THE TWO DOMAINS

The distinction between the domain of speech, language, and grammar on the one hand, and the domain of people, individually and collectively, on the other, is perhaps clear by now. It is a distinction that had its origin in antiquity. The Stoic split at the most basic level of philosophy between the the logical part and the physical separated the topics of dialectic and rhetoric in the logical part from biological, medical, and other topics, including the eight parts of the (material) psyche or soul, in the physical part (Diogenes Laertius VII 39–47, 132–3, 157–9). It thus separated the study of speech and language from the study of sound as vibrations in the air. It separated sensation and perception as part of how we know the truth from the five senses as parts of the psyche. It separated logic and the form of argument from the power of reasoning as part of the psyche. And most important, it separated language and grammar from the power of speech as another part of the psyche. Thus it separated the perceptual, logical, and grammatical as parts of a theory of knowledge from the physical and psychological as parts of the real world.

It is usual even today to separate the phonological from the phonetic, the logical from the psychological, and the epistemological and philosophical from the biological, chemical, and physical. The choice between a grammatical linguistics and a scientific linguistics is basically a choice between these two domains. It is a choice between a linguistics that sees itself in terms of concepts of language and grammar as part of the philosophical domain of logic, epistemology, and the theory of knowledge, and a linguistics that seeks to understand the phenomena within its purview as natural phenomena, as aspects of physical reality to be studied scientifically.

The nature of this choice may have been partly concealed from the founders of modern scientific linguistics because the very powerful tradition of normative grammar had intervened historically, and the choice between the traditional and the scientific was then seen in terms of a choice between fanciful and well-supported etymologies, between normative and descriptive grammars, between grammars grounded in logic and grammars grounded on usage, and between grammars influenced by Latin grammar or the native language of the linguist and grammars true to the languages being described. In each case the traditional philosophically based concepts of language and grammar remained assumed and not brought into question.

The Stoics also saw and discussed important issues in the conduct of investigation in the two domains. Their theory of dialectic provided a criterion of truth about physical reality. The criterion was evidence through the senses, and on this they agreed. But they also asked what we would see as a metatheoretical question. What criterion of truth could be used in studying incorporeal things like the objects of logic? They faced this question by allowing also a nonapprehending presentation, which was received not through the senses, but through the mind itself or through reason. But this way of knowing was more controversial with them, some appealing to right reason, others to innate ideas or gifts of nature (Diogenes Laertius VII 46, 49–54). Even today there is wide agreement in philosophy that the appropriate criterion of truth in the scientific study of nature in the physical domain is ultimately evidence from the senses, but there continue to be disagreements about what criteria of truth are appropriate in the logical or metatheoretical domain, for example the disputes between empiricism and rationalism. The issue of what criterion of truth can be used when tests against physical reality are impossible is still a nagging concern in linguistics and the subject of a rapidly proliferating literature. It has no satisfactory answer short of insisting that theories be testable against a physical reality.

Modern science has prospered by focusing on the physical, where evidence can be obtained through the senses, and trying to reduce to a minimum any assumptions in the logical, philosophical, or metaphysical domain that cannot be tested against evidence from the senses. If assumptions can be introduced freely without empirical support, then anyone can introduce any convenient fictions that he thinks may be plausible, and when such assumptions conflict there are no reliable criteria for choosing among them. Witness the well-known nonuniqueness of phonemic solutions of phonetic systems. The endless march of new types of grammar cannot be expected to converge on a satisfactory agreed-upon body of linguistic theory, for as we saw, there are no real objects for them to be a theory of and against which they can be tested.¹ Compare the point made in a philosophical context by Hutchinson (1974:72) that in a “descriptive” (nonmental) linguistics, grammatical theory cannot be empirically true, as there is nothing for it to be true of. We would have to add that grammatical theory in a mental linguistics founders on the psychological reality of grammar fallacy.

Bloomfield’s analysis appears to be an attempt to isolate explicitly in a single fundamental assumption all the nonempirical factors in linguistics not shared with the more highly developed sciences. This may have been an attempt to move linguistics toward science, but the assumption certifies its nonscientific nature at the very core. The recent rationalistic trend in linguistics under the stimulus of Noam Chomsky and others has actually increased rather than decreased the nonempirical content of proposed linguistic theories, which would move linguistics still further away from science. It is significant that philosophers themselves have recognized Chomsky’s theories as being more in the realm of philosophy than of empirical science, no matter how Chomsky himself may have represented them (Ringen 1975; Botha 1971, 1973).

The problem associated with the psychological and social reality of grammar fallacies is an epistemological problem—the attempt to explain aspects of objects in the physical domain, people, which can be studied by evidence obtained through the senses, in terms of theories of the assumed immaterial objects of language in the logical domain, which are not given in advance, cannot be observed through the senses, and in fact have to be introduced by a special assumption, created out of whole cloth, as it were.

Basically it is a question of levels of theory and metatheory. On the first alternative facing linguistics, theory would be at the metatheoretical level as developed by the ancients and elaborated through philosophy and normative grammar. Its objects of study would inevitably be introduced by assumption. On the second alternative, linguistic theory would be at the lower level of scientific theory and would study real objects. It would accept only the standard metatheoretical assumptions of modern science. A linguistics that inappropriately confuses the two levels by trying to accept both grammar and science is incoherent and contains serious fallacies.

It is ironic that grammar and the theory of signs, originating in ancient epistemology, should lead to epistemological problems when assumed as the theoretical foundation of modern linguistics. The fact that linguistics has made the progress that it has in spite of this handicap is a tribute to the philosophers, grammarians, and linguists involved. But if linguistics is to be a science, it cannot accept in its foundations any assumptions like Bloomfield’s fundamental assumption of linguistics.

We will finally have to choose between a philosophical linguistics and a scientific linguistics, that is, between a grammatical linguistics that studies language in the logical domain and cannot be scientific, and a scientific linguistics that studies the phenomena as aspects of nature in the physical domain, for efforts to do both at the same time are incoherent and lead to fallacies.

It is important to understand that the choice facing the discipline between the first alternative, the second alternative, and business as usual is of an entirely different order from the familiar problem of choosing between different approaches to grammar.

It is difficult to choose decisively among the many forms of grammar because they differ basically in the details of the assumptions they accept, and since the assumptions cannot be justified in any case by evidence from the senses, there is a rather large amount of arbitrariness involved. Thus arguments over the relative advantages of various positions are often inconclusive. Choices are often made instead on such extrinsic grounds as aesthetic considerations, allegiance to a school of thought, or familiarity with a particular position through accidents of schooling or training. Positions are sometimes maintained or defended by polemics and the personal authority of charismatic individuals who adopt an eclipsing stance. As a reaction against this exclusionary atmosphere, there has grown up in some quarters an ethic of tolerance for conflicting positions. A linguist can even hedge or waffle on his choice of theory, eclectically selecting a bit from here and a bit from there, and in this way new positions are often born.

But the choices with which we are concerned here are quite different. They are substantive choices allowing no middle ground or compromise. This may appear to some to conflict with the ethic of tolerance and eclecticism. But the whole point of science is to be able to find objective reproducible criteria for making choices. Scientific tolerance of a plurality of hypotheses stands in proportion to their reasonableness and the lack of scientific evidence suggesting their falsity.

The choice between business as usual and one of the alternatives is a choice between an inherently incoherent linguistics that contains fallacies and an alternative that has a chance of being coherent and free of fallacies. To choose a coherent position over an incoherent position cannot be seen as arbitrarily exclusionary. It is hard to conceive of a good reason to retain an incoherent position, and it is doubtful that it could be shown convincingly that business as usual is not incoherent.

The choice between the two coherent alternatives is a choice between two quite different sorts of linguistics. It is a choice between a philosophical linguistics and a scientific linguistics.

THE TWO COHERENT ALTERNATIVES

Let us consider what linguistics would look like under each of these alternatives, and examine some of the most obvious advantages and disadvantages of each.

A GRAMMATICAL LINGUISTICS

Under the first alternative, linguistics would be disciplined by grammar and would study the objects of language. It would eliminate the incompatibility with the scientific goal by turning its back on any aspirations to be scientific, and it would avoid the psychological and social reality of grammar fallacies by setting aside the goal of seeking explanations in terms of people. It would take the concept of grammar literally to its logical conclusion, a pure grammar associated with a pure logic in the first domain. And following the lead of many logicians and prescriptive grammarians, it would carefully distinguish itself from the merely psychological and from the faulty logic and careless speech of ordinary people. If the predictions from an assumed logical principle, grammatical norm, or Platonic ideal conflicted with what people are actually observed to do, this alternative would resolutely place theory over observation, for to allow theory to be corrected by observation, if consistently followed, would inevitably lead to rejecting the foundational assumptions of the first alternative and thereby move the discipline to the second alternative, since as we have seen, a coherent discipline cannot follow both alternatives at the same time.

This alternative has the not inconsiderable advantage of being hooked into a very long grammatical, logical, and philosophical tradition with a familiar and highly developed conceptual structure. A rather large literature already exists within this tradition that can support further research. With support from this strong background, only a few years are needed to produce a new version of grammar. It also embodies a conceptual structure related to normative grammar and to the folk theory familiar to the public. Thus advances in a linguistics of this sort might be more readily applicable to improving normative grammar and the traditional teaching of composition, foreign languages, and other communicative arts and skills through grammar (although the efficacy of teaching these skills through grammar is sometimes disputed).

A major disadvantage of this alternative is one that it shares with current linguistic theory. It incorporates a foundation of unsupported assumptions. To the extent that theoretical statements rest in part on these unsupported assumptions, they are untestable. There are no objective criteria for accepting or rejecting them. This leaves us with nothing more substantial than dogma or opinion in the more central and basic regions of linguistic theory. If different linguists hold to different sets of assumptions, there are no objective criteria for deciding among them. There is thus a temptation to resort to polemics, imposition of authority, arbitrary dicta, and appeals to simplicity, symmetry of patterning, elegance, personal revelation, or other sorts of inconclusive argumentation. Such unresolvable differences have already fractionated the discipline into camps or schools of thought, reducing the cooperative nature of the discipline by making it difficult or impossible to compare results. This has led to wasted effort as each camp reworks linguistic theory to suit itself. This disadvantage is connected with the incompatibility of this alternative with a goal to be scientific.

Another disadvantage of this first alternative is associated with the necessity that it relinquish the goal of seeking explanations in terms of people. It lacks contact with reality. Its relation, if any, to the observations and data of linguistics is tenuous and problematic at best. It provides poor or no support for primarily observational research and little or no insight into what may lie behind the data. Thus it is incapable of serving the functions of a scientific theory.

A pure grammatical discipline according to the first alternative is possible, but in abandoning the quest for a scientific understanding of the phenomena, it would reverse a century and a half effort to develop a scientific linguistics and go against the long-established historical trend of particular sciences crystallizing out of philosophy. It would make linguistics either a part of philosophy, a part of the normative arts, or both. This would be a position true to its origin in Stoic thought, but whether a discipline of this sort would be appropriate as a part of philosophy today is for one or another of the schools of philosophy to decide. It is not appropriate for linguistics. It could not provide adequate theoretical underpinnings for confronting the observations of interest in linguistics previously identified. Whether it is more suitable for application in the normative and practical arts than an approach through the second alternative is a separate issue, and one that cannot yet be decided one way or the other.

A SCIENTIFIC LINGUISTICS

Under the second alternative, linguistics would be disciplined by science rather than by grammar. But if we give up studying the traditional objects of language, of grammar, and of a sign relation between sound and meaning, because they are only created by a point of view, what objects are there for linguistics to study that do actually exist independently of theory and can be studied scientifically? There can be no other answer than people, individually and in groups—the people who communicate. They are the source of our data and the locus of our explanations. And of course we can study the sound waves and other forms of energy involved when they communicate, and those aspects of the physical environment that may be involved. These all have an external reality independent of theory and thus do not have to be created by a special assumption. Their existence can be confirmed by evidence from the senses. They can be and are studied from different points of view. In linguistics they can be studied scientifically from the linguistic or communicative point of view.

We can use the data obtained from people to understand people not indirectly through language but directly from the point of view of how they communicate. The unjustified assumptions of the traditional objects of language could then be given up. The psychological and social reality of grammar fallacies would be avoided. It would be possible to develop explanations of linguistic phenomena of the sort we have been striving for, in terms of people, directly on the basis of the evidence from people, not indirectly through language. Such a linguistics would be a human linguistics rather than a linguistics of language. It would seek a scientific understanding of how people communicate.²

Let us consider some possible objections to this alternative proposal.

Perhaps such a course would be open to charges of psychologism. Recall that Bloomfield criticized the earlier linguistics for insisting upon “psychological” interpretation. Statements about language were accompanied “with a paraphrase in terms of mental processes which the speakers are supposed to have undergone. The only evidence for these mental processes is the linguistic process; they add nothing to the discussion, but only obscure it” (1933:17). Bloomfield appears to be combating here the psychological reality of grammar fallacy. But of course the fallacy only follows if one accepts the route through language to the psychological reality. Although Bloomfield did search for psychological foundations for linguistics, he always retained the assumed objects of language, and in his major work he resolutely abandoned any program for achieving a psychological explanation. The inhibitions that were then set up in the discipline discouraged linguists from even thinking about people, and ushered in an era of linguistic Victorianism that is only now coming to an end as an interest in people overcomes an aversion to fallacy. But if we don’t assume the objects of language in the first place, we are not confronted with Bloomfield’s dilemma of either condoning a fallacy or abandoning the possibility of explanations in terms of people. Without language blocking the way, we are free to move directly from the data obtained from people to theories and proposed explanations in terms of people.

It has been claimed, following Bloomfield and under the influence of behaviorism, that we have no right to guess about the linguistic workings of an inaccessible ‘mind’, because they are quite simply unobservable. But this was put forth within the context of objections to defining the phoneme as a mental or psychological reality because such definitions would “identify an entity which is inaccessible to scientific methods within the frame of linguistic study” (Twaddell 1935:9), in other words, inaccessible to methods of study through the objects of language. The objection is thus directed at the psychological reality of grammar fallacy, and from this point of view it is well taken. But it would be a mistake or a perverse view of science to jump from this qualified statement to a flat claim that we can’t study people, an existing reality, by observing them, while yet maintaining that we can apply some of those observations of people to elucidating the nature of language, a collection of convenient fictions simply introduced by assumption.

Perhaps such a course would lead us toward infinity. Saussure (1916:25;1959:9) had worried that without placing both feet on the ground of language as the integral and concrete object of linguistics we would be confronted with a mass of heterogeneous and unrelated things, and the door would be open to several sciences—psychology, anthropology, normative grammar, philology, etc.—which might claim speech as one of their objects. Recall also that Bloomfield (1933:78) had worried that without a special assumption linguistics would require the near perfection of all branches of science, including, especially, psychology and physiology. He said that the situations which prompt people to utter speech include every object and happening in their universe, requiring for a scientifically accurate definition of meaning a scientifically accurate knowledge of everything in the speakers’ world. The actual extent of human knowledge is very small compared to this (1933:139).

The fear is overblown and unfounded. Omniscience is not required. In the first place, when people cooperate by means of sound waves we need not be concerned about a complete scientific knowledge of the things they talk about. We do not need the chemist to tell us everything about NaCl in order to handle Please pass the salt. Techniques are available in lexicography for deciding how much material to include. In the second place, the fear may stem precisely from the choice of an approach through language and its philosophical foundation in the theory of knowledge. The methods of grammar are not appropriate for studying the psychological and social aspects of how people communicate, so it was thought that to venture into these domains at all would embroil the linguist in a whole other discipline with completely different concepts, methods, and goals. One would recoil from such a prospect and insist on preserving the autonomy of grammar within its narrow borders even though that meant not being able to confront the communicative relevance of speech events. But if we study people from the point of view of how they communicate, there would be natural interfaces with the various other neighboring disciplines that study people from other points of view. It would become easy to set up convenient disciplinary boundaries and to have fruitful commerce across them. Linguistics would then lose the isolation imposed by grammar, and at the same time it would become more relevant to those adjacent disciplines where linguistic questions arise.

Perhaps such a course would get bogged down in complexities. It is sometimes suggested that we can never achieve a scientific understanding of people because they are too complex. It is true that people are complex, and their communicative abilities and behaviors are among their most complex aspects. This may be one of the reasons that the human sciences are not yet as highly developed as the other sciences. Differences in the complexity of the phenomena involved are perhaps a major reason why science had its first successes in physics, then in chemistry, and later in molecular biology. But science has been able to deal with complexity where it is found in these other sciences. Fear of complexity is no reason not to try a second-alternative approach in linguistics. Linguists have considerable experience already in dealing with complexity in the traditional approach through grammar.

Perhaps such a course could not be carried out because, as is sometimes supposed, you can’t do experiments on people for one reason or another, or you can’t get scientific evidence about their internal states and ideas. A similar objection regarding the difficulty or impossibility of doing experiments could be raised in the fields of astronomy, human biology, medicine, and elsewhere, but that does not prevent these endeavors from being scientific. Actually, observation and experiment have been extensively used in the other human sciences and even in linguistics. One can obtain experimental evidence about people in several ways. For example, one can observe people engaged in communicative behavior under controlled conditions, videotape what happens, and study it. Alleged difficulties in observation and experiment do not constitute a valid reason for discounting a scientific approach to the phenomena.

This objection sometimes reflects a first-alternative orientation where evidence from observation may not be obtainable. But the issue with which we are concerned is not the place of evidence from the senses in establishing the foundations of such first-domain areas as epistemology, logic, and pure grammar. Rather, the issue facing linguistics is a choice between a first-alternative and a second-alternative approach.

It is sometimes urged against a human science that people are different. Their behavior cannot be predicted. They will do whatever they please. One can never achieve experimental control. This objection has a certain Alice-in-Wonderland quality. Surely we now understand in linguistics that communicative behavior involves constraints. A person cannot in general say whatever he pleases and expect to be understood. Questions of the freedom of the will have been brought up before in linguistics, but this issue is a red herring that would draw our attention away from the contingent nature of prediction everywhere in science. The issue is only whether the study of these constraints will be disciplined by grammar and logic in the first domain or disciplined by science in the second. There is no reason to assume that people are not a part of nature or that they are somehow exempt from the laws of science. It is simply the task of science to find the relevant laws, whatever they may turn out to be.

Perhaps such a course would involve us in giving up too much of what has already been won through the study of language and grammar. This danger will have to be weighed against the advantages of ridding linguistics of what is probably its major source of difficulties and the stimulus for the repeated reworking of grammatical theory in scores or hundreds of attempts. The insights already won would not be lost: they remain recorded in the literature. Valid insights would be available, when needed, to be translated or reworked appropriately into a human-linguistic framework. Since the new framework would cover all the observable phenomena traditionally considered, one would expect that there would be a principle of correspondence roughly relating the two sorts of theories. The carrying over of insights into a new sort of theory would be more difficult than into yet another reworking of grammar, but the results of focusing on a reality would be sounder and more adequate, and thus prove to be more permanent.

On the second alternative, then, linguistics would take up the direct study of people, individually and collectively, from the point of view of how they communicate. Rather than being a linguistics of language disciplined by grammar, it would be a human linguistics disciplined by science. Under this alternative linguistics would not seek to achieve a scientific understanding of language, which we know to be impossible, but instead to achieve a scientific understanding of how people communicate. The goal that linguistics should seek explanations in terms of people is not only compatible with the goal that linguistics be disciplined by science, it is actually the only way that we can have a scientific linguistics.

The second alternative, retaining the goals of being scientific and of seeking explanations in terms of people, and giving up the goal of studying language, has the epistemological and methodological advantages of modern science now only minimally enjoyed by the discipline. In moving from the logical domain to the physical, linguistics would move from a domain where the metaphysics have always been and still are controversial and unsettled into a domain where the metaphysics are relatively settled and have proven over the last several hundred years to be spectacularly successful as a practical guide for the conduct of scientific research in a number of disciplines.

Extending the methods of science even and especially into the foundations of the discipline should yield in time a core body of observationally supported, tested, and tentatively accepted linguistic theory that could withstand attack by any attempted arbitrary promulgation of new and unsupported foundational assumptions. Since the ultimate criteria would be comparisons of the predictions of theory with reality through observation and experiment, theoretical statements would be subject to test by other linguists and could be accepted or rejected on the basis of objective criteria. Factionalism would be reduced. New observations and data could regularly be compared with the growing body of generally accepted theory, and if they call into question some aspects of that theory, the matter could be put to a test. As the body of generally accepted theory grows, the forefront of research would be widened and the cooperative nature of the discipline increased. Results of different investigations would count for more by being relatable to a wider body of accepted theory, which in turn would provide broader and firmer support for primarily observational research.

Linguistics under this alternative would be hooked firmly into the 400-year tradition of modern science and would be able to partake fully of the advantages of the most finely honed and effective system for the advancement of knowledge that is known. Being a true science, it would not be cut off or isolated from the physical and biological sciences, and as a part of science we can expect linguistics to be strengthened by close relations to these other sciences and to be able to cooperate closely with them in pushing back the edges of ignorance and expanding the area understood by science.

The major disadvantages of the second alternative are tactical. Little is known so far about the detailed architecture of a scientifically sound human linguistics. This disadvantage may only be temporary, however, as quite a bit is already known about its foundations. Another disadvantage is that linguistics would appear to be giving up much of what it has won in previous years and centuries. This disadvantage may also be only temporary if it proves feasible to carry over some of these results into the new structure in modified and carefully tested form. Since grammar has been reworked regularly and often anyway, the considerably larger effort to rework these results into a scientifically acceptable theory may not be out of the question, particularly since objective criteria are now available. If the reworking is done well, and carefully validated against the evidence, we may be able to get it right this time and never have to redo linguistic theory again.

WHERE WE STAND

Let us take stock of where matters stand. To continue with business as usual is ruled out because the goals involved are mutually incompatible: The goal of disciplining our studies by grammar is incompatible with the goal of disciplining them by science because the objects of language are not objects of nature given in advance; they are only created by a point of view. The goal of studying language is incompatible with the goal of seeking explanations in terms of people because of the psychological and social reality of grammar fallacies. Language and people exist in different domains altogether. Since a linguistics based on incompatible goals is incoherent, we are driven to choose either the goal of studying language or the goals of disciplining our studies by science and of seeking explanations in terms of people. Of these two coherent alternatives the first, pure grammar, is not appropriate for linguistics because it does not confront the evidence in its domain. It lacks contact with reality, and it does not lead to satisfying explanations. Thus we are driven to the second alternative, a scientific linguistics focused on people, rather than a philosophical linguistics focused on language.

Another way to approach it is in terms of the entities or objects studied. Both the first alternative and business as usual would study the assumed objects of language, while the second alternative would study only people. First there is the ontological argument. Since on the first alternative and in business as usual we are dealing with human language, which presupposes the existence of people anyway, and on the second alternative people are the only entities needed, any assumption of the objects of language would multiply entities beyond necessity and would be rejected by the principle of ontological parsimony (Ockham’s razor). But beyond this, and more decisive in ruling for the study of people in the second alternative rather than the study of the objects of language in the first alternative or in business as usual, is the empirical argument. In the test of theory against observation, the existence of people is confirmed by observation by the senses. They can be observed and studied from a linguistic point of view and from nonlinguistic points of view as well. But this test fails with the objects of language. They cannot be observed by the senses or studied from different points of view. They have no existence independent of theory. Thus again we are driven to the second alternative.

But this means giving up the traditional concepts of language and grammar around which the discipline has always been focused. This is a rather sobering prospect.

It is not surprising that linguistics would have to give up firmly held concepts and methods from an ancient philosophically based tradition if it aspires to become a science, for science does not knowingly tolerate unsupported assumptions. The more highly developed sciences, when they crystallized out of philosophy, also had to give up preconceptions deeply rooted in prior philosophical thought or long preserved in the folk culture. Recall, for example, the problems of a geocentric astronomy, the crystalline spheres, the immutability and perfection of the heavens, the prime mover, the earth as a living being, the flat earth with edges, astrology and divination, alchemy, special creation, pneuma or the breath of life, speaking from the heart, vitalism, spontaneous generation, the luminiferous ether, and many more. These all represented older assumptions that could not be justified by the evidence and had to be given up by science, a move that often met with considerable resistance from entrenched authority and received opinion but won out in the end. If linguistics opts for science and gives up studying the objects of language because they are only created by a point of view, it will not so much be giving up its object of study as finally acknowledging its true object of study, people.

The question immediately arises: Is the second alternative really feasible? It should be emphasized at this point that the conclusions already reached regarding the necessity of a second-alternative approach are independent of how easy or difficult it may be to develop an adequate human linguistics, and independent of the merits of any particular suggestions as to how this might be done. These conclusions establish the necessity of exploring and developing human linguistics if we are not to abandon altogether any attempt at a coherent approach to the phenomena of interest in the discipline. It becomes urgent to find out what kind of a body of theory could be developed for linguistics on the second alternative.

If linguistics gives up the study of language in favor of seeking a scientific understanding of how people communicate, what would it use as a theoretical structure in the place of grammar? That is, for the linguistics of language disciplined by grammar there is a well-known theoretical structure, learned by every student of linguistics during his normal course of study. Its shape is largely independent of whether it also includes an attempt to be scientific or to seek explanations in terms of people. It is the broad picture of language as a relation between sound and meaning, given by grammar and lexicon, conceived in terms of levels and components of grammar, involving paradigms or processes, structures or rules, and allowing a restricted variety of more detailed constructs that are variously introduced by different schools of grammatical thought. Developed over the centuries and millennia it is the conceptual foundation underlying every form of grammar, and it has contributed massively to our folk culture. It is so familiar to every linguist that he can comprehend a new form of grammar in a few weeks of study, or even develop one himself in a few years. It is the framework on which are hung the particulars of every new position. It furnishes the point of view or points of view that create the assumed objects of study in the linguistics of language—the forms, meanings, words, sentences, and the like that seem so real to us who are steeped in that ancient and powerful tradition.

If linguistics gives all of this up, what would take its place?