by their written counterparts (while sign languages have no such counterparts /at least not until some recent attempts to remedy this situation, cf Stokoe 1972, Baron 1981) seems to be a major factor. The gestural medium as such seems to have an inherent capacity for symbolization (reference and description)which is just as large as that of the vocal medium).

By way of conclusion we may say that the development of written language and literacy is no insignificant matter. On the contrary, it has far-reaching consequences for cultures, which become differentiated and specialized, for individuals, whose attitudes to knowledge and society are radically changed, and for the linguistic codes themselves, which are transformed in certain respects. Therefore, the investigation of the differences between spoken and written language is worth pursuing in linguistics too.

3.2 The Traditional Tasks Of Linguistics

In today’s highly industrialized and diversified Western societies many states have been able to use some of the economic surplus in order to hire numerous professional specialists who can pursue highly “theoretical studies” with the aim of finding out the true nature of various matters, independently of immediate practical, political or ideological goals. Thus, the field of theoretical linguistics studies general conditions of the structures and uses of natural languages. (At least, this is what we think we are doing.) However, this is something very new. Traditionally, the tasks of linguists have been more or less directly related to practical enterprises and political goals, and this is of course still true of a great many linguists presently working in many countries.

Traditionally, linguists have worked on tasks concerning the development of written standard languages (and this of course presupposes a sufficiently developed theory of the structures of the corresponding spoken languages), the promotion of literacy, and the improvement of the study and teaching of foreign languages. However, these activities have more often than not only been the means for attaining higher goals, especially that of serving the right religion and its various institutions and that of serving authorities of a more secular kind by establishing a language for governing and controlling people.

To a very great extent the history of linguistics is connected with the activities of religious institutions. In fact, linguists have most often been priests with the important tasks of taking care of God’s holy word, preserving and explaining it, providing the correct interpretation of the Scriptures and maintaining its proper form, and translating it to new languages, thereby making their contribution to the spread of the right religion to pagans and barbarians.

Linguists were priests during the era of ancient Indian linguistics, their task being to preserve certain ritual and religious, orally transmitted texts. The history of Western linguistics is very closely connected with Christian theology. This is true of the entire Middle Ages from St Augustine onwards. Hovdhaugen notes (1980:129) that “Christian theology is to a large extent exegesis of the Holy Scriptures, and exegesis is to a large extent a linguistic analysis”.

Linguistics in Classical Greece and Rome was in certain respects different, since grammar was at that time more directed towards the explication of the (written) language of the great authors. The same is therefore also true of the humanists of the Renaissance. One should note that this very much involved the study of foreign languages. Homer’s Greek was a foreign language for most scholars in the Hellenistic world, and classical written Latin was a foreign language throughout the Middle Ages, and of course later on too.

The connection between linguistics, or philology, and the church was also reflected in the school system of most Western countries. Language pedagogics was built upon classical and medieval linguistics, especially grammar, but also rhetorics, and logic or dialectics. The aims of the grammar schools were twofold: to instruct the pupils in the right faith, and to provide them with the correct language, i.e., Latin and (later on) the written standard national language (but not of course their own vernacular which was considered faulty and vulgar). These two tasks were united, and could not and must not be disconnected.

Even our 20th century history provides examples of the connection between religious aims and linguistics. American descriptive linguistics has largely been concerned with inventing alphabets for, and discovering the grammatical structures of, numerous American Indian languages. The ultimate goal has usually been the translation of the Bible and the integration of Indians into the society of White Americans.

In modern times linguistics has served more secular rulers. Among the most important tasks we find language planning and language cultivation (and the ancillary disciplines that might be needed for these activities); the goal was (or is) to create standard languages for new nations, i.e.,to develop and establish a language variant suited for administration, government and control, and for the development of a national literature. In practice, this means the standardization of a language, more exactly a written language, most often the language of the politically and economically dominant groups or tribes. Again we see linguists concerned with the establishment and legitimization of a correct language, that which should be the norm for the citizens. To some extent this may have served the imperialist goals of social repression:

“In this context recall the remark made by Elio Antonio de Nebrija, the first grammarian of the Spanish language. The year was 1492 and he had just presented the Spanish queen with the first copy of his Gramatica castellana. Her question to him was potentially disarming. What good is your grammar? What do you do with it? Fortunately for grammarians everywhere, Nebrija had just the right answer. You need it, he explained, to assure that all people’s subjected to the Spanish crown will have one voice with which to talk to the queen. Besides, he continued, ‘slempre la lengua fue companera del imperio’ — language has always been the companion of empire.”
(Di Pietro 1976)

As I pointed out at the outset, the economic surplus, which has been produced in some industrialized countries in the last century, has to some extent liberated some scholars from the immediate practical and political demands and goals which are normally set by society. We can, e.g., afford to have a limited number of “theoretical” linguists. In the same period we have observed a clear trend on the part of linguists to pay more attention to spoken language. In fact, it is part of today’s standard textbook ideology to assume that spoken language is the primary and most important medium of natural language (V). Surely it can not be denied that linguists have invested a lot of energy and ingenuity in the study of spoken language. Some well-established fields are dialectology, phonetics and anthropological linguistics. In recent years we have witnessed a formidable explosion of studies in discourse analysis, pragmatics of spoken language, and related areas.

Thus, the written language bias in linguistics now seems to be d1minishing. The very fact that I hit upon the idea of writing this book is a small but perhaps significant symptom of precisely this trend. However, traditions are not so easy to change. My aim is therefore to invoke some discussion around questions like the following: To what extent is there still a written language b1as in linguistics? In what ways must linguistics be modified in order for it to tackle the problems of spoken language in the best possible ways?

3.3 The status of written language

In most (all?) societies the written language has had, and still has, a very high status. It is regarded by the common man with respect, admiration and reverence. Originally, it was the language of religious documents, the laws (both religious and sear), and the great authors.

Of course, the traditional admiration for the written language is basically an admiration for those, the very few, who could use it properly. One should recall that in most societies there has been, and still is, no wide-spread literacy. Instead we have had oligoliteracy, i.e. a state in which literacy is limited to a small minority of educated people. No wonder writing is regarded as something special. Already among the Sumerians and Accadians, “writing was the pursuit of scribes and preserved as a ‘mystery’, a ‘secret treasure”‘ (Goody & Watt 1972:323) and this is still true of certain societies, e.g., in the Middle East.

Thus, the supremacy and mystery of the written language were established in societies with a low degree of literacy. Furthermore, the association with the church and its institutions added to its status. For a long time, priests and monks were a very important group among those who were able to read and write, and they also acted as school teachers and private tutors, and this gave them further opportunities to enhance the specific status of the written language. This is not to say that an admiration for the written language as a medium is unmotivated. In fact, literacy enables people to perform many important cognitive and social actions, which would be impossible without writing (cf III.1).

We noted in III.2 that nationalism and central state authorities may consolidate their power and strength by means of the written medium and its standardization. However, we must be aware that standardization is a natural and inherent feature of any attempt to “convert speech into writing”. No writing system can preserve all the variation present in natural speech. The crucial point is therefore not standardization as such but the level of abstractness at which the standarization of the written notation is established. In principle, the more heterogeneous the dialects one tries to cover by one common written standard are, the more remote will this written code be relative to the individual spoken vernaculars (and idiolects) (except possibly the dialect of the men in power, which is usually the basis of the norm). And in most parts of Europe there were indeed many often mutually very different linguistic varieties, which – as part of the formation of the new national states – were transformed into ”dialects of the same language” precisely through the establishment of a common superordinate written norm. In other words, this written norm came to constitute a separate “language” (e.g. German, Italian, Swedish, Russian), i.e. that language of which the original spoken varieties were to be regarded as “dialects”. Thus, by creating national written standard languages centralists strengthened their power and tried to impose a national identity onto a conglomerate of different tribes and ethnic groups, who had so far, in many cases, not had very much in common ethnically and/or linguistically. At the same time, this process served to reinforce further the idea of the written standard language as a superior linguistic medium.

The high status of written language still prevails in today’s highly literate and secular Western industrialized societies. The written standard language is the norm; it defines what people generally consider to be linguistically correct. The written word carries with it permanence, firmness, responsibility; only written statements are ultimately valid as legal documents and in judicial procedures. In fact, the written language belongs to the very foundation of the society and the national state (compare the function of national languages).

One of the most common ingredients in laymen’s outlook on language is the view that written language is the proper language, whereas the spoken vernaculars are crude, primitive, incorrect, and unsuited for higher functions. This view is traded down in school teaching. After all, the main task of the primary school is traditionally that of teaching children how to read and write, and how to use language properly, i.e. not as it is used in normal casual speech. It is obvious that schools in Western societies have communicated this view of written and spoken languages in surprisingly similar ways from antiquity, through the Middle Ages and further on right up to the 19th and 20th centuries, and it is debatable whether they are actually changing to a more moderate position today. (2) I have already alluded to the close connection between the traditional tasks of school education, the promotion of the right faith, and the study of the correct language. For a long time, the objects of study were the Holy Scripture and the great authors, and the analysis usually focused on formal aspects, especially the grammar, of the written varieties used there. Originally, the language studied was Latin, but the same attitudes and procedures were later transferred to the study of the national standard languages. The grammar of Latin was used as a model for the writing of the grammars of the various national languages, and these languages were analyzed and taught in much the same ways. In fact, the variants of the national languages to be used in religious practice, education, administration, and legislation were largely shaped, even invented or created, on the basis of Latin. (In Scandinavian countries, German partly played the role of an intermediary language). The deep-rooted attitude towards the mother tongues of the pupils has survived for centuries and millennia; the pupil comes to school without having a proper language! Later we shall study how such attitudes are reflected in contemporary linguistics (XI.2).

Thus, we have pointed out the position of the written standard language and the spoken vernacular as high vs. low status languages in the layman’s conception, in what could be called the ”linguistic folklore” or “ethnic linguistics” (3). But the same attitudes are found among professional linguists, and this is hardly surprising. On the one hand. linguists are of course members of the society at large, and the adoption of the unquestioned attitudes of the culture is part of the socialization of those who later become linguists just as much as it is part of other people’s upbringing. Secondly, the activities of the linguists themselves have served, confirmed and enhanced the very same attitudes and conceptions. We noted earlier (III.2) that linguists worked in the service of God and the King not only in describing and analyzing but also in cultivating, developing and establishing the structure and use of the proper written standards. Linguistic behavior is conventional, it is subject to rules and norms of various kinds, and since the written language was regarded as the correct language (language as it should be), that which should be aimed at and established, it was only natural that grammarians concentrated on trying to find out the rules underlying that variant. In today’s structural linguistics we can sometimes see how issues of grammaticality are solved by (implicit) reference to the inherited norms of written language (V.8).

3.4 Written language as metalanguage

The bias towards written language has been forced upon linguistics for two different reasons or at two different levels. On the one hand, our object of study has traditionally been the written language, and the reasons for this have just been outlined. On the other hand, the written language also serves as metalanguage, as the language of linguistic description. When we analyze language and linguistic phenomena (be they written or spoken), the analysis is performed by the use of written languages, and the various concepts, theories and results of analysis are formulated in that medium (plus pictures, diagrams, lists etc, but these are also writing in a broad sense). We may therefore suspect that our concepts and theories may be dependent on the inherent limitations and possibilities of the metalanguage. Even if we are in fact analyzing spoken language, we have to represent that language and the corresponding verbal behavior, and this we do by means of common writing (or some slightly modified variant of it, such as phonetic transcription or a symbolic notation like that of formal logic). This means that our data are transformed into written language, or at least into something that is heavily dependent on written language. I will discuss some of the consequences of this below (e.g. as regards phonology and the theory of communication). (4)

It should be noted in passing that today’s linguistics is beginning to use computers quite extensively, and we have already seen how theories, models and metaphors from the computer sciences are coming into linguistics. It would be worth while considering how much of these new conceptions depend on the inherent properties of digital computers (cf. e.g. VII.10 on the representation of knowledge).

Naturally, the two functions of written language within linguistics, the functions of being both object of study and language of description, interact in complex ways and thus add up to a state of affairs, in which it may be difficult to isolate the contributions of the one or the other.



1. It should be noted that these hypotheses about the cognitive consequences of literacy (cf. below) apply primarily to the historical development of cultures. It is probable that there are some similar differences in cognitive and social attitudes between literate and non-literate individuals in contemporaneous society, but, as Scribner & Cole (1981) point out, one cannot leap to the conclusion that what was necessary historically is necessary also for children, who are born into a society in which elaborate uses of literacy have existed for a long time and thus presumably have had repercussion on general modes of thinking (i.e. even among non-literate members).

2. I am only familiar with Scandinavian school systems in this regard, but I assume that most Western countries fit the same mould. The history and traditions of the teaching of Danish and Swedish in our countries have been thoroughly described by Diderichsen (1968) and Thavenius (1981) respectively.

3. It must be admitted, however, that we do not yet much systematic knowledge of the linguistic folklore and its functions. Apart from some sociolinguistic studies of isolated points, there seem to be few or no studies of this interesting topic. What we know is mainly what can be gleaned from subjective impressions of how laymen tend to talk about language in everyday conversations, how their opinions surface in e.g. the letters to the editor columns of daily newspapers and magazines, how teachers in elementary schools behave with respect to language, and what school books have to say about language.

4. If we narrow down our perspective a little, we may recall the unusually important role ascribed by transformational grammarians to the notational systems used in linguistic models. The form of the theoretical apparatus is considered to be of utmost importance, and the form is that which is provided by the metalanguage (e.g. in Chomsky & Halle 1968). “There has been a remarkable tendency for even the most eminent transformation grammarians to pay more attention to the typographical combinatorics of their notational systems than to the content that is expressed in them and to condone even drastic innovations as long as they stay within or close to familiar notation systems” (McCawley 1980:919).