|
|
|
|
This paper originally appeared in Applied Linguistics 16.3, 1995, 273-306. |
|
There is no normal pagination on a web page and if, as in this case, numbers indicative
of the page numbers in its original publication have not been interpolated into the text,
scholarly reference to passages in it must be made in another way. In this case,
paragraph numbers have been editorially assigned instead for this purpose: these
are bracketed and placed flush right, just above the paragraph.
|
|
[1]
[2]
PEIRCEAN THEORY
The foundation for the theory of schemata contained here is the semiotic theory of relations (originally the `logic of relatives') as developed by Charles S. Peirce (1883, 1897, 1898). Of course, every theory of discourse or cognition purports to explain some aspect of how meanings are represented and understood, but Peirce's unique brand of logic is superior to its potential competitors (both ancient and modern) in three respects. First, it brings a higher degree of mathematical exactitude to the subject and achieves a more rigorous consistency (cf. Ketner 1992, Houser 1986, and Nagel 1959). Second, Peirce's system is radically more comprehensive than prior systems (again see the references just cited) by bringing the materially existing objects, persons, events and other genuine objects of discursive relations into the theory and by extending the theory logically to all possible similar cases (Oller and Kennedy in press). In this second respect, Peirce's approach anticipated and answered certain weaknesses of Chomskyan linguistics as detailed by Reichling (1961), Uhlenbeck (1963, 1967), Rommetveit (1968, 1979), Oller, Sales, and Harrington (1969), Carswell and Rommetveit (1971), Makkai (1972), and Jakobson (1980). Even certain critical weaknesses of government and binding can be removed as shown by Oller and Kennedy (in press) by employing Peirce's logic of relatives. Third, Peirce's system is superior in its mathematical elegance and simplicity (cf. Putnam 1992). [3]
For a long time, it has been widely argued (cf. Leech 1969, Rommetveit 1968, 1979, van Dijk 1981, Johnson-Laird 1983) that the material contexts of ordinary experience must somehow be taken into consideration by the theories of linguistics and its cognate disciplines. Indeed, psycholinguists (e.g., Rommetveit 1968, 1979, Johnson-Laird 1983) and cognitive scientists (Kintsch 1988, Mandler 1992) have been moving more and more toward accounting for the relations between discursive representations and the facts of the material world. Yet, as Oller and Kennedy (in press) have shown with respect to Chomsky's government and binding theory (1982, 1988), Langacker's functional grammar (1987) and Givón's pragmatics (1984, 1989, 1990), the existential material world along with its persons, objects, etc., has generally been left on the outside. This is also largely true for theories of language teaching and acquisition in spite of the fact that it has been popular for about three decades to speak of communicative, social, and pragmatic aspects of these processes. For instance, the popular communicative and notional/functional approaches recommended in recent years by Widdowson (1990), Wilkins (1994), and others working in similar paradigms have notably come up short of incorporating the particular material contexts of the experience of real individual persons in a world of genuine risks (cf. Valdman 1992, Walz 1989). [4]
In fact, it still remains true today (see Oller and Richard-Amato, 1983) that the notional-functional vignettes, the sociocultural excerpts of discourse, or the illustrative conversations, activities, and games of the typical language classrooms of the 1990s often fail to connect fully with the genuine experience of real persons with actual objectives and at-risk plans for trying to achieve them. The vignettes of sociocultural experience as applied in language teaching often fail to connect with the real world for lack of any past or future. The conversational exchanges are often cut loose from any moorings that might connect them with the material world. Characters seem to appear from nowhere without a past into a present where they move forward briefly toward no future at all. While applied linguists everywhere, it seems, are writing about teaching languages in contexts (e.g., Walz 1989), about the need for sociocultural relevance and sensitivity (Wong-Fillmore 1989), and about genuineness and authenticity (Spolsky 1989, Valdman 1992), no adequate theory of how language is actually connected with experience has yet been articulated. Fortunately (and somewhat surprisingly) the means for such a development to be realized has existed since about 1883 when the logic of relatives was first introduced to modern thought by Peirce (cf. Hartshorne and Weiss 1931, pp. 195-209; also Ketner 1992). While regarded as an esoteric branch of logic by some, in fact, it generalized the binary system of George Boole from the dyadic, to the triadic, and to all higher polyadic relations-i.e., to all possible forms of predication regardless of the number of real, fictitious, or virtual arguments they might involve (Ketner 1992, Putnam 1992). Peirce (1897) described his new system as `nothing but formal logic generalized to the very tip-top' (in Hartshorne and Weiss 1933, p. 299). [5]
Yet, the Peircean theory which forms the starting point for this discussion has only been superficially noticed even by sympathetic semioticians (Jakobson 1980). For instance, Eco (1990, p. 214) barely devotes a single page to the logic of relatives and the subject is not even mentioned by Sebeok (1991). It has been noticed even less by linguists (though Leech 1969 is an exception), not to mention practitioners in applied linguistics. For instance, in many seminal works on applied linguistics, psychology, and psycholinguistics pertaining to schema theory, pragmatics, and discourse processing (e.g., van Dijk 1981, Mandler 1984, 1992, Flammer and Kintsch 1982, Kintsch, Miller, and Polson 1984, Kintsch 1988) there is hardly a mention of Peirce much less a discussion of his logic of relatives. Shapiro (1983) omits any notice of this watermark of Peircean thought as does Givón (1984, 1989, 1990). Although Artigal (1992, 1993) is an exceptional applied linguist who uses certain Peircean concepts to great advantage (especially the index and symbol), his application relies on only two of the main Peircean categories (omitting explicit consideration of the icon) and, for all of its utility, comes up short of the full power of the logic of relatives. What Peirce accomplished there was to show how indices linking general predicates to one or more particular arguments (whether existing or fictional) can be mathematically described-but not without all three of his major categories (icon, index, and symbol). [6]
Oller and Kennedy (in press) have shown why the linking of indices to actual objects and states of affairs in genuine experience is essential to determining the data of discourse so that they can be used to test grammatical theories. Rommetveit (1979) anticipated our findings in part in the following remarks:
[7]
In fact, Oller and Kennedy proved that if the indexing of actual logical arguments is not carried out all the way to the actual material persons, places, things, and states of affairs as known in ordinary experience (through icons, indices, and symbols), theories of grammar will necessarily remain untestable for lack of sufficiently determinate data. It had been proved earlier using a Peircean approach (Oller 1993b), that only true narrative representations can provide a satisfactory basis for the determination of any meaning whatever. Therefore, in spite of themes common to the vast literature on discourse processing and cognitive sicences that reverberate throughout the discussion to follow, the theory presented here nonetheless represents a profound departure from prior theories of text-linguistics, discourse-processing, language acquisition, and related topics. While the approach used here fully accepts the need to understand the pragmatic, social, and material contexts of language acquisition and use, it constitutes, nevertheless, a substantial theoretical departure from the methods employed up till now. Although many theoreticians have grappled with some of the same problems addressed here, and though many of them have alluded to Peirce's thinking in one way or another, as will be made evident in what follows, they have not applied the full power of Peirce's logic of relatives. On the contrary, Peirce's logic has been almost completely overlooked by applied linguists (with the exception of Artigal noted above) and experimentalists in the discursive sciences and, as a result, where Peirce's thinking is alluded to in our literature, it is often treated superficially and even incorrectly.[8]
For instance, Peirce's thought is often identified with that of Charles Morris as is done by Bentele (1985, p. 162) who merely follows Eco (1976). Or Peirce's theory is mixed together with de Saussure's notion of semiology as is done by Barthes (1967) and by Connel and Mills (1985, p. 33ff). As a result, the vagueness and other weaknesses of theories put forward by de Saussure and Morris are assumed equivalent to Peirce's semiotics which was in fact both vastly more comprehensive and much more exact than either of the others. Space would fail to document these claims fully, but it may be noted that de Beaugrande (1991) only mentions Peirce in his preface and does not in any way distinguish Peirce's thought from that of de Saussure and other more modern semioticians. Fine (1994), in the fifty-first volume of Advances in Discourse Processes (edited by Freedle), does not even mention semiotics nor Peirce much less the logic of relatives. While cognitive scientists such as Kintsch and van Dijk (1978), van Dijk (1981), Kintsch (1988), Mandler (1984, 1988, and 1992) all see the importance of relating discursive kinds of representations to the material contexts of experience, none has yet provided a comprehensive theory of how this occurs. Moreover, among these authors, only van Dijk (1981) mentions Peircean semiotics and only in a footnote. [9]
It is the purpose of this paper, therefore, to show the theoretical relevance of extending and applying Peircean theory in order to obtain certain experimental results hardly anticipated by prior theories. To accomplish this objective, it is useful to reflect on the definition the term schema. As construed in the exponentiating literature on the subject, a schema is a way of looking at states of affairs in experience including literature, films, and vicarious experiences. From Piaget (1947, 1952) forward, the term schema has come to describe the kind of organization that enables its user to handle certain kinds of tasks more efficiently than would otherwise be possible (see Rumelhart 1975, Schank 1975, Schank and Abelson 1977, Kintsch and van Dijk 1978, Mandler 1984). In her landmark work on the subject, Mandler (1984) discusses scripts, scenes, scenarios, plans, and frames of reference. All these bear resemblances to the abstracted commonalities of memories and fictions (including fantasies, dreams, etc.) and are nearly equated by Mandler. A commonly used example (Schank 1975) is a restaurant where we expect to order something to eat, and to pay for it. More recently a framework has been developed for differentiating content schemata-those which pertain to the particular facts of a situation-from formal schemata-those which pertain to the underlying structure of one or more similar situations. Kintsch and Yarborough (1982) and Carrell (1984) also used the term rhetorical schemata as a cover for various kinds. Carrell says, for example, that formal schemata can be defined as 'background knowledge of the rhetorical structures of different types of texts' (1984, p. 192). [10]
This paper expands the theory of schemata to incorporate fully general and abstract ones in addition to the traditionally recognized kinds. The aim is logical comprehensiveness. It is demonstrated logically that content, formal, and abstract schemata exhaust all the kinds that can possibly exist-not to say that other arrangements cannot be constructed, but that any other arrangement would have to include the subject matter of these three and would not add substantively to them. It is not claimed that the distinct kinds are always independent but that they do produce differing results under certain experiemental conditions, and that they are arranged as a strict logical hierarchy. Content schemata are less general (and thus less powerful) than are formal schemata which in turn are less general (and less powerful) than abstract schemata. The first step is to lay down some definitions. The arguments are essentially mathematical deriving deductively from definitions which cannot (it is argued) be reasonably doubted, but they can be subjected to empirical testing and in that vein are regarded more as hypotheses than may seem to be the case in the way they are stated. After the necessary theoretical distinctions are introduced, several experimental studies will show the empirical power of the theory.[11]
We begin with the true narrative case (cf. Oller 1993b). This case, as shown in Figure 1, was earlier called pragmatic mapping (Oller 1975, 1990) or valid abduction (cf. Peirce in Hartshorne and Weiss 1931-1935).
Figure 1 gives a diagrammatic view of the essential elements. It has been proved that all other classes of representations are either dependent upon true narrative cases or are degenerate precisely to the extent that they depart from or cannot be related to them (Oller 1993b, and in press). Every true narrative representation has each of the three elements shown in Figure 1. For example, if the author should say I attended the Baltimore meeting of the American Association of Applied Linguistics in the spring of 1994, this representation would qualify as a true narrative case. Let it be stipulated as a matter of record that this statement is a true narrative case as contrasted with a fiction, an erroneous report, or a lie. It is a trivial example, but it is all the more satisfactory for that reason to make clear the definition at hand. More momentous representations, that Caesar crossed the Rubicon, that Lincoln was assassinated, or what-have-you, add nothing new. But, if the reader does not like the proposed example, let any other example of a true narrative case be selected and it will be found to have the following properties. [12]
Each such case as narrated is linked to three kinds of evidence showing its material facts. First, there are perceptual evidences in the experience of one or more observers. Second, there are memories of actions underway at the time of the narrated events. Third, there are meanings of the narrative representation that could be paraphrased or translated into another language. Of course, the sights, sounds, and tactile impressions of the facts in any true narrative case are iconic as are the memories called up later in thinking about them (e.g., buying the ticket, boarding the plane, traveling to Baltimore, meeting and talking with the people there, etc.). That is, as Peirce made clear, the representations of the sensory kind are generally copies of the shapes, textures, densities, weights, colors, movements, resistances, chemical qualities, and so on of the logical objects of the true narrative representation. [13]
The facts are determined first (see the left side of Figure 1) by the sensory experience of those involved in the reported events. Secondarily (as shown in the upper middle part of the diagram), the facts are determined by being pointed out by the representations that single them out for attention-indexing them, as might be done with a pointing finger. Then, in a tertiary manner, as is shown in the right hand side of the diagram, the particular meanings of the linguistic forms in question, are determined by the material facts (that is the events, persons, relations, etc.) in that context then and there. Taken out of that context, the same terms become indeterminate. For instance, a phrase such as the person doing the talking or a pronoun such as I could refer to absolutely any speaker in any context. The word person could be used to refer any conceivable one whether real, imagined, or non-existent. And so on. That is to say, a representation without any particular facts to determine its meaning, by definition must be less determinate (exactly to the degree that it is separated from determinate facts) than a true narrative case whose meaning is determined by the very material facts that it singles out for attention.[14]
As a result of their structural peculiarities, true narrative representations have three logical perfections not found in any other representations (cf. Oller 1993b). True narrative cases are more determinate with respect to their meaning. They are the only kind that are connected to the material world of space and time, and they provide the only basis for achieving meaningful generalizations. Moreover, each of these propositions is susceptible of mathematical proof (cf. Oller 1993b, and in press), but the power of the theory and a thumbnail sketch of the proofs can be seen by examining degenerate representations and by comparing them to true narrative cases. There are just three main categories of degenerate cases that purport to represent particular facts. They are respectively, fictions, errors, and lies. Each is examined in turn. [15]
The underlying logical structure of all fictions is shown in Figure 2.
In a fictional case, e.g., someone imagines grabbing the tractor of an eighteen-wheeler by the front axle and lifting it into the air over his or her head. The facts as represented are fictional. They are virtual as is shown in Figure 2 by the broken-line of the circle and the outline printing of the word 'Facts'. What is missing is the material that would give the tractor weight, spatial dimensions, and that would, presumably, prevent its being lifted overhead. Still, the representation is properly linked with the purported 'facts' (that is, with the virtual facts), so the other two elements in the figure are just as they would be in a true narrative case. However, in a true narrative case, the meaning of the representation can be indefinitely refined by referring to the facts at hand-e.g., with respect to the example given above, the hotel where the author stayed in Baltimore might be examined, the flight numbers of the planes traveled on could be found out, the room number where the author stayed could be obtained, and the whole constellation of facts could be described in finer and finer detail by any suitably positioned observer operating with appropriate methods. But, with the imagined eighteen-wheeler, a fictional case, the possibility of consulting the material facts to refine the determination of the meaning is entirely absent. In any fictional representation, the interpreter is completely dependent on the representation itself to determine whatever facts it may purport to represent. If the color of the eighteen wheeler is not specified in the representation, it cannot be determined by examining the purported facts. Nor can we determine the dimensions of the thing, whether or not it has a snub-nosed cab, a sloped hood, a wind-spoiler, and so on. The tractor might be a toy or a drawing. In short, what is intended in any fictional case is indeterminate. [16]
The logical structure of fictions includes many useful propositional forms such as hypothetical inferences, predictions about future events, expectancies, contrary-to-fact conditionals (e.g., If X were the case, but it is not, then, Y would have to also be the case, but it also is not), and a host of others. Fictions are useful because in them only one degree of the triad is virtualized and some of that part is determined by the representation. However, the determination of the facts in the case of fictions comes exclusively from the representation. Therefore, for any language learner who does not already understand the representation, fictions provide a poor basis for finding out meanings and thus grammatical structure. For this reason, approaches I have recommended (Oller 1993a) are grounded in cases that have the true narrative structure (e.g., Asher's TPR, Rassias's dramatic approach, Oller Sr.'s pragmatic approach, etc.). Fictions can be made sufficiently similar to true narrative cases if they are instantiated with action, drama, or other sensory-motor representations to determine the meanings of the linguistic representations at hand in the way that true narrative cases would. But in fictions the missing element is the perceptual part (the iconic aspect of the material facts) and its spatio-temporal connectedness to the material world. For this reason writers of fiction must often be reminded by their teachers (Swain 1980, Peck 1980), editors, and would-be readers, to supply the sights, sounds, textures, tastes and smells so that the reader at least will have the illusion of being dragged through the events of the story.[17]
While any fiction involves a necessary incompleteness in one of its parts, an error involves a second degree of degeneracy that incorporates an inconsistency in two of its parts. With errors (which include the special cases of illusions, hallucinations, and the like), as shown in Figure 3, it is the matching up of the facts with the representation that has gone wrong.
As a result, the facts that ought to be represented are not and the representations given (or taken for true) are also not as they should be. A driver thinks he sees a ground-squirrel by the side of the road, but it is a piece of cardboard from a nearby construction site turned just so that it accidentally looks like a squirrel sitting up on his haunches. The facts are other than they are represented to be and when corrected the representation must be other than the one at first settled upon as true. But, the linking of the purported facts with the representation taken to be true is understood to be correct. Hence, an error is even less suitable than a fictional case to use as a basis for language teaching or acquisition. No doubt it is for this reason that error correction rarely results in great advances (as Krashen and Terrell 1983, among others, have argued). [18]
A third degree of degeneracy is found in the case of a deliberate deception (Figure 4).
Here the inconsistency is not only known in advance, but it planned into the representation deliberately with the intent to deceive. A lie is therefore degenerate in all three of its parts. As a result, here, the indeterminacy reaches a limit relative to all those cases that resemble a true narrative case sufficiently to be mistaken for one. The facts (as might be perceived or imagined iconically) are not as the representation at hand purports to make them out to be. The representation (as a symbolic description) does not truly correspond to the facts that do obtain. Further, the linking (the indexical part of the act of representing) has been deliberately corrupted by the producer. The serpent says Eve will not die, but before long she does. Now a lie, or any representational form that is degenerate in the manner of a lie, e.g., the pretense of asking permission to go somewhere from someone who has no right or desire to regulate our going there (cf. Brumfit and Johnson 1979) is a poor basis for language teaching. The only thing that is less determinate than a lie is nonsense. With nonsense there may be a resemblance to the surface form of a representation, but the rest of the structure of the true narrative case is missing. But, any kind of representation whose meaning remains inaccessible to the learner might as well be nonsense. [19]
One consequence of all the foregoing is that true narrative cases, or well-instantiated fictions (e.g., involving drama, role-play and the like) provide a greatly superior basis for language instruction. Just as Rommetveit (1979) argued years ago, in constructing any true narrative case, the bodily position in the space-time universe cannot be dispensed with. In fact, to the extent that there is any physical (material) existence it can be known ultimately only through the senses and the icons that they present to consciousness. Here we find a theoretical justification for Kintsch's argument (1988) in favor of an essential bottom-up basis for certain aspects of discourse processing. Our being bodily situated (as Dreyfus 1979, p. 62 argues; also McClelland and Rumelhart, 1986) in the material world is what enables the physical bodies and relations around us to be indexed (i.e., deictically pinned down by significant bodily movements and gestures). Only through bodily actions can indices be constructed so as to determine anything in one's present. In such acts of indexing, we also classify, name, and refer to perceived objective situations via symbols (mainly of the linguistic kind) that articulately represent those states of affairs as facts of experience. We interact with the physical world across time exclusively through actions that ultimately constitute true narrative cases (no matter how many fictions, errors and lies we may entertain along the way). This interaction involves an indissoluble linking of representations with the material world. It comes out from the theory, then, that fictions, errors, illusions, hallucinations, lies, false generalizations, or generalizations of undetermined meanings, are all dependent on true narrative cases. Not only do these degenerate cases not provide a basis for questioning the existence of true narrative cases (or the existence of a material world), on the contrary, they are prima facie evidence that some true narrative cases (and the material world to which they refer) must exist. Otherwise, we could not tell a fiction from an error or either of these from a lie, or any of them from a true narrative case. [20]
The principal mystery to be accounted for (Einstein 1941, 1944) with respect to true narrative cases is the connectedness of mental representations and the physical world. With respect to this connection, Einstein said in a 1941 radio broadcast that 'everything depends on the degree to which words and word-combinations correspond to the world of impression' (in Einstein 1956, p. 112). He insisted that there is a logically uncrossable gulf that separates these two distinct realms (1944, p. 287)-although he did not deny that two realms are connected, `miraculously' as he supposed, through the bodily actions of intelligent beings. Yet the connection between the realm of representations and that of the material world is dependent on the crossing of Einstein's gulf. Unknowingly, it seems, Einstein agreed in part with Peirce (1898 and elsewhere) who argued that continuity is not only a major mystery of physics but also the central problem of mathematics, and that both of these are tied to the fundamental mysteries of intelligence. With respect to particulars (that is, the unrepeated events of space and time-the hic et nunc of every person's individual experience-this body interacting with others in this place, here and now) we literally sense (or think we do) the continuity across time. And yet, it is easy to prove that in ordinary experience, the most distinctive characteristic is constant change. Permanence of objects, and thus, continuity itself remains a mystery (see comments on Peirce's mathematical theory of continuity by Ketner and Putnam 1992 and Putnam 1992). [21]
It seems that the present is transformed into the past exactly as the future arrives, but just what is it that enables us to make this connection? It cannot be the sheer physical sameness of the past and future or else we would not have the constant sense of motion and change that marks our present experience. But, on the other hand, how come it is that the new experience arising every waking moment of our lives is not altogether surprising and uninterpretable? It turns out that the present is partly determined by the past, and in its turn partly determines the future. Yet the determinacy is not complete or else the future would have to be precisely the same as the past. What is known about the present is a function in large measure of what has been experienced in the past (the true narrative representations already constructed), what we perceive to be going on now (true narrative cases presently under construction), and what we expect about the future (possible true narrative cases yet to be determined).1 Hence the first level of connectedness is perceptual. It can be diagrammed as shown in Figure 5.
[22]
Past, present, and future come together through representations as the intelligent organism rides through time actually straddling Einstein's gulf. The mind at least is chiefly occupied with representations while our physical body and our feet are planted in the material world. The realm of material facts meets the realm of representations precisely at the point where the present slides into the past. At the point where the present meets the future, the material world is (relative to our perceptions of it) not yet materialized. Therefore, as shown in Figure 5, the material world as known to us stops right where the present meets the future.
What lies on the other side of that point is known only through expectations which are mere representations of what may lie ahead (e.g., the road beyond the horizon or the step not yet taken). Those expectations are very real, as representations, but they remain fictions unless they become linked with material facts emerging into the present.[23]
Yet, if our world were totally determined only by percepts of objects immediately before us, there would be little hope of linking the present objects, events, qualities, etc., of any particular perceived fact with other facts like it that may have occurred at some point distant in time and/or space. How can it be known that a certain object is the same one encountered on one or more previous occasions? How can it be known that the object has had a continued existence in the interim while it was not being perceived? Since an object not presented to the senses, cannot be perceived, its existence must be inferred if it is to be known of while it is not being perceived. Following Piaget (1947, 1952) a whole paradigm of research, mainly with infants and young children, has flourished investigating mainly these kinds of inferences from perception (cf. Spelke, et al. 1994). Peirce subsumed this kind under the term abduction which involves the abstraction and generalization of an icon (i.e., of a perceptual representation).[24]
Abduction occurs at just the point where a particular fact, e.g., the Lord Baltimore Hotel, is linked with a distinct representation. This linkage involves all three of the elements of the true narrative case. It also involves crossing Einstein's gulf as shown in Figure 6.
At its most basic and elementary level, as it pertains to bodily objects extended in time and space, an abductive inference merely supposes the material existence in a consistent form of its logical object. It is an inference to the temporal continuity of the bodily object. However, the abductive inference enables a higher sort of generality and a higher degree of continuity than could be justified by perception operating in the here and now all by itself. [25]
But abduction alone will not enable the knowledge that an object perceived on a particular occasion is the same one that was perceived on some other occasion. To connect such separate and individual abductive inferences and to know them as pertaining to the same logical object, induction is required. Now, where abduction involves the abstraction of an iconic representation (a perceptual one) from a particular logical object in experience, induction involves abstraction of an indexical representation from one or more discontinuous perceptual cases appearing at different points in time. For instance, having once been to the Lord Baltimore, suppose someone were to go there again. The place will be recognized as the same hotel visited previously by virtue of indexing both occasions as pertaining to the same logical object. Yet the linking of the two distinct occasions will require an inductive inference over and above the separate abductive inferences which derive from the separate perceptions of the Lord Baltimore Hotel. The difference as shown in Figure 7 is that induction involves a linking of the past and present so as to form expectations about the future.
[26]
But, more is known about the logical objects of experience than either abduction or induction alone can reveal. For instance, if the Lord Baltimore really is a hotel, it must be a lodging place for travelers. This involves a different kind of inference identified as deduction. Here, reason works on the basis of an abstracted symbol which applies not only to the case in hand, but to all possible cases. As Peirce showed, this sort of inference takes us across all spatio-temporal barriers and must have its basis in the non-temporal realm of abstract meaning (Figure 8).
[27]
THREE KINDS OF SCHEMATAWith the foregoing as background, it is evident that there must be three kinds of schemata as shown in Figure 9.
The first two kinds are well-known in the applied linguistics literature, but the third is required by the theory even though it is not generally recognized in the literature, or at least not as related to the other two. First, as the theory shows (and as is well known, cf. Carrell 1984) there must be content schemata based on abductive judgments about particular facts and states of affairs. These schemata are concerned with particular arrangements of things (i.e., with facts) in the material world as known through perceptions. These schemata are the perceived relationships that obtain in a particular context [28]
of experience. They are structural to the extent that they involve complexes of particular things, qualities, events, etc., in particular relationships, e.g., a familiar person will appear such that his or her eyes remain, generally, in the same relation to his or her nose from one occasion to the next, and so forth for other bodily parts. [29]
Second, the theory requires that there must be formal schemata, and again, this is no surprise, as these are well documented in our literature. They are the result mainly of inductive connections established across distinct states of affairs that are indexed as being similar in some respect. For instance, a hotel lobby may look physically quite different from one hotel to another, and yet, there will be formal similarities that enable an observer to identify various hotel lobbies and to distinguish them from restaurants and so forth. The similarities of the indexed facts judged by induction are dependent upon structures and arrangements abstracted to some degree from the particular facts of any given context. Such formal structures remain relatively invariant in spite of the fact that the perceptual surrounds in which they are found may differ radically from one occasion to the next.[30]
Abstract schemata must constitute a third class and unlike the other two this class of schemata has not been recognized previously in schema theory as a distinct category (unless grammars themselves are taken to be schemata). Abstract schemata carry the inductive integration to the completely general (abstract, non-material, non-syntacticized) level of pure symbols (in Peirce's sense of the term `symbol'). Such a schema is necessary, for instance, if we are to draw inferences from representations that are independent of any particular case or any finite number of actual cases in the material world. For instance, if hotels are businesses that aim to make a profit they must generally charge more for their services than those services cost the owners.2 Thus, deductive inferences give us a great deal of information about all possible hotels that could not be acquired by merely examining or auditing the records of however many individual cases we might gain access to by whatever methods might be applied. Deductive inferences are not derived from the particulars of any given case, but are, to the extent that they are grounded in correct definitions of symbols, implicit in all possible cases to which the symbol might validly be applied. Thus, with deduction, a higher level of integration is achieved beyond the inductive level. Here the conceptualization reaches such a degree of integration and completeness that if the definition of the symbol is sound, the inferences drawn from it must be equally valid. Except for the definitions upon which they are based, deductions, unlike abductions and inductions, are relatively impervious to errors. [31]
The three kinds of inference and hence of schemata are, respectively: (1) only positive and accidental (neither probable nor improbable) at the abductive level; (2) bivalent (positive and negative) and with some degree of probability at the inductive level; and (3) completely abstract, general, and logically necessary at the deductive level. Thus, content schemata are owed mainly to the accidents of history. Where we were born, what culture or society we grew up in, etc. They just involve what is (or what appears to be). They are determined mainly by the positive accidents of history as known through the senses. They are for this reason mainly spatio-temporal judgments accounted for in the phenomenal present. Formal schemata involve a higher degree of determinacy. They are probabilistic. That is, they enable judgments about proportions of some range of facts observed in the past as contrasted with what can be expected in the future. They relate what is (and what is not) to what is likely (or unlikely) to be. As a result, induction is concerned mainly with existing states of affairs as contrasted with whatever states do not obtain. It is scientific thought as applied to experience. It is dualistic as concerning what is and what is not. Abstract schemata, by contrast, concern everything that is contained within the meaning or definition of a symbol (including propositions, arguments, and discourses). They take all that possibly could be (as known through the symbols used) and relate it to whatever must be (provided the symbols are used truly). As a result, the abstract level of the symbol (as Peirce showed) reaches from outside of time and space into the material world and yet is itself neither temporal nor spatial in its compass. It comes nearer to our ideas of eternity, infinity, continuity, and universality than to anything known through our senses in the material world. It involves what Peirce called a 'thirdness' beyond the duality of opposing forces clashing in the material realm.[32]
Hence in relative degrees of theoretical power, the connectedness afforded by deduction, is higher than any afforded by induction which in turn is higher than any afforded only by abduction. As a result some interesting hypotheses about language acquisition and use, and especially about discourse processing can be formulated. The theory predicts, for instance, that formal schemata will, other things being equal, have greater power than content schemata in facilitating discourse comprehension and hence communication, language acquisition, etc. Also, other things being equal, abstract schemata involving symbols and their definitions, will be more helpful still. In fact, as Peirce proved, they are at a limit of generality and abstractness.
[33]
EXPERIMENTAL APPLICATIONS
All of the foregoing expands upon the episode hypothesis and what was earlier called the expectancy hypothesis (Oller 1983, Richard-Amato 1988). In a vague way the theory proposed, along with a number of other theories, predicts the results of numerous experiments pertaining to the comprehension and/or acquisition of relatively well-organized representational forms as contrasted with less well-organized ones (e.g., Oller and Obrecht 1969, Oller 1975, Chihara, et al. 1977, Thorndyke 1977, Kintsch, Mandel, and Kozminsky 1977, Kintsch and Greene 1978, Kintsch and van Dijk 1978, Kintsch 1982, Kintsch and Yarborough 1982, Levin 1982, Mandler 1984, and so on). But there are many theories differing in important ways from the Peircean perspective under consideration here that can predict the fact that better organized discourse is generally easier to produce, comprehend, recall, summarize, and otherwise benefit from than less organized discourse. Therefore, in order to test the theory proposed here, it is desirable to derive hypotheses from it which have maximum surprise-value relative to competing theories and yet which can subsequently be put to suitable experimental tests which will sharply distinguish the proposed theory from its possible competitors. Beginning from the bottom (iconic level) and working upward (to the symbolic), some surprisingly subtle predictions can be made about the comprehension of icons. These hypotheses could not even be formulated within any system that did not rely on the triadic distinctions proved necessary in Peirce's logic of relatives and illustrated above. [34]
Giardetti (1992) studied still photographs appearing in National Geographic between 1974 and 1989 (see details in Giardetti and Oller in press) with 123 students at the University of New Mexico as his subjects. They were asked to categorize each of 30 photos (with captions removed) into one of six proto-typical categories in a multiple-choice format. Subjects were given a brief orientation with previously selected exemplars of photos judged by two different panels of experts to fall to the center of each of the broad proto-type categories. Three of the thematic categories pertained to commerce (production, transportation, and communication) and three pertained to human themes (emotion, thought, and physical skill). [35]
According to the schema theory described above, it was predicted that photographs without visible indices (e.g., someone in the photo pointing to a particular object of interest) and without any visible symbolic element (e.g., a printed word appearing on something in the picture or some other symbol such as a priest's robe) to guide the interpreter in categorizing the photo would produce less agreement among interpreters. According to the theory, photographs without any visible indices or symbols to guide their interpretation would be most difficult to categorize. In these cases, subjects would have to rely on content schemata almost exclusively. However, photos with formal (indexical) elements to help determine their content would be more determinate (as judged by the degree of agreement on the categorization of the photos) than ones without visible indices. However, they would be less so than ones with one or more visible symbolic elements (primitive abstract schemata).[36]
In several of the 30 photos, an object of interest is literally being pointed to, looked at, or touched by someone in the photograph. In these cases, determining what the photo is about (i.e., agreeing on its classification into one of the six thematic categories) is helped by the indexical act of pointing. In various other cases, the content of the photograph is symbolically determined either by some familiar element in the picture itself, such as a priest's robe (he is at the center and foreground of the photograph) or the space shuttle, or by words printed on something in the picture identifying the nature of the object in view (e.g., the word Speak visible on a 'Speak & Spell' keyboard). In these cases, either a symbolic element in a linguistic form or some icon with a symbolic value (the priest's robe) helps determine the thematic content. Each of the 22 photographs having one or more of these formal or abstract schematic elements to aid in identifying its theme was classed as 'transparent'.The remainder of the photos, a smaller set of just 8, were treated as a separate `opaque' subscale. These latter photos involved content for which no iconic vestige of a formal or abstract schema could easily be made out. Nothing in the photo is specifically being pointed to, or if it is, it is not in view of the camera. No especially familiar object is prominently centered in the photo so as to be specifically pointed to by the camera or by anyone in the scene. The activity or scene depicted may itself be indeterminate, e.g., an aerial shot with oil tanks in the foreground looks vaguely out to sea with more than one ship in view at various loading docks. There is no apparent movement in the photo (no moving tractor or stream of cars on a bridge at night as in two other photos where the movement serves as an indexical element) to draw attention. For example, in one of the opaque shots a disassembled model of some sort is spread out on the ground on canvass sheets. In another some women are sitting around on what could be someone's living room floor with some fabrics or skins on their laps and a TV in the background. In yet another, some men in bathing suits are launching a boat into heavy waves and pointing to something off to one side of the photo that is not within view of the camera. [37]
The mean agreement between the 123 subjects and the expert consensus on the categorization of the 'opaque' subscale was .416 as against a mean of .787 for the 'transparent' items. The contrast here between 'opaque' and 'transparent' photos was highly significant by an appropriate analysis of variance (p < .001). What the effect shows is a sharp contrast between iconic representations that are relatively well supported by formal (indexical) and abstract (symbolic, especially linguistic) schemata, and ones that are supported mainly by content schemata.3 Giardetti's experiment shows that the benefit gained from formal and abstract schemata can be measured even at the level of iconic representations. [38]
But there is a still more subtle effect predicted by the theory in hand and observed in the Giardetti data. Another variable examined was the impact of color versus black and white versions of the photographs. For the purposes of the experiment, color was actually subtracted from the photographs by reproducing them in high quality black and white versions. The theory predicts that in cases where indexical or symbolic elements are present to guide the interpretation and categorization of the photos, the contrast between matched color and black and white versions should be nil. But in cases where no formal or symbolic elements are present to guide the interpreter, color should produce a slight gain in agreement and this is exactly what was found. Color had no effect on the photographs with indexical or symbolic elements to help interpreters determine their categorization, but on the `opaque' photos, color contributed to a small but statistically significant gain in agreement (p < .031). [39]
Now color is an iconic aspect of the sensory information represented in the photographs. It clearly falls at the content level with respect to the above discussion of schema theory. The proportion of agreement gained when color was added was .078-that is, the 61 subjects randomly assigned to judge the thematic content of the color versions of the 8 opaque photos agreed with each other and with the expert judges that much more of the time than did the 62 subjects randomly assigned to categorize the same 8 photos in high quality black and white copies. This is, according to Peircean theory, a content (iconic) effect owed to enriching the sensory image itself. However, the theory predicts a much greater gain owing to the effects of formal and abstract schemata than owing to the addition of color to the 'opaque' photos. In fact, the observed gain of .371 attributed to the effects of formal (pointing) and abstract (linguistic and other conceptual) schemata is almost five times greater than the gain achieved by adding color to the relatively `opaque' photos. In fact, pictures that included a printed word or symbol (e.g., the name of a ship-line, NORDANA, the word Speak in the 'Speak & Spell' toy, the words Sportsclub and Darmstadt on a gymnastics coach's shirt, and the words Hepburn, Toronto, Canada, 405 Tonnes that could be made out on a piece of heavy equipment in one of the photos) produced the highest agreement scores on the whole of any subscale (.869). [40]
At any rate, distinct effects of content, formal, and abstract schemata can be discerned at the level of iconic representations conforming to subtle hypotheses derived from the proposed theory. Additionally, there is evidence from Tudor and Tuffs (1991) that comprehension of oral discourse associated with moving pictures (a voice-over with video) also conforms to the mainline predictions of the theory. Of course, in view of the substantial tradition of Peircean thought associated with cinematography (cf. Barthes 1967, Metz 1971, 1974, 1983, Kora_ 1988, Tagg 1988) it should surprise no one that a more elaborate Peircean theory should be relevant to video-tape comprehension. Tudor and Tuffs (1991) studied the comprehension by students of English as a second language of an 8 minute video-tape consisting of an off-air recording by the BBC on the possibility of privatizing roads of Britain. One group of subjects (n @ 41) received 20 minutes of advance preparation concerning the specific facts (i.e., the content) of the video, e.g., names of persons and places, while another group (n @ 33) received the same amount of advance preparation but concerning the formal nature of the problem to be addressed in the video, e.g., the pros and cons of privatization. The first group was identified as the 'content schemata' group since the preparation focused on the specific factual content of the video. The second group was identified as the 'formal schemata' group because their attention was directed to the logical character of the problem, e.g., what consequences would follow from privatization. Yet another group, identified as controls (n @ 34), received no advance preparation at all before seeing the video. After viewing the video all three groups were asked to summarize it, to respond to certain comprehension questions, and a week later to answer the same comprehension questions again. As predicted, the group given the content preparation did better than subjects given no preparation at all (the controls), and the group given the formal preparation did better than the content group. While not all of the contrasts between the formal, content, and control groups were significant, the overall pattern was as predicted. [41]
While Kintsch and others, as noted above, have studied some of the effects of `rhetorical structure' roughly characterized since the 1970s (also see Williams 1993 and Millis and Just 1994), tests of more specific contrasts between content, formal, and abstract schemata require the sort of elaboration of schema theory proposed here. Several relevant studies bearing on the issues at stake appear in Oller and Jonz (1994). A number of them involve two narrative texts (one about Joe going off to college, and another about Nicholas going to visit his uncle in Greece). Each text was converted into two rather different cloze tests. Two of the tests were created by deleting about every 5th or 6th word from each passage. Then, these tests were scrambled into two more cloze tests. The procedure was to divide the two texts roughly into thirds. Then, the first sentence of the last third was placed first, the first sentence of the second third, next, and so on until two new (scrambled) cloze tests were constructed for each passage. The prediction was that this procedure would make the cloze test more difficult. That is, subjects would benefit from the normal arrangement of the narratives in each case and this would show up in higher scores on the normally arranged cloze tests.[42]
Analysis showed that the normal arrangement benefited from the formal schemata of each particular narrative. For instance, when Joe is getting ready to go, his mother makes him do a number of things he prefers not to do. When he finally gets on the train and it pulls out of the station, the reader supposes correctly that Joe must be relieved to be on his way at last. The expectation that Joe will be relieved when he gets going would not necessarily be appropriate to another narrative, yet it is inferentially suggested by the facts of this one. The connections required, however, are formal ones. They involve mainly abductive and inductive generalizations. There are also some deductive aspects to comprehension of the narrative, but the main structural relations that make it a story are abductive and inductive. Expectations are formed about what will follow from what has preceded. This is the essence of a formal schema as defined above and the theory predicts that it must (other things being equal) improve comprehension (and thus cloze scores) on the intact passages as indeed it does.[43]
A more subtle prediction tied to the effect of the formal schemata underlying the two narratives was that as subjects advance in proficiency, they will gain increasing benefit from formal schemata. To test this hypothesis (and the contrast between the normal and scrambled variants), it was desirable to eliminate any effect that might come from prior acquaintance with the content of either text. Therefore, all the subjects were tested (71 beginning, 66 intermediate, and 64 advanced students of EFL at the Osaka YMCA, and 41 native speakers of English at the University of New Mexico) on one passage in its scrambled version and the other (a different passage) in its normal arrangement. This design enabled elimination of any effect of practice or learning owed to working through one text more than once and thus ensured distribution of the effects owed to the scrambling procedure equally over the two passages and over each of the four groups of subjects. This aspect of the design, though crucial, has often been overlooked by researchers attempting to replicate our findings (Jonz and Oller 1994).[44]
The outcomes were as predicted (see Figure 10) and the growth pattern across the four groups proved to be linear (cf. Oller and Jonz 1994, p. 143) with significant contrasts across each of the adjacent levels. That is, intermediates benefit from the formal schemata underlying the two narratives significantly more than beginners, advanced EFL learners benefit more than intermediates, and natives more than advanced EFL students. Therefore, the effects of the formal schemata were as predicted.
[45]
In addition, in the two decades following Chihara's original research in 1976, the basic design was replicated and extended in various applications to distinct groups of non-native speakers and to various groups of native speakers of English. One finding was that Japanese learners found the text about Nicholas going to Athens easier than the text about Joe going off to college. However, Thais, Haitians, and Florida Hispanics (all, like the Japanese subjects, speakers of English as a second language) generally found the Joe text easier or about the same as Nicholas. Since cultural background and experience can be expected to affect content schemata more than formal or abstract ones (the latter pair becoming more general and necessarily more universal), it was supposed that the differential performance on the two texts (Joe vs. Nicholas) of the distinct linguistic and cultural groups might be attributed to such factors. This hypothesis was followed up and confirmed by Chihara, Sakurai, and Oller (1989) and also by Al-Fallay (1994 discussed below). In each case, the latter studies manipulated texts positively or negatively according to experience-based cultural expectations and showed that cloze scores were raised or lowered accordingly. However, it should be noted that in all cases where comparisons could be made, the contrasts owed to content schemata were relatively smaller than those owed to formal schemata. [46]
Following up the work of Chihara, et al. (1977) various other researchers examined the information gained from working through a cloze test over a narrative as contrasted with that gained from working through a scrambled variant of the same text. The reason for examining this contrast is straightforward. As unfortunate as it may be, most language instruction is based not on the sort of formal structures that are found in narratives (or in all of ordinary experience), but on linguistic structures cut loose from nearly all of their material contextual supports. Isolated sentences, or in recent years, isolated sociocultural vignettes are commonly used. If there is a conversation, it is apt to be dropped out of the blue sky with no particular lead-in or follow-up. People may agree to go to the movies but without deciding which one, when, or where they will meet. That is, the typical language lesson is apt to be constituted by isolated, detached, and effectively scrambled elements drawn more or less at random from the kinds of texts and discourses common to ordinary experience. Therefore, demonstrating that language learners get more benefit from studying normally arranged narratives than from studying scrambled variants has immediate and obvious implications for language teaching. Spelling out the most obvious one, if students learn more from materials conforming to a story-line, then language teachers ought to use materials with a well-developed story-line. It turns out that procedural texts (telling how to do something or reporting on a game or activity) and expository reports (explaining something), provided they are grounded in facts, also have the requisite structure of the true narrative case (cf. Walker, Rattanavich, and Oller 1992).[47]
In the series of experiments in question, literate native speakers of English and literate non-natives were studied. In all the cases examined, provided only that the learners were well beyond the beginning stage (per the Chihara findings of Figure 10 above), a dramatic contrast was observed between the scrambled and normal version of the texts. However, to study the information gained from performing one or the other task, it was necessary to complete the Chihara Latin-squares-type design by giving each subject a second pair of tests to do on a second occasion (usually within an hour or a day of the first test period). On the first occasion, subjects completed the Joe or Nicholas passage in its normal narrative version and the corresponding text in the scrambled version. As predicted, with fairly advanced subjects, the normal narrative arrangement produced higher scores than the scrambled variants (averaging over the two texts). In fact, the scores differed by about 6% on the whole for the least proficient non-natives (favoring the normal narratives) and by a little more than 11% for the most advanced native speakers.[48]
On the second occasion, the same subjects completed the scrambled version of the text they had worked through on the previous occasion but now in its sequential version and the sequential version of the text they had worked through on the previous occasion but now in a scrambled form. If nothing had been learned on the first occasion, the second occasion should have looked exactly like the first. But this was decidedly not the case. The theory predicted that the information gained from the normal narrative on the first occasion would transfer almost completely to the second occasion. It also predicted that essentially nothing of significance would be gained by studying the scrambled version on the first occasion-that is, nothing significant would transfer from that experience to the testing on the second occasion. [49]
The findings are summed up in Figure 11.
In seven distinct empirical studies, as predicted, working out cloze items in a normally structured narrative produces a profound effect on the second occasion when that same narrative is presented in its scrambled form. While, on the other hand, working through a scrambled narrative produces no significant effect on cloze items in the same text arranged in its normal narrative order. The upshot is clear: in teaching languages it is generally a waste of time to have students study bits and pieces of nonsense dropped out of the blue sky. They will perform better and learn more from studying normally structured narrative-type discourse. Furthermore, Oller (1975) had shown that this benefit is not limited to narratives per se but applies about equally to expository and procedural texts of various types.4 [50]
With respect to the contrast between formal and abstract schemata, it would appear that the formal schemata of either narrative (the Joe or Nicholas text) has its impact on the first encounter as shown in the left side of Figure 11. However, this formal effect is so generalized that it applies to the same surface forms (sentences in this case) even if they are scrambled. In fact, in our design after the normal narrative has been encountered on the first occasion (see the upper mark at the left side of Figure 11) when the scrambled variant appears on the second occasion (the mark to the right side of Figure 11 connected to the first one with a solid line), subjects remember the story-line and make sense even of the scrambled sentences.5 [51]
That is to say, the benefit of the content schemata underlying the particular narrative (either Joe or Nicholas) is generalized even to the scrambled case by the formal arrangement of the particular factual elements of either case into a coherent narrative. The reader, apparently, produces a formal schema that corresponds to the story-line. Once it has been constructed (by comprehending the narrative), the formal schema can be applied to make sense of even a relatively nonsensical (scrambled) variant of the same text. [52]
Making perfect sense of a text that has been deliberately scrambled shows that subjects have formed a concept of the story that is quite abstract. They must have a concept or schema of the story that is not dependent on any particular syntactic arrangement of the surface sentences. In short, they must have created an abstract schema that is, in a sense, non-syntacticized. It is unordered relative to any particular surface forms, but nevertheless can be used to order any bizarre arrangement (or simply to produce a new one as in a paraphrase) relative to the facts of the story and their formal connections. Thus, there must be an abstract schema which has become relatively independent of the particular arrangement of the surface forms of the given text. However, this last result is shown more dramatically and more independently of the particular materials studied in each of the two remaining studies yet to be examined.[53]
Tatsuo Taira (1992, 1993) used a modified variant of the foregoing designs with computer assisted language learning. It had been hypothesized in 1904 by Otto Jespersen that episodically organized materials are bound to be superior to non-episodic variants of the very same materials. This idea was implicit in the later writings of John Dewey (esp. 1916) who got his ideas about the temporal connectedness of experience mainly from Peirce (cf. Dewey's preface in 1938). Such sources influenced Oller, Sr. (1963-1965) to suppose that a series of lessons organized in the manner of a soap opera, or the chapters of a novel, would be easier to comprehend, to learn from, and to gain a grounding for grammatical intuitions, than any non-episodic arrangement even of the very same structures (also see Oller, Sr. and Oller 1993). The reason is simple. An episodic arrangement affords a higher level of structure from which to launch comprehension. It provides a richer formal schema. The narrative itself, the story-line, is a schema imposed above all the other formal schemata that are contained within it. Remove that higher level and a great deal is lost-hence, the episode hypothesis that episodic organization will improve comprehension and learning.[54]
&nbp;
But the episode hypothesis (see Richard-Amato 1988), though incorporated implicitly into Oller Sr.'s Spanish program, had not been tested experimentally except with relatively brief texts (e.g., by Oller and Obrecht 1969, Oller 1975, Kintsch, Mandel, and Kozminsky 1977) until Taira (1992, 1993) developed an experiment assisted by computerized instruction. Taira aimed to test the effects of episodic organization over a somewhat longer narrative and over a larger time frame than had been done previously. His research employed 96 students of English as a foreign language studying in a computer assisted language learning context in Okinawa, Japan. The work required a series of lessons grounded in a narrative. Taira's story was about a Japanese man going off to New York and later being joined by his family. Taira also developed a parallel series of vignettes identical to the narrative series except that the story-line had been removed (cf. Taira 1992, 1993, Oller and Taira 1994).[55]
Subjects were pre- and post-tested on various measures and were tested on both variants of the materials used in the computer assisted language instruction after thirteen 90-minute lessons extended over a semester. The experimental group (n = 48) was exposed to a series of connected conversations forming a story (presented in various computer assisted cloze formats accompanied by still pictures on screen and various other learning aids) while the control group (n = 48) got essentially the same exposure to the same conversations but in a random order where the story-line was removed across the various conversational vignettes. To remove the story-line, about all that had to be changed were the names of persons and places. Also, the sequential arrangement of the conversational vignettes was different between the episodic and the non-episodic versions.[57]
The results showed that students in the experimental group outperformed the control group on essentially all measures in the post-test condition except that there was no significant contrast on the non-episodic materials studied by the controls but not studied beforehand by the experimental subjects. In that case, all else being equal, the controls should have excelled but did not. More interesting still, the experimental subjects performed significantly better than the controls on post-tests that were completely independent of the instructional environment. In particular, the experimental subjects outperformed controls on the post-STEP (a standardized English placement test widely used in Japan for college entrants) given at the end of the semester. This result, combined with the results of pre-testing, showed that the experimental subjects had evidently reached beyond any formal schema tied specifically to the episodes they studied and had generalized to English materials found in test content they had never seen before. While this effect of generalization resembles what Seger (1994) and others have called `implicit learning', until now, such an effect has not been specifically tied to schema theory. However, according to the theory presented here, it is clear that the experimental subjects in Taira's study achieved a higher level of grammatical control of English in a very general way than the control group. That is, the experimentals attained a better abstract schema than controls owing to the superior episodic organization of the materials studied by the experimentals.[58]
The last study to be examined was actually a follow-up of two prior research projects. It was a doctoral dissertation by Ibrahim Al-Fallay (1994) following up Chihara, Sakurai, and Oller (1989) as well as the Taira study just discussed. The point of Chihara et al. (1989) was to assess the impact of deliberately changing the cultural content of a text to see if this adjustment would affect cloze scores of a particular subject population (Japanese women in a junior college in Japan). The texts used were the Joe and Nicholas narratives. Only a few place names, names of persons, and one or two personal interactions were changed in either text to conform to the expectations of Japanese subjects. For instance, instead of Joe and Nicholas, male Japanese names were used, and instead of Joe kissing his mother (a nearly incestuous act in the eyes of Japanese), he hugs her at the train station, and so forth. As surprising as it may seem, these minor content adjustments, to conform to culturally-based content schemata of Japanese subjects, were sufficient to produce a gain of about 6% over the same texts presented in their unmodified variants. In fact, contrary to expectations generated by other theories, the Peircean schema theory propsed here shows that minor content changes can have relatively far-reaching consequences both of the abductive and inductive kind, e.g., even a name recognized as pertaining to a male referent generates expectancies that will be absent if the name is not recognized as having any gender bias.[58]
Al-Fallay sought to replicate results from both the Taira study and Chihara et al.(1989). His subjects were 74 Arab students of EFL in Saudi Arabia and his materials were two narratives. One was based on an Arab story (unfamiliar to his subjects). The other narrative was originally written in English in an American context. Each of these was adapted either positively (toward Arab expectations) or negatively (away from Arab expectations). The story originally told in Arabic was translated into English leaving the original context unmodified for one version and, then, an adjusted version was produced to fit American cultural expectations (changing names of persons and places and events, e.g., Ramadan to Christmas, to deliberately violate Arab expectations). The story originally written in English (which already violated Arab expectations in various ways) was adapted so as to fit Arab expectations (e.g., dinner at 9 pm instead of 6) and it too was produced in two versions. The five tests over the Arab story and the five tests over the Arabized version of the American story were presented sequentially in ten installments over a three month period to the experimental group (n = 37). The counterpart tests, consisting of five unmodified cloze tests over the American story and five tests over the American-ized version of the Arab story were all presented over the same three month period in matched installments to an approximately matched control group (n = 37). In fact, several independent pre-tests were used to compare experimentals and controls at the outset. Independent post-tests were also used at the end of the three month period. The ten corresponding paired means (consisting of an Arab-based and an American-based text) were contrasted on the hypothesis that the experimental group would obtain a higher mean score on all ten contrasts owing to the conformity of the textual material to their sociocultural expectations. Results were as predicted in every case (p < .05), this, in spite of the fact that the controls were somewhat more advanced at the outset as judged by pre-test scores. [59]
More importantly, the benefit gained by working through content arranged according to the formal schemata characteristic of the Arab-world generalized to material not studied. This result came out in the contrasts on post-tests adjusted by pre-test scores as covariates. There was, for example, a significant contrast favoring the experimental group on an independent reading test (p < .05) and also on a grammar test. There was no contrast on an independent measure aimed at writing. More importantly, Al-Fallay found, as in the Taira research, that the gains owed to his modest sociocultural adjustments in the two texts apparently produced a generalized effect on the post-test condition even in materials never seen before by his control and experimental subjects. That is, the experimental subjects who benefited from the minor sociocultural adjustments in the materials studied throughout the three month period acquired more English. This result suggests that the experimental group was developing a better informed abstract schema necessary to the comprehension and production of English texts in general.[60]
CONCLUSIONS
Content, formal, and abstract schemata evidently do have the predicted differential impacts on the processing of representations ranging from the iconic phase all the way to the processing of abstract discourse. Formal schemata, based in inductive reasoning and the indexing of particular cases linked to abstract concepts, have a greater impact than content schemata, and abstract schemata grounded in concepts generalized so as to transcend any particular context of experience, have the greatest impact of all. Furthermore, as noted, abstract schemata fall at an absolute maximal limit because, by definition, nothing can be more abstract than a representation that is completely so. Likewise, the particular, non-repeatable and strictly individual facts of any actual context of experience are as particular as any facts whatever can be because nothing can be more particular than the sort of particular that is completely so. In between these extreme limits, the abstracted generalizations on the one hand and the particular facts of someone's individual experience on the other, it can be rigorously proved that the only means of linking these disparate realms (on the opposite sides of Einstein's gulf) is through the sort of bodily action that indexes concrete entities of the material world against abstract concepts of the ideal realm. Therefore, the theory is (Figure 5 above), in a strictly logical sense, complete. [61]
Interestingly, it clarifies certain ideas of earlier theoreticians in a straightforward and explicit manner. In 1984, Jean Mandler wrote:
[62]
But, upon closer examination, all representations of any kind are mental structures extracted from the kinds of representations that are found in true narratives. And, as Kintsch (1988) has argued on a rather different basis (working upward from empirical data rather than downward from logico-mathematical proofs), it comes out here in a completely general way that all comprehension and learning must be grounded ultimately in bottom-up processing of perceptual representations linked to actual material facts.
[63]
Finally, the implications for language and literacy teachers, and for educators in general, are worth noting. Bits of discourse cut loose from the moorings in experience that give them meaning ought not to be used in classrooms anywhere. The discourse that we introduce, use, and create in the classroom ought to involve the reasonable motivations normally provided by episodes of significant experience involving real material persons, events, places, and sociocultural relations with which our students can identify and find some common ground. In fact, community and communication entail common ground in the material world, so classroom procedures need to involve facts that are richly embedded in the dynamic episodes of ordinary discourse and which conform to the requirements of the true narrative case. To acquire a new language, or to become literate, or to achieve proficiency in new subject matter, it is both necessary and sufficient that the student should be enabled to make the connection between abstract representations of the new material and his or her own personal experience. Only to the extent that such true narrative links are actively made by the student, is the desired learning assured.
This paper was presented in part at the meeting of the American Association of Applied Linguistics in Baltimore, Maryland on March 8, 1994 at the Lord Baltimore Hotel. The findings reported here were also discussed at the Arizona TESOL State Conference at the Thunderbird Graduate School of International Management in Glendale, Arizona, April 23, 1994 and at the Native Literacy and Language Roundtable sponsored by the Office for Bilingual Education and Minority Language Affairs on May 7, 1994 in Denver, Colorado. Subsequently, parts of it were presented at the International Conference on TEFL Commemorating the 30th Anniversary of the College English Teachers Association of Korea, July 22-23, 1994 at Inchon Memorial Hall, Korea University in Seoul. The author is grateful to the organizations and sponsors who helped encourage its completion. Appreciation is also due the three anonymous readers for Applied Linguistics in addition to Elise Estrin, Thomas Scovel, Tahereh Paribakht, Wilga Rivers, Jack S. Damico, Stephen D. Krashen, Carolyn Kennedy, Teresa Meehan, Teruo Ueno, Christine Monikowski, Neddy Vigil, Andrew D. Cohen, and John L. Omdahl who discussed one or more of the foregoing versions with the author (Department of Linguistics; University of New Mexico; Albuquerque, New Mexico 87131-1196; electronic mail to 'JOLLER@BOOTES.UNM.EDU' or call 505-856-6078 home or 505-277-6353 office at UNM).
1 Some have argued that this is a merely a western point of view, but that argument fails. The facts under consideration, if facts at all, are universals of physics, physiology, and psychology and are little affected, if at all, by any cultural or experiential overlay. They are materially determined by physics, biology, and logic.
2 Of course, if the hotel were merely a front for money-laundering, the generalization would not apply because the ‘hotel’ would be something else and not really just a hotel.
3 It is important to note that Peircean theory shows iconic representations to be degenerate with respect to their qualities and those of their logical objects. A picture of a baby, for instance, could be almost anyone. It is not qualitatively faithful to the adult that the baby may grow up to be. Also, even perceptions (the most faithful of icons) are always incomplete. Parts of any scene are obscured by other parts. For instance, when sitting at a desk inside a room, it is impossible to see most of what lies outside the room. Further, it is a proved mathematical result of topology that any surface can be transformed into any other by degrees and in an infinitude of distinct ways. For this reason, as Peirce pointed out, all icons more or less resemble absolutely all possible states of affairs. Thus, icons need to be supported by other kinds of representations if they are to be made determinate with respect to what they mean.
4 Of course, as Jonz (1994) has demonstrated there are some descriptive texts (ones Cziko 1983 had termed ‘encyclopedic’) which are essentially unordered lists of facts, and these, as a result, are not very susceptible to any effect owed to scrambling.
5 A similar effect was found with scrambled paragraphs by Kintsch and Yarborough (1982), though the ability to unscramble paragraphs was limited to ones later judged to be incorporated into well-organized essays (less well-organized essays could not be unscrambled). Yet on the same essays cloze scores on intact and scrambled texts did not differ in their study. But see Jonz (1994) and note 4 above.
Al-Fallay, I. 1994. Limiting Bias in the Assessment of English as a Foreign Language: The Impact of Background Knowledge on the Proficiency of Saudi Arabian Students Learning English as a Foreign Language. Ph.D. thesis, University of New Mexico, Albuquerque.
Artigal, J. M. 1992. 'Some Considerations on Why a New Language Is Acquired by Being Used.' International Journal of Applied Linguistics 2/2:221-240.
Artigal, J. M. 1993. 'The L2 Kindergarten Teacher as a Territory Marker, in J. E. Alatis (ed.), 1993: Georgetown University Round Table on Languages and Linguistics 1993: Strategic Interaction and Language Acquisition: Theory Practice and Research (pp. 452-468). Washington, D. C.: Georgetown University Press.
Barthes, R. 1967. Elements of Semiology. New York: Hill and Wang.
Bentele, G. 1985. 'Audio-Visual Analysis and a Grammar of Presentation Forms in News Programs: Some Mediasemiotic Considerations' in van Dijk 1985:159-184.
Brumfit, C. J. and K. Johnson. (eds.) 1979. The Communicative Approach to Language Teaching. Oxford: Oxford University Press.
Carrell, P. L. 1984. 'Evidence of a Formal Schema in Second Language Comprehension.' Language Learning 34/2:87-112 also in J. W. Oller, Jr. (ed.) 1993: 191-205. [Page references in the text are to the latter.]
Carswell, E. A. and R. Rommetveit. 1971. Social Contexts of Messages. London: Academic Press.
Chihara, T., T. Sakurai and J. W. Oller, Jr. 1989. 'Background and Culture as Factors in EFL Reading Comprehension.' Language Testing 6/2:143-151.
Chomsky, N. A. 1982. Some Concepts and Consequences of the Theory of Government and Binding. Cambridge, Massachusetts: MIT Press.
Chomsky, N. A. 1988. Language and Problems of Knowledge: The Managua Lectures. Cambridge, Massachusetts: MIT Press.
Connel, I. and A. Mills 1985. 'Text, Discourse, and Mass Communication' in van Dijk 1985:26-43.
Cziko, G. 1983. `Another Response to Shanahan, Kamil, and Tobin: Further Reasons to Keep the Cloze Case Open.' Reading Research Quarterly 18: 361-365. Also in Oller and Jonz 1994:197-203.
de Beaugrande, R. 1991. Linguistic Theory: The Discourse of Fundamental Works. London: Longman.
Dewey, J. 1938. Logic: The Theory of Inquiry. New York: Holt.
Dreyfus, H. 1979. What Computers Can't Do. New York: Harper and Row.
Eco, U. 1976. A Theory of Semiotics. Bloomington, Indiana: Indiana University Press.
Eco, U. 1990. The Limits of Interpretation. Bloomington, Indiana: Indiana University Press.
Einstein, A. 1941. 'The Common Language of Science' in Einstein 1956:111-113. Also in Oller 1989:61-65.
Einstein, A. 1944. 'Remarks on Russell's Theory of Knowledge' in P. A. Schilpp (ed.) 1944: The philosophy of Bertrand Russell (pp. 277-291). New York: Tudor. Also in Oller 1989:21-29.
Einstein, A. 1956. Out of My Later Years. Secaucus, New Jersey: Citadel.
Fine, J. 1994. How Language Works: Cohesion in Normal and Nonstandard Communication. Norwood, New Jersey: ABLEX.
Flammer, A. and W. Kintsch. (eds.) 1982. Discourse Processing. Amsterdam: North-Holland.
Giardetti, J. R. and J. W. Oller, Jr. in press. 'Testing a Theory of Photographic Meaning.' Semiotica.
Givón, T. 1984. Syntax: A Functional-Typological Introduction, Volume I. Amsterdam, The Netherlands: Benjamins.
Givón, T. 1989. Mind, Code, and Context: Essays in Pragmatics. Hillsdale, New Jersey: Lawrence Erlbaum.
Givón, T. 1990. Syntax: A Functional-Typological Introduction, Volume II. Amsterdam, The Netherlands: Benjamins.
Ketner, K. L. (ed.) 1992. Reasoning and the Logic of Things: The Cambridge Conferences Lectures of 1898: Charles Sanders Peirce. Cambridge, Massachusetts: Harvard University Press.
Ketner, K. L. and H. Putnam. 1992. `Introduction: The Consequences of Mathematics' in Ketner 1992:1-54. Ann Arbor, Michigan: University of Michigan Press.
Johnson-Laird, P. N. 1983. Mental Models: Towards a Cognitive Science of Language, Inference, and Consciousness. Cambridge, Massachusetts: Harvard University Press.
Jonz, J. 1994. `Textual Sequence in Linguistic Semiosis' in Oller and Jonz 1994:287-316.
Jonz, J. and J. W. Oller, Jr. 1994. `A Critical Appraisal of Related Cloze Research' in Oller and Jonz 1994:371-408.
Kintsch, W. 1982. `Memory for text' in Flammer and Kintsch 1982:186-204.
Kintsch, W. 1988. `The Use of Knowledge in Discourse Processing: A Construction-Integration Model.' Psychological Review 95:163-182.
Kintsch, W., J. R. Miller and P. G. Polson. (eds.) 1984. Method and Tactics in Cognitive Science. Hillsdale, New Jersey: Lawrence Erlbaum.
Kintsch, W., T. S. Mandel and E. Kozminsky. 1977. `Summarizing Scrambled Stories.' Memory and Cognition 5:547-552.
Kintsch, W. and E. Greene. 1978. `The Role of Culture-Specific Schemata in the comprehension and recall of stories. Discourse Processes 1:1-13.
Kintsch, W. and T. A. van Dijk. 1978. `Towards a Model of Text Comprehension and Production.' Psychological Review 85:363-394.
Kintsch, W. and J. C. Yarborough. 1982. `The Role of Rhetorical Structure in Test Comprehension.' Journal of Educational Psychology 74:828-834.
Kora_, N. 1988. `Functional, Cognitive, and Semiotic Factors in the Development of Audio-Visual Comprehension. Educational Communication and Technology: A Journal of Theory, Research, and Development 36:67-91.
Krashen, S. D. and T. Terrell. 1983. The Natural Approach: Language Acquisition in the Classroom. San Francisco, California: Alemany Press.
Langacker, R. W. 1987. Foundations of Cognitive Grammar: Volume 1, Theoretical Prerequisites. Stanford, California: Stanford University Press.
Leech, G. 1969. Toward a Semantic Description of English. Bloomington, Indiana: Indiana University Press.
Levin, J. R. 1982. `Pictures as Prose-Learning Devices' in Flammer and Kintsch 1982:412-441.
Makkai, A. 1972. A nyelvészeti részletjelentést_l az átalános emberi jelentésig, mint stratifikációs és össz-ökológiai problémacsoport. [From partial linguistics meaning toward general human meaning as a stratificational and pragmoecological problem.] Paper presented at the Second International Congress of Hungarian Linguists, August 1972, Szeged, Hungary.
Mandler, J. M. 1984. Stories, Scripts, and Scenes: Aspects of Schema Theory. Hillsdale, New Jersey: Lawrence Erlbaum Associates.
Mandler, J. M. 1988. `How to build a baby: On the Development of an Accessible Representational System.' Cognitive Development 3:113-136.
Mandler, J. M. 1992. `How to Build a Baby: Conceptual Primitives.' Psychological Review 99:587-604.
McClelland, J. L. and D. E. Rumelhart. 1986. Parallel Distributed Processing: Explorations in the Microstructure of Cognition, Volume 2: Psychological and Biological Models. Cambridge, Massachusetts: Bradford/MIT Press.
Metz, C. 1971. Langage et cinéma. Paris: Librairie Larousse.
Metz, C. 1974. Film Language: A Semiotics of the Cinema. New York: Oxford University Press.
Metz, C. 1983. `Photography and fetish.' October 34:81-90.
Millis, K. K. and M. A. Just. 1994. `The Influence of Connectives on Sentence Comprehension.' Journal of Memory and Language 33:128-147.
Nagel, E. 1959. `Charles Sanders Peirce: A Prodigious but Little Known American Philosopher.' Scientific American 200:185-192.
Oller, J. W., Jr. 1983. 'Some Working Ideas for Language Teaching' in Oller and Richard-Amato 1983:3-19.
Oller, J. W., Jr. (ed.) 1989. Language and Experience: Classic Pragmatism. Lanham, Maryland: University Press of America.
Oller, J. W., Jr. 1990. `Semiotic Theory and Language Acquisition' in J. E. Alatis (ed.) 1990: Georgetown University Round Table on Languages and Linguistics 1990. Washington, D. C.: Georgetown University Press.
Oller, J. W., Jr. (ed.) 1993a. Methods That Work: Ideas for Literacy and Language Teachers. Boston, Massachusetts: Heinle and Heinle.
Oller, J. W., Jr. 1993b. 'Reasons Why Some Methods Work' in Oller 1993a:374-385.
Oller, J. W., Jr. in press. 'Semiotic Theory Applied to Free Will, Relativity, and Determinacy: Or, Why the Unified Field Theory Sought by Einstein Could Not Be Found.' Semiotica.
Oller, J. W., Jr. and J. Jonz (eds.) 1994. Cloze and Coherence. Cranbury, New Jersey: Bucknell University Press.
Oller, J. W., Jr. and C. Kennedy. in press. `The Determinacy of Linguistic Data.' Manuscript University of New Mexico, Department of Linguistics.
Oller, J. W., Jr. and D. H. Obrecht. 1969. `The Psycholinguistic Principle of Informational Sequence: An Experiment in Second Language Learning.' IRAL 7:169-174.
Oller, J. W., Jr. and P. Richard-Amato (eds.) 1983. Methods that Work: A Smorgasbord of Ideas for Language Teachers. Rowley, Massachusetts: Newbury House.
Oller, J. W., Jr., B. D. Sales, and R. V. Harrington. 1969. A Basic Circularity in Traditional and Current Linguistic Theory. Lingua 22:317328.
Oller, J. W., Jr. and T. Taira. 1994. 'Cloze and Episodic Organization' in Oller and Jonz 1994:345-369.
Oller, J. W., Jr., G. K. S. Yü, L. A. Greenberg and R. Hurtado de Vivas. 1994. 'The Learning Effect from Textual Coherence Measured with Cloze' in Oller and Jonz 1994:247-268.
Oller, J. W., Sr. 1963-1965. El español por el mundo. Chicago, Illinois: Encyclopedia Britannica Educational Films.
Oller, J. W., Sr. and J. W. Oller, Jr. 1993. 'An Integrated Pragmatic Curriculum' in Oller 1993b: 50-62.
Peck, R. N. 1980. Secrets of Successful Fiction. Cincinnati, Ohio: Writer's Digest.
Peirce, C. S. (ed.) 1883. Johns Hopkins Studies in Logic. Boston, Massachusetts: Little, Brown and Company.
Peirce, C. S. 1897. 'The Logic of Relatives.' The Monist 7:161-217. Also in Hartshorne and Weiss 1931:288-345.
Peirce, C. S. 1898. `Reasoning and the Logic of Things' in Ketner 1992:105-268.
Piaget, J. 1947. The Nature of Intelligence. Tottowa, New Jersey: Littlefield Adams.
Piaget, J. 1952. The Origins of Intelligence in Children. Madison, Connecticut: International Universities Press.
Putnam, H. 1992. `Comments on the Lectures' in Ketner 1992:55-104.
Reichling, A. 1961. `Principles and Methods of Syntax: Cryptanalytical Formalism.' Lingua 10:1-17.
Richard-Amato, P. 1988. Making It Happen. London: Longman.
Rommetveit, R. 1968. Words, Meanings, and Messages. New York: Academic Press.
Rommetveit, R. 1979. `Words, Contexts and Verbal Message Transmission' in R. Rommetveit and R. M. Blakar (eds.) 1979: Studies of Language, Thought and Verbal Communication. London: Academic Press.
Rommetveit, R. and R. M. Blakar. 1979. Studies of Language, Thought and Verbal Communication. London: Academic Press.
Rumelhart, D. E. 1975. `Notes on a Schema for Stories' in D. Bobrow and A. Collins (eds.) 1975: Representation and Understanding: Studies in Cognitive Science (pp. ). New York: Academic Press.
Schank, R. 1975. 'The Structure of Episodes in Memory' in D. Bobrow and A. Collins (eds.) 1975: Representation and Understanding (pp. 237-272). New York: Academic Press.
Schank, R. and R. Abelson. 1977. Scripts, Plans, Goals, and Understanding. Hillsdale, New Jersey: Lawrence Erlbaum.
Sebeok, T. A. 1991. Semiotics in the United States. Bloomington, Indiana: Indiana University Press.
Seger, C. A. 1994. `Implicit Learning.' Psychological Bulletin 115:163-196.
Shapiro, M. 1983. The Sense of Grammar: Language as Semeiotic. Bloomington, Indiana: Indiana University.
Spada, N. and P. M. Lightbown. 1993. `Instruction and the development of questions in L2 classrooms.'Studies in Second Language Acquisition' 15:205-24.
Spelke, E. S., G. Katz, S. E. Purcell, S. M. Ehrlich, and K. Breinlinger. 1994. `Early Knowledge of Object Motion: Continuity and Inertia.' Cognition 51:131-176.
Spolsky, B. 1989. Conditions for Second Language Learning. Oxford: Oxford University Press.
Swain, D. V. 1980. Techniques of the Selling Writer. Norman, Oklahoma: University of Oklahoma Press.
Tagg, J. 1988. The Burden of Representation. Amherst: University of Massachusetts Press.
Taira, T. 1993a. `Episodes on the Computer' in Oller 1993a:309-319.
Taira, T. 1993b. Episodic Organization and CALL: A Pragmatic Approach. Ph. D. thesis, University of New Mexico, Albuquerque.
Thorndyke, P. W. 1977. Cognitive structures in comprehension and memory of narrative discourse. Cognitive Discourse 9:77-110.
Tudor, I. and R. Tuffs. 1991. 'Formal and Content Schemata Activation in L2 Viewing and Comprehension.' RELC Journal 22/2:79-97. Also in Oller 1993a:247-259.
Uhlenbeck, E. M. 1963. `An appraisal of transformation theory.' Lingua 12:1-18.
Uhlenbeck, E. M. 1967. `Some further remarks on transformational grammar.' Lingua 17:263-316.
Valdman, A. 1992. `Authenticity, Variation, and Communication in the Second Language Classroom' in C. Kramsch and S. McConnell-Ginet (eds.) 1992: Text and Context: Cross-Disciplinary Perspectives on Language Study (pp. 79-97). Toronto, Canada: D. C. Heath.
van Dijk, T. A. 1981. Studies in the Pragmatics of Discourse. The Hague: Mouton.
van Dijk, T. A. 1985. Discourse and Communication: New Approaches to the Analyses of Mass Media Discourse and Communication. Berlin: Walter de Gruyter.
Walker, R. S. Rattanavich and J. W. Oller, Jr. 1992. Teaching All the Children to Read. Buckingham, England: Open University Press.
Walz, J. 1989. Context and Contextualized Language Practice in Foreign Language Teaching. Modern Language Journal 73:160-168.
Widdowson, H. G. 1990. Aspects of Language Teaching. Oxford: Oxford University Press.
Wilkins, D. 1994. `Recent Issues in Syllabus Design.' Forum lecture presented at the International Conference on TEFL Commemorating the 30th Anniversary of the College English Teachers Association of Korea, July 22, 1994 at Inchon Memorial Hall, Korea University, Seoul, Korea. In English Teaching No. 49: 41-56.
Williams, J. N. 1993. 'Processing of Anaphoric Nouns in Extended Texts: The Role of Surface Information.' Language and Speech 36:373-391.
Wong-Fillmore, L. 1989. `Language Learning in Social Context: The View from Research in Second Language Learning' in R. Dietrich and C. F. Graumann (eds.) 1989: Language Processing in Social Context (pp. 277-302). Amsterdam, The Netherlands: Elsevier.
|
|
|
![[*]](../../../../backgrnd/divider.gif){kind=small}
|
|
|
Do you think the author has it wrong? If so and you want to contribute a critical comment or commentary, brief or extended, concerning the above paper or its subject-matter, or concerning previous commentary on it, it will be incorporated into this webpage as perspicuously as possible and itself become subject thereby to further critical response, thus contributing to Arisbe as a matrix for dialogue. Your contribution could also be of the nature of a corroboration of the author, of course, or be related to it or to some other response to it in some other relevant way.
|
