49, 1994, 1-40. Coherence in ..... representations can be divided into those that purport to relate ... The subclass of particulars can be further divided into four ...
'. Oller, J. W., Jr. (1994). Coherence in comprehension,communication and
language acquisition: transcending the post modern condition. English Teaching, in English Teaching: Korea Association of Teachers of English 49 49, (3), 161-168. 1994, 1-40.
Coherence in Comprehension, Communication, and Language Acquisition: Transcending the Postmodern Condition John W. Oller, Jr. University of New Mexico Traditionally, coherence has been construed roughly as meaning at the level of discourse. The theoretical, experimental, and practical literature of language teaching has shown for a long time, in a vague way, that coherence makes a difference to the comprehension of discourse and thus to the acquisition of language and literacy (just as Krashen, 1980, 1984, 1985, 1989, 1994 has been arguing). Now, owing to theoretical advances of a logico-mathematical kind, it is possible to say more explicitly what coherence is, but first it is necessary to answer the modern and postmodern skepticism about whether we can know any facts of the material world. Such skepticism, it is demonstrated, is inherently self-contradictory and leaves itself no room to deliver a case against valid knowledge of the material world. Further, it has been proved (Oller, 1993b and in press a-b) that absolutely any kind of meaning can only be determined through true representations of particular facts known through the experience of some observer. So representations that determine particular facts must be of the true narrative kind. Such cases differ from fictions, errors, and lies and form the only basis for determining the meaning of any true general representations, or in fact, any representations whatever. With these theoretical advances in hand, evidence is examined showing that coherence makes a difference in discourse processing in all of its aspects-iconically with respect to perception-based representations, indexically with respect to grammatical structures (more explicitly defined as deictic relations determined by intentional actions), and symbolically with respect to abstracted and completely general meanings. Certain experimental hypotheses are formed and tested that could not have been formulated without the elaborated
theory. Finally, implications for the teaching of languages and literacy are considered. PRELIMINARY CONSIDERATIONS
Before launching into the substance of the argument, a few preliminary remarks must be made about modern and postmodern skepticism. In the next section we discuss the nature of mathematical proofs (Peirce, 1883, 1897, 1898; Ketner, 1992; Ketner & Putnam, 1992) as contrasted with the scientific method of demonstration as detailed, for instance, by Popper (1959) or Platt (1964). The point is to differentiate a thoroughly consistent, comprehensive, and mathematically precise theory of representations from the sort of willy-nilly approach that looks first to this empirical result, then to that, and affords no adequate basis for distinguishing which experiments ought to be given the greater or lesser weight. The point here is to show that any number of theories built up like patchwork quilts from the ever-growing and never-to-becompleted series of ever-so-many experiments with this, that, or the other syllabus, method, or what-have-you must lead with algebraic certainty to the confused state of mind in the profession of language teaching described by Kumaravadivelu (1994) as "the postmethod condition" (playing off the popular notion of "postmodernism"). After the fundamental premises of postmodern skepticism have been reviewed and shown to be untenable, a number of consistent proofs of the logico-mathematical kind will be summarized showing the only basis there can possibly be for any meaning or coherence in discursive representations of any kind. Following the summaries of these proofs various experimental studies are also summarized showing that some of the subtlest and most unexpected hypotheses generated by the theory are consistent with empirical evidences. Implications for teaching languages, literacy, and for education in general are presented along the way and certain conclusions are
reached pertaining to the moral obligations of theory-builders, researchers, and practitioners. MODERN AND POSTMODERN SKEPTICISM REVIEWED
Penny cook (1990), Widdowson (1990), Stern (1992), Wilkins (1994) and others "counsel us"-or so says Kumaravadivelu (1994)-"against the very concept of method itself" (p. 27). Or as Wilkins (1994) argues in regard to "syllabus design", we have now reached the stage where there are so "many factors" (cf. Spolsky, 1989), that there just cannot be "any definite answers" to questions about the ideal syllabus for teaching a language-because not even "authenticity" can be assured. On the contrary, authentic discourse in the target language is, according to Wilkins, unachievable in the classroom. He supposes that whatever might have been "authentic discourse" before we brought it to the classroom becomes no longer so once it is made the object of study. Thus, he says, we are reduced to suspend disbelief such that we must really just "pretend" we are working with authentic discourse in the classroom. He supposes in the final analysis that applying the former epithet of "nonsensical" is no longer a condemnation of any classroom procedure as it was once construed to be-this, owing to the fact that all discourse in the classroom is really consigned to the realm of a more or less implausible fiction in any case. Thus, by his reckoning, all classroom discourse is more or less nonsensical. Similarly, Pennycook (1994) urges that different interlocutors understand the term "discourse" itself in such different ways that they may not even be on the same subject. Indeed the subject, he argues, is not really out-there to be known as the "pragmatists" have supposed, but rather, he contends "meaning is produced by discursive regimes that are related to a 'will ~o know' in diverse areas of social life" (p. 133). Thus, Pennycook recommends "a postmodern stance" and that we should "abandon teleologies" (p. 134). He does not seem to realize how nihilistic a teleology he himself 3
proposes nor how all-knowing and unquestionable he represents it to be. Nevertheless he asserts absolutely that "we cannot know ourselves or the world around us in any objective fashion" and in the same sentence he asserts that we should "nevertheless" tackle a more difficult problem: we should "ask how it is that we come to think as we do" (p. 134). Actually, even Pennycook would have to admit that history is relevant in working at the grand question he poses, and the fact is that the sort of skepticism Pennycook espouses can be traced back to Bertrand Russell and before him to David Hume. Russell (1951) ended up arguing that there is no logical basis for supposing that there is a real world of experience that is common to different observers. That argument has been thrashed through in other contexts (cf. Oller 1989) however, and Einstein (1944), Dewey (1916), and Peirce (1865, 1906) showed that the position taken by modern and postmodern skeptics is inherently self-refuting. Peirce gave a complete argument against skepticism though the same view in essence was expressed independently (though much less comprehensively) by Einstein. Dewey's argument with Russell, of course, was grounded exclusively in the views already enunciated by Peirce. Peirce's argument, in a nutshell, was that all justifications of complete doubtfulness about know ledge of the external world must be formulated by a person who stakes off some ground somewhere within that self-same world. Indeed, it turns out that to deny absolutely that any valid knowledge whatsoever can exist concerning the material world, the skeptic must effectively possess all possible knowledge in order to be able to show that no possible know ledge can exist. The position taken, then, is inherently and irremediably self-contradictory. The entire position is mounted on the patently absurd premise that we can be absolutely certain that nothing can be known for certain. For this reason, Einstein (1944) pointed out that if any skeptics doubting perceptual knowledge of the world, as Hume and
4
[Numbering is incorrect. There is no page 6. All pages are here.]
Russell both claimed to do, were merely consistent (i.e., morally responsible in their actions), it would be necessary for them not only to rule out all propositions claiming knowledge of the world, but it would also be necessary to rule out absolutely any proposition whatever. Einstein's case was simple. If one doubts the sort of perceptual impressions that are the ones most closely associated with material facts, how can one not doubt even more strongly all other kinds of representations whose very existence depends on those of the perceptual kind? Hence, if perceptual evidences are ruled out of bounds, all other evidences are a fortiori also excluded, and thinking itself is abolished (Einstein, 1944, in Oller, 1989, p. 23). Peirce's development of the case (1865, in Fisch, et aI., 1982, p. 257) was a little more rigorous. He began by defining representations, things, and forms. Now all that we know absolutely and directly are representations. Clearly, we cannot reasonably doubt the existence of representations because any such doubt would itself constitute an undoubtable evidence of the supposedly doubtful existence. Hence, we cannot say, I doubt that any representation exists, because the argument would itself be a forcible example of what we claim to doubt the existence of. An argument of that form would by its very nature contradict itself. But what about the abstracted forms of things? Can their existence reasonably be doubted? And what about the things themselves? Cannot their existence legitimately be questioned? It turns out that we know things and forms only through representations. The form of a thing is what would be left if all of its constituative material could be removed. For instance, consider a material car as an example of a thing. Now, remove all of its physical material but leave behind its virtual image. The virtual image of the car is the form of the thing. Thus, things, forms., and representations can be defined sharply, but do they all three really exist? Peirce argued that since we know of material things and their forms
only by virtue of representations of them, if the representations that we have access to are consistent with the things and forms represented, admitting the existence of those things and forms cannot possibly introduce any inconsistency into our theory. Carrying his argument forward just one step from where he left it, to deny the existence of things and forms would be to deny the consistency of the representations and would introduce an inconsistency. Hence, both forms and things must be admitted into our thinking in just the manner and for the same reason that we must admit the existence of representations. In this way, pragmaticism (as distinct from the pragmatism of William James and John Dewey) utterly annihilates any ground on which Pennycook's position, or any of the postmodern alternatives like it, might pretend to be erected. Will Rogers offered a bit of cowboy pragmatism that is relevant here. He said, "It ain't what you don't know that hurts you. It's what you know that ain't so. That's what hurts you." The hurtful part in Pennycook's philosophy (and in all similar varieties of postmodern skepticism) is in the claim to know for sure, i.e., with objective certainty, just what no one is supposed to be able to know at all. More explicitly, as soon as Penny cook (1994) asserts with absolute certainty that "we cannot know ourselves or the world around us in any objective fashion" (p. 134), he cannot escape proposing to abolish teleologies (all of them) apparently without realizing that his own position is absolutely annihilated by its own claim that there can be no determinate meaning in discourse. Thus, to propose that we should instead seek to know how it comes about that "we think as we do" amounts to undertaking what his absolutely general teleology asserts to be an even more impossible task (as if that were possible) than finding out what and where we are in the material world. His methods and conclusions are all patterned after the same error and have their origin in the skepticism that has
5
7
spawned postmodernism. It is also the source for what Kumaravadivelu (1994) calls "the postmethod condition" in language teaching. But all such skepticisms, however much they may tout their humility, at the same time pretend an absolute know ledge that shows the falseness of their pretended humility. If their quest could be salvaged by sheer weight of numbers, they might prevail, because there are evidently a great many people who are willing to cast aside all of their knowledge of the material world and to believe that questions of fact really are meaningless to start with. Yet, like Pennycook, they prove that they do not really believe their own skepticism fully (not consistently) because they come around eventually to be-moaning the notion that all issues connected with such questions can be resolved by sociopolitical force, or by what Pennycook (1994, p. 133) calls the "will to know" which magically manifests itself in the "discursive regimes" that "produce" meaning "in diverse areas of social life". It is essential to say "magically" here because the skeptics such as Pennycook claim not to know anything objective at all-e.g., they claim to know nothing of the world of guns, dollars, and votes. But if the meaning in a discourse were in some way linked to the kind of clout in social life that comes from the barrel of a gun, from the purchasing power of a Federal Reserve Note, or from the political weight of a vote, it could only be precisely the sort of thing dependent on knowledge of objective facts that Pennycook wishes to exclude from the discussion as teleological. The upshot, as is rigorously demonstrable, is that any known facts whatever are excluded from any method of reasoning such as Pennycook employs. So, any such skepticism cannot bear up under its own weight. Having dispensed with all such thinking, then, including its so-called "postmodern" variants, it becomes at. least reasonable to consider the possibility of developing a well-grounded, factually based theory of representational
8
systems along Peircean lines. However, before launching into the theory of true narrative representations (cf. Oller, 1993b) and their peculiar logical properties, it may be useful, especially to those of us trained in the humanities, to consider the profound difference between empirical tests of an experimental kind and general proofs of a mathematical kind. COMMENTS ON THE METHOD TO BE APPLIED
Logically it is impossible to prove a universal proposition (regardless whether it be positive or negative) on the basis of any number of experiments. For example, suppose someone were to assert that it is impossible to translate a text into a new language without changing the target culture and that, as a result, all translations must be demonstrably imperfect (Givan, 1989 expresses this idea). Now the italicized part of the preceding sentence is a universal proposition. But how could such a proposition be proved by empirical studies of actual translations? All possible translations would have to be examined and each one of them would have to be shown, by some method, to be imperfect or to produce some radical change in the target culture. Yet this is patently impossible. All possible translations cannot be examined because there are inexhaustibly many of them. Even if we could examine all those that now exist and that have ever existed in the past, there would still be all those translations that are possible but that have not yet been constructed which would remain to be examined. Moreover, the same sort of difficulty exists for any conceivable universal proposition because all of them purport to apply not merely to some finite number of cases, but to all possible cases of a given kind. Therefore, owing to their very definition-i.e., in the Peircean sense of mathematically precise logic-it is utterly impossible to provide an empirical proof, that is, by accum~lating results from any number of experiments, for absolutely any universal proposition. However, a universal proposition can often be disproved in the empirical way. For instance, Xiao & Oller 9
'.
(1994) disproved Givan's universal claim about translation by showing that two cross-translations of English and Chinese narratives were not in fact measurably distinct by a rigorous measurement technique that examined every single word of the translations and originals. This result held up in the face of the profound differences across the languages and cultures of the U.S.A. and mainland China. But if just one relatively perfect (non-culture changing) translation is possible, then, Givan's universal claim is false. So, based on the empirical evidence, his claim is shown to be false. Besides, the claim must be false on purely theoretical grounds (that is by virtue of strict logical reasoning) because the contrary can be proved to be true by perfectly generallogico-mathematical reasoning. As Peirce proved, any symbol is perfectly translatable in a strict logical sense by virtue of its very definition. Further, since any icon or index can be symbolized, any representation whatever must be perfectly translatable. Now the latter proposition could not be proved empirically no matter how many translations might be examined, but the converse (that all translations are measurably imperfect) which is also a universal could be disproved empirically by just one counter-example. In respect to the logical impossibility of proving a universal by empirical methods, Popper (1959) and Platt (1964) were correct in arguing that experimental science advances only by disproofs. That is, if science is limited to empirical demonstrations of the experimental kind, it is necessarily limited by strict logic to advance only by disproofs. But, if we ask what is disproved in any given advance of science, we discover that the only kind of proposition that any experiment can possibly disprove conclusively is the sort that purports to be a universal. But where can such a universal originate if not from a theory purporting to cover all the possible cases of some kind? The universal proposition cannot arise from any particular experiment which purports to test it because the universal proposition must pre-exist the
experiment. Indeed, Peirce showed for all possible representations that universals can only arise from general predicates and are never themselves created by particular facts. The universal may be manifested and determined in its meaning by particular facts, but those particulars (as Immanuel Kant argued, followed by Peirce and Einstein; cf. Oller, 1989 for references) could not even be known as particulars of a given kind if the universal categories to which they pertain were not known in advance. That is, a particular whose kind were not found among the universe of kinds that could be known would be a kind of particular that could not be known at all. It could not be experienced. So, universal categories cannot be produced in the mind of any observer by any particular fact or facts. Therefore, to achieve positive advances, the sciences must resort to mathematical (strictly logical) proofs. Only proofs of the mathematical kind can show that a given universal proposition must be true. This is the method upon which the next part of this paper depends. Afterward, however, a number of empirical demonstrations will be summarized showing that certain interesting positions contrary to those outlined here can be ruled out.
10
11
TRANSCENDING POSTMODERNISM
As an example of a true narrative representation consider the following: On July 19, 1994 at about 12:20 pm, I boarded Korean Air Lines Flight 17 in Los Angeles bound for Seoul and I arrived there at about 4:28 pm Korean time on July 20, 1994. This case is a true representation because it conforms to the articulately balanced structure as summed up in Figure 1. The facts of the reported experience conform to the representation so that the representation is true of the facts. The case is a narrative representation because it purports to be about the particular experience of one or more particular observers. It is a true narrative representation because the particular facts report~d conform to the particular
representation of them and this is possible because, evidently, the observer is both intelligent enough to make the observations in question and is competent enough, and sufficiently sincere and energetic, to report them correctly in intelligible English.
2 Linking
/
~ 3 Representation
1 Facts
Figure 1.1. The logical structure of the true narrative case.
Now it turns out that representations of the true narrative kind have certain logical peculiarities (Oller, 1993b, and in press a-b). In fact, they are more perfect in three ways than any other conceivable representations. First, relative to all other kinds, true narrative representations are more perfectly determined with respect to what they mean. Second, they are the only kind of representations among all that are possible which are connected in a relatively well-determined way to the space-time continuum. Third, owing to the fact that only true narrative representations are well-determined with respect to their meaning, only they can provide a determined basis for any general meaning whatever. But it would be superfluous to reproduce here the proofs of the foregoing theorems that have appeared elsewhere
12
(Oller, 1993b, in press b), so only a summary will be given. The method (following Peirce's approach of "exact logic" or mathematicized logic) consists of first setting certain absolute limits on the universe of all possible representations. Such limits can be found in the definition of a representation-i.e., what a representation must be to qualify as such. For instance, it must be noticeable in some way or other. One that could not be noticed in any way at all would be one that could not serve to represent even itself. Moreover, in order to perform its central function a representation must relate to things other than itself. If it related only to itself, it could not be noticed in anyway by anyone and could not represent anything at all, not even itself. It would as Krashen observed (personal communication in 1994) "implode". Or, putting the requirement in its most abstract form, no representation can be completely meaningless. Hence, a limit or boundary can be established so as to exhaustively define the universe of possible representations as shown in Figure 2. The edge of the circle marks a limit or boundary. Anything that is altogether unnoticeable, altogether unrelated to anything other than itself, anything altogether meaningless passes beyond the absolute limits of the universe of representations.
The Universe of Representations
Figure 1.2. Lgically limiting the universe of all possible representations.
13
The next step, then, is to examine all possible representations in comparison to the earlier defined true narrative case. Now the whole universe of possible representations can be divided into those that purport to relate to particular facts (as do true narrative cases) and those that purport to relate in general to all possible facts of a given kind. As for pluralities of particular facts, they are always particulars by definition regardless of the number associated with the plurality. Hence, nothing falls between the two parts of the whole universe of possible representations as defined: on the one hand there are representations that purport to relate to particulars and on the other there are those that purport to be general. So, it becomes possible to ask separately for each subcategory whether there are any possible elements in either one that are as well determined, connected, or as generalizable as those of the true narrative kind.
The proofs of all three theorems hinge ultimately on the proof of the first. So we begin with it. First we show that only true narrative cases are relatively perfectly determined with respect to what they mean. To do so we must exhaustively contrast representations of the true narrative kind with all other possible kinds. Now the class of all possible representations can be exhaustively divided into those that purport to be about particulars and those that purport to be general (cf. Figure 3). First we exhaust the subclass of particulars-that is, of representations that purport to be about particular facts. Then, we exhaust the class of generals-i.e., of representations that purport to be about all possible facts of a given kind. The subclass of particulars can be further divided into four non-overlapping classes of representations: (1) True narrative representations and imitators of these consisting of (2) fictions, (3) errors, and (4) lies.
The universe can be perfectly divided into two distinct and non-overlapping parts: (1) those representations purporting to be particular, i.e., about particular facts; and (2) those purporting to be general, i.e., about all facts of a given kind.
Particulars I Generals
Figure 1.4. The exhaustive subclasses of Figure 1.3. The exhaustive subclasses of representations purporting to be about particular facts and all possible facts.
14
representations purporting to be about particulars. It comes out that we can exhaustively divide the subclass of. particulars into true narrative cases plus tbree imperfect imitations of them -namely, fictions, errors, and lies (Figure 4). Each of these classes of imitators differs in an
15
explicitly definable way from the true narrative case. In fact, they differ by degrees of degeneracy from the true narrative case- fictions are degenerate in one degree, errors in two, and lies in three. A more degenerate case than a lie would merely devolve into utter nonsense which cannot purport to be about anything in particular, so there are no additional kinds of representations purporting to be about particulars beyond the relatively perfect true narrative cases and the three degenerate kinds found respectively in fictions, errors, and lies. The degeneracy of any fiction has to do with the fact that the particular facts to which it purports to relate are not material facts. For instance, the raft built by Tom Sawyer and Huckleberry Finn in the story written by Samuel Clemens exists only as a form described in the fictional story. It has no material weight or substance to it and it never displaced any water anywhere on the Mississippi River. The river and Clemens are real facts of the space-time continuum with material characteristics, but the raft, Tom, and Huck, are fictions that merely resemble real cases in their form (but not in their substance). Thus any fictional representation lacks the material facts of Figure 1 above and is degenerate in just that one degree relative to the more perfect true narrative case. Also, fictions are dependent exclusively on the representational tokens giving rise to them (e.g., the words used by Samuel Clemens in our example; element 3 in Figure 1) to determine their meaning. The particular facts that fictions purport to represent cannot be appealed to in order to make their meaning any more determinate than it already is based on the representational tokens used in the story (or whatever genre of fiction). Therefore, all fictions differ in this respect from absolutely any true narrative case. In the true case, the particular material facts may be referred to independently in order to further determin~ the meaning of the representation. For instance, if someone points out a mongoose to someone who has never seen one before, the particular fact
(specifically, the mongoose pointed out) can be examined to discover more about (i.e., to determine more explicitly) what a mongoose is like. But if someone should merely ask an alien from outerspace to imagine a mongoose, the alien could not benefit at all from looking at any material mongoose because none would be present other than whatever the alien might be able to conjure up based on its fictional representation. With errors the case is more extreme. Every error (viewed as a representation purporting to be about particular facts) is degenerate in two degrees relative to the true narrative case. On the one hand it is degenerate with respect to the material facts which are not as they were at first represented to be (e.g., he thought he was about to take a drink of Dr. Pepper but it turned out to be Diet Coke) and on the other hand, they are degenerate with respect to the representation of those facts (e.g., he should have supposed it was Diet Coke rather than Dr. Pepper). Hence, errors are even less determinate than fictions. Errors are degenerate in elements 1 and 3 of Figure 1. Deliberate deceptions, by contrast, are degenerate in all three of the elements of the true narrative case shown in Figure 1. Lies are degenerate with respect to (1) the material facts and with respect to (2) the correspondence between them and (3) their representation (e.g., Bill says Sam committed the murder, but it was Bill himself who did it). Now, anything more indeterminate than a lie (which resembles a true narrative case less than either a fiction or an error) is so indeterminate as to become nonsensical. That is, any representation that purported to be about particular facts but that did not qualify as either fictional, erroneous, or a lie, would have to be nonsense and could not represent anything in particular at all. Moreover, the three degeneracies form a strict hierarchy relative to the true nar.rative case. For instance, a fictional representation does not purport to represent a real, existing state of affairs, but rather a possible state of affairs. A
16
17
fiction merely describes the way particular facts might have been, or might become. All hypotheses, expectations about the future, and speculations about the past, except to the extent that they happen to coincide with true narrative cases, partake of the hypothetical, virtual aspect of fictions. But since a fiction does not purport to be true, rather it only purports to be possible, it is potentially less harmful than an error which is taken for the truth. However, the mistaking of a false representation for a true one is different from a lie. An error is innocently taken to be the truth, but a lie is known to be false and yet is nonetheless represented as if it were true. Clearly, all fictional, erroneous, or deliberately deceptive representations purporting to be about particular facts lack the kind of determinacy with respect to what they mean that is found in every true narrative case. To complete the proof of the determinacy perfection of true narrative cases, it only remains to show that true, false, and indeterminate general representations (i.e., ones that purport to be about all possible facts of a given kind) do not possess it (cf. Figure 5). Now false generals (false abstract propositions, such as washing of the hair causes male pattern baldness) owing to their falseness can be ruled out by definition as less determinate than true narrative cases. Also it can be shown that a false general is like a fictional particular with respect to determinacy. Whatever facts might be examined to determine the meaning of either one would have to be found out by representing those facts in the manner of true narrative cases. Otherwise, for a false general just as was found out for any fiction, only the representation itself (the third element of Figure 1 above) is available to determine meaning. This leaves only true generals (e.g., some representation such as all human beings are mortal) as possible exceptions to the proposition that only true narrative cases are relatively perfectly determined with respect to their meaning. But true generals purport to be about all possible particulars of
a given kind and to determine their meaning requires at least one true narrative case instantiating the kind. Suppose that no such true narrative case can be found to instantiate the kind of facts that the true general purports to be about. Then, the general will be exactly as indeterminate with respect to its meaning as any fiction. Otherwise, the true general (e. g., take as an example the proposition that there are super-heavy particles emitted only rarely by the nucleus of atoms) is not fully determined with respect to what it means. Hence, only true narrative representations are so determined. Returning then to the two other unique properties of the true narrative case, we find that connectedness to the space-time continuum absolutely requires the sort of determinacy only found in the tru.e narrative case and that generalizability of any meaning is similarly dependent (as just explained in the previous paragraph) on the determination of the meaning to be
18
19
The subclass of generals can similarly be divided into three non-overlapping classes of representations: (1) True generals, (2) false generals, and (3) generals of indeterminate meaning.
Generals (1) True, (2) False, and (3) Indetermmate ones
Figure 1.5. The exhaustive subclasses of representations purporting to be about all possible facts of a given kind.
generalized by at least one true narrative case. Thus the case is made. Only true narrative representations have the three pragmatic perfections of relatively perfect determinacy, connectedness, and generalizability. The implications are many and varied (cf. Oller, in press a-b). Among the ones of interest to educators-especially, language teachers, and elementary school teachers responsible for helping children over the literacy hurdle-are certain universal consequences for the comprehension and production of intelligible discourse in general. As argued elsewhere (Oller, in press a), three main kinds of schemata can be posited to undergird the intelligibility of ordinary discourse. The first two of these are well-known in our literature (cf. Kintsch & van Dijk, 1978; Carrell, 1984; Mandler, 1984; Kintsch, 1988; Tudor & Tuffs, 1991) but the third has only recently been proposed to complete the triad (Oller, in press a). If this third kind of schema is taken to be nothing but grammar in the Chomskyan sense of the term, then, it is true that its role in the coherence of ·discourse has been known about for a long time. However, Peircean theory shows how this third level is linked to the other two kinds of schemata and how it is logically essential to the completion of the triad. Thus, the connection between a strict hierarchy of content, formal, and abstract schemata, it would seem, had not been spelled out previously. In any case, based on the perfections of true narrative cases, exactly three kinds of discourse schemata can be shown to exist. First, there are content schemata which are derived in Peircean terms by abductive reasoning from the first element of the triad shown in Figure 1. Formerly, content schemata were defmed somewhat loosely as pertaining to the facts of discursive representations (cf. Carrell, 1984; Tudor & Tuffs, 1991). From the .Peircean perspective we can be more explicit. Content schemata are abstracted from true narrative iconic representations of particular facts. Second, there are formal
schemata that are derived in Peircean terms by inductive reasoning from the second element shown in the triad of Figure 1. According to recent research these schemata have vaguely been associated with the logical structure of the propositional relations contained within a particular discourse (cf. Kintsch & Greene, 1978; Kintsch & Yarborough, 1982; Mandler, 1984; Carrell, 1984; Tudor & Tuffs, 1991). In fact, they have, owing to the inexplicitness of the theories on which they were based, been commingled with abstract schemata. More explicitly, from the Peircean perspective, formal schemata are true narrative indexical representations obtaining between similar facts across a plurality of particular contexts. Third, there are abstract schemata which are derived (according to Peircean theory) from the generalization of iconic and indexical representations. For instance, an icon abstracted from its perceived object can be symbolized in such a manner as to suggest the continuous existence of the perceived object even when it is not being perceived. An index abstracted from its logical object (the one or ones to which it points; each of which must be known iconically in some way) can be symbolized in such a manner as to suggest a possible plurality of logical objects of the kind from which it has been abstracted. A symbol abstracted from any particular context, however, comes to represent all the possible logical objects to which it might be applied. Hence, it also comes to a logical limit (as Peirce showed 1883) of perfect generality. From the theory of schemata just outlined in the previous paragraph certain general principles can be stated from which many more detailed hypotheses follow. In general, the theory shows why true narratives are bound to be superior to other kinds of representations as the basis for language and literacy curricula. Plausible fictions are next best, but should be converted into true n.arratives as actualized in dramatizations by the teachers and students in classroom contexts. That is, in the dramatization, the discursive
20
21
representations will at least achieve the structure of true narrative representations with respect to the actions performed in the drama. On the whole, materials that conform to the schemata that learners have already internalized or that learners have a better basis for constructing will work better (other things being equal) than materials that violate learner expectations in fundamental ways. By and large, representations that are presented in a richer variety of modalities (i.e., sensory, motor, and abstract) will give better results than ones presented in just a single modality and/or representational form. For instance, as shown in Methods That Work (especially the 1993 edition), dramatization can help learners achieve the necessary connections of facts with representations as recommended by J. J. Asher. In fact, his total physical response method involves physically carrying out commands, e.g., stand up, sit down, etc. This kind of actualization of language into the facts of .one's own experience can be enriched through episodic organization along the lines of Kalivoda, et ai. (1971). They recommend a sequence of actions coordinated with either commands or descriptive statements (eventually both will be developed) arranged like a narrative. For example, the teacher might illustrate the procedures for making French onion soup: I look in the refrigerator. I take out an onion. I chop the onion. My eyes begin to water, so I wipe away the tears. And so forth. Variations on this theme are uncountable per the recommendations of Kalivoda, et ai. (1971). Rodrigues & White (1983) recommended incorporating the former kinds of preparation into actual experiences. John Rassias uses all of the foregoing but extends his dramatizations to the full cultural richness of the target language and culture up to and including delivery of the learner to the target culture for a survival experience. Oller, Sr.(1963-1965) showed how all of the foregoing could be incorporated into a richly developed and
fully articulated curriculum carrying students form the earliest level of language development all the way to relatively mature native competence. Other global approaches involved in bilingual education, content-based instruction, FLES, immersion, and sheltered subject-matter are also supported in a general way by the theory. That is to say, in general, approaches which involve a higher density and greater accessibility of true narrative representations, ceteris paribus, will have to work better than others. Unfortunately, many traditional approaches, some enlightened communicative, notional-functional approaches, and many of the recommended "postmethod" eclecticisms (per Pennycook, 1994; and Kumaravadivelu, 1994) fall short of the required grounding in true narrative cases. Similarly, certain approaches to the development of literacy in children and adults are favored above others. For instance, as shown by Walker et aI., (1992) activity-based and story-based approaches to the initialization of literacy in young children, and especially in children of minority language backgrounds, produce results markedly superior to those attained by methods that are less well grounded in true narrative representations. With respect to clinical applications, again, the theory of true narrative cases supports certain kinds of intervention procedures above others (e.g., cf. Damico & Damico, 1993). All of the foregoing facts are accounted for, albeit vaguely, by the theory of true narrative cases and the derived theory of schemata. However, as noted at the outset, the fact that meaning and coherence enhance the processing of discourse and language acquisition is not news to anyone and does not, in itself, contribute any significant advance to our understanding of how to go about teaching languages, literacy, or specialized subject-matter. To test the theory experimentally mQre explicit hypotheses will be required relative to particular research designs. In the section to follow, results are summarized from several experiments beginning with the
22
23
iconic level of discursive representations and working up to symbolic representations of the linguistic kind. Peircean theory enables the prediction that still photographs containing written symbols will be easier to interpret, other things being equal, than ones not containing any such symbols. Similarly, the theory leads to the prediction that pictures with an index included, e.g., a man pointing to a satellite dish in the foreground of the picture, will be easier to interpret, other things being equal, than pictures without any indices to guide the interpreter. Pictures containing only iconic information about undetermined states of affairs, other things being equal, will be more difficult to understand. More importantly, the theory also leads to the prediction that adding color to photographs supported by symbols or indices will provide no significant benefit over the same pictures in black and white, while adding color to photographs dependent more or less entirely on icons for their interpretation will produce a significant advantage. However, the gain contributed by color to the interpretation of photographs which are dependent mainly on iconically supplied information will be relatively small by comparison to the difference between classes of photographs attributable to the presence of determining symbols (or indices) . Indeed, all of these hypotheses were borne out. Skipping over details (see Giardetti & Oller, in press), in a test devised by Giardetti (1992), agreement on the thematic interpretation of photographs involving symbols was 45 % greater than on photographs involving merely iconic information (and the contrast was significant at p < .001), and this difference was more than five times as large as the difference (7.8 %'. a contrast significant at p