This chapter has two objectives. The first is a critical examination of current psycholinguistic views on the nature of language and on the process of second language acquisition. The second objective is to propose a program of research designed to offer solutions to major problems in language didactics. The point of view adopted in this analysis is that of a psychologist interested in human communica-tion. The qualification of the analysis as a psycholinguistic one is intended to indicate such a focus and to distinguish it from other possible and potentially useful perspectives, particularly linguistic, ethnographic, and sociological.
More often than not, a "critical analysis" ends up being a nega-tive piece of work. Yet the meaning of "critical" is not solely "fault-finding or carping"; it also means "disposed to judge with care and precision." This review will find many things at fault, but it will spend more time in analyzing issues and suggesting constructive alternatives than in carping. I am also aware of a third meaning of "critical," namely, that which is "of doubtful result," even "risky" and "perilous." The extent to which this review avoids this latter charac-teristic, it is for the reader to judge.

Section 1 considers the complexities involved in the answers to three basic questions. What does it mean "to know how to use a language"? What does it mean "to be a bilingual"? What does it mean "to teach a second language"?
Section 2 presents an outline for the analysis of language teach-ing and specifies the factors to be considered from three points of view: the learner, the instructional process, and the sociocultural environment.
Section 3 is a critical examination of the role research can play in increasing our understanding of language teaching and language learning processes.
Section 4 outlines a major research program which is guided by the problems and principles discussed in the first three sections.

3.1 SOME QUESTIONS ABOUT DEFINITIONS

3.1.1  What Does It Mean to "Know How to Use a Language"?
     The changes which have occurred during the past decade in the field of linguistics are familiar to anyone who possesses a professional interest in the study of language and there is no advantage in review-ing these developments here. The question which is posed here re-quires an answer not from the field of linguistics but rather from its sister discipline known as "psycholinguistics." It is a question that pertains to language performance rather than language competence; or, more accurately perhaps, it is a question about communicative competence and performance.
    One possible answer to this question is provided by an examina-tion of language tests for they are designed and used for the expert purpose of assessing an individual's knowledge of a particular lan-guage. The most common and widely used language tests are com-posed of items which attempt to sample the significant units in specific linguistic fields such as vocabulary, phonetic contrasts, and grammatical relations. The rationale behind these attempts has been that in order to know how to use a language one must know a minimum number of discrete facts about that language and the major problem in test construction consists of making up an adequate sam-ple of discrete units which would be representative of the minimum number that defined "knowing" the facts of that language.
     The discrete point approach is still the major one in the language testing and teaching fields today; it can be shown to be seriously inadequate on both theoretical and practical grounds. From the practical point of view, its inadequacy is clearly demonstrated by the fact, well known to all teachers, that students who obtain relatively high scores on such language performance tests (also: achievement tests) do not necessarily at the same time also have the capacity of using the language in a communicative setting such as carrying out a free conversation with a native speaker. Similarly, individuals who are quite capable of communicating with natives in a variety of settings do not necessarily obtain high scores on such performance or achievement tests. Thus, performance on these language tests and the ability to make use of the language for communicative purposes are not necessarily related, indicating that the former is not a good measure of the latter. This lack of intrinsic relationship is clearly accounted for by the conceptions currently in vogue in linguistics, thus also demonstrating the theoretical inadequacy of the approach under consideration. According to these views, knowledge about a language does not consist of a finite set of facts that pertain to the surface structure of a language (which these tests attempt to sample). Current linguistic theory has clearly shown that a significant (if not major) portion which constitutes knowledge about a language is not even expressed at the surface level but lies hidden in its deep or underlying structure, and thus far no standard tests have been developed which measure this knowledge.
     One does not need a great deal of formal linguistic sophistication to convince oneself of the superficiality of surface features for an understanding of sentences. Consider just a few examples (based on Rutherford, 1968). The difference between the following two sentences

(1) It's a shame he never wins
(2) It's a game he never wins

is felt to be much deeper than what is indicated by the surface difference between "shame" and "game." Sentence (1) is obviously related to "That he never wins is a shame" in a way that (2) is not, and sentence (2) is related to "He never wins the game" in a way that (I) is not. The surface "it" in the two sentences are furthermore quite different as shown by the fact that one can say "Tennis is a game he never wins" but not "Tennis is a shame he never wins"; in (1) "it" is the extraposed "it," whereas in (2) it is the pronominal
To take another example, contrast the following two phrases:
 (3) the keys to the house that he bought
(4) the keys to the house that he brought

No information in the surface structure of these two phrases allows the correct deduction that the buyer did not bring the house and, most likely, did not buy the keys. Finally, consider the case where a "syntactophonemic" fact is realized by nothing more profound than a slight phonological contrast as in

(5) They're going to battle with their allies. /
(6) They're going to battle with their allies. /

In (5), "battle" is a noun, part of "go to battle," and "with" is the commutative "with"; in (6) "battle with" is verb + preposition. The result, semantically, nowhere to be found in the surface structure, is that in (5) the "allies" are truly comrades-in-arms, while in (6) the "allies" become the enemies to be defeated.
    The linguistic facts illustrated in these examples are explainable by generative transformational grammar, but many more facts about English are not. There is no guarantee that generative transformational grammar, or for that matter any other linguistic theory, will be able to account for all the facts about language which native speakers possess. This consideration, coupled with the reasonable assumption that what is not known cannot be explicitly taught, leads to the sobering conclusion that language teaching may never become an exact science, and of course, it is not now, or likely to become one in the foreseeable future. Similarly, and for the same reasons, it is not now possible to test adequately a person's knowledge about a language.
     It is quite a different matter, however, to ask the more specific question of the extent to which an individual is capable of using a language for this or that purpose. Can he carry out a conversation on the weather and on sports with a monolingual native speaker of language X? Can he write a scientific article in language Y about the biochemical synthesis of amino acids? Can he serve as an interpreter to a military officer engaged in the pacification program of villagers in country Z? And so on. Note that it would be a straightforward matter to assess an individual's performance on such specific tasks. Note, too, that such tests would reflect much more than the strictly linguistic component of the person's competence. It would include, to an unknown but not necessarily unknowable extent (this remains to be determined) his knowledge about the foreign culture, his ability to reason in communicative settings, and other things as well.
     The various considerations just discussed clearly show that the original question posed at the beginning, namely what is it to know how to use a language, is indeed a very complicated one for which no clear answer is possible at this time, if ever. The most reasonable strategy as a matter of practical business would seem to be that of defining specific communicative goals and of assessing the extent to which an individual is capable of meeting these goals. Clearly, these communicative goals will need to be specified in sufficiently detailed manner, since there is no guarantee of transfer effects across modes, settings, and topics.
     The implications of these arguments for the strategy of language teaching and language testing are profound and need to be spelled out. First, what is the function of present-day language tests, i.e. tests that are designed to sample knowledge about discrete units? From one point of view, these tests accomplish fairly well what they are designed to do: as long as FL courses restrict themselves to imparting a given amount of discrete surface facts about a language, these tests meet reasonably well the purpose of assessing how many of these facts an individual has acquired. One may take the point of view that at the beginning stages of FL study, this is a reasonable goal and, thus the available tests are adequate from that standpoint. Then, at more advanced stages of study, communicative competence and performance become the goals, at which time one can develop and use behavioral tests of the type discussed above. Of course, the content of these later courses would have to change in accordance with these different goals and presumably they would lay emphasis on cultural facts and on paralinguistic and kinesic features within sociolinguistically relevant settings. This kind of over-all strategy may be said to be representative of current ideas and practices in the FL teaching field. There are grounds to question the correctness and wisdom of this policy and we shall now turn to an examination of such criticisms.
     The first and perhaps most damaging counter argument is the notable lack of success that such an approach has had thus far. By the early 1940's the language teaching field had achieved two highly significant goals which amount to no less than technological breakthroughs. One was an advanced state of knowledge of the surface facts of language made possible by structuralist linguistics, which yielded among other things highly efficient methods of contrastive analyses between the native and the target languages as well as the ability of breaking up the known facts into small units for their ordered and preprogrammed sequencing. The other breakthrough came with the availability of tape recorders, visual displays, and mechanical devices for presenting the surface facts of language in an efficient manner. If these two advances had represented the "correct" method of teaching a FL, then the didactic problem would have been solved. Nothing further of great significance remained to be solved, and indeed, the twenty years since that time have brought only improvements of a minor sort chiefly consisting of cheaper and faster machines. What has been the result?
     Although an over-all assessment of the FL teaching program is not an easy business, its lack of success in imparting meaningful knowledge in sufficient proportions to generations of school children and college students is quite well known and hardly needs documentation (but see Carroll, 1968). Faced with this obvious failure, the FL teaching profession has been searching for an explanation. It has apparently found one in the "lack of student motivation" for learning FL, a topic which has become increasingly fashionable at conferences and in the literature. The cry of "how to motivate" a reluctant student body is now pervasive. But is this the answer to the problem?
     It may be instructive to reflect upon the fact that the one instructional setting where FL teaching has had a considerable amount of success is the "intensive" course as developed by the Army Language Schools in Monterey, California and still used there today as well as in the U.S. Foreign Service Institute run by the State Department and in the Berlitz Schools throughout the world. The intensive language course differs from the school type course in three important respects. One is the amount of concentrated daily individual instruction, which for intensive courses may run up to the limit of ten to twelve hours a day (but no less than four hours); another is the emphasis placed on the communicative use of the language which is begun immediately upon beginning study. The third factor is that intensive courses are typically taught to individuals who have been preselected for relatively high FL aptitude. Let us consider each of these differences, taking the last one first.
     FL aptitude as measured by a short (two to four weeks) trial course or by the Modern Language Aptitude Test (MLAT, see Carroll, 1965) is indeed an important factor affecting success, but its chief influence appears to be in determining the time required for attaining a given level of achievement. For instance, according to the tables developed by the Foreign Service Institute (see Cleveland, Mangone, and Adams, 1960, 250-251), it takes an individual with "average aptitude" 50 per cent more time to attain "sufficient proficiency in speaking a FL [French, Spanish, Swedish, German, etc.] to satisfy routine travel requirements" as compared to an individual with "high aptitude"; it takes him 33 per cent more time to attain "fluency" in a "difficult" FL such as Arabic, Vietnamese, or Thai. In general, given more time (in weeks or months) with some minimum sufficient daily instruction, the person with lower aptitude can attain the same level of achievement as the person with higher aptitude. Thus, the fact that FL courses in the schools do not preselect the students for high aptitude is not a sufficient factor by itself to account for the difference in success between that setting and the intensive course. Furthermore, even where such preselection for aptitude (and high motivation) likely takes place, and in addition a considerable amount of cumulative time is spent in language study, as is the case for FL majors in colleges, the success achieved in communicative ability is considerably less than in the intensive course setting (see Carroll, 1968). Nor is the total cumulative study time sufficient to account for the difference: the typical college graduate majoring in a non exotic language has accumulated in the order of 1,000 hours of study throughout his career as a student, as compared to 700 hours for the high aptitude student in an intensive course.
     Thus neither aptitude nor total accumulated study time can account for the superiority of the intensive type course over the school type course.' There remains the matter of the differential emphasis placed upon communicative skills, and it is here that the significant difference likely resides. Carroll (1963a, p. 1069) has stated what he terms a "rather commonplace conclusion" that "by and large, students learn (if anything) precisely what they are taught" and this is precisely the point. Time spent in learning the discrete surface facts about language, in habit drills, in "automatization" of mechanical skills, in reading classical literature, etc., is time lost in learning and developing the communicative use of language. The irony of the matter lies in the fact that the "New Key" approach, unlike the traditional methods in which reading classical literature was considered a worthwhile activity in itself, does not attach an intrinsic value to grammatical knowledge per se, but views habit drills as a means toward achieving communicative skills, yet it seems that these very activities are the chief roadblocks to attaining meaningful skills ("liberated expression" as it has been called).
     There may indeed be a problem in "student motivation," but to what extent is this problem caused by the method of teaching itself? Today is the generation of the "New Student" (see Chapter 2) who demands that his studies be "relevant" to his educational goals and aspirations. Habit drills do not apparently fall in the "relevant" category and the FL teaching protession must meet this new challenge if it is going to survive. Fortunately, the ultimate interests of teacher and student coincide. The question is a matter of change for the better. In subsequent sections we shall examine the possibilities for such an adjustment and the form it ought to take.
 

     An additional possibility not considered in the text lies in the difference between "massed" versus "distributed" practice. In general, the hitter is more advantageous than the former, thus favoring the school type situation, but this is by no means a simple matter since some type of massed practice may be superior to some type of distributed practice and there is the further complication of summer breaks during which considerable forgetting may take place. It should also be noted that intensive courses typically involve individuals that are more advanced in age than students in school, but this difference is not considered decisive; see Chapter 2.

3.1.2 What Does It Mean "to Be Bilingual"?
     One of the facts of "folk bilingualism" is the following curious no-tion that native speakers have: a foreigner who is capable of uttering a few mechanical sentences with good pronunciation and accurate syntax impresses them as being "bilingual" whereas someone who speaks their language with a strong foreign accent and lacks fluency, does not, despite the fact that the latter's knowledge is considerably greater than the former's. Accent, pronunciation, and fluency are given a disproportionate degree of importance by the nonprofessional judge. So is accurate syntax, and a foreigner who speaks a "bookish" variety of speech, even where the native speaker would normally use a colloquial style, represents to them a more impressive accomplishment than the vastly more difficult achievement of the bilingual who can adapt his style to the sociolinguistic context of the communicative setting. The professionals linguists, psychologists, teachers have also their own version of "folk bilingualism." Some require equal facility in two languages ("balanceness") before they are willing to talk of "bilingualism"; others consider any occurrence of linguistic interference as evidence of bilingualism even if knowledge about the second language is minimal (cf. Mackey, 1956); still others prefer to talk of an intermediary stage as "incipient bilingualism" (see Diebold, 1961).
     The question when is a person a bilingual is actually not a very interesting one and there seems to be no particular advantage in setting arbitrary limits for a definition. A more relevant concern is that of specifying the extent of an individual's knowledge about a second language. This specification, to be meaningful, would have to be in terms of the discussion in the previous section concerning what it is to know how to use a language. Such specification must include what Fishman (1 966a) has called the "bilingual dominance configuration": that is, in terms of media (listening, speaking, reading, writing), roles (comprehension, production, inner speech), formality levels (intimate, casual, formal), and domains (art, government, home, business, etc.). Mackey (1962) has provided an even more complex description which includes degree at various levels (phonological, lexical, semantic, etc.) and in different modes (listening, speaking, etc.), function (external contacts: home, school, correspondence, mass media, etc.; internal: counting, praying, diary writ-ing, etc.), alternation (rate and proportion over topics and inter-locutors), and interference (type and locus).
     It is clear that in view of such complex descriptions, two "bilingual" persons may possess quite different dominance configurations and what each of them knows about the second language may be quite different with conceivably little overlap. Statements about "over-all bilingual proficiency" are likely therefore to be quite misleading. This problem is particularly acute in situations where "bilingualism" is a prerequisite for employment. Hiring an individual on the basis of his performance on the usual language proficiency tests can be quite disappointing to the employer since performance on these tests may be irrelevant to the individual's capacity to function in the particular setting required by the job. Clearly what is needed here is a specification of the minimum functional skills demanded by the job requirements and an assessment of the bilingual's capacity to meet these needs in the second language.
     Psychologists have expended considerable effort in developing so called "indirect measures" of over all bilingual proficiency (see the review by Macnamara, 1967a). This work, while of some theoretical interest, has not so far proved to be very useful for the problem that is being considered here. In general, these studies have not paid serious attention to the validation of the psycholinguistic measures that have been used, in some cases relying on their "face validity" (e.g., latency of giving translation equivalent words), in others relating them to language-proficiency tests of the usual kind. Spolsky's (1 968a) work is unusual in this respect in that he shows the proper awareness of the problem of validation. While he recognizes that discrete point proficiency tests may be useful in the control of instruction, whether for diagnostic purposes or as measures of achievement, he is clearly in agreement with the assessment given in the earlier section that such an approach is inadequate to characterize overall proficiency:

"All of this suggests the impossibility of characterizing levels of knowing a language in linguistic terms, that is, as mastery of a criterion percentage of items in a grammar and lexicon. A more promising approach might be to work for a functional definition of levels: we should aim not to test how much of a language someone knows, but test his ability to operate in a specified sociolinguistic situation with specified ease or effect. The preparation of proficiency tests like this would not start from a list of language items, but from a statement of language function; after all, it would not be expected to lead to statements like 'He knows sixty percent of English,' but 'He knows enough English to shop in a supermarket' "(Spolsky, 1968a, p.93).
     Selinker (1969) has recently focused on a most interesting characteristic of bilingual proficiency which he calls "fossilization." This refers to interference phenomena which permanently characterize the speech of bilinguals irrespective of the age at which the second language is acquired or the amount of instruction or practice in it. These bilinguals thus speak a kind of "interlanguage" which is different from the speech of native speakers of the target language:

"It is my contention that the most interesting phenomena in IL [inter-language] performance are those items, rules, and subsystems which are fossilizable, and that investigators should concentrate on them, trying to relate them to the five processes listed above. [Also below.] If it can be experimentally demonstrated that fossilizable items, rules, and sub-systems which occur in IL performance are a result of the NL [native language], then we are dealing with the process of 'language transfer'; if these fossilizable items, rules, and subsystems are a result of identifiable items in training procedures, then we are dealing with the process known as the 'transfer-of-training'; if they are a result of an identifiable approach by the learner to the material to be learned, then we are dealing with 'strategies of learning'; if they are the result of an identifiable approach by the learner to communicating with native speakers of the TL [target language], then we are dealing with 'strategies of communication'; and, finally, if they are the result of a clear over generalization of TL rules, then we are dealing with the 'reorganization of linguistic material.' I would like to hypothesize that these five processes are central processes to second language learning, and that each process forces fossilizable material upon surface IL utterances, controlling to a very large extent the shape of these utterances" (Selinker, 1969, p.8).
     The suggestion that "interlanguage" is the relevant data for the study of bilingualism merits serious consideration even if one dis-agrees with Selinker's classification of established dialects as in-stances of it. I have proposed (see Chapter 4) a description of bilin-gual proficiency which is consonant with this suggestion. My "four dimensional model of bilingual description" includes the following three-way relationship as one of the dimensions:
 

Equation (1) refers to the dominance relation between the two languages (L1 and L2) and is given by a comparison of the bilingual dominance configuration, perhaps in such terms as discussed above. Equation (2) refers to the relation between the bilingual's skills in the second language (XL) relative to an idealized monolingual speaker of the target language (XL) and is coterminous with the concept of interlanguage. Equation (3) refers to backlash interference and focuses on the changes that may occur in the bilingual's native language as a result of his contact with the second language. The latter, when severe, may also develop into a kind of inter language so, perhaps, we should be speaking of an "inter language rela-tive to the native language" as well as an "inter language relative to the target language.
    Parallel to an inter language we must also consider the probability of an 'inter culture' (again two kinds, relative to the native and to the target culture). Concerning the latter, Peal and Lambert (1962) and Kolers (1963) speculate on the existence of a conceptual supra structure peculiar to the bilingual which enables him to draw upon experiences from two cultures. Peace Corps volunteers returning to the United States from overseas service have been reported to experience various degrees of feelings of estrangement as they found that "things back home just didn't seem the same any more."
    The Peace Corps experience does not disentangle the bicultural from the bilingual experience as producers of the inter cultural individual. However, there is other evidence to support the contention that bilingualism entails biculturalism. Lambert and his co-workers have for several years now gathered evidence which shows that as an individual becomes increasingly more proficient in a second language he experiences attitudinal changes involving his own as well as the culture of the target language, and that furthermore, these psychological factors may influence the individual's further progress in mastery of the language. I have suggested that an individual's atti-tudes towards the native and target cultures influence his bilingual proficiency in the manner shown in the figure below:

     Ethnocentrism" refers to the belief in the superiority in one's own culture. "Orientation" refers to the individual's stated purpose for studying the second language, and may be "instrumental" (e.g. use-fulness in work or travel) or "integrative" (e.g. getting to know the people better). "Compound" and "coordinate" bilingualism refer to the dominance configuration in terms of the three-way distinction discussed above in relation to inter language and inter culture. There are several hypothetical relationships depicted in the figure one of which is that very strong cultural attitudes (either "ethnocentric" [High] or foreign culture oriented [Low] produce bilinguals with strong inter lingual and inter cultural characteristics (Compound). "Coordinate" bilingualism which more nearly corresponds to the linguistic and cultural features of the idealized monolingual within either language culture community is attainable, according to the figure, only under conditions of moderately held attitudes. This type of individual can be said to be characterized by "bilingual schizophrenia" whereby he is virtually "two different persons in one. Ervin-Tripp (1967) has adduced evidence showing the personality changes such bilinguals undergo when switching from Japanese to English, and Howell (1968) has vividly described the interpersonal difficulties some bilinguals have in interacting in the "American way" while using Korean, and, of course, vice versa.
     This discussion on the meaning of bilingualism, what constitutes bilingual proficiency and the interrelations between bilingualism and biculturalism has pointed up the complexity of the problem viewed psychologically (the sociological aspects as discussed by Weinreich (1968) or Lieberson (1969) were not even mentioned) and under-scored the ludicrousness of a simple definition. Much of the discus-sion was based on "hypothetical" notions and there is clearly a critical need for empirical data based on careful observations.

3.1.3 What Does It Mean "to Teach a Language"?

     Hayes (1964 p.148) has stated that "one cannot teach what has not been described." Can we attribute the lack of success of the FL curriculum in our schools to the fact that as shown in Sec-tion 3.1.1, we have a long way to go before we understand in explicit terms what knowledge about a language consists of? Children learn their native language despite the fact that parents certainly do not know more about language than linguists or language teachers, but are children in fact being "taught" to speak? Our knowledge about the underlying structure of simple mathematics is quite explicit, certainly much more so than that of natural languages, yet school teachers often experience difficulty in teaching arithmetic, algebra, and geometry
     Thus it appears that one can learn without being taught and one may fail in trying to teach that which is well known and explicitly understood. The meaning of "teaching" as commonly understood can have associated with it two different claims, one of which is a strong one, the other weak. The strong version claims that the activity that is being acquired by the learner is contingent upon the specific environmental conditions arranged by the "teacher." Classical "conditioning" falls in this category, since the "response" of the subject is supposed to be directly elicited by the "unconditioned stimulus" provided by the experimental setting and the "conditioned stimulus" becomes attached to this response (or a slight variant of it) by virtue of its contiguous presentation within a highly restricted temporal span. "Operant conditioning" also falls within the strong claim category since it is asserted that a specific response (when it occurs) is strengthened by virtue of the presentation of a specific "reinforcing stimulus" presented in close temporal contiguity with the emitted response. Note that proponents of the strong claim prefer to use the term "conditioning" rather than "teaching" although it is usually asserted by them that the latter is to be understood by means of the former process.
    When dealing with complex activities, the strong claim has been maintained with the help of the notion of "chaining" or of "shaming" whereby the complex activity is broken up into simple identifi-able responses which are then said to be tied together by further conditioning in the case of "chaining" (e.g., programmed instruction, or learning to play a piece on the piano) or by "successive approximation" in which the reinforcement is provided only with changes of the original activity that bring it increasingly closer to criterion until it is finally reached (e.g., mastery of articulation of the sounds of a foreign language). It is clear that in the strong sense of "teaching" one could not teach something unless one knew exactly what it was. In principle, it is not essential that one have an explicit understanding of the subcomponents of a complex activity; one need only wait until the desired activity is emitted (or a reasonable approximation of it) and then reinforce it. Thus, in principle, it is not essential to know explicitly the structure of a language in order to be able to teach it (in the strong sense), but only to recognize appropriate instances of it (e.g., well-formed sentences, meaningful sounds, etc.). Thus it is conceivable that parents could teach language (in the strong sense) to their children even though they are not applied linguists. Similarly, it is conceivable that language teachers could teach (in the strong sense) a second language despite the fact that their explicit understanding of its structure and subeomponents is inadequate or even inaccurate.
      In practice, two highly crucial factors impose definite limitations on what is possible in principle, and in many situations these limitations reduce it to nil or close to nil. One of these is the fact that the desired activity is so complex and improbable that one may have to wait an indefinite length of time for it to be emitted spontaneously or solely through the efforts of the learner. Conditioning cannot therefore take place. The other factor is that when the structure of the activity is not well understood, and yet the activity is complex enough so that subgoals or approximations to it must be chosen for reinforcement, the subcomponents chosen to be reinforced may be either irrelevant or mutually contradictory, so that no significant progress will be made toward the ultimate goal. This may be the case in attempts at language teaching when the learner is provided with contradictory or erroneous rules or when irrelevant subgoals are chosen (e.g., translating skill, or knowing how to repeat learned sequences of questions and answers in the classroom). This is the sense in which the "good teacher," the "good textbook," the "good lan-guage program" are superior to their poorer equivalents, for they will avoid to a greater extent the pitfalls of irrelevancy and mutual con-tradiction. There is wisdom therefore in Mackey's (1965) insistence that effective language teaching must be based on accurate "language teaching analyses," for L is here that one must find the desired subcomponents that are to be reinforced positively, or negatively when the responses are erroneous such as incorrect hypotheses and generalizations which the student provides on his own. "Desired sub-components" will have several characteristics: (a) they should be simple enough so that the learner can acquire them within a reason-able amount of time without frustration or loss of interest; (b) they must be explicitly defined so that the learner knows exactly what he is to acquire; (c) they must have demonstrated relevance to the criterion goal and this relevance must be perceived by the learner; (d) they ought to have as wide a range of applicability as possible and be practiced within a sufficiently variable setting; (e) they must be internally and externally consistent so that they do not involve contradictory principles and habits; and so on. Many language teaching procedures can be shown to violate principles of this kind. Let us mention a few: (a) some pattern drills are made up of sets of utterances that are superficially alike, but which because their under-lying structure is very different, confuse the learner by suggesting similarities that aren't there; (b) some teachers avoid stating gram-matical rules and relations explicitly on the assumption that it is better if the learner discovers them for himself. The assumption may be correct in some instances, but in others it does not seem to be, as in cases where the learner forms an erroneous hypothesis (which then must be unlearned), a weak hypothesis (does not cover many of the facts), or none at all (because it is too difficult for him); (c) many activities in the classroom that appear superficially similar to the criterion goal are in fact not so, such as the question-answer pattern which involves only minimally, if at all, the requirement to formulate one's free thoughts; (d) some teachers insist on subgoals that have no demonstrable relevance to the desired goal, as for instance the insistence on accuracy in pronunciation beyond the demands of comprehension, or the insistence on well formed sentences in free speech (even natives do not meet this standard), or the use of particular dialectal or style forms that are unsuited for the learner's ultimate goals.
     It is evident that there are aspects about the strong claim that appear reasonable and useful for rendering language teaching an effective process, even if, in practice, such a stage may not have been reached in any particular program. Particularly useful are the notions derived from the concept of "successive approximation" whereby subcomponents and subgoals are chosen that have characteristics which promote the ultimate goal in an effective manner, as discussed. But there are other aspects of the strong claim which are not viable and these are the notions associated with conditioning and reinforcement. Conditioning assumes that there occurs some identifiable response for reinforcement, but when we deal with "higher mental" or cognitive processes the relevant response is not identifiable, in large measure because it is not observable by the teacher. It would be a serious mistake, bordering on folly, to assume that the overt response the teacher can observe is the knowledge or principle the learner has acquired (see Chomsky, 1959). Furthermore, principles the learner acquires which underlie his overt activities do not seem to be very responsive to reinforcements doled out by the teacher, and certainly are not "controlled" by him in any real sense of "control." The truth of the matter is that "what goes on in the learner's head" proceeds in highly mysterious ways completely unknown and unknowable to the teacher, and yet it is certain that it is these mental processes and not the overt response that represent the significant aspects of what is being learned. Therefore the concepts of conditioning and reinforcement, in the sense in which they apply to classical stimulus-response analyses, become in these situations completely gratuitous and useless.
     It is for these reasons that the strong claim, as presented, is untenable and a weaker one must be substituted for it. This concession should not be construed as a relinquishment of the belief in the ultimate effectiveness of the teaching process or even as a dilution of the teacher's role in teaching, for it will be seen that all the valid and useful aspects of the strong claim also apply to this weaker claim, and the only aspects which have been relinquished are those which in the strong claim were ineffective anyway (besides being invalid).
     In order to understand the implications of what it means to "teach a language," in the present sense of "teaching," we must introduce a few additional terms and definitions. This analysis is specifically based on Carroll (1 963b) and may be considered consonant with current views in educational psychology, even though these definitions are only approximate.

Learning the acquisition of a task whose performance could not be previously accomplished; also: Arriving at the understanding of a concept, rule, or principle previously not understood or known about.

     Transfer: The process whereby something learned in one situation manifests itself in another situation due to the inferred commonality between the two situations. "Transfer effects" are to be understood as "hypotheses about transfer," since in most situations the elements in common to two situations cannot be explicitly identified (see Chapter 4).

     Learning time: Amount of time spent in the act of learning. This should not be confused with "elapsed time," which includes time spent on such activities as sitting at the desk dreaming, "wasting time," looking for a book or sharpening a pencil.

      Aptitude: An inferred capacity which determines learning time under the most favorable conditions; the shorter the learning time, the higher the inferred aptitude. If the task is difficult and the learner's aptitude is sufficiently low, learning time may be indefinitely long. Aptitude is conceived to be specific to particular tasks and depends on possession of certain characteristics by the learner. These characteristics may be either innate (or genetically determined) or they may be dependent upon prior learning of a specific sort or prior exposure to certain situations.

       Ability to understand instructions: This is conceived of as dependent upon two factors: general intelligence and verbal ability. The first enters into the ability of the learner to infer the concepts and relationships needed for mastery of the task, especially when these are not explicitly spelled out for him (which, one might add, is the usual case in most complex learning situations). The second comes into play in the understanding of the language used in the instructions.

      Quality of instruction: The extent to which it is made clear to the learner what it is that he is supposed to be learning. This is to be understood to refer to highly specific elements within the overall learning task. Telling the learner that he is supposed to acquire say "a reading knowledge of this foreign language" says very little about just what he is supposed to be learning: does he learn the writing system first, then vocabulary items (which ones?), is phonology necessary (how is he to learn it?), which sentence patterns (from "simple" to "complex"?) What contrasts is he supposed to be paying attention to? How does he know whether he has understood the correct principles? It is evident that "quality of instruction" deals with such highly complex (and unsolved) problems as the identification of relevant contrasts, sequencing and choice of materials, the amount and type of practice needed at each level, and so on. It should also be noted that quality of instruction interacts with learner characteristics so that it does not have an absolute criterion. The question is to what extent has the task been made clear to a particular learner: his aptitude, previous knowledge, ability to infer relationships and understand instructions, all will influence the specific requirements for making it clear what he is supposed to be learning. It follows that a "standard" method of instruction which does not adjust to the learner's characteristics cannot be of high quality.

     Perseverance:  The time the learner is willing to spend in learning. This definition is not adequate since it is not clear what is involved in "willing to." Thus, are factors such as distractibility, frustration, anomie, disinterest, all of which may reduce perseverance, pertain to "willingness to spend time in learning" or are they beyond the control of the individual? What about the type of motivation for learning: is one kind of orientation (say "integrative" or "intrinsic") more conducive to perseverance than another kind (say "instrumental" or "extrinsic")? These are unresolved issues that need further study and elucidation.

     Opportunity to learn: Learning time allowed by the method of instruction and the environmental conditions. (Applies to substeps as well as to the overall task.)  Degree of learning can now be expressed as a function of the ratio of two time factors related to these definitions, as follows:

Degree of learning =  time actually spent/time needed

The numerator in this equation will be equal to whichever of the following terms is smallest: aptitude, perseverance, or opportunity to learn. The denominator will be a function of the ability to understand instructions and the quality of instruction. Let us explore some implications contained in this equation.

       Situation X Student motivation is adequate (perseverance), opportunity to learn extensive, and ability to understand instructions high (general intelligence and verbal ability): degree of learning will then be a function of aptitude/quality of instruction. Of two students placed in the same program, the one with the higher aptitude will learn more than the one with the lower aptitude. Or, alternately, the one with the lower aptitude will need more time to reach the same criterion.

      Situation Y: Student motivation is adequate, aptitude high, and ability to understand instructions high: degree of learning will then be a function of opportunity to learn/quality of instruction. Of two students placed in the same classroom, using the same text, practicing the same exercises, the one that is given more time will learn more than his colleague. Or, alternately, given the same amount of time for learning, the student who has the better text will learn more.

     Situation Z: Opportunity to learn is fixed and ability to under-stand instructions equal: degree of learning will be a function of perseverance-aptitude/quality of instruction. Two students placed in the same program may learn the same amount despite differences in aptitude since the less capable student may compensate by higher perseverance. The interaction between perseverance and aptitude on the one hand and quality of instruction on the other needs to be explored. For instance, which student is better able to cope with a poor quality of instruction, the one with high perseverance and lower aptitude, or the one with higher aptitude and lower perseverance? Or which of the several other factors is more effective in compensating for poor instruction: aptitude, ability to understand instructions, perseverance? Do students who are low in ability to understand instructions suffer more from poor quality of instruction than students with higher ability to understand instructions?
 
    Quite apart from the problem of interdependence, can the factors be measured independently? What constitutes "opportunity to learn" for students with differential ability to understand instructions? Can one specify in sufficient detail the task to be acquired when transfer effects are not well understood? How is one to assess the extent to which it has been made clear to the student just what he is supposed to be learning? How do we weed out time spent in active learning from "listening without paying attention"?
     There have been various attempts at predicting success in second language learning. On the basis of these studies (Carroll, 1965; Carroll and Sapon, 1959; Flaugher, 1967; Gardner and Lambert, 1961; Pimsleur, Sundland, and McIntyre, 1964), we can draw up the following table that shows the variance contribution of the various factors involved:

Grade-point average which includes intelligence and perseverance contributes up to 40 per cent of the total variance. Multiple correlations between grade-point average and aptitude can account for up to 52 per cent of the variance (Pimsleur, Sundland, and McIntyre, 1964). It should be pointed out that this table does not represent exact data but is rather an approximation based on various studies reported in the literature. Furthermore, only student vari-ables are considered here. These predictions do not assess the contribution of the quality of instruction; they allow the latter to vary randomly or assume that it is "constant." There are two reasons for this: one is that educators find it useful to be able to predict success on the basis of individual differences independently of the variations in instructional practices. The other is that assessing the quality of instruction has proved to be a refractory undertaking without notable success.
 
 
 

back to top

The learning task to be acquired (defined in terms of some criterion goal), the quality of instruction, and the opportunities provided for learning. Let us assess the degree of control the teacher has over these variables.

     The opportunities provided for learning consist mostly of the amount of time available for study (in months, years, or total cumulative hours) and the availability of opportunities for learning outside the classroom such as laboratories, travel to a foreign country, con-tact with native speakers, etc. These conditions are determined by the overall school curriculum, the FL program as a whole, and the sociocultural context. Thus, teaching English as a second language in the United States has associated with it very different conditions from teaching Spanish; teaching French as a second language in Montreal is quite a different matter from teaching French in Toronto or Calgary. Some colleges and high schools provide "a year abroad" programs, while others do not. Some students are willing to try out outside opportunities when they are available, others are not. And so on.
    The definition of the criterion tasks, the goals of the second- language course, are dependent to some extent on educational, eco-nomic, philosophical, and political factors outside the direct control of the teacher. But here the teacher can increase his influence by pluralizing the goals of the curriculum and offering various courses which have different purposes. The unifunctional character of the "New Key" approach in FL teaching with its exclusive emphasis on audio-lingualism has contributed to a significant extent to the decline of support of the FL curriculum on the part of students and parents (see arguments in Chapters 2 and 5). In the first place, such a single-purpose outlook is unresponsive to the needs of students, many of whom are either unsuited to develop meaningful speaking skills within the school program (given low aptitude and insufficient opportunity to learn), or have interests of a different sort (such as reading scientific literature) whose neglect by the FL program (if not outright deprecation) not only demonstrates a certain snobbishness on the part of some FL teachers, but also makes them the unwitting accomplices of those who would reverse the present trend towards universal FL instruction in our schools (since they contribute to the general dissatisfaction with what the FL curriculum accomplishes or fails to accomplish). By organizing a multipurpose curriculum composed of specialized courses each with a specific and limited goal, the FL teacher would increase the effectiveness of his teaching by the fact that he will thereby be able to make adjustments for those factors over which he has no or little control: perseverance, intelligence, aptitude, and opportunity to learn. Rather than try to "motivate the student" within the "standard" course (a futile and impotent attempt in most cases), the teacher can offer a course which the student wants-in which case there will be no problem of perseverance. Rather than be frustrated with the lack of success of many students in present courses the teacher can take pride in accomplishing more limited or different goals in courses suited for given aptitudes, intelligence, and opportunity to learn as determined by the composition of his students and the educational environment. Some fundamental rethinking on the definition of goals of particular language courses needs to be effected. It may be that we should not be thinking about language skills per se (listening, reading, etc.) but rather about limited communicative skills. Thus we might develop courses designed to enable an individual to "converse with a native on travel and shopping," or "to understand foreign movies," or "to be able to read newspapers," or "to listen to radio broadcasts," etc. These courses on "how to do something in a FL" may be as short or as long as their complexity warrants and the aptitude of the students requires, and students can take as many of these as their interest, aptitude, and time allow. For purposes of educational (e.g., college entrance) or employment requirements relevant combinations of such skills can be specified in accordance with the rational and justified (rather than arbitrary) demands of the situation. Finally, the quality of instruction is the major factor under the control of the teacher and it is here that the meaning of "teaching a language" assumes its most relevant aspect. As discussed previously, the essence of quality lies in adjustment the adjustment of the instructional activities to the student's aptitude, intelligence, and ability to understand instructions within the defined goal of the course and the available opportunity to learn. Clearly, such adjustment is most effective under individuated instruction. Self-instructional programs based on careful "language teaching analyses" approximate this feature, and their current development represents one of the most exciting prospects for the future of second language teaching. Small classes with a high teacher-student ratio also permit individuated instruction, and this characteristic is no doubt a major contributor to the success of intensive courses. From the previous discussion it should be clear that quality of instruction by itself does not determine the success of the course or the overall program. One may end up teaching in a highly effective way, but one may teach the wrong thing. One must distinguish what is being taught from how it is being taught. The latter deals with quality of instruction, the former with the definition of the criterion task. The success of the intensive courses is made possible by high quality of instruction, but what determines it is the choice of the criterion task which, as discussed in Section 3.1.1, is defined in these types of courses in terms of specific communicative skills rather than in terms of some nebulous references to "knowing the structure" or "speaking ability."
      We started out this chapter by referring to Hayes's statement that one cannot teach what has not been described. To this limita-tion we must add still another restriction, which is that one cannot teach without understanding the factors involved in the learning process. Our sketchy knowledge of what language is and our lack of understanding of how people acquire complex cognitive processes therefore represent very serious limitations to our ability to teach language. Either limitation alone would be sufficient to defeat the teaching process. Thus although we understand in complete detail how a computer learns (since we build the components and write the programs), yet because we do not know what language is, we have completely failed to teach computers how to use language. Similarly, although we know what simple mathematics is, yet because we do not understand how the brain functions, we sometimes fail to teach arithmetic or algebra. When it comes to teaching a second language we are working under a double handicap. And yet people do learn (second languages and mathematics and culture and other highly complex things), and although the amount and rate of learning are not always determined by the teacher, they nevertheless often are.  Compared to lower animals or to man-made automata, the human brain is an amazingly versatile and efficient learning device. Its ultimate limit is not known, or even that it has limits. The exact contribution of teaching to learning is not known. The capacity of the brain to learn to extract knowledge from the environment and organize it by far outstrips our capacity to teach. It is only when conditions to acquire knowledge are unfavorable (because of inadequate access to facts or lower intelligence or the presence of distractors) that the teaching process assumes significant importance. This importance stems from the fact that under such conditions the function of teaching is to counteract the disturbing factors by improving the accessibility of the facts (e.g., grammatical principles, useful vocabulary, contrastive presentations), by circumventing lower intelligence (e.g., clarifying instructions, giving more opportunity to learn), or by eliminating some of the distractors (e.g., identifying erroneous generalizations, rewarding the desired goals, changing attitudes). Thus, in the last analysis, teaching is a remedial or compensatory process, nothing more nor less.

3.2  LANGUAGE TEACHING ANALYSIS

This section bears the title of Professor William Mackey's book (1965), which is a careful and detailed analysis of the language teaching process. In the Introduction, Mackey describes his views on the problem as follows:

"Here we are chiefly concerned with the factors involved in language teaching, and only with language learning to the extent that it is a factor in theories of language teaching and language analysis. Good teaching must take the learning process into account since its very purpose is to promote good learning; but the one can and does exist without the other. They must therefore first be analyzed separately, for each contains its own complex of factors.
      "In the analysis of language teaching, it is essential to maintain a distinction between the method and the teaching of it, without forgetting the obvious relationship between them since one of the purposes of a language teaching method is to direct the teaching of the language.
      "A second distinction has to be made between the language and the method, between the description of the language as presented in grammars and dictionaries and the way this material is used in a particular language teaching method. Again it is important not to forget the relationship between both. All language teaching methods must be based on some knowledge of and about the language to be taught. The more that is known about the language, the more complete the method may become. But there are different ways of finding out about a language and of describing what it is made of; many of the differences rest on different ideas of what a language is.
     "We therefore have three distinct but related fields of inquiry: I Language, II Method, and III Teaching" (Mackey, 1965, Introduction).

The notions that were discussed in Section 1 are entirely consonant with Professor Mackey's thesis. For language teaching to be effective, in the compensatory sense in which the concept of teaching was defined, the teacher must have knowledge about what language is (linguistic description), about what it means to know how to use a language and to be bilingual (criterion measures and goals), and about the factors that affect the learning process. Mackey devotes a third of his book to language analysis, but here we shall not be concerned with that aspect of the problem. The purpose of this chapter is to introduce in outline form an interaction model between three sets of factors with which the language teacher must concern himself in order to be an effective teacher. This will set the stage for the discussion in the last chapter, which will propose a program of research designed to increase our knowledge of the language teaching process. The three-way interaction model is outlined below:
 

Let us specify, still in outline form, some further notions that must be considered under each of these headings:

LF. Learner factors

1.   Ablity to understand instructions
1.1  General intelligence
1.2  Verbal ability

2.   Aptitude
2.1 Components of FL aptitude
2.2 Factors that determine aptitude

3.Perseverance
3.1 Need achievement
3.2 Attitude toward teacher
3.3 Interest in second-language study
3.4 Attitude toward foreign culture
3.5 Ethnocentrism
3.6 Anomie and its resolution

4.  Learning strategies
4.1 Tendency to assimilate new knowledge
4.2 Resistance to transfer effects A.4.2.   Structural
                    A.4.2.2 Attitudinal
4.3 Study habits
4.4 Self-evaluation criteria

5.  Consequences
5.1 Intellectual functioning
5.2 Educational achievement
5.3 Personality
IF. Instruction factors

1.  Quality of instruction
1.1 Adequacy of linguistic description
1.2 Adequacy of cross-cultural description
1.3 Adequacy of paralinguistic and kinesic description
1.4 Strategies of presentation of facts
1.5 Teaching aids
1.6 Degree of individuated instruction and responsiveness to learner factors
1.6.1 Aptitude
1.6.2 Motivation and interest
1.6.3 Feedback
1.6.4 Ability to understand instructions
1.7 Teacher observation and evaluation

2. Opportunity to learn
2.1 Time available for study
2.1.1 Number and distribution of hours for the course
2.1.2 Cumulative total amount of time
2.2 Opportunities outside the regular classroom activities
2.2.1 Use of the language for specific purposes
2.2.2 Direct contact with natives

3. Transfer effects
3.1 Structural
3.1.1  Areas of transfer
3.1.2 Level of operation
3.2 Environmental
3.2.1 Decoding aspects
3.2.2 Encoding aspects

4. Criterion evaluation
4.1 Achievement tests
4.1 .1 Linguistic competence
4.1.2 Communicative competence
4.2 Definition of goals
4.2.1 Degree of competence
4.2.2 Area of competence
4.2.3 Nature of competence
4.2.3.1 Dialectal variations
4.2.3.2 Code switching
4.3 Strategies of gradation
SF. Socio cultural factors
1. Language loyalty
2. Linguistic composition
3. Biculturalism
4. Consequences
Let us now turn to a brief examination of these various topics.

3.2.1  Learner Factors

     Psychologists and educators have known for a long time that "active learning" is by far superior to "passive learning" and in the dis-cussion in Section 3.1.3 we have rejected the notion of teaching language through some automatic conditioning process. Both of these considerations point to the crucial role of "learner factors in language acquisition and to the importance of knowing just what the learner contributes to the learning process so that it can be taken into account in the teaching process.
     Intelligence is usually conceived of as the ability to learn and thus it is to be expected that I.Q. will be related to second-language learning as well, and indeed the evidence we have corroborates this expectation (see earlier discussion). This relationship, under condi-tions of school learning, appears to be in the order of 15 to 20 percent of the contributing variance. The importance of intelligence in second-language learning can be interpreted as stemming from the fact that the teaching process is incapable of making it completely clear just what the learner is to acquire. Thus the variance contribu-tion of intelligence can be expected to increase under conditions where instructional procedures are weak and amorphous and decrease under conditions where they are effective and well integrated. In this sense, the correlation between intelligence and achievement can be viewed as an indication of the quality of instruction. The 20 percent variance contribution just referred to thus indicates that within the present conditions of language teaching in schools, given the criterion definitions and goals, and distribution of aptitude and perseverance factors that now prevail, the quality of instruction is quite high and the amount of possible improvement fairly limited (not more than one-fifth). This conclusion, if valid, is quite significant, since it suggests that in order to increase the success of the FL curriculum in schools the major changes and improvements will have to come in the area of criterion definition and implementation, i.e., a change in what is being taught under particular conditions. Note that quality of instruction can be high that is, it is being made quite clear to the student just what he is supposed to learn while the success of the overall program can remain inadequate owing to lack of perseverance on the part of the student or to the choice of a criterion goal for the course that turns out to be not what was wanted (e.g., teaching a certain number of surface facts about language as opposed to some particular communicative skills). It should be realized that a change in criterion goal may affect the quality of instruction, since corresponding to a change in what is being taught there may have to be a change in how it is being taught. With these new goals, quality of instruction will have to be reevaluated, since there is no guarantee that our present knowledge will be equally effective in teaching these newer skills.
     Aptitude is another major factor and under present school conditions it appears to be more important than intelligence, contributing roughly one-third of the total variance. The two best known measures of FL aptitude for native speakers of English are the Modern Language Aptitude Test developed by Carroll and Sapon (1959; Carroll, 1965) and the Pimsleur Language Aptitude Battery (Harcourt Brace & World). According to Carroll, FL aptitude consists of at least the following four identifiable abilities: (a) Phonetic coding, which is "the ability to 'code' auditory phonetic material in such a way that this material can be recognized, identified, and remembered over something longer than a few seconds." Evidence shows that this is not a matter of auditory discrimination ability per se. "Most normal people apparently have enough phonetic discrimination ability serve them in learning a FL, and in any case, it is more a matter of learning the discriminations over a period of time than any fundamental lack of auditory discrimination which can readily be tested in an aptitude battery" (Carroll, 1965, p.96). Thus it appears that the popular notion of "having a good ear for languages" is an ability that does not depend so much on one's "ear" as on the brain's capacity to code and store for later recall auditory information of a phonetic type. (b) Grammatical sensitivity, which is "the ability to handle 'grammar,' i.e., the forms of language and their arrangements in natural utterances" and is measured by a subtest of the MLAT which requires the individual to recognize the grammatical function of words in various contexts using English sentences. (c) Rote memorization ability, which is measured by a "paired-associates test" that simulates vocabulary learning. (d) Inductive language learning ability, which is the ability "to infer linguistic forms, rules, and patterns from new linguistic contexts with a minimum of supervision and guidance." Unfortunately, this ability is not measured by the present commercial version of the MLAT
     The other commercially available aptitude test is the Pimsleur Language Aptitude Battery (Pimsleur, Sundland, and McIntyre, 1964), which differs from the MLAT chiefly in that it includes "interest" tests designed to index how eager the student is in studying a FL and the value he sees in such study.
      The factors that determine FL aptitude are not known. Apparently, however, it stabilizes around puberty and continues without much change throughout adult life. There have been few attempts at improving foreign language aptitude through coaching, but in at least one area, that of phonetic coding, they have been unsuccessful so far (Yeni-Komshian, 1962), although clearly this is an area that needs to be further investigated. Certain kinds of early language experience or innate capacity, or both, may be the determining factors, and in that case it is not expected that coaching later in life would have significant effects on aptitude. Nevertheless, if research can identify the kinds of early experiences that lead to the full development of innate potentialities, the educational system can better insure full exposure to such needed experiences.
     Even if FL aptitude proves to be refractory to improvement through special training, knowing its components can be used to great advantage in the FL teaching enterprise. It will be recalled that aptitude was defined as the amount of time needed to learn under best learning conditions. The subcomponents of FL aptitude are to a large extent independent, as shown by the low correlations between the scores on the subtests of the MLAT. Thus, an individual may be high on phonetic coding ability but low on grammatical sensitivity or rote memorization ability. This means that he will need relatively more time understanding and internalizing grammatical rules or acquiring vocabulary. The teacher who has information about the aptitude configuration of his students can adjust his instructional activities accordingly, speeding up the rate in one area, slowing down in another, or assigning remedial exercises as required. This type of "diagnostic" use of aptitude tests assumes the possibility of individuated instruction, but where this is not possible, selection procedures can be used where students with similar aptitude configurations can be placed in the same class. Furthermore, information about the aptitude configuration of students provides rational criteria for decisions about criterion goals: for instance, faced with a student who has low phonetic coding ability but adequate grammatical sensitivity, the teacher may decide to place him in a course that emphasizes reading skills unless quality of instruction and opportunity to learn are sufficiently high to compensate for the lack of aptitude in the audiolingual area.
     Once more it appears quite clear that effective teaching consists of adequate compensatory activity on the part of the teacher. This principle applies equally to the area of perseverance. Anxiety about "how to motivate the student," which is a question that an increasing number of frustrated FL teachers ask the psychologist expert, is really not a well motivated concern. It shows a lack of understanding of the factors that contribute to perseverance, i.e., the student's willingness to spend time in learning. What are these and are they under the control of the FL teacher? (a) Need achievement:

    The degree to which the student strives for accomplishing goals in life. This characteristic is deeply intertwined with the socialization process that is linked to parental attitudes and demands, their education, religion, and class, and has probably stabilized long before the student takes a FL course. (b) Attitude toward teacher: not much detailed knowledge is available here, but questionnaire surveys on student attitudes show that the student can and does make a differentiation between the instructional activity per se and the instructor: he may like one but not the other. In our current educational system, the old idea about the positive force of "emulating" the popular teacher sounds rather improbable. (c) Interest in second-language study.' it was mentioned above that the Pimsleur Language Aptitude Battery includes "interest" tests and indeed, they have predictive value. But the relevant question here is, again, to what extent is this under the control of the FL teacher? A student who is not interested in developing audiolingual skills, but might be in learning reading, is unlikely to have his mind changed by the teacher. In fact, the evidence available on this (Politzer, 1953-1954 and Chapter 5) shows the students' negative attitudes are actually strengthened, not weakened, by being forced to learn aspects of language in which they are originally disinterested. (d) Attitude toward foreign culture.' these pertain to the larger sociocultural context, the attitudes that prevail within the community. The study of German in the United States decreased markedly during the years of World War II and thereafter; Spanish has increased in importance with the influx of Mexican and Cuban immigrants; so has Russian in the post-Sputnik era, and when the figures are in, it would not be surprising if French has declined, or at least its ascendancy checked, by what Americans perceived as the anti-Americanism of the DeGaulle era. Similar analyses would no doubt held up for bilingual countries such as Canada. (e) Ethnocentrism and anomie.' the arguments indicating that these factors are influential in FL study have been discussed in Section 3.1.2. As indicated there, much research is needed to disentangle hypothesis from fact, but it is at least possible that the teacher might be able to influence the outcome of these feelings were he sufficiently aware of them.
     It can be seen that the problem of how to handle the unwilling" student is not one that the FL teacher is likely to solve by his own activities in the classroom. There is one important area where the teacher does have some influence on perseverance and that is the extent to which he is responsive to the factors just considered. The politician's time tested adage "If you can't lick 'em, join admirably expresses the attitude of the FL teacher who is "aware and "with it" in this age of the New Student.
     The final set of "learner factors" to be considered deal with the various learning strategies that an individual uses in any learning task. Bugelski (1956), among others, has advanced the plausible argument that much of an adult's "learning' is really finding out for himself strategies of responding on the basis of what he already knows. "Child learning" is mostly concerned with building up "behavior elements," "reaction systems," or cognitive schemas; adult learning consists of setting up connections between these "building blocks." The process of generalization or transfer is the use of previously available strategies in new situations. In first language learning the child uses strategies made available by innate capacities and prelinguistic experiences to organize the linguistic facts to which his environment exposes him. In second language learning the available strategies the individual can call upon to organize the new set of facts are even more numerous, but this is a sword with two edges: some of these strategies will prove helpful in organizing the facts about the second language but others, perhaps owing to superficial similarities, will be misleading and inapplicable. In the former case we can speak of "positive transfer," in the latter case, "negative transfer." Contrastive analyses between the native and target languages represent explicit attempts at identifying the points of positive and negative transfer, although it is not yet clear whether surface features or deep structure features are the more relevant comparisons for assessing the operation of transfer effects.
The problem of analyzing transfer effects in language learning, sometimes considered under the topic of "interference," is a very involved one and has been reviewed in detail elsewhere (Chapter 4). Some aspects of this problem relate to activities of the teacher, some to the environment, some to the relation between the two languages, some to the learner. With respect to the latter, individuals differ in the extent to which they tend to "assimilate" new knowledge and facts to what they already know, and thus they may exhibit different patterns of positive and negative transfer effects. In addition, there appear to be attitudinal factors specific to language learning and usage that affect the extent to which assimilative (as opposed to separative) strategies will be used. For instance, individuals with low tolerance for ambiguity would tend to resist linguistic interferences that might lead to errors of confusion in the target language (e.g., failing to render an articulatory feature which might confuse two phonemes in the target language). Similarly, loyalty to the recipient language and an intolerance adopted from the community for linguistic importations ("linguistic purism") may increase the individual's resistance to certain interferences.
    Another learning strategy that varies among learners relates to the pattern of their learning activities outside the immediate supervision of the teacher. This is sometimes considered under the topic of ''study habits'' and includes such things as regularity, ''cramming order of carrying out exercises, note taking, etc.
    Finally, the question of self-evaluation criteria can be viewed as falling within the topic of learning strategies. Experience has shown that students have definite ideas about how well they are progressing in a course and when asked to give themselves a grade their estimate correlates reasonably well with the teacher's grade. In language courses, however, students appear to have certain misconceptions about what it is to know a language (misconceptions often shared by the teacher!) and these erroneous beliefs may lead to some difficulties. We have already discussed a few of these misconceptions in Sections 3.1.1 and 3.1.2, some of which were referred to as "folk linguistics" or "folk bilingualism." An individual who is used to the facile and apparently effortless flow of speech in his native language may underestimate the tremendous complexity of knowledge that underlies speaking and view with great frustration the halting and excruciatingly difficult productions in the target language. Once he gains a proper appreciation of the complexity of the knowledge he is being asked to acquire, he is more likely to accept the difficulty he is experiencing in the learning task.
 
 

3.2.2 Instruction Factors
 
    We have already discussed in Section 1.3 and elsewhere many of the variables listed in the outline under this heading and only a few additional comments will be made here. It is important to realize that the quality of instruction in any course is not under the full control of the teacher, and in language courses the contribution which the teacher makes towards quality might be negative rather than positive. This may seem a rather harsh statement but consider the argument: the primary determinant of quality of instruction the extent to which the learning task is made clear to the learner is the adequacy of description of the criterion task and its subcomponents; for this, the language teacher is almost totally dependent on the available texts and teaching materials, and these in turn reflect the state of scientific knowledge in that discipline. Even if we had available today which we have not adequate descriptions of grammars, of semantics, of culture, etc., it would still be a long way from that point to knowing how to teach these facts to students. At any rate, the quality of instruction (as defined on p.95) can never be higher than the state of knowledge as contained in the best texts available at any one time. The probability of improving on these texts on the part of the classroom teacher is slight indeed. His contribution to quality, therefore, is problematic: at his best he faithfully reflects what is in the best texts; at any other time, he simply detracts from quality by departing from the text or by supplying his own description of the facts. It is true that there is a timing factor also to be considered and the best available texts may not reflect the latest advances in a scientific discipline; in that case, the knowledgeable teacher has the opportunity of making a positive contribution to quality. Nevertheless, the argument presented here remains essentially correct for the majority of cases. (See 3.4.3.5).
     There is an indirect contribution to quality which the teacher can make and which will now be considered. This has already been expressed earlier when it was argued that teaching is essentially a compensatory activity. To be at his best and to reach the quality contained in the best available texts, the teacher must be responsive to the learner factors considered in the previous section. This is clearly the function of the individual teacher, since textbooks and teaching materials are geared to the "ideal" or "average" student, and except for self-instructional texts or programs, they do not take into account individual variability. What is involved in "being respon-sive to learner factors" has already been discussed in the previous section.
     There are a number of additional variables that are included under the heading of "instruction factors" in the outline; some of these have already received attention throughout this and the pre-vious chapter, others will be considered further in Section 3.4.

3.2.3 Sociocultural Factors

     The focus of the present analysis is not sociological or sociolinguistic. There are excellent treatments of the problem from that point of view in Weinreich (1968), Fishman (1966b), and Lieberson (1969) among others. Nevertheless it behooves us to consider the implications of the sociocultural environment for learner and instruction factors.
      Language loyalty and maintenance activities of a community can have important effects on a number of variables: early exposure to certain language experiences that might affect FL aptitude; interest in second-language study, attitude toward the target culture, ethnocentrism, and anomie, all of which may affect perseverance; attitudes toward language "purity" which influence learning strategies such as resistance to interference, inter language, and interculture; ideas pertaining to "folk linguistics" and "folk bilingualism" which affect self-evaluation criteria and criterion goals; pressures towards bilingual education and bilingual requirements for employment; available opportunities to learn; environmental conditions that affect the nature of attained bilingualism (e.g., along the compound-coordinate continuum); the nature of the communicative competence that is set as a goal and which includes dialectal variations, code switching, and the bilingual dominance configuration; and perhaps still others. All of these were discussed in one form or another in the preceding pages, and although each of them merits further consideration, this is not possible within the limited confines of the present analysis.

3.3 STRATEGIES OF RESEARCH

     An important purpose of the present analysis is to outline a program of research designed to increase our understanding of the language acquisition process and of the phenomenon of bilingualism. Such an understanding is a prerequisite to effective teaching. But because it seems that the nature and function of research in the social and behavioral sciences have been widely misunderstood, there is need for a critical examination especially in the light of recent develop-ments in the philosophy of science.

3.3.1 The Functions and Limits of Research

      The revolution in linguistics that was sparked by Chomsky's publication of Syntactic Structures in 1957 is now a dozen years old and in the modern scientific world this represents a substantial amount of time. (It is said that scientific knowledge doubles every decade!) We can now view this revolution with some historical perspective. In the field of linguistics proper, almost all of Chomsky's original proposals have now been challenged by his students and colleagues. "Generative transformational linguistics," the name by which the new field is known, offers today a picture of fantastic confusion, marked by "in-groupishness" and highly esoteric debates in which only a hand-ful of people are the active participants, while an army of devotees sit by the side line waiting for the newest pronouncements from half a dozen centers that come almost on a monthly basis in hard-to-get preprints and dittoed papers. The ultimate outcome of these activities cannot be predicted at the moment, but there is no question whatsoever that they amount to a scientific revolution of major historical importance that has reverberated throughout the "social sciences" and continues to do so unabated.
       It is now clear that it is not the detailed proposals advanced by Chomsky and his followers to which this revolution must be attributed, but to the philosophical implications of those proposals for the conduct of scientific research in the social and behavioral disciplines. Historians of science will no doubt refer to this period as the waning of "The Age of Behaviorism" and the waxing of "The Age of Rationalism." Scientists outside the United States (and per-haps the Soviet Union) may be much less impressed by this "revolution" since the rationalist position has continued to permeate their philosophical outlook and it is chiefly in the United States that behaviorism has had such staunch and rigid advocates.
      A major distinction between the behavioristic and rationalist schools lies in the status of the primary explanatory concepts. In psychology, behaviorism is based on the notion that all behavior is analyzable into small units called "responses" and all of the brain's environment (external and internal) into "stimuli." Responses are allegedly built up into larger behavior repertoires by a process of conditioning whereby they become associated with stimuli (or rein-forcing stimulus conditions). Various subschools within behaviorism exist and there is disagreement among them on many points: how associations are formed and forgotten, the nature of the reinforcing stimuli, the degree of willingness to speculate on higher order conditioning and inferred mediation processes, the role of neurophysiological explanations, and so on. These differences, however, do not affect the basic empiricist position to which all behaviorists subscribe, that all that the organism knows is viewed as (a) stemming from its interaction with the environment and the stimulus-response contingencies therein and (b) describable in terms of responses either observable or inferred.
     It is a most curious fact, one which will no doubt puzzle future historians of science, that the weakest aspects of behaviorism lie precisely at the most obvious places. Any general theory of human behavior, if it is to be taken seriously, must attempt to explain the most widespread and obvious of human activities: speaking and thinking in all their commonplace manifestations: describing, reporting, joking, lying, planning, conceiving, realizing, willing, refusing, and so on. The most typical fact about human behavior is that it is not controlled (in any true sense of "control") by external stimuli but follows rather from highly mysterious. And unobservable processes in the mind (or brain, if you prefer). Behavioristic speculations whereby these mental processes are redefined in terms of stimulus-response contingencies within the brain have been not only totally ad hoc and gratuitously designed to retain "consistency" of explanations for overt and covert behavior, but have also totally failed. None of these attempts had any measure of success, as has been demonstrated by a number of reviews (e.g., Chomsky, 1959; Katz, 1966; Fodor, 1968; Jenkins, 1968; Deese, 1968).
     Despite the obvious theoretical shortcomings of behaviorism it has occupied a position of dominance in American psychology since the second decade of this century. This is puzzling enough; but even more amazing is the fact that it continues in this position of dominance now that its limitations are understood. One would have expected a collapse of the total enterprise, nothing short of a debacle; but, aside from a few defections from some noted Behaviorists (Jenkins, 1968, Deese, 1968) and some spirited polemical exchanges (Fodor, 1965, Osgood, 1966, Berlyne, 1966, Bever, Fodor, and Garrett, 1968), the position of behaviorism has not changed either in theoretical outlook or in disciplinary influence. How can that be?
      Some ready explanations immediately come to mind, but none of these are really profound and convincing. It is said, for instance, that American Behaviorists are too rigid and entrenched in their views to change their position, or that no competitive theories are available to replace the vacuum that would be created if behaviorism were abandoned. The first hypothesis, besides being offensively condescending, is quite contrary to fact: the history of American psychology, is characterized by functional eclecticism, which is in harmony with the larger American ethos of placing value on change and eschewing traditionalism. The second hypothesis is also contrary to fact for although behaviorism has been the dominant school, it has not been the only one and the existence of these other approaches, both American and European, have been known.
      There is another sort of explanation, one which is more profound and goes deep into the nature of the social sciences, and which if valid, has very serious consequences not only for Behaviorism as a school within one discipline but for the totality of the scientific enterprise itself. According to this view the continued strength of behaviorism is attributable to its practical success, the power it has given to the manipulators of behavior: the educator, the animal trainer, the psychotherapist, the advertiser, the propagandist. The achievement of behaviorists in these areas is undeniable: programmed instruction, the Skinner box, behavioral modification based on operant conditioning principles, attitude change, to name those that are best known and most significant. How can a school of thought which appears on theoretical grounds naive, simplistic, and inadequate in the extreme lead to practical accomplishments in the area of human behavior that are so powerful and effective?
     We are faced here with a tremendous paradox wherein might lie the bankruptcy of the modern social scientific enterprise: what seems to work best on practical grounds, should not, it seems, on theoretical grounds, while the more sophisticated and "powerful" theories do not enable us to develop effective practical approaches. There is obviously something very wrong with our notions of "theory," "science," and "research."
      Careful analyses of the specific claims of one subschool of behaviorism, that of operant conditioning, has shown that they are not applicable to precisely those areas where that school has achieved its most notable practical achievements. In other words, "theory" seems to be an incidental adjunct to practice, and if the latter is effective it appears to be so despite the theory rather than because of it.2 This argument is strengthened by a parallel observation in another area, that of FL teaching. The development of highly successful techniques for teaching audiolingual skills in the intensive wartime courses of the U.S. Army Language Schools was achieved, as far as is known, without the theoretical precepts of any school in psychology. Subsequently, a term, the "New Key Approach," was invented as these practices were adopted in our FL curriculum and the "theorists" went to work to justify these practices from the behavioristic perspective. Now the behaviorists view these practices as an offshoot (and proof) of their theoretical claims. Similar arguments can be offered for the successful practices of the advertiser and the politician.
    What all this amounts to is this: human beings are tremendously more clever in doing things than in explaining how they do them. Scientific research is the organized and culturally sanctioned method of arriving at explanations. Here we find a difference in quality, not quantity, between the natural sciences and the social or behavioral sciences. The scientific concepts and explanations in physics or chemistry have led to degrees of control over the physical environ-ment which was by far superior to the degree of control afforded by man's ingenuity without the benefit of these theories. The "scientific" concepts and explanations in psychology (sociology, education, etc.) have led to far less control of the environment than the nonscientific intuitions of practical men. This difference leads one to question the equivalence of the scientific enterprise in the physical and social (or behavioral) sciences, with the resultant suspicion that the meaning of "research" in the two areas is fundamentally different.
The architect of the revolution in linguistics has this to say about the problem:
 

"Note that successful animal trainers have existed for centuries before Skinner ever invented the concept of "shaping by successive approximation."
 

"Modern linguistics shares the delusion-the accurate term, I believe-that the modern 'behavioral sciences' have in some essential respect achieved a transition from 'speculation' to 'science' and that earlier work can be safely consigned to the antiquarians. Obviously any rational person will favor rigorous analysis and careful experiment; but to a considerable degree, I feel, the 'behavioral sciences' are merely mimicking the surface features of the natural sciences; much of their scientific character has been achieved by a restriction of subject matter and a concentration on rather peripheral issues. Such narrowing of focus can be justified if it leads to achievements of real intellectual significance, but in this case, I think it would be very difficult to show that the narrowing of scope has led to deep and significant results. Furthermore, there has been a natural but unfortunate tendency to 'extrapolate,' from the thimbleful of knowledge that has been attained in careful experimental work and rigorous data processing, to issues of much wider significance and of great social concern. This is a serious matter. The experts have the responsibility of making clear the actual limits of their understanding and of the results they have so far achieved, and a careful analysis of these limits will demonstrate, I believe, that in virtually every domain of the social and behavioral sciences the results achieved to date will not support such 'extrapolation' "(Chomsky, 1968, Preface).
     Integrity dictates that we clearly recognize the inherent limita-tions of the social and behavioral sciences as we know them today and not claim for them more than is reasonable. Practitioners, such as teachers and therapists, need not feel that they must take sides in theoretical controversies and should desist from their attempts to justify their practices by appealing to particular theories. It is more relevant and important to have a functional attitude concerning the effects of their methods which leads them to developing and making constant use of evaluation criteria upon which they can base their decisions for maintaining or altering their activities in accord with the results they achieve. Researchers should not insist on the use of traditional experimental techniques in the analysis of real life situations where the term "control group" assumes a totally different meaning. Neither is it wise to restrict the scope of the research in deference to the maintenance of experimental techniques, for then the necessary extrapolation procedure vitiates the rigorous techni-calities of the laboratory.3
     The program of research outlined in the next section will attempt to steer clear of these obvious pitfalls. Given the limitations inherent in research of this kind, these activities should perhaps be referred to as "observation techniques" rather than "research." At any rate, the proposals are justified by the belief that systematic observation is a better technique for the discovery of facts than more intuitive approaches. Even this may not be true in some situations and for some gifted individuals but it is, at the moment, a plausible assumption for the majority of cases.

3.3.2  The Tactics of Research

     The figure on the following page outlines seven kinds of approaches in research which can usefully be differentiated.
     The first major tactical decision relates to whether the researcher intends to assess the effects of an attempt to manipulate conditions or to describe the variability of particular factors under existing conditions. There are two ways in which manipulation may be attempted: selection of naturally occurring phenomena and conditions and induction of some change that ordinarily would not be expected to occur in particular settings. An example of manipulation by selection is the control of the composition of a class in terms of some factor such as aptitude (cf. selection of students in intensive language courses). An attempt to improve grammatical sensitivity (a subcomponent of I~L aptitude) by teaching transformational gram-mar is an example of manipulation through induction of change.
      There are three basic approaches to description: the use of standardized tests, questionnaire techniques, and field observations.

" Postman has stated that "thus far, there is little empirical evidence in support of [the] assumption [that] forgetting outside the laboratory is a function of the same variables and represents the same processes as are involved in formal studies of interserial interference"
(1961, p.166).

     Standardized tests are as a rule commercially available, have known reliability, and the distribution of scores for defined populations (norms) allows transformation of scores attained by an individual into percentiles. The Modern Language Aptitude Test and the French Cooperative Tests are well known and widely used. Questionnaire techniques vary in structure and mode: the oral interview technique is usually not highly structured, although it may be, in which case it resembles written questionnaire forms. Both may vary in terms of "objectivity" which refers to the degree of bias that is likely to affect the scoring procedure. The nonobjective interview technique is useful in situations where the researcher does not have definite ideas about the information he is seeking and hopes that the unstructured interview will permit the spontaneous emergence of significant facts. On the basis of such information he may then proceed to the preparation of written questionnaires. If the variability of responses is limited and known, objective written questionnaire forms may then be prepared.
      Field observations may be either descriptive or evaluative. The degree of systematization of the observation technique can vary widely. In general, the more systematic the observation technique is, the more it can be reliable, unbiased, and specific. Hayes, Lambert, and Tucker (1967) have recently developed a highly systematic check list technique for the description of FL curricula along several dimensions (teacher activity in the classroom, administrative support, etc.). The Absolute Language Proficiency Ratings described by Rice (1959) (see also Chapter 4) is an example of a systematic evaluation technique for rating oral proficiency along a six-point scale.
     The focus of the observation refers to the primary goal the researcher has in mind and this will determine the kinds of statistical treatment imposed upon the data. The search for covariance relations most often occurs in situations where a large number of descriptive measures are obtained simultaneously. Usually, some suitable multi-variate technique (e.g., factor analysis) is used to reduce the large number of correlations to a few clusters whose intercorrelations are interpretable by means of an underlying mediating factor common to a set of variables. The correlation approach is also used in attempts at prediction by means of regression techniques. Basically, a correlation coefficient is an expression of the degree of common variance between two variables. (The percentage covariance equals the square of the correlation coefficient multiplied by one hundred.)
     The statistical technique known as "analysis of variance" enables the researcher to make statements about the amount of variance to be allocated to each of a number of variables that act simultaneously in a situation. Covariance information is also obtained.
      The preparation of norms for standardized tests involves descrip-tion of the distributional characteristics of scores within a defined population. Carroll's (1968) survey of foreign language proficiency levels attained by language majors near graduation from college is another example of a concern with the distributional characteristics of a variable, in this case proficiency level, in a given population.
      A focus on corrective measures is unlike the previous three con-sidered in that it represents a normative rather than a descriptive concern. Such a focus is particularly important in the light of our interpretation of teaching as a compensatory and remedial activity. The term "corrective measures" is intended to cover both perspectives in the teaching learning situation: What adjustments can the teacher make in his instructional activities to take into account learner factors and teacher evaluation results?, as well as, What kind of direct manipulation on the part of the teacher can lead to changes in learner factors?
     These are then the tactical features that make up the research strategy of the program outlined in the next section. It will be noticed that the problems that will be defined vary greatly on a number of dimensions: scope and range, specificity, objectivity and degree of systematicity, and the configuration of tactical features. It is obviously impossible to include all the problems that deserve the attention of the researcher, but an effort will be made to render the overall program comprehensive in terms of the three-way model summarized in the outline on pages 105-106.

3.4  A PROGRAM OF RESEARCH

The discussion in the first three sections of this analysis forms the background for the program of research now to be outlined. Therefore there is no need to justify in detail the rationale for each problem. Rather, the reader will be referred to the relevant parts of the previous discussion whenever this is deemed useful. Questionaire forms marked by an asterisk can be found in Chapter 5.

3.4.1.  Problem: To Assess the Learner Factors That Influence the Study of a Second Language

3.4.1.1 Reference

See discussion of "learner factors" in Section 3.2.1.

3.4.1.2 Tactics

3.4.1.2.1 Standardized Tests

(a)  Intelligence.. Stanford-Binet (verbal and quantitative parts), Primary Mental Abilities Test Battery (verbal, space, reasoning, num-ber, word fluency), and others.
(b)  Aptitude: Modern Language Aptitude Test and Pimsleur Language Aptitude Battery.
(c)  Perseverance.' Thematic Apperception Test (need achieve-ment), California F~Scale* and E~Scale* (ethnocentrism).
(d)  Criterion Measures: Most of these are of the discrete-point variety (e.g., the French Cooperative Test series) and are well known to language teachers.
(e)  Personality.' The MMPI is widely used, but there are many others well known to psychologists and there would be little ad-vantage to name them here.

3.4.1.2.2 Interview Technique

(a)  Management of Anomie.' Lambert has developed and used a questionnaire for the measurement of anomie in second-language study (see 4.1 .2.3.(a) below), but there is need for the development of questionnaire techniques to probe into the learner's attempts at handling anomie: When does it develop? Under what conditions? To what degree? How does the learner resolve it? etc. Frequent interviews in depth at different periods during language study can throw
light on this phenomenon and would permit the development of written questionnaires. (See Section 5.8.2).
(b)  Learning Strategies: * Because tests are not available and, as an aid in their preparation, frequent interviews with the learner can help determine his typical mode of learning: Does he use an assimilatory or a separative strategy? Is he consistent in this in various areas (phonology, syntax, vocabulary)? Does he switch strategies as his proficiency increases? Is he a linguistic purist? Is he a perfectionist? etc. (See Section 5.8.11).

3.4.1.2.3 Written Questionnaires

(a)  Attitude and Anomie: * We know enough from surveys used in previous studies (see Chapter 5) to prepare questionnaires designed to assess the learner's attitudes toward FL study; see also forms used by Lambert and co-workers on "Francophilia," "orientation," and "anomie," as well as indirect measures of "evaluational reactions" reviewed in Lambert (1967). The semantic differential technique (Osgood, Suci, and Tannenbaum, 1957; Snider and Osgood, 1969) is also well suited as a measure of attitude.
(b)  Study Habits.' * There may already be such questionnaires in "how to Study" books designed for students, but at any rate, their preparation should be a straightforward matter. (See Section 5.8.11).
(c)  Self-evaluation Criteria.' Interviews with students may need to precede the preparation of questionnaire forms to assess the learner's ideas pertaining to folk bilingualism.

3.4.1.2.4 Descriptive Field Observations

(a)  The School-type Setting: This approach may be particularly useful as a check on learner reports obtained through the questionnaire techniques: How does the student conduct himself in the classroom? Does he ask questions when the task is not clear to him? Does he show shyness in using the language? To what extent does he make use of the various opportunities to learn?
(b)  The Natural Setting.' How does the individual get by with little knowledge of the language and culture? Does he seek out or avoid contact with natives? What developmental patterns in language use are evident? etc.

3.4.1.2.5  Field Evaluations

     Because available tests of proficiency and achievement are mostly of the discrete-point variety, the development and use of field evaluations becomes highly crucial. The Absolute Language Proficiency Ratings (Rice, 1959) and the standards established by the Foreign Service Institute of the U.S. Department of State are examples of this type (see also Chapter 5, Section 1). But none of these are sufficient in themselves and a great deal of work remains to be done in this area (see suggestions in Chapter 4, Section 1; Spolsky, 1 968b; and Upshur, 1968).

3.4.1.2.6  Manipulation by Selection

      This approach can be useful when the researcher is interested in the effects of certain learner factors in special settings. Selection can be achieved statistically by separating out the data for those individuals possessing the desired characteristics or, naturalistically, by collecting data only in those settings where the desired features hold. For example, one may wish to assess the range of communicative abilities acquired by natural bilinguals as a function of the degree of anomie that develops in the early stages of immersion in a foreign culture. Or one may wish to assess educational achievement as a function of instruction in a second language in a cultural setting where the second language is not normally used as was done by Macnamara (1966) in Ireland.

3.4.1.2.7  Manipulation by Induction of Change

     This may be attempted whenever there is reason to believe that a learner factor can be changed through some action by the researcher, educator, or parent: coaching to improve aptitude, providing remedial practice exercises, using propaganda techniques to change attitudes, giving feedback inforrnation on learner progress, etc. The assessment of the effectiveness of the manipulatory attempt may be done in a number of ways: the "before-after" technique described in the attitude change literature, the use of control groups, the induction of opposite effects, etc.

3.4.1.3 Procedure

      The implementation of the research on the present problem must proceed through a number of stages that will now be outlined.

3.4.1.3.1  Selection and Preparation of Observation Techniques

      Where standardized tests are available, a decision will have to be made about selecting those that are best suited for the particular sample of subjects to be included in the study. The administration of some tests (e.g., TAT, MMPI) must be done by trained individuals having proper credentials. Appropriate steps must be taken for the recording of interview data, and questionnaire forms should be prepared, preferably after pilot versions have been tried out. The individuals that are involved in field observations should be properly trained and the reliability of their observations and evaluations assessed. Decisions about selection should be carefully considered and provisions made for including necessary comparison groups and conditions to determine the effectiveness of the induction manipulation.

3.4.1.3.2 Selection of Setting
 
      There are many factors to be considered here: a sufficiently large sample of subjects with the desired characteristics, the representativeness of the sample vis-a-vis the population to which generalizations are to be made, the probability of getting good cooperation from the subjects, their parents, teachers, and school officials, and so on.

3.4.1.3.3 Analysis of Data

      Decisions about this ought to be made prior to the start of the project and they will depend on the focus the investigator has in mind. Consultation with measurement experts and computer statisticians is of course highly desirable.

3.4.1.4 Overview
 
      The overall problem of this particular project concerns the identification of learner factors that influence the study of a second language. The strategy that has been described consists of obtaining a large number of diverse observations in a setting where individuals are engaged in the study of a second language. Some of the observations are restricted to naturally occurring circumstances, others are designed to record the effects of a manipulated change in these conditions. Statistical treatment of these data will yield information on a number of important problems, some of which are listed below.

3.4.1.4.1  Problems of Covariance Relations

     How many learner factors should be kept track of in a second language learning situation? Intercorrelations between the many variables recorded and subsequent factor analysis may show that there is no need in future studies to pay attention to all of them, since many variables may be highly intercorrelated and some can then be estimated on the basis of others. For example, high general intelligence, high need achievement, an assimilatory strategy in learning syntax, good study habits, and realistic self-evaluation criteria may all form a tightly intercorrelated cluster, and another cluster might be made up of a favorable attitude toward the teacher, low ethnocentrism, high phonetic coding ability, and high interest in FL study. These are pure speculations, but if the data were available hypothesis and fact could be separated.
      The particular configuration of correlation clusters yields infor-mation about what might be the underlying factor that accounts for their intercorrelations. Thus, the first cluster in the example above suggests a general learning ability that also manifests itself in second-language learning, and the second cluster might suggest a general attitudinal component in FL study. These higher order mediating factors may subsequently be assessed by a single observational technique thus making for economy in both explanation and testing.

3.4.1.4.2  Problems of Allocation of Amount of Variance

     Given the specific criterion measures that are included (proficiency tests, evaluation of communicative abilities, course grades), what amount of variance is contributed by each of the learner factors also measured (aptitude, intelligence, attitude, etc)? Are some learner factors more important for some criterion measures? (e.g., intelligence may be more important for course grade, while attitude may be more important for acquired range of communicative abilities). Can learner factors compensate for each other? (e.g., whether positive attitudes could overcome the disadvantages of low aptitude). Are certain learner factors more crucial in the presence of certain others? (e.g., self-evaluation criteria may be important only under the assimilatory strategy; or aptitude may become more crucial with advanced age). And so on.

3.4.1.4.3 Problems of Distributional Characteristics

     What is the distribution of attitudes toward FL study in a particular school district? Is the variance comparable across grades or does it become restricted in a particular direction after a year or more of study? Are there extremes in feelings of anomie and do these change with age or number of years of study? How do study habits vary, change? etc.

3.4.1.4.4 Problems Related to Corrective Measures

      What learner factors can be manipulated? Which ones resist change more than others? How long does induction of change last? Is induction of change as important in variance contribution as the same characteristic that develops spontaneously?
      These are then some of the problems for which this project would provide hard information. It may not be feasible to obtain all the measures suggested in any one study, but it should be realized that the more restrictions are imposed the less the "pay-off" value is likely to be.

3.4.2 Problem: To Assess the Instruction Factors That Have a Significant Effect in the Teaching of a Second Language

3.4.2.1  Reference
      See discussion of "instruction factors" in Sections 3.1.3 and 3.2.2.

3.4.2.2  Tactics

3.4.2.2.1  Standardized Tests

     There are no standardized tests for the assessment of instruction factors except those related to criterion evaluation (as in 3.4.1.2.1 (a) above). The development of standardized tests for the evaluation of communicative competence is a critical need in the language testing field today and this problem is examined in Section 3.4.3 below. The checkAist form discussed under 3.4.2.2.4 below has the capability of developing into a standardized test for the evaluation of a FL program, but at the moment no norms are available for it.

3.4.2.2.2  Interview Technique

      The area where this technique could be useful is that of "teacher observation and evaluation." If viewed in narrow terms, this approach would take the form of an interview between the instructor and his supervisor in which the latter would obtain information on a number of matters: the degree to which the instructor is aware of recent advances in language pedagogy, his satisfaction with the administrative support he feels he is getting, special problems with students on which he might need help, the degree to which he is responsive to learner factors, innovations in the instruction program that he might wish to bring about, his attitudes toward the native language and culture of the students when the latter have a different ethnic background, and so on. Defined in broader terms, the "interview technique" can be conceived in terms of "discussion groups in which instructors and supervisors (and perhaps other "experts") examine jointly the problems they share and the solutions that might be available for them.

3.4.2.2.3  Written Questionnaires

     The teaching inventory profile discussed below can be used as a questionnaire to obtain responses from teachers, supervisors, and students. This procedure yields information on what instructional activities are considered good or bad by each of these groups.

3.4.2.2.4  Descriptive Field Observations

     Hayes, Lambert, and Tucker (1967) present a 324-item inventory that describes in fairly specific terms the principles and policies which language teachers consider relevant to the instructional setting. The inventory is grouped under six headings: (a) Administration and policy; (b) Teachers and teacher competence; (c) Course design; (d) Testing; (e) Classroom and classes; and (f) Language laboratory. The authors of the inventory propose that independent assessors visit schools and attend classrooms and make direct observations along each of the inventory items. This would yield a profile for each particular setting. Different profiles can be correlated with achievement measures on students matched on relevant factors (aptitude, intelligence, attitude, etc.) and statements can then be made about the effectiveness of specific instruction activities. Such results can be used in decisions about changes in practices and policies.
     This approach is undoubtedly superior to previous attempts at comparing "methods of second4anguage teaching" in which a great deal of variability was subsumed under a single "method." However, there are two basic weaknesses with the proposal of Hayes et al. (1967) which ought to be remedied. The first relates to their suggestion to use matched groups of students. This would be practically impossible given the large number of learner factors that would have to be matched (see section L.F. of outline on page 105), and in any event, the sociocultural setting in each school is likely to be unique, and hence, would not be matched. The second weakness of their proposal lies in inadequate provisions being made for assessment of proficiency and achievement of communicative skills. This is extremely serious since the evaluation of the effectiveness of any instructional activity is dependent upon the availability of adequate criterion measures. It is important, therefore, that the teaching inventory be used in conjunction with a program of research that includes the identification of learner factors, as described in Problem 3.4.1 (above) and the development of adequate criterion measures, as discussed in Problem 3.4.3 (below). Statistical techniques can then be used to assess the interaction effects between specific learner and instruction factors as they affect particular criterion goals.
     Some additional comments are in order. The inventory developed by Hayes et al., useful as it undoubtedly will be, neglects certain important instruction factors which ought to be incorporated in subsequent versions of it, or obtained by some other means. The most important of these is the degree to which the instructional program is responsive to learner factors as discussed in Sections 3.1.3, 3.2.1, and 3.2.2. What provisions exist for adjusting criterion goals to aptitude and interest? What corrective measures are being taken against inappropriate student self-evaluation criteria, learning strategies, and study habits? and so on. In other words, what is needed is not only a description of the "statics" of the instruction program, but its "dynamics" as well, viz., the built-in measures for change (direction and rate) as required by the existing demands.

3.4.2.2.5 Field Evaluations

     The descriptive measures just discussed can serve as input to state-ments of evaluation about specific practices, policies, and characteristics, or about the overall program. Teachers can be rated on competence in terms of the degree to which they display desired characteristics. Such evaluations can serve a diagnostic function and suggest areas where they need remedial training. Schools can be given ratings for the effectiveness of their foreign language program which can serve the basis for administrative changes and a guide to be used by teachers and students in deciding about employment and attendance. Ultimately, national standards can be adopted which can have significant sociopolitical and cultural implications.

3.4.2.2.6 Manipulation by Selection

     In the context of the problem under discussion, there are many occasions for using this technique to assess instruction factors in special settings. The sociocultural setting may result in certain peculiarities unique to that situation (e.g., schools in New York with a heavy proportion of Puerto Ricans, or schools in Montreal vs. Toronto, etc.) and this kind of "natural selection" allows for interesting comparisons. Intentional selection may be desirable in assessing experimental programs where the selection may occur at the level of teacher, student, or policy.

3.4.2.2.7 Manipulation by Induction

     Teacher training programs combined with pre- and post-observations of effectiveness represents an example of this approach. Others include special programs such as "a year abroad," summer language camps and the "French House," and varieties of special techniques which the teacher might want to try out (e.g., language games) in combination with comparison groups or pre- post-designs.

3.4.2.3 Procedure

     The same kind of considerations as discussed in Section 3.4.1.3 apply here as well with respect to selection and preparation of observation techniques, choice of setting, and analysis of data.

3.4.2.4 Overview

     The problem reviewed in this section concerns the description and identification of instruction factors that have a significant effect in the teaching of a second language. Th