1. The Diminishing of Classical Cognitivism
Cognitivism has been a major theoretical framework for organising the data of applied linguistics. By ‘classical’ cognitivism I will understand the model that is traced to Chomsky and developed by authors like Krashen and Long, which states roughly that language acquisition is the product of adequate comprehensible linguistic input. This basic model is further elaborated to include more variables required for acquisition to take place. For example, Schumann raised serious questions about the adequacy of the cognitivist model in the 1970s and proposed social variables as causal. Subsequently, these social variables were found to be difficult to measure and Schumann moved on to consider motivational and interactive factors and their neurobiological bases. More recently, approaches drawn from parallel distributed processing models have suggested a way of conceptualing a language emergence model that might account for the lack of linearity observed in language development.
The problem with these alternative approaches is that they do not really offer an alternative to classical cognitivist theory. If we consider the matter historically, cognitivist theory itself replaced earlier approaches such as behaviourism, which replaced even earlier approaches. Since the history of psychology is a sketchy subject that is known to relatively few in psychology and linguistics it is correspondigly little known what theories of language and its psychology were replaced by behaviourism. However, we can posit that a major reason for the move to behaviourism and then to cognitivism was precisely that they offered, or promised at any rate, the sort of relatively simple and predictive explanations of behaviour and competence. Cognitivism is effectively a meta-theory for generating psychological explanations that conform to the model of scientific explanation developed by logical positivim, that is, the hypothetico-deductive method.
Viewing the matter from this perspective, cognitivism gains explanatory power when it can explain a wide range of complex phenomena in terms of relatively few variables and causal laws. Behaviourism offered to explain all competence in terms of stimulus-response and laws of reinforcement. Its success or failure rested on its ability to carry out this reduction, that is, to explain all the behavioural complexity in terms of simple laws of conditioning. Chomsky argued that this is reduction is not feasible and that we need to posit a functional architecture whereby the system takes in inputs and carries out computational operations to arrive at a competence conceived as the knowledge of rules. Connectionist or parallel distributed processing models fit within the cognitivist framework but they posit that they underlying processes are non-linear, that is, we cannot expect predictable results given a set of inputs because of the distributed nature of the processing. There has always been a question whether such models can be reconciled with the classical cognitivist scheme (Fodor 1983), or whether they are in effect giving the game away, that is, the project of predicting output given input.
We can assume that a cognitive model constitutes progress when some phenomnon, such as learning a language or some part of language, is reduced to some identifiable skill that is a funtion of specifiable inputs, and that is generalisable. In other words, it cannot be that we can find some effect generated under laboratory conditions which is no way generalisation outside of the labortory. Thus, while certain psychological effects such as the size of short term memory being 7 plus or minus 2 is established as being universal, it is questionable whether psychological phenomena generalise beyond the groups of ‘subjects’ that they are tested on, these typically being an undergraduate student in an
Anglophone university, which is not a truly random population. In other words, it is arguable that psychological and linguistic research can conform to the experimental framework only by narrowing down the sample and also by framing its hypotheses in ways that are not generalisable. Furthermore, this only comes up in fields like applied linguistics where application in the classroom across culturally diverse student populations is inevitable and raises the question whether, as the number of variables increases to account for the real-world phnomena which are unavoidable in the field, the cognitivist model is watered down and diminished to the point whereby it is questionable whether the whole cognitivist framework in its entirety should not be seriously questioned.
The promise and attraction of cognitivism was precisely simplicity and predictability, because these features offered the desirata of the theories of psychology as natural science. As the model adds more variables and its predictions become ever more narrow in scope, thus failing to generalise, the question we must ask is why retain the model at all? The answer to this question is precisely the desire to retain the status of science, and this basic motivation imposes what Danziger has called the ‘methodological imperative’ of psychological theories generally, namely, the requirement that such theories resemble in certain respects the methods and explanations characteristic of the methods and explanations in the natural sciences. In other words, the positivist philosophy of science has postulated that natural sciences employ the hypothetico-deductive method and that for psychology and the social sciences to attain the status of science at all their statements must conform to this conception of the logic of explanation.
Within this framework, as developed by Krashen, language acquisition takes place as a consequence of adequate comprehensible input and low affect filter. This in turn has been a development of Chomskyan idea of an innate universal grammar. A number of arguments currently considered in applied linguistics signal a shift away from the orthodoxy of Chomskyan cognitivism, eg., the idea of language acquisition as the unfolding of an innate grammar elicited by external input. Thus, some major figures in the field have been expressing scepticism that the cognitivist framework is able to produce the promised predictive variables. As a consequence, alternative models are now being explored, signalling a diminishing in the role of the putative cognitive processes. Ideas that have been explored include the role of social proximity (Schumann), interaction, gesture (Schumann), and emergence (Larsen-Freeman) in the search for (a) what is innate, and (b) the causal factors in language acquisition. This is interesting because it raises the question whether the field has attained the status of normal science. Cognitivism provided for theoretical stability and offered the promise of a predictive framework, and the question is whether these develoments signal a phase of abnormal science and paradigm shift.
While this appears to signal a paradigm shift there are two respects in which it is not. First, it seems to be more of a movement back in history in that the approaches being considered closely resemble pre-Chomskyan approaches so that we might be really dealing with a re-consideration of approaches that Chomskyan nativism displaced. Second, nonetheless, the way in which these earlier approaches are being reconsidered is in a sense Chomskyan in that background philosophy of science of these earlier approaches is replaced with the positivism that characterises cognitivism. In that sense we can be said to be dealing with a new synthesis.
Pre-Chomskyan approaches were non-positivist and developed their ideas about language and psychology in within a non-positivist framework. Positivism and the search for causal explanations in the social sciences has generated behaviourism and subsequently cognitivism. The inability of cognitivist theory to be predictive in the required ways has led to the reconsideration of these earlier ideas but without abandoning the positivistic philosophy, that is, the idea that we are searching for explanations of a certain sort.
Kurt Danziger (ref) has called this the ‘methodological imperative’ in psychology. It is the imperative, motivated by the desire to attain the status of ‘science’, that theories and explanations in disciplines like psychology and applied linguistics take a particular form that resembles the form of the natural and experimental sciences, that is, they should seek to provide causal explanations and utilise quantitative research methods. From this perspective, while the pre-Chomskyan approaches may have generated some useful ideas, they lack the status of scientific explanations because they appear speculative and a priori, and in that sense are pre-scientific and philosophical.
As Danziger points out, in order to render ideas scientific they need to be reworked and elborated such that they ‘fit’ into a particular methodological framework. The question raised by Danziger et al is whether this is legitimate, that is, whether the objects of inquiry ought to be augmented to fit the methodology, or whether methodology should be adapted to the objects. The alternative view is that when these phenomena are augmented to fit the methodology the original theory or object is no longer recognizable. Privileging methodology over the object means that the object and the form of explanation are fundamentally altered.
The implication here is that pre-Chomskyan approaches are not merely philosophical precursors to the final stage of rendring them scientific, but actually provide an alternative view of science to the positivistic reduction. On this view, the idea of science as generating predictive hypotheses will never generate an approach that is adequate to the object, and such explanations inevitably generate reductions that eliminate too much from the object to render these explanations adequate to the object under investigation. In other words, if we reduce enough predictions can be generated, but in achieving the desired predictions we have hollowed out the object beyond recognition and rendered our explanations inapplicable outside of the narrow, decontextualised laboratory conditions.
The pre-Chomskyan view, most commonly associated with Wilhelm Dilthey but held widely in 19th and early 20th century German academic world (Farr), is that the human sciences deal with phenomena that are too complex to be amenable to the techniques of the natural sciences (Makkreel), and so a different set of scientific methods must be developed. In other words, these approaches were developed specifically in reaction against positivism and the generalisation of the methods of the natural sciences to the human sciences. So in order to compare and evaluate these pre-Chomskay, pre-positivist approaches vis a vis their recent reincarnation we need to have some conception of the basic conception of the mode of inquiry appropriate to the human and social sciences.
2. The Methodological Imperative in Applied Linguistics
The methodological imperative means that the objects of psychology have to be elaborated or reshaped so as to fit the methodology rather than the other way around. In other words, the methodology forms the limiting factor and the objects of psychology need to be elaborated to fit that rather than the methodology being adapted to the objects (Danziger, …). Thus, psychological and linguistic theories that do not fit the methodological requirements are reformulated or reconstructed to do so. The assumption must be that as they stand these theories are inadequate. However, the question is raised whether the reconstruction does not render the objects or the theories unrecognisable (Danziger, …).
We can see this process in applied linguistics in the attempts to formulate alaternative approaches to classical cognitivism that appear to visit pre-Chomskyan, social, ecological and evolutionary, approaches to linguistics. However, in doing so they reconstruct these early approaches in precisely the positivist framework that was not only absent in these early approaches, but when we look at their history were self-consciously opposed to positivism. It is arguable that this actually reflects the general trend in later 20th century analytic philosophy, and cognitive and social sciences, namely, to produce constructs that are amenable to the sort of inquiry characteristic of positivist science.
Thus, in the case of applied linguistics, at each stage of critique of classical cognitivism the alternatives considered are framed in term of causal factors and predictive hypotheses. The idea of language emergence seems to have given the game away but not quite, for it offers predictability at the microlevel. Yet this is in effect giving the game away because the anti-positivist approaches accepted that there is a level of local, narrowly circumscribed responsiveness at which behaviour can be predicted (Makkreel on Dilthey). However, in order to make psychological generalisations the view was that we need to dispense with the idea of causal explanations, and instead offered accounts which were environmental, evolutionary, developmental and process-oriented. But such accounts, while they may allow for quantiative studies, are not committed to the sort of methodology characteristic of positivist science, but rather select and adapt methodology to the phenomena in question. This puts into question the entire program of elaborating and recontructing the objects of the cognitive sciences.
The idea of language emergence offers a choice. We can continue with the assumption that we are doing natural science and abide by the methodological imperatives of positivism. Alternatively, we can accept that at best we can find regularities but not predictive laws, and that these regularities will be a function of local affordances, and that therefore we must accept that in language learning and language teaching there are no psychological laws but understanding and expertise that has the form of Gestalts rather than deterministic rules. This fundamentally affects how we approach linguistic research and our commitment to a scientistic conception of psychology constrained by the methodological imperative. It might also lead us to reconsider the pre-Chomskyan approaches to psychology and linguistics.