+ All Categories
Home > Documents > THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

Date post: 03-Oct-2016
Category:
Upload: barry-mclaughlin
View: 217 times
Download: 2 times
Share this document with a friend
20
THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM' Barry McLaughlin University of California, Santa Cruz The answers we give to the questions we ask are limited by the methods we use The argument of this paper is that theory and research in second language learning are leading us to ask questions for which we need a new research paradigm-one that focuses on the interaction of person and situational factors. This paradigm has been used in some areas of research in education and psychology, although only recently has the application of regression analysis allowed adequate testing of the model. Advantages and problems in the use of the interaction research paradigm are discussed and implications are drawn for research on second lan- guage learning. The field of second language learning has experienced a tremendous expansion in the last decade. Indeed, the field can be said to have developed as a coherent research area in the 1970s. Prior to this time researchers con- cerned with understanding the process of second language learning were a relatively rare breed. Now all that has changed, and there are hundreds of research reports appearing each year on various aspects of language learning. As we end the decade of the 70s and enter the 80s, two themes recur in the literature and at professional meetings: 1. The importance of the social situation. Increasingly researchers have come to recognize that language occurs in a social context. A shift in interest from form to function is especially apparent in the work on discourse analysis and on the structure of conversations. At the level of application, this trend is reflected in the growing concern with special purpose components of ESL programs. Furthermore, there is a greater recognition that language learning occurs in various situations, not simply in formal classroom instruction: The classroom is only one of many possible language learning contexts. 2. The importance of individual differences. Recent research in both first and second language learning has revealed consistent differences in learning styles. This work suggests that individuals vary in the strategies ' The author wishes to thank the Alexander von Humboldt Foundation for its support during the writing of this paper. 33 I
Transcript
Page 1: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING:

AN EMERGING PARADIGM'

Barry McLaughlin University of California, Santa Cruz

The answers we give to the questions we ask are limited by the methods we use The argument of this paper is that theory and research in second language learning are leading us to ask questions for which we need a new research paradigm-one that focuses on the interaction of person and situational factors. This paradigm has been used in some areas of research in education and psychology, although only recently has the application of regression analysis allowed adequate testing of the model. Advantages and problems in the use of the interaction research paradigm are discussed and implications are drawn for research on second lan- guage learning.

The field of second language learning has experienced a tremendous expansion in the last decade. Indeed, the field can be said to have developed as a coherent research area in the 1970s. Prior to this time researchers con- cerned with understanding the process of second language learning were a relatively rare breed. Now all that has changed, and there are hundreds of research reports appearing each year on various aspects of language learning.

As we end the decade of the 70s and enter the 80s, two themes recur in the literature and at professional meetings:

1. The importance of the social situation. Increasingly researchers have come to recognize that language occurs in a social context. A shift in interest from form to function is especially apparent in the work on discourse analysis and on the structure of conversations. At the level of application, this trend i s reflected in the growing concern with special purpose components of ESL programs. Furthermore, there is a greater recognition that language learning occurs in various situations, not simply in formal classroom instruction: The classroom is only one of many possible language learning contexts.

2 . The importance of individual differences. Recent research in both first and second language learning has revealed consistent differences in learning styles. This work suggests that individuals vary in the strategies

' The author wishes to thank the Alexander von Humboldt Foundation for its support during the writing of this paper.

33 I

Page 2: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

332 Language Learning Vol. 30, No. 2

they employ in learning a second language because of differences in cognitive style. There is a growing consensus that more flexibility is needed in teaching methods as individuals differ in how they approach the target language.

There is a third theme, which has received less attention, but which

3. The interaction between person and situation variables. If we concep- tualize the situational factor as the context of language learning and the person factor as those cognitive and personality factors that influ- ence learning, we find that situation and person variables interact: Different individuals react differently to the same language learning situation and different situations have varying effects on the same individual. In the language of experimental design, it is not simply the main effects of situation and person that matter, but the interaction between the two.

The argument of this paper is that (a) the interactionist perspective is the logical outgrowth of developments in theory and research in second language learning and (b) if we are to understand the complex relationships between situation and person variables, we need a new paradigm for research. After looking briefly at where I believe our main theories are leading us, I will identify new directions in second language research that, like developments in theory, raise questions that require for their answer a new research para- digm. I will discuss this paradigm in the second part of the paper. By para- digm is meant a style of conducting research and corresponding methods of analysis-in this case a focus upon person x situation interactions and the use of methods appropriate to their analysis. In discussing this paradigm I will relate it especially to the work of Cronbach and Snow (1977) on Aptitude x Treatment Interaction methods.

follows from the first two and is equally important:

DEVELOPMENTS IN THEORY AND RESEARCH

Obviously not all developments in theory and research-not even all of those bearing on the present theme-can be reviewed here. In considering theoretical developments, I will limit my remarks to what I believe to be the three theoretical formulations that have in recent years most influenced our thinking about the processes of second language learning: Schumann’s pid- ginization theory (1978), Selinker’s notion of interlanguage (1972), and Krashen’s Monitor Model (1977, 1978).

Page 3: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

McLaughlin 333

The impetus of theory

Recent discussions of pidginization theory (Andersen in press, Bickerton in press) have been concerned with similarities and differences between pidginization and early second language learning in a natural situation. Bickerton (in press) has made the important point that speakers of pidgin languages use pidgin with other members of the pidgin-speaking community and receive feedback from them, not from native speakers of the source language. As a result, pidgins develop in isolation from the source-language community and may even be unintelligible to speakers of the source language. In contrast, the second language learner gets feedback, no matter how limited, from target-language speakers. In the natural setting individuals learn a second language so that they may interact with members of the target- language community.

This difference in function suggests that there may be differences between pidginization and early untutored second language learning that have escaped the notice of researchers t o this point. The development of the theory re- quires comparing cross-sectional and longitudinal studies of pidgin speakers with the findings from similar research with second language learners. Fur- thermore, understanding the process of “depidginization” (Andersen in press) requires attending to those situational and person variables (and their inter- action) that affect the pidginization and depidginization processes. The work of Andersen (in press) and Meisel (1979) are important steps in this direction.

Similarly, recent work on interlanguage theory has looked at both person variables and situational factors. Individual learning styles (Adjemian 1976, Selinker, Swain, and Dumas 1975) and communication strategies (Virgil and Oller 1976) have been seen to affect the interlanguage and the process of “fossilization” (Selinker 1972). The importance of situational factors was stressed by Tarone (1979), who has argued convincingly that style shifting occurs in the interlanguage when the same person responds in different experimental contexts. Some variables that affect style shifting are the atten- tion the learner is paying to speech and whether speech is formal or informal and casual. In formal speech and when the learner is attending to speech, the interlanguage rule system is likely to be “invaded” by the target-language rule system (Tarone 1979).

Tarone goes on to urge that closer attention be given in future research on interlanguage to precise description of situational and experimental variables so that data are interpretable and replicable. This is an important point, not only for the development of interlanguage theory, but for research on second

Page 4: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

334 Language Learning Vol. 30, No. 2

language learning generally: Only if care is given to defining situational variables can we gain systematic understanding of how these interact with the communication strategies and cognitive styles of individual learners.

Finally, Krashen’s Monitor Model (1977, 1978) stresses the importance of situational factors-the language classroom versus the “naturalistic” environ- ment-and the role of individual differences in language learners’ styles- over- and under-users of the Monitor (Krashen 1976). At the present stage of its development, however, the theory suffers from an important defect, which I pointed out in my critique (and which Krashen, in his reply (1979), curiously ignored)-namely, the theory’s central distinction between learning and acquisition “rests ultimately on whether the processes involved are conscious or subconscious” (McLaughlin 1978, p. 330). Appeals to conscious versus subconscious experience are notoriously unreliable. Similarly, judg- ments as to whether conscious “learning” precedes subconscious “acquisi- tion” or vice versa cannot be resolved-no matter what Krashen or I believe to be the case-until the meaning of these terms is established.’

Krashen argued that the Monitor Model provides “testable hypotheses about a wide range of phenomena in terms of a general theory” (1979, p. 160) and cited “evidence” from research on grammatical morphemes and on age differences between language learner^.^ The difficulty with this

Krashen (1979) pointed out that “special experimental conditions” are necessary to establish differences between cognitive phenomena such as “learning” and “acquisi- tion,” but his one attempt in this direction (Krashen, Butler, Birnbaum, and Robertson 1978) was a study in which subjects had to say whether their judgments of the gramma- ticality of English sentences were based on “rule” or “feel.” The deficiencies of this study were pointed out in McLaughlin (1978).

Krashen’s explanation (1979) of the results of various morpheme studies is con- trived and ad hoc. It appears to be rather difficult to demonstrate the operation of the Monitor in such studies and Krashen’s repeated failures to do so are explained away by changing the requirements (focus on form as well as time, use of “extreme” discrete- point test, etc.). However, Farhady’s finding (1979) that there is no practical difference between discrete-point and integrative tests greatly reduces the force of this argument for the Monitor Model. Similarly, Seliger’s recent research (1979) testing Monitor Model predictions led him to conclude that the Monitor is limited to such specific output modalities and requires such carefully confined conditions for its operation that it cannot be thought to be representative of the learner’s internal, conscious knowledge of the grammar.

The research on age differences between language learners is no more convincing. Krashen (1979) maintained that studies that focus on eventual attainment rather than rate of acquisition show that children are superior to adults. This is seen to support Monitor Model claims in that children do not use the Monitor (which interferes with communication), whereas adults, having passed the stage of formal operations, do. The evidence Krashen cited, however, comes from studies dealing with pronunciation, with

Page 5: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

McLaugh lin 335

argument is that unless we know what “learning” and “acquisition” refer to, it is impossible to test the theory. I noted that the phenomena Krashen talked about in terms of ‘‘learning’’ and “acquisition” could just as well be described in other terms-i.e., in terms that do not have recourse to conscious or subconscious experience (McLaughlin 1978). I suggested the use of the terms ‘‘controlled’’ and “automatic” processes as these terms are used by Schneider and Shiffrin (1977). Another possibility is to refer t o “intentional” and “incidental” learning processes (McLaughlin 1965). Both of these sets of terms can be tied to observable experimental conditions. My point is simply that whatever conceptualization is employed to account for differ- ences in learning processes, the requirement of falsifiability must be met. Appeals to “learning” and “acquisition” have no meaning unless these terms are tied to observable experimental conditions. Lacking this, we are in a night where all cows are black.

Nonetheless, Krashen’s work has called our attention to the importance of situational factors and person variables. In this sense his writings fit well with (and have contributed to defining) the contemporary Zeitgeist in second language research. In particular, his work raises questions about the different learning styles that characterize individuals and the possibility that the same individual uses different styles in different situations. Like pidginization and interlanguage theory, the Monitor Model has posed questions that can only be adequately answered by the study of person x situation interactions.

New directions in research on second language learning

If anything characterizes recent research and theory on child first language acquisition, it is the movement away from an emphasis on children’s grammar to a concern with the role of communication and dialogue. Whereas earlier research was directed at such questions as how the child’s grammar differs from adult grammar and how various grammatical forms develop in the direc- tion of adult performance norms, recent research has been focused on the situational context of language acquisition and the nature and function of communication exchanges. The data base for the earlier research was the iso- lated utterances of the child; the data base for recent research is the parent- child dialogue.

This orientation has led researchers to examine the characteristics of

the exception of a study by Oyama (1978), which deals with comprehension. Children do seem to be superior to adults in acquiring the sound system of a second language, but the Monitor Model makes no prediction about this. Nor does the theory deal with comprehension, as it is a theory of speech output.

Page 6: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

336 Language Learning Vol. 30, No. 2

linguistic input to children and the effects that “caregiver speech” has on lan- guage development. Attention has also been given recently to individual differences in caregiver-infant dyads that may be related to the wide range of individual differences displayed by children in both speed and method of language development. Researchers in second language learning have similarly become increasingly concerned with linguistic input and individual differences.

Research on linguistic input has been motivated by the growing recogni- tion that the goal of second language learning is communicative competence and that learning takes place through communicative exchanges. Like the child acquiring a first language, the second language learner is exposed to a specialized input, both in the classroom and in naturalistic settings.

In spite of the progress that has been made in understanding the charac- teristics of foreigner talk (Freed 1978, Hatch, Shapira, and Gough 1975) and teacher talk (Henzl 1979, Wagner-Gough and Hatch 1975), we do not yet know enough about the nature of the input second language learners receive. There have been no factor-analytic studies of the features of foreigner or teacher talk. Hence we do not know whether the various features constitute a single factor or whether, as Brown (1977) has suggested for caregiver talk, there is more than one dimension to the register. We do not even know to what extent native speakers use foreigner and teacher talk: There are prob- ably large differences between individuals and within individuals in different circumstances.

At the level of explanation our knowledge is even more limited. We know little about the effect that input has on the language learner. Research on first language acquisition suggests that caregiver speech assists in the com- munication of meaning, not in the acquisition of syntax. Tlvs is most likely the case for foreigner and teacher talk as well (although the data are not yet in). Since teaching is essentially the manipulation of input, these issues have instructional repercussions. It would be helpful to know, for example, which features of the foreigner talk and teacher talk registers are more important for linguistic growth. In research on first language acquisition Cross (1978) found that maternal elaboration of child-initiated topics facilitated the child’s language development. Does the teacher’s elaboration of the second language learner’s speech assist learning? How far beyond the learner’s corn- petence can the teacher go and still have the input accepted? What input is not processed or, if processed, not utilized? Work on these questions is beginning and will inevitably push us in the direction of a more complex research paradigm, one involving both person and situation variables.

As in first language research, interest in individual differences in learning

Page 7: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

McLaughlin 337

styles in second language learning is fairly recent. Hatch (1974) had identified two types of second language learners: “rule-learners,” who showed orderly stages of acquisition with little variability of forms produced at any stage, and “data-gatherers,” who imitate a great deal and use a mixture of correct and incorrect forms. Krashen (1976) distinguished between over-, under-, and optimal users of the Monitor. Fillmore (1976) discussed the manner in which individual differences in social skills affected the application of those cogni- tive strategies that ultimately lead to language learning. In her analysis the interaction of social, personality, and cognitive factors led to variations in the pace at which children learned a new language in a natural setting.

Recently Snow and Hoefnagel-Hohle (1 979) reported that factor analysis of the language skills of naturalistic second language learners revealed two dominant factors: phonological ability and grammatical ability. If the ability to assimilate and imitate chunks of auditory information relates to phono- logical ability and the ability to avoid syntactic and morphological errors re- lates to grammatical ability (as the authors suggested), then these two factors are quite similar to the types of dichotomies in language learning styles observed by other investigators (see Krashen and Scarcella 1978).

In spite of these consistencies, however, many questions remain. More research is needed to determine the extent to which learning styles interact with situational demands. Again one is reminded of Tarone’s argument (1979) that individuals use different language varieties in different situations and that no systematic data are possible until the characteristics of the situa- tion are clearly specified.

My argument to this point has been that the unanswered questions arising out of theoretical formulations concerning second language learning and generated by new research developments-especially on language input and individual differences-have focused attention on the individual and the situa- tion and the interaction between the two. Now I would like to discuss a research paradigm for dealing with person x situation interactions. This research paradigm is not new; it has been used in psychological and educa- tional research for some years now. But it is new to the field of second lan- guage learning, where most research employs a single, univariate model. Theory and research, however, are beginning to require more complex conceptualizations, and it would be unwise not to utilize a research paradigm so well suited to our present needs.

Page 8: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

338 Language Learning Vol. 30, No. 2

THE INTERACTION RESEARCH PARADIGM

The interaction research paradigm is a particular style of conducting research, one that focuses on the problem of interactions between person and situational variables. This paradigm has its own methods of design and statis- tics, most adequately discussed by Cronbach and Snow (1977) in their book on Aptitude x Treatment Interactions (ATI).

The AT1 research model

By “Aptitude” Cronbach and Snow meant any characteristic of a person that forecasts the probability of success under a given treatment. “Aptitude” does not refer exclusively to characteristics predicted by “aptitude tests,” though such variables are included under this rubric. “Aptitude” is broader in the sense that it includes all person characteristics, including those I have been discussing-individual differences in learning styles, cognitive styles, and the like.

By “Treatment” is meant situational and manipulatable variables, such as pace, method, or style of instruction. An interaction exists when a situation (Treatment) has one effect on one kind of person and a different effect on another (see Figure 1). Cronbach and Snow saw the identification of inter- actions to be the prime goal of science:

The whole process of seeking order in behavioral and biological science is one of partitioning a grand matrix of organisms and situations into blocks in such a manner that a single generalization applies to all the organisms and all the situations classi- fied within a block. The science of human behavior is built up by identifying a class of persons who respond similarly to some particular range of situations (Cronbach and Snow 1977, p. 3).

Much applied educational and psychological correlational and experimen- tal research has ignored interaction effects. For example, correlational research in which an aptitude test or a battery of tests is used to predict success in an instructional setting ignores the possibility that the instructional setting or treatment may not be undifferentiated, but may affect different individuals differently. In most experimental research the researcher tries to remove the effects of individual differences by randomly assigning individuals to treatment conditions. The goal of such research is to determine which treatment effect can be applied to optimalize the gain for all individuals- though in point of fact there is probably no one best treatment for all individuals.

Page 9: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

McLaugh lin 339

High

Outcome measure

Low

Low High (Person A) (Person B)

Person variable

Figure 1. Example of Aptitude x Treatment Interaction. A person low on the Person (Aptitude) variable (Person A ) does better in Situation (Treat- ment) Y than in Situation X ; the reverse is true for a person who scores high on the Person variable (Person B).

Page 10: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

340 Language Learning Vol. 30, No. 2

The interaction research paradigm is intended to avoid these shortcomings by directing attention to the interaction per se. Rather than making cutoff decisions on the basis of the main effect of the treatment variable, interaction research aims at finding interactions and capitalizing on them. This difference can be seen in Figures 2 and 3. In Figure 2 the investigator’s decision is to use Treatment A for all persons since the mean outcome for Treatment A is higher than for Treatment B. An example might be scores on a performance test (Outcome) after experiencing two different teaching methods (Treat- ments A and B). Since those persons receiving teaching method A did better, the assumption is that that method is better for all persons.

In Figure 2 the investigator assumes that the regression lines for Treatment A and Treatment B are parallel (dotted lines). If in fact the regression line for Treatment B is steeper than for Treatment A, as in Figure 3, a different decision should be made. In this case those persons scoring above the point X (which is the point at which the two regression lines cross) should be assigned to Treatment B, since persons scoring above X do better in this treatment condition. Those scoring below X should be assigned to Treatment A, since persons scoring below X do better in the Treatment A condition.

The advantage of this last procedure can be seen by comparing the Out- come scores for the two cases. When all persons are assigned to Treatment A on the assumption of parallel regression lines, as in Figure 2, the mean for A becomes the score for all persons. If some persons are assigned to Treatment A and some to Treatment B on the basis of scores on the person variable, as in Figure 3, the mean for the group assigned to A is slightly lower than it would be if all persons were assigned to A, but this is more than compensated for by the mean of the group assigned to Treatment B.

In other words, the interaction research paradigm aims at selecting those treatment methods that optimize the outcome for the persons involved. The assumption underlying this procedure is that by matching learner traits to learning methods one produces significantly better learning. There is no simple and sovereign solution for all learners-instead, our instructional proce- dures should be varied enough to take individual differences into account.

To investigate interaction effects Cronbach and Snow (1977) urged the use of generalized regression analysis. This technique of data analysis permits analysis of the multiplicity of influences of a number of variables dn a depen- dent variable, since it yields measures of the unique contribution of individual variables, or interactions between variables, t o total variance (Cohen and Cohen 1975, Kerlinger and Pedhazur 1973). The availability of computer programs for regression analysis puts the statistical techniques for interaction

Page 11: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

McLaughlin 341

Outcome measure

Gain

Mean of all persons

Person variable

Figure 2. The figure shows the gain that results when all persons are as- signed to Treatment A rather than to Treatment B on the assumption of parallel regression lines (Adapted from Cronbach and Snow 1977).

Page 12: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

342 Language Learning Vol. 30. No. 2

Outcome measure

I Mean for persons assigned to B

I

assigned to A

7

I Mean for persons

I I

Treatment F A

Mean for X Mean for persons below X persons above X

Person variable

Figure 3. The jigure shows the situation where the regression lines for Treatment A and Treatment B are nonparallel. Point X is the point at which they intersect. Persons below point X on the person (Aptitude) variable are assigned to Treatment A and those above point X are assigned to Treatment B. The points inside the triangles represent the means for all subjects in Treatment A and Treatment B and have the same values as in Figure 2 (adapted from Cronbach and Snow 1977).

Page 13: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

McLaugh lin 343

research within the reach of almost all investigators. One advantage of regression analysis is that it is in certain respects more

powerful than conventional procedures. Whereas analysis of variance proce- dures require independent variables that are classified into nominal scales, regression analysis permits the use of quantitative scales as independent variables. That is, the researcher is not forced to group or “block” subjects into high and low categories on the basis of scores on some person variable, but can treat the independent variable as a continuous quantitative scale, thus not losing information and statistical power by setting up arbitrary categories.

An example may make this clearer. Researchers often measure person variables on the basis of aptitude tests. In order to carry out an analysis of variance, subjects are then divided into high and low groups on the basis of test scores, e.g., high or low ability in language learning, and are given separate treatments, e.g., instruction method A or B. Note that this procedure groups some individuals whose scores on the person variable differ little into separate groups and puts into the same group some individuals whose scores differ greatly. Regression analysis tests the difference in regression slopes and does not require bloclung subjects into groups. Hence regression analysis will identify as significant some interaction effects that are nonsignificant when analyzed by a blocked analysis of variance design.

Another advantage of regression analysis is that it allows the investigator to get at the complexity of relationships that characterizes actual behavior. The statement that a person high in Aptitude C does better on Treatment A than on Treatment B, while the reverse is true for a person low on Aptitude C, does not tell the whole story. It may be that only persons high on Apti- tude C and low on Aptitude D do well on Treatment A. Regression analysis makes possible the use of statements of the form Y = f (A, B, C, D, E, A x B, A x C x D, etc.), where Y is the criterion variable, A and B are situational variables, and C, D, and E are person variables. In this way a researcher is not forced to sum up a person on the basis of one or two variables, but can begin to approximate the complexity of persons, situations, and their interactions.

Uses of the interaction research model

Cronbach and Snow (1977) reported extensively on the use of the inter- action research paradigm in educational research. I will briefly give a few examples here and will mention some problems that arise in attempting to use this research paradigm. In the final section I will turn to implications for

Page 14: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

344 Language Learning Vol. 30, No. 2

research in second language learning. In a study of the effects of student- vs. experimenter-paced programmed

instruction, Gallegos (1968) manipulated tape-recorded instructions on the Spanish passive voice to 110 high school students. Students were either given slow or fast taped instructions or could go at their own pace. There was a nonsignificant trend in the data for students with high IQs to do better with self-pacing and those with low IQs to do better with slow, externally regu- lated pacing.

Maier and Jacobs (1966) investigated the effect of instructional variation on achievement in Spanish in a study in which 17 classes had a small-step ordered program of instruction and 22 classes had a “scrambled” program with no orderly progression in instruction. While the mean improvement was not significantly different for either program, there was a trend for classes with students of higher IQ to do better on the scrambled version and for classes with students of lower IQ to do better on the ordered version. Classes initially in favor of the idea of learning Spanish tended to do better on the scrambled version.

Two hundred eight sixth grade children were taught an artificial language by Carroll and Spearitt (cited in Cronbach and Snow 1977). The manner in which the rules of the grammar were presented was varied, with either several rules presented at once in a random manner or one rule at a time in an order- ly fashion. In the “random” group, errors were identified without explana- tion; in the “ordered” group, errors were explained. When the criterion of mastery learning-i.e., time to work through a 60-page instruction booklet- was used, students of high IQ were found to do better on the ordered version, whereas other students did better on the random version.

While these studies, and other studies like them, examined interaction effects, they are not ideal examples of the interaction research paradigm. They used only a single person variable-IQ-and did not employ regression analysis (because the techniques of regression analysis have only recently been refined for educational and psychological research). Carroll and Spearitt, for example, dropped out subjects at random within cells so as to achieve a balanced analysis of variance design. Regression procedures make this squan- dering of data unnecessary.

In spite of the logical appeal of discovering relationships between learner characteristics (Aptitudes) and learning situations (Treatments), empirical evidence for the utility of the Aptitude x Treatment Interaction approach for improving educational procedures is weak (Cronbach and Snow 1977). In addition to design problems and the failure to use powerful enough

Page 15: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

McLaughIin 345

statistics, Cronbach and Snow identified three problems in past research: (a) small sample size and, hence, chance statistical results, (b) the lack of adequate theoretical grounds for hypothesized interactions, and (c) ill-defined treatments of too short a duration.

Another particularly acute problem is that of measuring person variables. interaction studies are based on human traits for which the assessment tech- nology is quite primitive. Research has tended to identify people on the basis of socioeconomic status, sex, ethnicity, or IQ, rather than by functional characteristics such as cognitive style, motivation, and temperament. Perhaps the most important future development is the determination of those func- tional characteristics underlying human diversity that, interacting with specific treatments, influence learning.

Implications for second language research

The task of identifying functional characteristics of learners is as impor- tant for second language research as it is for other areas concerned with human learning. We are beginning to understand something about the nature of individual differences in second language learning and about language learning styles. Objective instruments to measure these styles are lacking, however. This means moving beyond impressionistic descriptions (e.g., of under- and over-users of the Monitor) and instead developing operational assessment procedures to measure differences in individual learning styles. The factor-analytic approach of Snow and Hoefnagel-Hohle (1979) is a good example of how reliable assessment techniques can be developed.

Identifying individual learning styles does not necessarily imply commit- ment to a static trait model. There is reason to think that generalized abilities that cut across situations are stable and enduring personality traits-logical ability, reading comprehension, spatial ability. Other abilities, however, are more task-specific and may in fact be changed by instructions and other situa- tional variables. It may be that some learning styles are of this second type. That is, some learners may use a particular learning style because it has served them well in the past, but may switch to another style when confronted with novel instructional techniques. For example, students who have approached language learning by memorizing the rules of grammar may adopt new strategies when the instructional situation stresses communication. In this case the treatment has worked back on the learning style, so that a dynamic model becomes more appropriate than one that views learning styles as static personality traits.

Page 16: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

346 Language Learning Vol. 30, No. 2

I have mentioned Tarone’s plea (1979) that second language researchers take care in reporting how their data were gathered and what precisely was involved in the experimental situation. This means, among other things, identifying the demand characteristics of situations-whether they require attention to form, whether they allow subjects enough time, whether they have a motivational effect. We also need to develop a typology of learning situations that takes into account communication pressures, social relation- ships, and the nature of input. Controlled research on teacher talk and on foreigner talk is indispensable if we are to further our understanding of the impact of input variables.

As progress is made in understanding the social situation(s) of second language learning and individual differences in language learning styles, research on their interaction becomes possible. The goal is a practical one: to use the findings of research to develop selection methods that optimize learning. This implies belief in the interactionist credo-that not all situa- tions (Treatments) are ideal for all individuals (Aptitudes) and that the task of research is to identify those methods that maximize the outcome for different individuals.

Some research of this nature has been conducted. For example, Zam- pogna, Gentile, Papalia, and Gordon (1976) found that second language achievement was positively correlated with a high level of conceptual develop- ment in a second language program that used individualized instruction but not in a more conventional, uniform program. Hamayan, Genesee, and Tucker (1977) reported that measures of conformity and control were associated with second language achievement in a conventional FSL program but not in a French immersion program. While it is premature to make policy recommendations on the basis of such research, consideration of the inter- actions between learner characteristics and instructional method is important in evaluating any bilingual effort. As Gonzalez (1977) has noted, one of the reasons bilingual education programs for Mexican-American children have mixed results is that the children are considered to be a homogeneous group for whom instruction in a prestige variety of Spanish is considered uniformly appropriate. The success of bilingual education programs will depend ulti- mately on identifying relevant person variables and determining how they interact with various treatment methods.

Swain (1977) has pointed out that theoretical models that take into account the variables that affect second language learning have become more complex as we have come to understand more about the processes involved and the different factors affecting learning. An adequate model, she argued,

Page 17: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

McLaughlin 34 7

must distinguish between the input situation, learner variables, the learning process, and what is learned. Each of these different aspects of second lan- guage learning involves sets of variables, so that the theory must portray language learning as the product of a multiplicity of interacting factors. As we have seen, unanswered questions arising out of our present theories require consideration of interacting person and situational factors.

The advantage of the interaction research paradigm is that it approximates more closely than present research procedures the complexity required by our theoretical models. In particular, the use of regression analysis allows us to make sense out of the multiplicity of relationships between independent variables and dependent variables. By skillful use of regression procedures we can determine whether a set of data for a dependent variable, Y, and two independent variables, A and B, are consistent with any of the following possibilities: (1) A and B bear causally on Y, ( 2 ) A is a surrogate for B in the relationship with Y-that is, when B is partialled from A, A retains no vari- ance in Y, and (3) B suppresses the effect of A on Y-that is, when B is par- tialled from A, the unique variance of A in Y is greater than the proportion it accounted for when B was ignored (Cohen and Cohen 1975).

Furthermore, using regression analysis we can look at complexity in the dependent variable per se. Language learning is not a monolithic concept- learning is multivariate. A person’s performance at one point in time can be represented by a set of scores describing various aspects of learning-vocabu- lary, auditory comprehension, pronunciation, discourse rules, and so on. If we are to avoid oversimplification, multivariate studies of the learning process are called for.

A good example of such research has been provided by Genesee and Hamayan (in press) in a study of individual differences in English-speaking children’s second language learning in an early immersion program. These authors used multiple regression procedures to assess the association between a number of predictor factors-including students’ attitudes, personality traits, nonverbal reasoning ability, degree of field independence, and school- related behavior-and achievement in French language arts, listening compre- hension, and oral production. The results indicated that different factors were associated with different aspects of language learning. For instance, field independence and nonverbal reasoning were found to be significant predictors of achievement in French language arts and listening comprehension, but not of oral production. This research supports the notion that learning involves multivariate processes and has important educational implications.

Such multivariate studies begin to get at the complexity of the phenomena

Page 18: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

348 Language Learning voz. 30, No. 2

involved in second language research. Regression analysis is especially helpful in this respect because it can be used with experimental or nonexperimental data. Regression analysis also relates closely to other methods, such as path analysis, structural relations, and related techniques used by sociologists and economists to test “causal models” for non-experimental data. These methods are becoming increasingly more frequent in educational and psycho- logical research and may be useful in unraveling the thread of causality in second language research.

In concluding, I would argue that the use of regression techniques and multivariate analysis is necessary if the field of second language research is not to become methodologically obsolete. This does not mean that all research need fit into this mold. Obviously we still have much to learn from careful longitudinal research with single cases, or small numbers of cases, where the application of statistics is impossible or limited. Nor am I advo- cating a single statistical method. Regression analysis is one approach to interaction analysis, but traditional statistical approaches-especially factorial analysis of variance designs-also test interaction phenomena. The research paradigm for which I have been arguing here directs our attention to the interaction between person and situation variables regardless of how this interaction is analyzed. The point is that we attend to these interactions and move beyond a univariate research paradigm inadequate for the complexity of the phenomena we are concerned with. Interaction analysis provides an orientation, a frame of reference, without which we cannot answer the questions that developments in theory and research are forcing us to ask.

REFERENCES

Adjemian, C. 1976. On the nature of interlanguage systems. Language Learning 26:297- 320.

Andersen, R.W. In press. Two perspectives on pidginization as second-language acquisi- tion. In Andersen, R.W. (ed.), New Dimensions in Research on the Acquisition and Use of a Second Language. Rowley, Mass.: Newbury House.

Bickerton, D. In press. Discussion of two perspectives on pidginization as second-lan- guage acquisition by Roger W. Andersen. In Andersen, R.W. (ed.), New Dimensions in Research on the Acquisition and Use o f a Second Language. Rowley, Mass.: New- bury House.

Brown, R. 1977. Introduction. In Snow, C.E., and C.A. Ferguson (eds.), Talking to Children: Language Input and Acquisition. Cambridge: Cambridge University Press.

Cohen, J., and P. Cohen. 1975. Applied Multiple RegressionfCorrelation Analysis for the Behavioral Sciences. Hillsdale, N.J. : Erlbaum.

Cronbach, L.J., and R.E. Snow. 1977. Aptitude and Instructional Methods: A Handbook

Page 19: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

McLaughlin 349

for Research on Interactions. New York: Irvington. Cross, T.G. 1978. Mothers’ speech and its association with rate of linguistic development

in young children. In Waterson, N., and C. Snow (eds.), The Development of Com- munication. New York: Wiley.

Farhady, H. 1979. The disjunctive fallacy between discrete-point and integrative tests. TESOL Quarterly 13 :347-357.

Fillmore, L.W. 1976. The second time around: cognitive and social strategies in language acquisition. Ph.D. Dissertation, Stanford University.

Freed, B.F. 1978. Talking to foreigners versus talking to children: similarities and differ- ences. Paper presented at 2nd Annual Los Angeles Second Language Research Forum.

Gallegos, A.M. 1968. A study and comparison of experimenter pacing and student pacing of programmed instruction. Journal of Educational Research 61 :339-342.

Genesee, F., and E. Hamayan. In press. Individual differences in second language learn- ing. Applied Psycholinguistics.

Gonzalez, G. 1977. Teaching bilingual children. In Bilingual Education: Current Per- spectives. Vol. 2. Arlington, Va.: Center for Applied Linguistics.

Hamayan, E., F. Genesee, and G.R. Tucker. 1977. Affective factors and language ex- posure in second language learning. Language Learning 27:225-241.

Hatch, E. 1974. Second language learning universals? Working Papers on Bilingualism

Hatch, E., R. Shapira, and J. Gough. 1975. Foreigner talk discourse. Working Papers

Henzl, V.M. 1979. Foreign talk in the classroom. IRAL 17:159-167. Kerlinger, F.N., and E.J. Pedhazur. 1973. Multiple Regression in Behavioral Research.

New York: Holt, Rinehart, and Winston. Krashen, S. 1976. Second language acquisition. In Dingwall, W. (ed.), Survey ofLin-

guistic Science. Stamford, Conn. : Greylock. Krashen, S. 1977. The Monitor Model for adult second language performance. In Burt,

M., H. Dulay, and M. Finocchiaro (eds.), Viewpoints on English as a Second Lan- guage. New York: Regents.

Krashen, S. 1978. Individual variation in use of the Monitor. In Ritchie, W. (ed.), Prin- ciples of Second Language Learning. New York: Academic Press.

Krashen, S. 1979. A response to McLaughlin, “The Monitor Model: some methodologi- cal considerations.” Language Learning 29:151-167.

Krashen, S., J. Butler, R. Birnbaum, and J . Robertson. 1978. Two studies in language acquisition and language learning. ITL: Review o f Applied Linguistics 39-40:73-92.

Krashen, S . , and R. Scarcella. 1978. On routines and patterns in language acquisition and performance. Language Learning 28 :283-300.

Maier, M.H., and P.I. Jacobs. 1966. The effects of variations in a self-instructional program on instructional outcomes. Psychological Reports 18539-546.

McLaughlin, B. 1965. “Intentional” and “incidental” learning under conditions to learn and motivation. Psychological Bulletin 63:359-376.

McLaughlin, B. 1978. The Monitor Model: some methodological considerations. Lan- guage Learning 28:309-332.

Meisel, J.M. 1979. Strategies of second language acquisition: more than one kind of simplification. Paper presented at Linguistic Society of America Annual Meeting, Los Angeles.

Oyama, S. 1978. The sensitive period and comprehension of speech. Working Papers on Bilingualism 16:l-17.

3:l-17.

in English as a Second Language, University of California, Los Angeles.

Page 20: THEORY AND RESEARCH IN SECOND LANGUAGE LEARNING: AN EMERGING PARADIGM

350 Language Learning Vol. 30, No. 2

Schneider, W., and R.M. Shiffrin. 1977. Controlled and automatic human information

Schumann, J. 1978. The Pidginization Process: A Model for Second Language Acquisi-

Seliger, H.W. 1979. On the nature and function of language rules in language teaching.

Selinker, L. 1972. Interlanguage. IRAL 10:209-231. Selinker, L., M. Swain, and G. Dumas. 1975. The interlanguage hypothesis extended to

Snow, C.E., and M. Hoefnagel-Hohle. 1979. Individual differences in second-language

Swain, M. 1977. Future directions in second-language research. Paper presented at the

Tarone, E. 1979. Interlanguage as chameleon. Language Learning 28:181-191. Virgil, N.A., and J.W. Oller. 1976. Rule fossilization: a tentative model. Language Learn-

ing 26:281-295. Wagner-Gough, J., and E. Hatch. 1975. The importance of input data in second language

acquisition studies. Language Learning 25 :297-308. Zampogna, J., R.J. Gentile, A. Papalia, and R. Gordon. 1976. Relationships between

learning styles and learning environments in selected secondary modern language classrooms. Modern Language Journal 60:443-448.

processing. I. Detection, search, and attention. Psychological Review 84:l-66.

tion. Rowley, Mass.: Newbury House.

TESOL Quarterly 13 :35 9-370.

children. Language Learning 25: 139-152.

ability: a factor-analytic study. Language and Speech 22:151-162.

First Annual Los Angeles Second Language Research Forum.


Recommended