(Return to the Table of Contents)

Chapter 7

The Paradoxical Nature of Scientific Truth 476
The Whorfian Phenomenon 478
Containment vs. Differential-Enablement Hypotheses 484
Whorfian Effects 485
Creativity Effects 487
Long-Term Performance Effects 490
An Experimental Design 492
The Experimental Treatment 498
Measuring the Host Culture Effect 500
Some Possible Consequences of the Several Outcomes 501

Notes 507

7.1 The Paradoxical Nature of Scientific Truth

Many readers of earlier editions of this book have been charmed by the notion that the Sapir-Whorf hypothesis may actually turn out to be true, and that their own personal commitment to learning Loglan may contribute to that outcome. Most of the discussion of the Whorfian prospect in the loglandical literature takes for granted the truth of the central Whorfian thesis, namely that the structure of individual languages does in some way shape the thought of monolingual speakers of those languages. The potential role of Loglan in mobilizing such effects is also seldom doubted. For such discussions immediately move on to what is evidently the more interesting question: how Loglan may best be used to exploit the "Whorfian effect". What is clearly presupposed by all such discussions is that there is a pool of such effects, and that some or all of them are salubrious.

These are cheerful opinions and they lead to cheerful prospects. They may even be the stuff of which human commitment is made...for example, to the not inconsiderable investment one has to make to learn a sutori language. But while I welcome this action-supporting optimism, I would also like in this final chapter to take a more sober, scientific view of the Whorfian prospect.

In doing this, I will consider several possible hypotheses--all Whorfian in one sense or another--that might be tested by future work with Loglan. I will try to explain exactly what these hypotheses claim, and what we can expect to observe--either in nature or in the laboratory--if their claims are true. More importantly, I would like to show how we can expect to know--not merely to conjecture--that those hypotheses are false should they turn out to be so.

There is an apparent paradox here. The scientific enterprise appears to proceed most vigorously through efforts to falsify ideas rather than to prove them. This is apparently because of both the local and the selective nature of all organismic knowledge. Our sensory apparatus, like that of any other organism, does not allow us to observe nature either completely or directly. So we can never know that any given explanatory model we contrive is true of it. The paradox is that we can, and often do, make the opposite discovery: that some theory that we hold is false. The history of science is replete with instances of a well-established theory turning out to be in some unexamined corner false just as we were getting used to it.

The methods of science are based on this curiously asymmetric relationship of human understanding to objective truth and falsity. We have learned in the last five or six centuries that we and nature are apparently constructed in such ways that we can discover the vices but not the virtues of our theories about her. If hypothesis H is true, we reason, then observation O will surely follow. So we run an experiment, or we go half way round the world to make an observation, having prepared ourselves very carefully beforehand to measure O or to record it should it occur. If we fail to observe outcome O after having invested so much care and attention on its possibility, then we can say clearly and flatly that H is false...and turn back to the drawing-board to repair our model. We have found a defect, a "vice" of our theory; and this is useful, for we then have useful work to do. And it is by such work that scientific models grow and prosper. But if O is observed, if our prediction that it would come about is upheld by the facts, then, paradoxically, we have nothing to do. We can rejoice, of course, and often do. But no intellectual progress has really been made. Confirmation is the absence of something, the failure to find out something definite, namely that our model is false to nature in some concrete respect.

Confirmation is, of course, reinforcing in the psychological sense; it is fun and therefore makes us work to achieve it. But the actual event of confirmation makes very little addition to the fund of knowledge. 1 A confirmed model is an unchanged model, and so all we can be is happy. Confirmedness, moreover, is a temporary condition of nearly every model. We know as scientists that we must not be confident that what even the most abundantly confirmed model says about the universe is true. We have, for example, virtually abandoned the word 'theory' in modern science; for it tempts us to believe. So we use the word 'model'. It is obvious that it is we, not nature, who make our models. It is equally obvious that nothing so simple-minded as a humanly-constructed model will ever provide a permanently true picture of the majestically unfolding universe. We know even after the most resounding series of successes that we could run a new experiment tomorrow and find that some new O which it also predicts, but which we didn't think to look at, doesn't happen, thus learning that our model, though historically confirmed, is now not true to nature after all.

What this curious until-further-notice character of our scientific explanations seems to lead to in the working scientist is, one, skepticism about even the most well-supported theories and models, two, a readiness to overhaul one's own beliefs when new facts emerge, and three, a methodological resolve to try as earnestly as one can to disprove even one's own theories--that is, to find their defects--rather than try to prove them. The modern scientist, unlike those of earlier centuries, apparently harbors a suspicion that attempting to find "proofs" of any natural principle is, increasingly, a foolish enterprise. Proofs deal with certainty; and, as the late British philosopher Bertrand Russell once put it, we only know with certainty what is not about nature, and what we know about nature is never certain. So demonstration, while it may be a useful tool to a teacher or a civil servant, has an increasingly small role to play in science.

Such a Spartan resolve is difficult to live with. Moreover, it is more than just a touch inhuman. It counsels us to feel happy only when we fail to disprove our most cherished beliefs. In fact, one could argue, the happiness that one deserves to feel about any successfully predicted observation is in strict proportion to the earnestness and the cunning that one had invested in the effort to make it not appear. If, after so much effort to defeat it, the predicted outcome emerges anyway, then we are entitled to a little jubilation. And therein lies the paradox. Science is an enterprise whose logic demands that we try to fail, and then be happy when that effort fails.

Let's apply this curiously inhuman logic--no wonder science was such a late emergent in human cultural evolution!--to the Sapir-Whorf hypothesis, or rather, to the cluster of hypotheses that can be deduced from the several existing models of how those "Whorfian effects"--if they exist--are generated.

7.2 The Whorfian Phenomenon

Sapir and Whorf thought they had seen a pattern of relationships between the cultures of certain peoples and the structures of the individual languages with which those cultures were associated. That repeated pattern seemed to be containment. It was, they thought, as if human cultures were contained in their own languages, that each language set limits on the minds of its monolingual speakers so that each culture was constrained in its development by the very structures of the language in which it was expressed. Thus individual human cultures seemed not to develop in certain directions but to develop quite freely, even luxuriantly, in others. The observations on which Sapir and Whorf based these conclusions were originally confined to a small set of exotic peoples and languages; but Whorf later applied this containment model by analogy to several European languages and peoples.

Let us suppose that what Sapir and Whorf thought they had observed in nature actually exists. For brevity, let us call it the "Whorfian phenomenon". What other data do we have that might help is to understand the Whorfian phenomenon? There are two other fairly impressive bodies of data that relate human cultural enterprises to linguistic structures. One is provided by the history of mathematics; the other, by the everyday experience of travelers who learn foreign languages.

The history of mathematical progress over the last six centuries is a curiously repeated story. From time to time, the invention of a new notation--plus, of course, the system of calculation that quickly develops to explore its boundaries--ushers in a flood of new results and problems. These seem always to have been results and problems that had not been foreseen--indeed, were scarcely expressible--in the epoch before the new notation appeared on the scene. From its earliest beginnings the story of mathematics seems to have been segmented by notational innovations of this kind: the invention of zero and, what is more, its expression as a manipulable digit; the invention of negative numbers and their expression; of irrational numbers and theirs; of imaginary numbers; complex numbers; modular numbers; groups; matrices; lattices; transfinite numbers...each new conceptual advance having been accompanied by its separate, new notation.

One explanation of these historical phenomena is that it was, in each case, the new notation itself that allowed successively wider domains of mathematical phenomena to be explored. Indeed, this seems to be so obvious a feature of the history of their discipline that few mathematical workers seem to have any doubt that some kind of Whorfian mechanism has control of their intellectual lives. The power of notation is such a banal truth among them that it seems hardly worth mentioning. Computer scientists seem also to be of this opinion. Again it is the power of notation--often, in their case, as expressed in a new programming language--that seems to set limits on the elegance and power, in the end the productivity, of the programmer's work. Again it is by expanding those limits by inventing new notations that progress in this field, too, seems at least in part to be occasioned.

We might well surmise that in all those fields of human creativity in which inventions are expressed in written symbolisms--mathematics, computer science, logic, poetry, among many others--it is the resources of the language or notation in which the worker is working that seems to free da's mind. Thus long before Sapir and Whorf wrote of the possible linguistic constraints on thought, mathematicians were aware that it was the elegance of their notations that was "liberating minds". Before imaginaries, for example, the mental domains of mathematicians had been in some sense "restricted by" the absence of a "suitable notation". In effect, and except as they were then invested with a prior existence by retrospection, imaginaries did not exist. But those same minds were in some sense "liberated" by that "suitable notation" when it at last appeared. (Historians of mathematics tend to have a curiously impatient view of these matters...as though notations, like continents, were waiting to be discovered.) Thus the domain of what is thinkable mathematically has grown explosively over the last six centuries; and it seems obvious that it has done so by a series of definite, discrete pulsations, each one associated with a new mathematical idea, and a notation in which to express it.

But there's the rub. Is each domain-augmenting "pulsation" a result of the notation? Or of the invented idea itself? We cannot be certain. As in other Whorfian phenomena, there is no disentangling the substance of these new mathematical ideas from their form. At least there seems to be no chance yet of separating the electrochemical traces in the brain of the mathematician that represent the new mathematical concept about which da is thinking, from the effects of the new "squiggling rules" that guide da's hand on paper and so reveal new relationships to da's eye. So, obvious or not, we must put the histories of mathematical invention and notation in the same category of unexplained phenomena as the culture-language phenomena observed by Sapir and Whorf. But this is the question we must somehow answer about both: Is it the human mind, independent of the languages and symbolic notations in which it expresses itself, that is evolving in these culturally distinct or historically pulsating ways? That is, are all such developments implicit in the mind-stuff itself, whatever that may be? Or do the linguistic/notational garments in which the mind dresses its thoughts, and in which it appears to do much of its work, at first somehow "constrain" those minds, and then, when new costumes of previously undreamed-of symbolic cut are invented, they somehow "liberate" the mind from just those earlier "constraints"?

This, then, is the puzzling problem that faces all serious investigators of Whorfian phenomena: in nature, at least, there seems to be no way of disentangling the form of the thought from the thought itself. Only experiment gives any hope of separating linguistic form from either cultural or intellectual content...unless, of course, nature obligingly runs some disentangling experiment for us.

But what about the third pool of phenomena which we thought might possibly be Whorfian? The experience of foreign-language speaking travelers? Again the data are banal. Again they seem to lead to conclusions that are "indubitable" to the travelers and expatriots who experience them but distinctly doubtable to the scholar. One seems to "know", as a traveler, that when one is speaking Spanish with a Spaniard, or French with a Frenchman, or Arabic with an Arab, that one is in direct, immediate contact with a mind of strikingly "different flavor" from the minds one left at home. The differences one senses are (for want of a better word) "philosophical"...not more substantial than that. Yet the more widely one has traveled, and the more intimately one has known the minds of foreigners while traveling--minds that were forged originally by other languages, one senses--the more deeply one is convinced of the existence of powerful linguistic (or only cultural? or are they gene-pool?) forces that must be cunningly at work everywhere on the planet to give shape and flavor to the human soul. One "knows" this, as a traveler, as certainly as one knows the differences between the overt styles of cuisine, work, friendship, art, religion, architecture, romance, manners and decor that decorate the countries one is moving through.

Most anthropologists would disagree with this travelers' intuition. Most would brand it as superficial, confusing the messenger with the message, the language with the culture it expresses. For the bulk of things one notices when one travels, this must surely be true. It must, we feel, be the culture, not the language, that is responsible for most of the cleavages we notice in the human fabric when we travel. Some anthropologists would say that it is the internal dynamics of cultures in reaction to their habitats that creates those deep discontinuities in beliefs and customs in the first place. But most would confess they don't know what creates the gullies in the human landscape, but there they are.

But isn't this the same problem of form vs. content that we encountered earlier? Indeed, how can we disentangle the culture from the language in which it is expressed? Isn't this, too, a case of disentangling--at first conceptually, and later experimentally--two forces that are in nature tightly wrapped? As soon as we begin to consider these travelers' data, and the facile explanation all travelers seem to invent for them, again it is clear that we are conjuring up forms as possible causes of mental and social behavior that are in fact disentangleable from the substance of that behavior itself. Like the ethnologist, we are led to ask, Is it the cultural dynamo itself we are sensing when we speak with foreigners in their own tongue? That is, is it something quite independent of the linguistic garments in which we find it (fascinatingly) garbed? Or is it the very garments, their cut and style, that have shaped (as we seem always to believe they have) the myriad forms of human philosophy which we encounter when we travel?

Sapir and Whorf are not the only scholars who have believed that they'd found convincing evidence that the shape of language shapes the culture and the mind. The American philosopher F. S. C. Northrop, in his (1946) book The Meeting of East and West, examines the cultural gulf that lies between the Chinese civilization and the cultures of the West, and offers us a wide variety of historical, geopolitical, and cultural explanations. But prominent among them are the linguistic differences between Chinese and the Indo-European languages, according to Northrop. Indeed, he adduces the deeply esthetic orientation of the Chinese civilization directly to the metaphor-richness of this remarkably long-lived and grammatically permissive language...a possibly important feature of the grammar of Chinese which we have had occasion to note more than once in this book.

Northrop's hypothesis is yet another Whorfian hypothesis, this time a facilitory one. Indeed, so is the most obvious explanation of the apparent effect of notational advances on the development of mathematics; for whatever else they are, notations that replace other notations are facilitory. This is a very different matter from talking about grammatical obligations and their restrictiveness. Restriction takes place in present time. But no one can seriously argue that the absence of a "suitable notation" was restrictive before it was adopted. How can it have had any effect before the event? An absent thing can have no effects, neither restrictive nor facilitory.2 Yet the travelers' data seem not to be better explained by either kind of explanation, neither the facilitory nor the restrictive one.

Whorf's containment model for culture-language phenomena is definitely restrictive, however...at least that is most frequently the thrust of his arguments. Whorf uses the restrictive argument especially effectively with respect to obligatory grammatical rules: the fact that all Indo-European sentences, for instance, must bear some mark of tense. Also, he observes that I-E predicates are sorted into unmixable "metaphysical" categories: nouns for things and substances, verbs for actions and processes, adjectives for qualities, and so on. All these structural features of languages are seen by Whorf as limiting the domain of the possible for minds shaped by them. Yet Northrop speaks of the facilitation of poetry by the grammar of the Chinese language, which, with its virtually limitless interplay of verbal categories, is a lot like Loglan in its freedom to combine predicates. And certainly there is no historian of mathematics who would balk at the word 'facilitation' to describe what has happened linguistically in mathematics...if, indeed, anything linguistic has. On the other hand, 'enablement' might be a better word to describe what has happened in mathematics.

So let us distinguish between two quite distinct kinds of Whorfian hypotheses: those that offer containment by domain-restriction as their explanation of all or some of the Whorfian phenomena, and those that offer facilitation or enablement. The obligatory grammatical arrangements of the natural languages are instances of the kind of cause to which restrictive hypotheses attribute Whorfian phenomena, while the invitation to pour old wine into new notational bottles--to rethink old problems in new formulations--is an instance of the kind of cause to which enabling hypotheses attribute some or all of the same phenomena.

What kind of hypothesis is Loglan good for testing? Well; both kinds. Loglan is both enabling and domain-enhancing. At certain points the design of Loglan incorporates restrictive hypotheses by removing many of the domain restrictions which Whorf described, or at least by pushing them outward in domain-augmenting directions. The loglanist will know what these are; I hardly need to mention them. But chief among them are the absence of any obligatory inflections on the Loglan predicate, and of any grammatically significant distinctions among them. At the same time, in the same design-for-a-language, enabling hypotheses are also served. For example, the intricate but instantly decipherable connective apparatus of Loglan is plainly meant to facilitate the logical transformations. And the fact that Loglan predicates, much like Chinese ones, may be freely combined in literally any order is meant not only to remove domain restrictions of a metaphysical kind on metaphor-making, but also, and not altogether incidentally, to facilitate the poetic richness of metaphor in the language as well.

The question is, Will Loglan do any of these things? Will it do all of them? Will the "removal" (in the sense of not requiring any) of obligatory grammatical arrangements (like tense inflections) actually increase the domain of the sayable, and hence of what is thinkable in Loglan? Obviously we don't know yet. But these are formal features of the language, not substantial or cognitive ones; so they can be invoked by both kinds of Whorfian hypotheses. Some matters of substance are expressed in the design of Loglan, too, of course, but these, like its accommodation of the vocabulary of science, are robustly international in bias and provincial only in the planetary sense of belonging to the human "global village". In fact except for science, Loglan as it stands is almost entirely a set of content-free linguistic forms.

7.3 Containment vs. Differential-Enablement Hypotheses

Let me first make clear that the containment vs. the differential-enablement interpretations of the Whorf hypothesis are not rival hypotheses in the usual scientific sense of that phrase. That is, the two models of the origin of Whorfian phenomena are not incompatible with one another in the sense that they may not both be true at once. As applied to cultural differentiation, for example, as this is sensed by the traveler and described by the ethnologist, the restrictive interpretation of the Whorfian hypothesis (Whorf's own interpretation, in the main, although enablement is a recurrent minor theme of his analyses) generates an hypothesis that language A of culture A' is sufficiently different in some structural feature--say its tense system--from language B of culture B', that the domains of the grammatically permissible in the two languages A and B differ in both size and extension, and in ways that explain at least some of the content differences--that is, differences in beliefs, techniques and values--that characterize the two cultures A' and B'. In other words, the containment hypothesis argues that the observed A'-B' cultural differences are at least partly due to the differences in the "shapes" of the two linguistic bottles, A and B, in which the two cultures have been--and still are--contained. But it must be emphasized that there is nothing in such a containment-oriented explanation of the observed phenomena that is logically inconsistent with, or dependent on there not being, language-related facilitations also at work in the mental lives of speakers of A that are not at work in those of B, and vice versa and that these also explain some of the differences between A' and B'.

So it is a question of mix. We may eventually know what the mix of causal mechanisms in any given A-B/A'-B' language/culture comparison is. But for the present, we have no idea how the ball will bounce between the two kinds of hypotheses. Both kinds interest us. And both kinds of models generate detailed predictions of experimental outcomes. It becomes the task of the early experimenter, then, to measure a broad enough range of outcomes to include the predicted products of both kinds of mechanisms. For example, enhanced facility with metaphor might be an expected (Whorfian) outcome of Loglan's undifferentiated predicate-structure, should Loglan be the A in some A-B comparison, while enhanced awareness of the varieties of possible existences might also be an expected consequence of the lifting of metaphysical restrictions as this might be attributable to the same undifferentiated lexical arrangement on the containment model.

So at the same time that certain enablement effects were being generated by what we might think of as the "logical" parts of a language, certain "liberation effects", such as measurable increases in speakers' senses of the possible, might be simultaneously attributable to a restructuring of its "metaphysical" parts, for example, to a reduction in scope of the obligatory arrangements of the same language. The point is that both kinds of effects could be exhibited historically by the same population during the same period--presumably one in which such linguistic changes were occurring naturally--as well as by the same subjects in the same experiment. To put the experimental point more concretely, there is nothing (that we know of) that would make it psychologically impossible--or even awkward--for a subject to become both more aware of existential possibilities and more adept at coining metaphors; and for both events to happen during the same experimental treatment. So we do not have to choose between the containment and the differential-enablement models in the early experiments. In fact, our best bet is to provide for the measurement of both kinds of effects in the earliest experiments. It is all but inevitable, of course, that in later experiments, more sophisticated models will be developed, and that these will predict a different mix of outcomes for different language-pairings, and so require a different battery of instruments to record them.

7.4 Whorfian Effects

By a "Whorfian effect" I will mean in this chapter that subset of Whorfian phenomena that can be regularly released in experimental settings. Of course we don't know what these effects are yet, or even whether any can be released. But in order to measure any Whorfian effect whatever, we must obviously be able to predict it, or at least to predict phenomena which are very similar to it, in order to prepare the instruments that will measure or record that effect...in a word, that will be capable of "catching" it. We must, in short, have some clear ideas as experimenters about the kinds of effects we are likely to release in our language-learning experiments in order to be able to capture them as observations when we run those experiments.

Loglan was designed to release Whorfian effects in subjects who learn it as a second language under conditions of laboratory control. It was meant to do this in two ways: (1) by increasing the domain of the grammatically possible over the formal size of that domain in the subjects' primary languages, we expect some of the subjects to occasionally show signs of using or having used the larger domain in their thinking...that is, of having been "liberated" from the smaller one "imposed by" da's native tongue; and (2) by providing the subject with notational facilitation of certain language-mediated performances, such as logical and causal inference, we again expect at least some subjects to show sips of increased competence in just those areas of da's mental life as involve those kinds of performances. Let us call the first type liberation effects--those that, according to the containment model, arise from the removal of constraints--and the second type enablement effects those that arise from supplying the subject with what amount to notational improvements, in the mathematical or computational sense, over corresponding structures in the subject's native tongue. In short, we hesitatingly predict that there will be signs--perhaps faint signs at first, but definite signs, and observable in the most receptive subjects if not in all--of both domain transcendence, as permitted by the lifting of constraints, and enhanced mental performance, as enabled by what may then be seen as Loglan's notational advances over the state of these same matters in the subjects' primary tongues...and I trust I need not add 'if Whorf is right.'

Of course we would expect the enablement effects--if they exist at all and if any given exposure to Loglan lasts long enough for them to develop--to include various kinds of long-range historical effects, such as enhanced rates of discovery and innovation, in the Loglan-speaking portions of some affected population; for this result, too, is predictable on the Whorf hypothesis given linguistic enablement over long intervals in large populations. In fact, there are those who will say that the "real" confirmation of the Whorf hypothesis will not take place until and unless such historically visible results do emerge. This may well be; but such historical outcomes obviously cannot happen until Loglan has been adopted as a sutori language by a considerable number of people for some years. Nothing is more likely to lead to widespread adoption than clear experimental results that its use is likely to be both individually and socially salubrious. So with this acknowledgement of the importance of long-term effects--but as qualified by their probable dependence on some useful short-term ones--let us return to our experiment and its necessarily short-term focus.

We must now address the matter of the time it takes to develop a response, which is so crucial for a real-time experiment. Cutting across the field of liberating and enabling effects, is the question of the interval required for generating a measurable expression of some effect. We can expect (a) short-range creativity effects or those verbal expressions of both domain-transcendence and enablement which might be emitted early in the second-language learning experience; but we may also expect (b) long-range performance effects, or those relatively permanent alterations in the subject's verbal/mental competence which may not appear in measurable magnitudes until deep familiarity with an enabling and/or domain-expanding second-language has been achieved. Obviously only the first of these are suitable for assessment by a short-term experiment, say a Summer Workshop. But the second are more interesting, and certainly more germane to the ultimate assessment of Loglan as a second language. So we must keep in mind the design of longitudinal studies to track the slower-developing performance effects of the Loglan-using experience, as well as the design of instruments to trap the short-range and perhaps even ephemeral creativity effects we will with luck observe even in the earliest experiments.

Let us now consider these several kinds of possible Whorfian effects in some detail, concentrating first on the creativity effects we must be prepared to measure in short-term experiments, and looking later and more briefly at the longer-range performance effects we might want to track in longitudinal studies of our once-experimental subjects after we release them.

7.5 Creativity Effects

Linguists and other travelers have long been aware that certain playful, and often utterly novel, uses of one's native language often accompany sutori-language learning. Indeed, they accompany many non-linguistical varieties of formal learning as well, professional logicians, computer programmers, and masters of other essentially formal disciplines often exhibiting the same sort of verbal playfulness (Lewis Carroll having been, perhaps, the most conspicuous example of a linguistically playful scholar). I have noticed that these effects are particularly salient in students of Loglan.

Creative novelty in the use of one's own language is, of course, explicable on either version of the Whorf hypothesis, but especially on Whorf's own version of it, the containment model. We should expect such creative play to occur whenever a new and language-transcending symbolism is being acquired, whether that symbolism is strictly linguistical or not. For what one learns through the manipulation of a new and powerful symbolism--like symbolic logic, for example--is precisely that one's native speechways are both imprecise and unnecessarily constraining, that they are both convention-bound and logically incomplete; and it is this cluster of discoveries that invite the extension of those speechways by "illegal" means into new domains of meaning or existence. The interesting possibility is that these random excursions beyond the conventional usage-boundaries of one's native tongue may be precursors to genuine, long-range performance changes in the minds so touched.

I suggest that those who conduct the first Whorfian experiments measure the emission rates of these presumable early effects of second-language learning by counting their instances in recorded verbal protocols obtained in response to a set of standard stimuli administered in interview situations. Each stimulus must be usable in both "pretest" and "post-test" measurement sessions, that is, both before and after the experimental "treatment" has been applied, a treatment which in this case is the learning of a second language. On the other hand, no stimulus should be reused on the same subject. One way to do this is to divide the set of stimuli into two demonstrably equivalent sets. Then a random half of the subjects who receive the pretest will receive the two stimulus sets in one order while the other half will receive the other. The Whorfian prediction is, of course, that the "Loglan experimentals"--the subjects who learn Loglan--will exhibit the largest net increase in verbal creativity as so measured.

It is essential that the creativity-evoking stimuli be found to release high rates of such behavior in the subject population; and this can best be done by reducing the variance in verbal responsiveness between subjects. A type of stimulus that is known to have such properties is the "how-many-uses" question. Here is one: 'Suppose you were dropped by parachute into the middle of an Amazon rain-forest. How many uses could you make of this brick?' '.... this newspaper?' '.... this tennis shoe?' Procedures such as this are well-known in creativity research; see Parnes (1962).

As to the specific content to be abstracted from such open-ended protocols, and counted as instances of creativity in them, thirty years of work with Loglan has put me in touch with numerous categories of creative verbal behavior in the friends, co-workers and learners who have been exposed to Loglan at various stages of its development. It is possible that none of this behavior was actually evoked by exposure to Loglan but, given the circumstances in which I observed it, all of it could have been. For the benefit of other future experimenters let me record here the several types of this behavior that I have observed: (1) richness and oddity of metaphor; (2) unusually frequent designation of previously "unheard of", or unthought about, individuals and phenomena; (3) increased awareness of ambiguity as evidenced by jokes or other usages that call attention to it; (4) a taste for neologisms or for bizarre or over-literal usages; (5) the invention of inflected (or de-inflected) forms that do not exist in ordinary usage in the speaker's native language but are in principle possible in it (e.g., 'coolth' 'idiosyncrat', 'ert', 'qualitiedly', 'therapped grouply', 'encomiast', to list a recent sample); and finally (6) a heightened sense of fun with one's own and other people's English, e.g., with the often comic contrast between what people actually say and what they think they are saying...or have said.

Admittedly, none of these effects need have been genuine consequences of either learning or working with Loglan. It is true that they are often thought to be "Whorfian effects" by the people who experience them, but that won't do as science. They may be verbal/mental behaviors that develop quite spontaneously in the life-histories of whatever special kinds of persons are attracted to learning or working with Loglan...or, indeed, with its inventor. Even so, familiarity with these quasi-Whorfian effects among the aficionados of the language has probably given us at The Institute a number of concrete insights into how to measure the incidence of at least one species of "verbal creativity", one, moreover, that may be argued to be a direct expression of the expansion of the speaker's "ontological domain"...by which I mean the domain of possibly existent entities as da perceives them. I have shared these insights with my readers in the hope that among them will be the next generation of Loglan scholars, on whose shoulders the responsibility for designing and running the Whorfian experiments may eventually come to rest.

Such creativity measurements, however tentative, will allow the early experimenters to raise the question--both by additional experiments and by longitudinal studies later--of whether these playful verbal effects are persistently related to any more substantial kind of creativity. In our present corps of loglanists, they seem to be. But at present, this observation provides us with no more than a preliminary hypothesis. For the causal influence may very well run in the other direction. Thus it may be that only productively creative people have managed to learn, or even desire to learn, Loglan through their own boot-strapping efforts. This reminds us of the sobering fact that the experimenter must not draw his subjects from the population of persons who are spontaneously attracted to Loglan, as da might be inclined to do, but instead from some other, more broadly defined pool of available subjects, say university students, in which the association between willingness to participate in linguistic research and personal creativity is not likely to be so strong, and hence so troublesome. This is a matter we will take up in the section after next, the one on experimental design.

7.6 Long-Term Performance Effects

At its most precise, the assessment of the long-range effects of learning Loglan requires that performance measures be made at an early stage in each subject's acquisition of Loglan, and again at the stage of da's full linguistic competence. As a diverse set of controls on such longitudinal studies, it would be desirable to track both the growth of linguistic competence and the performance effects that might be associated with it not only among (1) the Loglan-learning experimentals from some short-term experiment (see next section), long since over, who had continued their studies of their own volition, and had eventually achieved full competence (measured at some realistic level after some substantial interval, say five years), but also among the following three additional kinds of subjects: (2) the Loglan experimentals who had not achieved competence by that time; (3) the other experimentals, that is, those who had learned other second-languages as part of the same experiment; and (4) at least some members of the control group; see the upcoming section on experimental design for the source of these four kinds of experimental subjects, some of whom might also be longitudinally tracked in long-term studies.

A problem with all longitudinal studies, of course, is the high loss rate of subjects over time. People move away, lose interest, fail to respond to inquiries. I will assume that this problem can be solved, that a reasonably high proportion of the original pool of subjects can be tracked and kept interested in what is happening to themselves and others linguistically. A further assumption of such a study design is that some substantial number of the Loglan experimentals, as well as the experimentals who were exposed to other second-languages, will continue their language studies voluntarily and ultimately achieve respectable competence in their respective second-languages. Only then will we be in a position to compare the performances of the several groups of experimentals with each other, and with the controls.

There is a second, even more troublesome problem with longitudinal studies of this kind. There may be--indeed, there probably will be--what statisticians call a strong first-order interaction between whatever causes a subject voluntarily to pursue da's studies to competence and the particular language in which competence is being achieved. Thus, subjects of one type may persevere in French to competence, while quite a different type of person may win through to full speaking competence in Loglan. If such an interaction exists, and it is strong, then any difference between the mental performances of the two linguistically competent groups could be as reasonably attributed to the differences between, let us say, their genomes or their personalities as to the formal differences between the two second-languages they had happened to learn. There are, fortunately, ways of solving such problems of longitudinal studies statistically; but it would take us too far afield to discuss them here.

I shall now list a number of specific verbal and/or mental skills whose enhancement might be arguably related to the specific structure of Loglan vis-a-vis English, and on which we might expect, therefore, the performances of competent Loglan speakers to be higher than, say, those of the subjects who had learned French on the same measures: (1) the understanding and the use of metaphor, as in the appreciation or composition of poetry, or in the acquisition of new vocabulary; (2) competence in achieving clarity in communication, as in the stylistic avoidance, or the editorial detection and repair, of both syntactical and lexical sources of ambiguity; (3) logical competence, as revealed by (a) the correct use and/or understanding of quantifiers, explicit and implicit ones; (b) drawing correct inferences from stated premises and/or detecting and correcting incorrect ones; (c) supplying the missing premises of incomplete arguments; and (d) the correct use and/or understanding of the logical modalities (actuality vs. potentiality, etc.). Finally, metaphysical awareness may also be argued to increase while acquiring competence in Loglan, and this might be measured by (4) the ability to understand and/or appreciate non-Western cultural assumptions about reality, or by (5) the subject's ability to reexamine da's own assumptions. Inductive and abductive competence may also be expected to increase, as expressed in (6) the frequency and originality of new insights and hypotheses.

The selection of standard psychometric measures of such human performances, where they exist, or the construction of new ones where they do not, would indeed be a challenging task. Nevertheless, if post-experimental studies of a subject-tracking, longitudinal type are ever to be performed, their initial conditions must be met in the earliest experimental designs. One possibility would be to include such a performance battery in the pretest of the original experiment. This strategy, however, might so increase the magnitude of the pretest/post-test interactions of that experiment as to mar its sensitivity to short-term, creativity effects. Fortunately, such destructive interactions can be investigated beforehand. But another, safer course would be to administer the performance battery along with the experimental post-test, where it would have no chance to influence the outcome of the experiment itself. This second strategy assumes that no very substantial changes in these performances will have taken place in, say, a two- to six-months' exposure to a second language, or that, if substantial, they will appear as clearly identifiable performance differences between each of the several groups of experimentals and the controls. Another design consideration is that, since the Loglan experimentals will be the crucial subjects in all future longitudinal studies, and since the number who will eventually achieve linguistic competence will inevitably be less than all, their original number should, accordingly, be larger than the other groups in the original experimental design.

7.7 An Experimental Design

Probably the most fruitful field in which to investigate the constraining effects of monolinguality, as predicted by the containment model of the Whorf hypothesis, is second-language learning. According to this model, some expression of the Whorfian phenomenon takes place whenever anybody learns a second-language. This is because of the predictable widening of grammatical constraints that must, if Whorf is right, take place during any instance of second-language learning. No two languages have identical structural constraints under a Whorfian analysis. Therefore the logical sum of grammatical/ontological/semanticaI possibilities in the mind of a person who has learned any second language will always be larger than the set of those same kinds of structurally determined permissibilties when da. was monolingual. The effect will be large and noticeable when, other things being equal, the second language has a domain of grammatical, or lexical, or designational possibilities that is far larger than, or differently grained than, that of the subject's first language; it will be small and unnoticeable when the two languages are so closely-related historically that there are only small differences between their grammatical, lexical and ontological domains. On this same hypothesis, the reason we have not noticed these effects before as scientists is that, until the development of Loglan, we had not found a way of bringing the experiences of the mind-teased traveler and the notation-enabled mathematician into the laboratory.

But suppose we do bring natural sutori-language learning into the laboratory, let's say on a Summer Workshop basis for a "total immersion" language-learning experience. Suppose we use before-and-after measuring of relevant behavior with a view to capturing Whorfian effects. If Whorf is right, they'd be there. But we would probably find that the magnitude of such changes in behavior after that brief experience--especially with any of the languages commonly learned as second-languages by the English-speaking university students (whom I expect to be our first subject population)--would turn out to be so small that we would not be able to detect any changes at all with our "first-generation" instruments...by which I mean the instruments we had prepared for these earliest experiments. For those early instruments would necessarily be crude ones compared to later generations of the same devices. Not only that, but even if we were lucky enough to capture--even in our first experiments--effects large enough to consider Whorfian, we would not be entitled to attribute them exclusively to linguistic structures. For whether we plan it or not, whole cultures would have gotten involved in those totally-immersing Summer Workshops.

Suppose French had been the language taught in one of these workshops, and German in another. Suppose further that our subjects had all been (previously) monolingual English-speaking North American university students, and that the teaching and measuring had been performed in both workshops under substantially similar conditions. Would we be entitled to say that the French and German languages were responsible for whatever "Whorfian effects" we thought we had found? Certainly not; one cannot teach French by immersion methods without bringing in the French, nor German without bringing in German life, and culture. So even if we did find (as is unlikely) that solid and persistent changes had taken place in the mental lives of both experimental groups, and also that equally solid differences separated the sets of measurements made in the two workshops, even then we would not be entitled to attribute those changes and those differences to linguistic structures as any test of the Whorf hypothesis would require us to do. French and German cultures would also have been involved. There is no practical way to keep them out.

It is at this point in our thinking that the strategy that led to Loglan arises. I discussed in Chapter 1 how using an invented and therefore culture-free language like Loglan might help us "disentangle" the linguistic from the purely cultural effects of second-language learning. But this is only part of the experimental strategy, and, in fact, the smaller part. A more important reason for using a small, (relatively) culture-free language like Loglan is that such a language can be designed to turn on Whorfian effects in large amounts. Loglan--if it works--will be a kind of large-bore linguistic tap for turning on Whorfian effects. The hope is that we will be able to release them in such magnitudes that our necessarily gross first-generation instruments will have no difficulty in measuring them. 'Later, when we learn more about Whorfian phenomena in all their subtlety--and there will be subtle ones if there are any at all--and have had the experience of observing the full range of the effects that we have learned to produce experimentally, then we will become more skillful in measuring them. Scales will be developed. We may eventually become capable of detecting even minute amounts. When that happens, then the differences between the Whorfian effects of learning even two closely-related natural languages may well be studied...even though even the first-order effects may be unmeasurable now.

So the centerpiece of our design is the predictably large effect of learning Loglan. Still, a good experimental design to test the Whorf hypothesis should provide a comparison between at least two sets of linguistically-produced results. One advantage of having two sets that were presumably linguistically-produced would be that they could be examined for signs that they were. For example, if the effects of learning French on English-speaking subjects were qualitatively similar to those of learning Loglan, and all or most of them showed movement in the same direction as the Loglan-associated effects, then we could argue that they were at least the same kind of effects even if strikingly different in magnitude. If there were no qualitative similarities, then that would warn us of some problem with our method. So it is not enough, methodologically, to measure only the presumed Whorfian effect of learning Loglan on, say, English-speaking minds. Even if that effect were large, we could doubt that it was a genuinely linguistic effect we were observing.

Let me be concrete here. Let us say that a "significant Whorfian effect", as we will now use this phrase, is a statistically significant difference between a pair of before-and-after measures on some Whorfian parameter, say the creativity effect discussed in Section 7.5. What is important, if our experiment is to submit our model to a searching test, is that we have at least two such Whorfian effects to examine at the end of the experiment: one that we predict will be small, possibly even insignificantly different from zero, and attributable to French, say (and to France!), and another that we predict will be much larger, significantly non-zero, and attributable to Loglan (and to a not yet existent Loglandia). The point is that if the French-on-English effect is significantly different from zero as well, then it must be qualitatively similar to the Loglan-on-English effect if the results are to be regarded as supporting the Whorf hypothesis.

But a tendency established by two points is much less meaningful than one established by three. In fact, we can practically double the amount of information we get from our experiment if we provide for a third set of Whorfian effects, from a third second-language. To be most informative, the language we choose as our third experimental language should have a predictable Whorfian effect that is intermediate in magnitude between the predictably large Loglan-on-English effect and the small one we expect to get from French-on-English. With what sorts of languages would we expect to observe such intermediate effects? Well; obviously not one that is in the Indo-European family, as both French and English are. It would also be desirable for the intermediate language to be more Loglan-like than French is. Chinese is a suitable candidate for the third language on both counts. It is completely unrelated to English; and, from a Whorfian point of view, it has a grammar that is marvelously different from English. It is also similar to Loglan in several interesting respects. It is not as "marvelously different" from English as Loglan is, of course, but it is in that direction.

With three such second-languages in the experimental design, we will be able to make even more exact predictions. We will now be able to say that the Chinese-on-English effect will be smaller than the Loglan-on-English one but substantially larger than the French-on-English one...if Whorf is right. So with a three-experimental-group design there are twice as many ways in which the Whorfian predictions can be wrong; and that is precisely the kind of improvement that we seek.

We are now envisioning a classic three-group, trend-assessing experiment in which all three kinds of experimental subjects--i.e., all those who were going to receive an experimental treatment of some kind, either with French, Chinese or Loglan--would learn in a Summer Workshop setting the particular second language to which they had been randomly assigned. It is obviously important that the assignment process be a completely random one. This would minimize--indeed, it could well eliminate--the effects of differential attraction between given subjects and the second-languages they were assigned to learn. If such effects were allowed to exist, they would align themselves with language selection where they would then exert large, unknown influences on the experimental outcome. In order to avoid them by using a random assignment procedure, however, we must evidently appeal to the kind of subject who would agree beforehand to accept the luck of a random draw. Such individuals might be fairly rare--rarer, surely, than those who could be tempted to volunteer to learn some specific one of our three languages--and to encourage them might require other, even monetary, incentives. If this turns out to be the case, financial incentives will have to be supplied; but these would probably add only a modest amount to the already considerable cost of such an experiment. Please note that it would spoil our experiment beyond salvation if we were to allow only those who were strongly drawn to Loglan to learn Loglan, only those passionately drawn to Chinese culture to learn Chinese, and only those with a secret desire to read French poetry in the original to learn French. Yielding to such powerfully differentiating forces among human beings would immediately reduce our study to a study of three, possibly sharply different pools of human tastes and temperaments; and none of the outcome differences among them could possibly be regarded as having been linguistically caused.

Let us now consider the all-important question of statistical controls. One would think that all of the experimental subjects would have to be exposed to a pretest on the Whorfian measures so as to establish their current, pre-experimental levels of creative behavior. It is true that some subjects will have to be so-measured in order to provide good estimates of the levels of creative behavior with which the experiment begins; but it would be a mistake to expose all our subjects to such pretesting. It turns out that pretests interact strongly with experimental treatments of this kind, and the unavoidable effects of that interaction must, therefore, be estimated and, in that way, statistically controlled. To control such interactions--that is, to take them into account statistically while evaluating our results--we must measure the final levels of the outcome variables both when the pretest was present and again when it was not. The difference between the two measurements will then give us our estimate of the strength of the pretest effect, an effect that can then be withdrawn mathematically from the main effects of the experiment.

The most efficient way to get an accurate estimate of the pretest effect is to divide each experimental group into random halves. To one half of each group the pretest is then given; to the other half, it is not. As the two halves of each group will then be mixed back together again in the workshop, there to live and work together while learning their particular second-language socially, it is essential that the people who have experienced the pretest--and who will thus have data from which any alert observer will be able to infer at least some of the reasons for the measurements they had undergone (it is of just such inferences and their usually cooperative effects on subjects' performances that the pretest effect is in fact composed)--be warned not to discuss their pretest experience with anyone else...especially-not with those who did not experience it and who are being kept, to that extent, in the dark about the Whorfian measurements that will be made on everybody at the end of the experiment. Experience has shown that it is not difficult to get the cooperation of the pretested subjects on this vital matter...precisely because their cooperation is so vital to the outcome of any before-and-after experiment. In general, subjects wish their efforts to be unwasted.

A final kind of disturbance which a wise experimenter will make an effort to control is what might be called the "ongoing world effect". Controlling it requires that an estimate be made of "what would have happened to our subjects had the experiment not been run". Every human experiment takes place at a particular time of year, at a particular historical moment, in a particular culture, in a particular matrix of surrounding events, and during a particular developmental season in the lives of the people from whom its subjects have been drawn. To measure what is happening to a group of persons who did not receive any experimental treatment, therefore, but who are similar in every other way to the experimentals, is just as crucial to a treatment-assessing experiment of this kind as to measure what happened to those who did. To control for this ongoing-world effect is, in fact, very simple. All we need is a fourth group, called the "controls", in addition to the three we have already planned. In size the control group need be no greater than any one of the experimental groups, but no smaller either; for a random half of the control-group, too, will be administered the pretest.

We have already explained why the partition of the total group of subjects into the four quadrants of this design should also be done at random. But the actual allocation of the four random quarters of the subjects to their four experimental destinies should not, in fact, be done until after pretesting has been concluded. This will prevent other potentially disturbing interactions from forming. It is quite conceivable, for example, that subjects who knew before the pretest that they were going to be controls would not perform as well on it as they would have done had they known that they were to be experimentals. Similarly, subjects assigned to the Loglan group might also do less well on a subsequently-administered pretest, but for a different reason. To eliminate all such possible influences, then, of the subjects' prior knowledge about their alloted roles in the experiment--influences caused, in the main, by the subjects' genial intention to "cooperate" with what they infer to be the "desires" of the experimenter--the pretest should be administered to a random half of all subjects before any of them are allocated to the various cells of the design.

What will the controls be doing while the experimentals are in their Summer Workshops? Anything except studying languages, or, for that matter, such other disciplines as philosophy, anthropology, or logic which might be thought to have quasi-Whorfian effects. Extra-cultural summer traveling, too, should probably be excluded as too likely to leave its own pattern of Whorfian traces. As for the rest, the controls could be free to spend their summers as they chose, reporting back at summer's end for their share of post-testing along with the experimentals.

Certain motivational problems are likely to be created by a design of this kind...for the controls, perhaps especially. Also, perhaps, but in weaker measure, for the non-Loglan experimentals. Inaction toward some valued end, or reduced levels of action toward it--such as the non-Loglan experimentals might well imagine they were being required to deliver--are intrinsically unpleasant for the human organism. How shall we compensate our subjects for this unpleasant cost of scientific rigor? (For that is precisely what it buys us.) There is one answer that seems to hold some promise of restoring a sense of fairness to the proceedings; and that is to make available to both controls and non-Loglan experimentals the same experience that the Loglan experimentals had--complete with curiosity-satisfying measurements--but during the following summer.3

Willy-nilly Loglan and the Whorf hypothesis will have become the public focus of this scientific enterprise. One would hope that publicity would be minimized until the results were in hand; but it could not be kept at zero. Speculation about its purposes would spread; and no one involved in it could be blamed for hoping for a positive outcome...never mind the inhuman logic that will be telling the investigators that they shouldn't do so...or rather, that they shouldn't let their doing so stack the cards in favor of that outcome.

7.8 The Experimental Treatment

Little space in this work need be devoted to describing how modern intensive language-teaching works. The "live-it, eat-it, sleep-it", total-immersion method of learning a modern language is well-known. We hear of the preparation of spies and diplomats through it, and of its use in the training of business and military people for overseas assignments. By the time the Whorfian experiments come along, immersion-teaching will almost certainly have been applied to teaching Loglan, too. We need only discuss here the scientific problems that may be generated by the effort to apply such a method concurrently to three such different languages as, say, Chinese, French and Loglan. The very largeness of the differences among these three languages is, of course, what makes them attractive as concurrent treatments in a Whorfian experiment. But the character of those differences may lead to some knotty problems, too, for the experimenter.

For example, a serious problem for the experimenter would be created by any large difficulty-differences among the three languages for any particular nationality of subjects. For example, for three groups of English-speaking learners to reach substantially the same level of competence in Chinese, French and Loglan, the three languages may well require substantially different learning times. Very likely a group of monolingual Americans undertaking to learn Chinese will require more time than a similar group learning French will; and while we do not yet have quantitative comparisons between these two natural languages and Loglan, we do know that Loglan learning rates are very high, so the Loglan group is likely to require even less time to reach any given level of competence than the French one will. Suppose the intervals required are in the ratios of 4-2-1. How do we conduct concurrent Summer Workshops on that basis?

Two solutions present themselves. One, we introduce calculated amounts of "idle time" into the Loglan and the French group's workshop experiences, and run the Chinese workshop flat out. Two, we start the Chinese group's workshop first, then the French group's, then the Loglan group's; and, having obtained good estimates of the total times required beforehand, we arrange for them all to reach approximately the same competence level at approximately the same time. In other words, we handicap the three groups as if we were running a fair footrace among runners of very different talents.

The "idle time" idea seems flawed to begin with. No time is really idle, and if the Loglan group's fascination with their new language reaches the same pitch we have observed in other Loglan learners, there will be no practical way of keeping their minds really idle about Loglan once they have been exposed to it. The different-starting-times proposal seems to be the only practical way of running a fair learning-race.

How much total time do we have? The typical language workshop lasts about two months...let us say 8 weeks. If, at the ratios given, the Chinese group were to use all of those weeks, then the French workshop would have to start at the beginning of Week 5, and the Loglan group at the beginning of Week 7. But this is absurd. Two weeks is not long enough to get comfortable with any language, not even with a charmer like Loglan. We will have disabled our fountain of Whorfian effects before we even connect the plumbing to it.

Clearly, it is better that we think in months. If 4-2-1 really are the expected multipliers, let's let the Chinese workshop run 8 months, the French one 4, and the Loglan one 2. But, if we did this, we would, in a certain sense, be penalizing the Loglan group by making the total time it spends with Loglan inversely proportional to the swiftness with which it can be learned! This sounds unfair. Definitely uncompetitive, and in the market-place, we wouldn't allow it. But isn't that stifled competition just what we want in science? Isn't it the learner's competence in da's second language that both Whorfian models predict will produce Whorfian effects? And not the time da takes to reach it? Besides, with such a schedule for the administration of our three treatments, we are stacking the cards against Loglan, not for it. That's not quite true, of course. We are actually handicapping the three languages as fairly as we can. So if, even after handicapping, the Whorfian effects of learning Loglan are still greater than those of learning Chinese, and the Chinese ones are in turn greater than the French effects, and if the magnitudes of these effects are not correlated with the duration of the respective exposure intervals--as this particular set of multipliers would have arranged for them not to be--then there must be some other explanation. It could be Whorf's.

So again there is every scientific advantage in making it tough for our hypothesis to emerge unscathed. To put the matter more precisely, we wish to arrange matters so that a chance, or inadvertent, confirmation of our model when it is in fact not true is as unlikely as it can be. A scientific "truth", i.e., a confirmed hypothesis, is apparently valuable in proportion to the unliklihood of its having been confirmed by chance.

I do not for a moment believe that the learning intervals required for English-speaking subjects to reach speaking competence in Chinese, French and Loglan will be as disparate as the sample multipliers 4-2-1 suggest. But if they are, we should live with it. Provided they are solidly based on preliminary studies of learning times, such multipliers will provide fair handicaps. To do the opposite, to allow the three groups to race without handicaps through equal learning times, would almost certainly be to settle for lower levels of terminal competence from the French- and Chinese-learners than the Loglan-learners had achieved. This would stack the cards in favor of an outcome that favored Whorf and Loglan, and so doom our experiment to triviality.

7.9 Measuring the Host Culture Effect

Even an experiment as carefully controlled as the one just described could not be regarded as being decisive...not, that is, if its outcome was corroborative. It could of course refute the Whorf hypothesis, and would do so if no discernible Whorfian effects emerged, or did not arrange themselves in the predicted pattern: largest among the Loglan experimentals, second largest among the Chinese-learning subjects, and smallest among the French-learning subjects. In this (happy) event, we could still not regard the Whorfian thesis as having been established...even temporarily. For a very large and potentially powerful factor would remain unexamined; and that is what might be called the "host effect": the causal influence of the host language and culture on the effects we had succeeded in capturing. In our case, of course, that would be the North American English one, assuming that only monolingual English-speaking American and Canadian subjects had been enlisted for the first experiment.

The truth is that there is a large unmeasured main-effect in such nature-embedded experimental designs, as well as unknown interactions with other factors. Who knows how North American English-speaking life prepared, biased, or even seduced our subjects into displaying pseudo-Whorfian effects under just such circumstances? What complex pattern of so-called "Hawthorne effects" might we have created?

But we need not speculate about these mysteries. The next step in any series of experiments meant to be decisive is, scientifically, a standard one: we replicate. We repeat our experiment in "another laboratory", or we get others to do so; that is, we repeat our four-group experiment in a setting in which such transient, uncontrollable causes as might have produced false positive results in North America are bound to be different ones. We will get the most information from our replications if we repeat our experiment in China and France.

Then, if we observe the same corroborating pattern--the same prominence of the Loglan-induced effects over the two natural-language ones, and the same rank order of effects in France as we observed in North America (replacing French, of course, with its cousin, English, to make the prediction), and a predictably different pattern of results in China (and much, much more that can be predicted for what would then become a twelve-group design) then we may, perhaps, claim to have run a decisive experiment.4

7.10 Some Possible Consequences of the Several Outcomes

Suppose that despite all such vigorous experimental efforts to refute it, the Sapir-Whorf hypothesis survives. Suppose an explanatory model emerges from these efforts that does successfully relate Whorfian linguistic causes to Whorfian psychological effects, that is, to a measurable expansion of the mind. Suppose as well that the effects we discover are as wholly salubrious as the predicted ones were: mind-expanding, thought-facilitating, and yet completely harmless to the persons who experience them. Such a pattern of scientifically induced effects could only be seen as augmenting human life. Probably all who learned about them would want to feel and see them happening in their own and others' lives. Suppose, further, that Loglan were the instrument of this extraordinary discovery, that it actually was the instrument that the experimenters used to turn on Whorfian effects. Wouldn't learning Loglan in the living-room promise to produce approximately the same effects? Wouldn't the entire experimental program, in fact, now be seen as a successful assessment of a proposed new educational experience, one that was available to everyone? It might even be seen as a treatment of a disease we didn't know we had! LLL, the disease of "local language limitation", or UNM of "unnecessarily narrowed minds". A disease, moreover, that had been pandemic on the planet until the studies of Sapir and Whorf began, nearly fifty years ago, to sniff it out? And wouldn't Loglan itself then be seen as the gentle new cure for that ancient human malady? An antidote for the antique provincialism of the human spirit? For what the late American psychiatrist-anthropologist Erik Erikson tellingly called our extraordinary tendency to "pseudo-speciate"? To behave like different species toward one another even though we're not? An antidote for the bigotry with which even "civilized people" tend to view their neighbors in the global village?

A fancy? Not entirely. This is what is very likely to happen given what the journalists will call a "positive" outcome of our Whorfian experiment, that is, a failure to refute this powerful notion. The Whorfian thesis is a powerful one, and it will be seen as powerfully salubrious if it is found to be "correct", that is, if it resists falsification by a well-designed experiment. It is for this positive outcome, perhaps, that we are allowed quietly to hope. As experimentalists, we are not, of course, allowed to be partisan, to unfairly help the fragile truth emerge, or even to try to demonstrate it. But as humans who see the practice of science as a service in the cause of life we are certainly allowed to hope for useful outcomes. This one surely would be.

Given that useful outcome, Loglan, and perhaps other engineered languages, could have a bright future in human affairs...even a utilitarian one. Backed up by such a result, Loglan would probably be seen as ideal in the role of that international auxiliary, for example: the first language to be taught to the world's school children, the one slated to become everybody's second tongue. The international language movement itself, now a century old, would receive a tremendous boost from a successful outcome of a Whorfian experiment. The choice of the best language for use in the international arena would be seen as an important choice again. At the same time, first-languages would continue to be seen as they are today: rich in nuance, history, literature, poetry, ambiguity, and humor; and dying first-languages would perhaps be strengthened everywhere by the appearance of an ally, the international auxiliary.5 But our engineered new second-language would be seen as the mind-expander, the instrument of thought, reason, invention, and exposition...and perhaps also as the medium of intercultural mediation, a culture-spanning bridge to a more tolerant and peaceful world. Loglan, which started life as Whorf's instrument, might well be seen as something very like a suspension bridge, ready to be flung over cultural gorges, its linguistic antecedents (the natural languages) having been the mossy logs on which humans had been crossing local streams for millenia....the instruments of local traffic, surely, but of human provincialism as well.

We would continue to use those stream-crossing logs in the cosy provincial corners of our minds, of course, and love them. Their massiveness, their intricacy, their almost organic wholeness and wanton "over-design", would continue to charm and mystify. But where wide distances were to be spanned, where both accuracy and efficiency in communication were required, we would turn to our new linguistic implements: the engineered languages. Perhaps language and the language arts would begin to be seen in an entirely new light: like symphonies, space-ships and computers, we would see them, finally, as they had always been: the products of human invention if not exactly of human design. We would begin to realize what had apparently been true all along, namely that human languages were artifacts...evolved artifacts, to be sure, but, like culture itself, artifactual: delicate fabrics of local inventions woven over centuries; composed of ancient metaphors and other tiny implements of human understanding, but as surely and intentionally made by human tongues and minds as roads are made by human backs and hands. Human languages, we would perhaps be surprised to learn, are as alterable as city-streets.

This entire tissue of consequences, of course, rides on the supposition--for that is all it is--that the Sapir-Whorf hypothesis will be experimentally confirmed. But it is perhaps just as likely that it will not be, that it will be thoroughly refuted, even that no Whorfian effect will show up to be related to Whorfian or any other kinds of causes. What then? It will repay us to look at this second set of outcomes, too, and just as carefully.

The negative outcome, if I may call it that, does not have so clear a skein of consequences; nor, at first glance, do they seem so useful. Still, some consequences of refuting Whorf seem clear. One of these is that the intellectual rivals of the Sapir-Whorf hypothesis would certainly be fortified even if not established. There are principally two of these: the Kantian view of knowledge, and the Chomskian view of grammar. The 18th century German philosopher Emanuel Kant, you may recall, decided after several decades of reflection that the forms of reason, pure and practical, were innate. So innate, in fact, that a certain rather precious variety of truth could be obtained, as it were, by the human thinker without leaving his armchair; in effect, by consulting da's own mind and not looking at the external world at all. These truths he called "synthetic a priori" truths, and they have mainly to do with reasoning itself. But this innate kind of knowledge also includes some ethical principles and some metaphysical ones, according to Kant, and together they constitute a sort of biologically accumulated and transmitted pool of practical wisdom that can be known to be true about the world before one has even looked at it. This innatist view, as we might call it, of the nature of at least some kinds of human knowledge, and therefore of the human thinking that discovers it, would certainly be revived by the failure of a rigorous experimental test to confirm the Sapir-Whorf hypothesis. At bottom, we could now again believe, thinking apparently runs on its own internal tracks, in ways quite independent of the linguistic forms and other notational devices that we find waiting for us when we enter this world.

The view of grammar associated with the 20th century American linguist Noam Chomsky is a close intellectual relative of this Kantian view, but nevertheless a distinct one. Chomsky has argued vigorously over the past thirty years--over almost the same decades that Loglan has been developing--that all essential linguistic structures are universal, given innately by the very structure of the human brain, and therefore fixed in the human gene pool. As a consequence, given the Chomskian view of things, the real grammar of any language is the biologically determined "deep grammar" it shares with all other human languages. It follows that all the surface variability we observe in natural language is epiphenomenal, making no real difference in how we generate or understand utterances, and therefore contributing to no or only shallow differences in how we think. It is clear that, to the extent that this Chomskian view is correct, and also to the extent that human thought is in fact shaped by the forms of utterance, the surface variability we--and Whorf--think we observe in language can make no, or very little, difference in how we think.

There is little direct, descriptive evidence to support the Chomskian notion of a universal human grammar, at least not one with a scope wide enough to account for any but a tiny proportion of human utterance forms. But the failure of the Whorf hypothesis in an experimental setting, such as the one we have just been discussing, would certainly renew interest in the possibility that, in some new and experimentally sharpened sense, Chomsky might well turn out to be right.

Still, looked at closely, neither of these two rival views of the relationship between language and thought--in Kant's case, of the relationship between experience as a whole and thought--actually contradicts the Sapir-Whorf hypothesis. Certainly they are very different accounts of the same phenomena. But that does not mean that all three of them could not be true. For example, there could be, and probably are, some rules of human information-processing (i.e., "mentation") that have been rooted so deeply by evolution in the human genome--using a computer-science metaphor, we could call them "hard-wired"--that neither linguistic forms nor any other kind of post-natal experience has any influence on their expression. When these hard-wired cognitive rules or mechanisms are fully understood, as I am confident they soon will be, there will then be every justification in calling them "Kantian"; for this is exactly the kind of knowledge that Kant was talking about. So the Kantian model of these matters would be to that extent confirmed. But nothing in such a result would in itself defeat the Whorfian claim that the content of the mature human mind--its beliefs, its sense of possibilities, its convictions about the underlying structure of reality--had not been in some way shaped by the structure of the language or languages in which it has been entertaining those beliefs and arriving at those metaphysical and ontological conclusions. So the Kantian and the Whorfian models of human knowledge are compatible, even though each may be said to limit the possible scope, or the domain of action, of the other. Thus if there are proportionally many genomically-transported Kantian rules, there must be proportionally fewer linguistically-shaped Whorfian ones; and vice versa.

An entirely parallel observation may be made about the Chomskian view of grammar. This model, too, does not contradict the Whorfian one, although it certainly opposes it. For to discover that the scope of universal grammar--that is, the size of the genomically determined portion of all human grammatical arrangements--was proportionally very large over the set of human languages would limit the domain in which Whorfian mechanisms could be at work in language; and again vice versa. Why is this? Because such a discovery about the depth to which genes determine human grammatical behavior would mean that at least some of the notational/structural differences we observe between natural languages were in some sense "unreal", that they were apparent only, and so unlikely to be robust enough to perform the duties of a Whorfian cause. This does not mean, of course, that a superficial difference in linguistic arrangements can have no effects on mentation. But it does mean that such a difference is unlikely to have any lasting effects on human minds. Thus, just as with Kant, the degree to which Chomsky turns out to be right will limit the degree to which Whorf may be right about the effects of linguistic differences on human mental life. But the opposite is also true. If Whorfian effects are large, and experimentation shows them to be so, and the grammatical features of the concerned languages which appear to produce those numerous effects also turn out to be numerous, then it follows that the extent of the universal grammar is probably small. So a positive outcome of the Whorfian experiment will occasion a reassessment of the Chomskian position without, of course, refuting ft.

Thus, while the collapse of the Sapir-Whorf hypothesis during a program of experiments designed to test its claims directly would not give automatic support to these two genetic theories of human knowledge and grammar, respectively, it would corroborate them. Indeed, such a robust but negative result could give many unexpected and yet useful redirections to the group of sciences that deal with human knowledge, culture, and language.

In sum, the Whorfian experiment is very likely to be useful however it turns out. And those who contribute to its running by becoming adepts at the Loglan language, and therefore able to offer themselves as models of its speechways to learners taught in laboratory settings, will make an especially large contribution to that outcome. For while English-speaking teachers of French, German, Spanish and even Chinese exist in satisfactory numbers now in North America, teachers of Loglan do not. Therefore the first item on any agenda to test the Whorf hypothesis with Loglan must be to grow its teachers.

Developing a corps of fluent speaker/teachers of the language will therefore be high on The Institute's agenda for the 1990's. So I may end this chapter where I began it by addressing those who believe that the chance of the Whorf hypothesis's being true is sufficiently high, and the human benefits of finding that out with Loglan are sufficiently great, that we are justified in learning it. So, I confess, do I. Indeed, it is only if sufficient numbers of us do accept this calculated wager that we will ever find out if Loglan is Whorf's instrument...perhaps even whether the Whorf hypothesis is, in that qualified scientific sense, true or not.

The long period of patient engineering is over; the bridge is ready to be used. So let us now learn the language, and teach others to do so...in our own countries first, and then in others. Above all, let us try to become fluent. Only as we become fluent in a sutori language can we expect to feel its effects in the deeper chambers of our minds. Fluency can, of course, only be achieved in the company of other speakers. But this is not a unique requirement. After all, it is only when there is a living, breathing and, above all, visible community of Loglan-speakers that an experimental test of this powerful idea will seem worthwhile to others. Conversely, whenever there is such a community, our decisive experiment will probably not be difficult to arrange.

One of the first successes of the Loglan-speaking community may very well be to bring about that decisive test of the Whorf hypothesis. Even if the experiment fails, however, and the decision goes against this grand idea, the test itself can only further our understanding of that mysterious triad of evolutionary novelties that emerged sometime in the Pleistocene and now seem to dominate human life: human languages, human cultures, and the not quite infinitely varied shapes of human minds. On our understanding of the interplay among these large planetary forces our very survival as a species probably depends.


1 Of course confirmation contributes something, and early ones, quite a lot. But the contribution even they make is more psychological (heartening, encouraging) than evidentiary. That is to say, the probability of an hypothesis being true does not rise much with each new confirmation; and once the number of confirmations is already large, the addition of a new one is often trivial. Observing the 597th black crow does not do much to allay the suspicion that there may be an albino somewhere...that the model "All crows are black" may be an over-simple one. It is the discovery of the albino crow that adds materially to the fund of knowledge. For whatever the probability was that all were black, it has now plummeted to zero. A great change in the body of evidence has been wrought.

2 Comparativists may apply the word 'restrict' to features absent from one culture or time and present in another; but in my opinion they do so improperly. The Amazonian Indian trotting through the rain forest with darts and blow-gun in search of food-animals in the canopy is in no meaningful sense restricted by da's non-possession of a shotgun. Similarly, the Roman arithmetician suffered no restriction from his zerolessness. That de's work was awkward, slow and clumsy by later standards--once zero had been invented (in India, as it happens, in about 600 A.D., from which it later spread throughout the world...although the Mayans, we've learned more recently, had invented a zero that didn't spread about five centuries earlier)--is a fact about the facilitating effect of zero, not about the "non-facilitation of non-zero", whatever that curious expression might be taken to mean.

3 Such a gesture would not be entirely wasted scientifically. For from this second lot of Loglan experimentals, subjects could be tracked for the on-going longitudinal studies to be completed later; see Section 7.6. They would be as useful for such studies as the experimentals that had been created by the "true experiment" of the previous year.

4 Professor John B. Carroll, Whorf's editor, and in fact the assembler of Whorf's posthumously published (1956) collection of essays, Language, Thought and Reality, as well as a distinguished psycholinguist in his own right, suggests that a more piecemeal experimental approach to the Whorfian hypotheses, using Loglan as its instrument, might make a suitably convincing overture to the grand twelve-group design I have just described. What Prof. Carroll has in mind is laboratory tests of "specific hypotheses about the comparative performance of Loglan and non-Loglan speakers on certain specified tasks" (personal communication). I would be glad to prepare a catalog of such "specific hypotheses" for any scholar interested in devising such experiments. The main problem, of course, with using existing loglanists in such experimental designs is matching them on all other relevant characteristics with non-loglanists. Spontaneously self-generated loglanists tend to be rather distinctive people and difficult to match. But even with this problem somehow solved, there remains the confinement of such designs to the exploration of enablement effects. Liberation effects, it seems to me, must be addressed globally: that is, by actually creating loglanists in the laboratory, and measuring the influence of such treatment in before-and-after experimental designs.

5 Dying first-languages are, in the current nature of things, almost always those associated with subdominant cultures, and with peoples now struggling for identity, and even existence, in the shadow of the world's dominant peoples. The languages associated with those dominant cultures--the once-imperial languages such as English, Russian, French, Dutch, Spanish, Japanese, Portuguese, German (some of which are still "imperial" in a now-commercial sense) are the languages which must now be learned by native speakers of dying languages; and it is always in the interests of the dominant cultures to diminish the importance of the local languages in every civil way they can...using the undoubted usefulness of education conducted in the dominant language, for example, as one of their justifications. The appearance of a world auxiliary language--the language that would eventually be everybody's second tongue--would change that asymmetric relationship immediately and completely. The instrument of communication between dominant and subdominant cultures would then be intermediate to them both; and, as people in the international language movement have long pointed out, there would then be no reason, neither commercial nor educational, to hasten the disappearance of the tiny languages of the Earth. Instead, there would be every reason to preserve them as repositories of the marvelous, the precious and the old.

(Return to the Table of Contents)